Вы находитесь на странице: 1из 133

Factors Influencing the Reproductive Efficiency of

Dairy Herds in the Dominican Republic

Angela Renea Billings

Virginia Polytechnic Institute and State University

Master of Science
in
Veterinary Medical Science

Dee Whittier, Chair


François Elvinger
Ernest Hovingh
Michael McGilliard
Leonardo Rafael Tineo Madera

February 4, 2002

Blacksburg, Virginia

Keywords: dairy, Dominican Republic, heat stress, lameness, reproductive efficiency,


reproductive outcomes
Factors Influencing the Reproductive Efficiency of Dairy Herds

Within the Dominican Republic

Angela Renea Billings

(ABSTRACT)

Despite an increase in domestic milk production, the dairy industry in the Dominican Republic

(D. R.) has not been able to adequately meet the demand of the ever-growing Dominican population, prompting

the government and milk processing plants to sponsor programs which will increase the national production of

milk and eventually lead to milk self-sufficiency. One obstacle to this goal is a very low reproductive rate in

cows and heifers due in part to the tropical setting. Year-round heat stress may result in abnormal follicular

dynamics and decreased oocyte and sperm quality causing reproductive efficiency to decline drastically. The

specific purpose of this project was to examine and characterize the reproductive practices and outcomes of the

Dominican dairy industry by region and to attempt to identify factors that influence dairy reproductive

efficiency.

During the course of the study, 43 farms were visited over a 10-week period and evaluated. Farms

were chosen throughout 4 of the 5 regions of the country and were selected based on their size (preferably 40

adult cows or more), availability of data, and demonstrated motivation by the owner in improving the herd.

Individual herd evaluation was broken into four major components: owner interview, farm evaluation,

collection of individual cow reproductive data, and adult cow evaluation.

Once observations had been made and catalogued, all data were summarized on the herd level and

analyzed descriptively. In addition to descriptive analysis, multiple regression techniques were used to select

independent variables which explain most of the variance for each of four reproductive outcomes: days to first

service, services per pregnancy, projected calving interval, and service rate.

In general, reproductive management practices varied depending on region and farm size. Average

lameness within the herd was the most important factor in explaining the variability within services per

pregnancy and projected calving interval. As average herd lameness increases by 1 point (based on a 1 to 4

scale), services per pregnancy and projected calving interval increase by 0.65 services per pregnancy and 61.1

days respectively according to the model formulated. Increase in the number of employees involved in estrus

detection resulted in higher days to first service. The Santo Domingo region had lower days to first service

ii
possibly due to widespread reproductive hormone use within the region. The percentage of Holsteins within the

herd was associated with increased services per pregnancy and projected calving interval. Increases in service

rate were most closely associated with the type of record category used, indicating that a larger sample

population with thorough insemination records may be needed to adequately assess this outcome.

In part, reproductive efficiency in the Dominican Republic can be potentially improved by enhancing

methods for estrus detection. Mechanical aids to estrus detection (tailhead chalk, K-mar patches, etc.) may

help increase estrus detection efficiency in herds currently only relying on visual observation. Assigning 1-2

people primarily to estrus detection and increasing the frequency of hormone usage may also improve estrus

detection efficiency.

The main emphasis for Dominican dairy producers, however, should be on preventing new lameness

and culling chronically lame cows once it is economically feasible to do so. Reducing the incidence of

lameness could, in itself, dramatically improve reproductive efficiency in the Dominican Republic.

Keywords: dairy, Dominican Republic, heat stress, lameness, reproductive efficiency, reproductive outcomes

iii
Grant Information

This study made possible by grant support from:

-The Dominican Association for Milk Self Sufficiency, Ciudad Ganadero, Santo Domingo,
Dominican Republic

-Leche Rica, S. A., Av. John F. Kennedy, Santo Domingo, Dominican Republic

-VA-MD Regional College of Veterinary Medicine, Virginia Polytechnic Institute and State
University, Blacksburg, VA 24061

iv
Acknowledgements

I would like to thank Dr. Julio Brache and Sr. Alfredo Rios of Leche Rica for your consistent financial support
and for ensuring I was well cared for while in the Dominican Republic. Thank you for providing me the
opportunity to further my education. I hope these results are very useful to the Dominican dairy producers.

I would also like to thank the Dominican Association for Milk Self Sufficiency and the VA-MD Regional
College of Veterinary Medicine for aiding in funding this project.

A special thanks goes to Dr. Leonardo Tineo for working very hard to arrange my visits and interrupting his
schedule to see that I collected the data I needed. I would also like to thank his family (Petra, Patricia, Heidi,
Elaine, and Leornardo) for taking me into their home, treating me like part of the family, and doing everything
possible to make my stay pleasant. You all are wonderful people and I will miss you!

A special thanks also goes to my advisor, Dr. Dee Whittier, for his support, guidance, and hard work in helping
me to achieve this goal. I could not have finished without your help! Thanks for not only being my advisor but
my advocate as well.

I would like to thank the rest of my committee members: Dr. Ernest Hovingh, Dr. François Elvinger, and Dr.
Michael McGilliard for their advice, encouragement, and evaluation of this thesis.

Many thanks and much credit are also due to the following:

To Dr. Ramon Martinez for his assistance in arranging farm visits and providing transportation while I was in
the Dominican Republic.

To the Dominican dairy farmers for their hospitality and cooperation. You are wonderful, hard working people
and I hope this study will help you in your efforts.

To the MEGALECHE staff in San Juan and Monte Plata for all of their assistance in data collection. I hope the
information I provided has helped you.

To all the other Dominican veterinarians and professionals that assisted me during my stay.

To Dr. Pedro Pablo Peña for loaning me the resources about the Dominican dairy industry.

To Mr. Daniel Ward for all your statistical guidance. I would have been lost without your help.

To Ms. Anne Clapsaddle for all the faxes to the D. R. you worked so hard to send.

To my mom and dad, Louis and Teresa Billings, for their constant prayers, support, love, and advice.

To my fiancé and best friend, Justin DeRosa, who has encouraged me to finish what I have started, even when it
meant months of separation, and who has been a source of encouragement and strength.

And finally and most importantly, to the Lord Jesus Christ, without Whom I would have never made it this far.
Thank You for giving me strength when I was weak, companionship when I was so far away from home, and
help when I needed it most. Everything good that I am, I owe it to You.

v
Table of Contents
Abstract ii
Grant Information iv
Acknowledgements v
List of Tables vii
List of Figures viii

Chapter Page
Chapter 1. Introduction 1
Chapter 2. Review of Literature 4
Reproductive Outcomes and Their Measurement 4
Climate 11
Nutrition 16
Chapter 3. Materials and Methods 19
Initial Observational Visit 19
Survey Design and Observational Data Collection 19
Statistical Preliminary Analysis 24
Chapter 4. Descriptive Analysis 35
Reproductive Outcomes by Region 35
Herd Descriptors by Region 45
Reproductive Management Descriptors by Region 50
Nutritional Management by Region 64
Conclusions 66
Chapter 5. Days to First Service 69
Abstract 69
Introduction 69
Variable Selection and Description 70
Results 71
Discussion 71
Chapter 6. Services per Pregnancy 74
Abstract 74
Introduction 74
Variable Selection and Description 74
Results 75
Discussion 76
Chapter 7. Projected Calving Interval 78
Abstract 78
Introduction 78
Variable Selection and Description 78
Results 79
Discussion 80
Chapter 8. Service Rate 82
Abstract 82
Introduction 82
Variable Selection and Description 83
Results 84
Discussion 84
Chapter 9. Summary and Conclusions 87
Literature Cited 91
Appendix A. Results from First Visit 97
Appendix B. Survey Forms 102
Appendix C. Calculation of Dependent Variables 117
Appendix D. Conversion Tables 118
Appendix E. Additional Descriptive Statistics 119
Vita 125

vi
List of Tables
Table Page
Table 1. Breakdown of Provinces of the Dominican Republic by Region. 1
Table 2. Effect of Standing to be Mounted in Combination With Other Signs of Estrus on
Error of Estrus Detection Based on Milk Progesterone Concentrations. 9
Table 3. Relationships Between Estrus Detection Efficiency and Culling Rate Based on
Failure to Conceive. 9
Table 4. Relationship Between Days to First Service and Estrus Detection Efficiency. 11
Table 5. Breed Differences in Effect of Season on Production of Embryos Via in vitro
Maturation, Fertilization, and Development in Louisiana. 14
Table 6. Initial Candidate Dependent (Outcome) Variables. 25
Table 7. Initial Independent (Regressor) Variables. 26
Table 8. Pearson Correlation Coefficients and p-values for Potential Dependent Variables. 30
Table 9. Number of Observations Recorded for Dependent and Independent Continuous
Variables. 31
Table 10. Breakdown of Reproductive Outcomes by Region. 36
Table 11. Comparison of D. R. Means with U. S. Optimum Values and
Averages for the State of Virginia for Select Reproductive Outcomes. 44
Table 12. Continuous Herd Descriptives by Region. 46
Table 13. Reproductive Hormone Use by Region. 56
Table 14. Procedure Determining the Amount of Feed per Cow per Day. 64
Table 15. Names and Descriptions of Variables influencing Days to First Service. 70
Table 16. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence 71
Intervals for the Final Model of the Response Variable Days to First Service.
Table 17. Extreme Predicted Values from Days to First Service Model Building. 72
Table 18. Names and Descriptions of Variables Influencing Services per Pregnancy. 75
Table 19. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Services per Pregnancy. 76
Table 20. Extreme Predictive Values from Services per Pregnancy Model Building. 76
Table 21. Names and Descriptions of Variables Influencing Projected Calving Interval. 79
Table 22. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Projected Calving Interval. 80
Table 23. Extreme Predicted Values from Projected Calving Interval Model Building. 80
Table 24. Names and Descriptions of Variables Influencing Service Rate. 83
Table 25. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Service Rate. 84
Table 26. Extreme Predicted Values from Service Rate Model Building. 85

vii
List of Figures
Figure Page
Figure 1. Effects of Energy Imbalance and Acidosis on Health and Reproductive
Performance in Dairy Cows Both Before and After Calving. 18
Figure 2. Mallow’s CP Plot for Projected Calving Interval. 33
Figure 3. Types of Record Keeping Systems by Region. 46
Figure 4. Distribution of Cooling Systems by Region. 51
Figure 5. Methods of Estrus Detection by Region. 53
Figure 6. Estrus Detection Efforts in Relation to Milking or Feeding Activities by Region. 54
Figure 7. Summary of the Number of Employees Involved in Estrus Detection by Region. 54
Figure 8. Nighttime Estrus Detection by Region (6 p. m. to 6 a. m.) 55
Figure 9. Summary of Hours the Milking Herd Spends on Concrete by Region 57
Figure 10. Interval Between First Observed Estrous Activity and Insemination by Region. 60
Figure 11. Bull Use by Region. 61
Figure 12. Groups of Cows Inseminated With Natural Service by Region. 61
Figure 13. Farms conducting Bull Breeding Soundness Examinations by Region. 62
Figure 14. Average Amount of Grain Fed to Lactating Cows per Day 64
Figure 15. Daily Access to Pasture by Region. 65
Figure 16. Farms Depending on Pasture for a Significant Proportion of Forage Dry Matter
Intake for Lactating Cows 65

viii
Chapter 1

Introduction

The Dominican Republic (D. R.) occupies the eastern two-thirds of the island of Hispaniola bordered

to the west by Haiti, to the south by the Caribbean sea, and to the north by the North Atlantic Ocean. It covers a

land area of about 48,380 square kilometers, or roughly half the size of the state of Virginia. Mountains and

fertile valleys characterize the terrain and about 52% of the land use is classified as either permanent crops or

pasture. A tropical maritime (warm, moist) climate results in little seasonal variation in temperature but

variations in rainfall do exist by region and by season [1].

The 29 provinces and 1 district making up the D. R. are divided geographically into 5 major regions

which are summarized in the following table [2]:

Table 1. Breakdown of provinces of the Dominican Republic by Region.

Region Province names

East Cibao Espaillat, Salcedo, La Vega, Monseñor Nouel, Sánchez Ramírez,


Duarte, María Trinidad Sánchez, Samaná

West Cibao Puerto Plata, Santiago, Valverde, Santiago Rodríguez, Dajabón,


Montecristi

East San Pedro, Hato Mayor, El Seibo, Altagracia, La Romana

Santo Domingo Distrito Nacional, Monte Plata, San Cristóbal, Peravia

Southwest Azua, Barahona, Pedernales, Bahoruco, Independencia, Elías Piña,


San Juan

The majority of Dominican residents (~73%) are a “mixed” race, mostly of Spanish and African

descent. There is a large income disparity between the D. R.’s 8 million (2001 est.) people, with the country’s

richest 10% controlling 40% of the national income and the poorest 50% controlling only about 20%. Until the

last few years, agriculture has been one of the main sources of employment for most Dominicans. By 1999,

only 11.3% of the population was employed in the agricultural sector, with 56.5% being employed in the

services sector (i.e. tourism, free trade zone industries). Significant in the decline of the agricultural labor force

has been the decline of the sugarcane industry, part of an international phenomenon. Agricultural exports

include, for example, sugarcane products, coffee, cotton, rice, cattle, pigs, dairy products, beef, and eggs. In

2000, the D. R., however, imported $3.8 billion more total goods than it exported, with foodstuffs being the

1
primary import. As such, efforts are underway to decrease imports, especially foodstuffs, and increase exports

[1].

The dairy industry stands as a key component of the Dominican agricultural economy. In 1998,

agriculture had begun to rebound by diversifying from sugar production. The agricultural sector had seen a

9.1% increase in its contribution to the gross national product (GNP) when Hurricane George hit on September

22. Although the rest of the agricultural sector saw a substantial loss in production, most of the damage

incurred to the cattle industry was infrastructural and not a result of lost animals. Therefore, domestic milk and

meat production declined only minimally, acting as a source of stability for the agricultural sector as a whole.

In 1997, the dairy industry from milk sales alone produced RD$1,597 million of the country’s RD$ 212,644

million GNP. The dairy industry serves as a promising segment of future economical growth for the Dominican

agricultural sector [2].

The Dominican dairy industry has its share of economic problems, however. In 1974-75, the

importation and re-hydration of powdered milk by processing companies became legal. Prior to this time, all

imported powder had to be sold as powder. Powdered milk, a subsidized import of the European Union and

United States, has enabled domestic milk prices to remain relatively stable over the years, despite changes in

domestic production. Its importation has, however, hurt the domestic dairy producer. Re-constituting

subsidized, imported powdered milk is cheaper than collecting milk from producers throughout the country,

especially given the D. R.’s poor road infrastructure [2]. By keeping the milk price artificially low, there is a

threat of undervaluing the domestic price of milk [3]. Therefore, producers not only have to be concerned about

maintaining a market for their milk; they also do not get paid at a level consistent with the increasing market

demand.

In 1997-98, the government established a school breakfast program that supplied milk to all public

school children. Since most of these children are from families living below the poverty line, it was desired that

this milk be of the highest possible quality and nutritional value. Large Dominican dairy associations, such as

The Dominican Association of Milk Producers (APROLECHE), began pressuring the government to increase

the standards of milk processing, which meant decreasing the proportion of powdered milk within the milk

supply. Other technologies, such as imported canned milk and UHT milk, were considered but were either too

expensive to use on a wide-scale basis or the technology did not exist within the country. Also, importing more

dairy products only served to counteract the goal to export more and import less. The most practical solution

2
given the potential contribution which the dairy industry could make to the country’s GNP and the most cost

efficient solution, was to increase domestic production.

The dairy industry within the Dominican Republic is poised for improvements. Despite an 84%

increase in national milk production from 1991-1998, the level of imported powdered milk also increased by

about 40%, indicating that national production has not been adequate to meet the demand of the ever-growing

Dominican population [2]. The government and milk processing plants have worked closely with dairy

producers to sponsor programs which will increase the national production of milk and eventually lead to milk

self-sufficiency.

Meeting the national dairy demand is no easy task considering the obstacles faced by the typical

Dominican dairy producer. One of the major challenges facing Dominican dairy producers is the tropical

climate of the country. Since there is little variation in temperature, cattle have almost no chance of escaping the

physiological stresses of high heat and humidity. Reproductive performance is a special challenge to dairy

producers because recent droughts have lead to forage shortages, causing cows to lose body condition and

further drop in milk production. Most Dominican dairies have a very low level of technology compared to

developed countries, such as the United States. Management practices vary widely and are often influenced by

the training of the owner, owner interest in the farm, employees, etc. Producers complain about low pregnancy

rates and “repeat breeders” almost year round. The role climate and/or specific management decisions play on

the over-all reproductive performance of dairy cows within the Dominican Republic has not been well defined

historically.

The goal of this study was to examine and characterize the reproductive practices and outcomes of the

Dominican dairy industry by region and to try to identify factors that influence dairy reproductive efficiency.

Until reproductive efficiency improves, the D. R. has little hope of meeting the ever-increasing domestic

demand for milk and dairy imports will continue to rise.

3
Chapter 2

Review of Literature

The current level of reproductive performance is well below the optimum in most countries of the

world [4]. Most authors have concluded that significant economic returns can be achieved by improving

reproductive performance [4-8]. One simulation model [5] predicted a $36.75 (1982 dollars) increase in

profitability per cow by decreasing the calving to conception interval by 30 days associated with a 30%

improvement in estrus detection.

The most inclusive measures of reproductive outcomes attempt to assess the interval that the average

cow in a year-round breeding herd has between calvings. Common indices include the “Intercalving Interval”

(calving interval, historical), the “Projected Calving Interval” (calculated by adding the length of an average

gestation to the period from parturition to conception) and the “days open” or “calving to conception interval”

[4]. The use of “days open” focuses on the critical time period between calvings, the time between calving and

conception. Since there is generally little variation in the time from conception to parturition, management

emphasis to improve the time between calvings must be placed on the calving to conception interval. This time

period can be divided into three major areas of emphasis:

- The time period between when the cow calves and when she becomes eligible for insemination.

This period is typically called the “Voluntary Waiting Period” (VWP).

- The efficiency with which cows are detected in estrus and inseminated following the vwp. The

emphasis for this factor is estrus detection efficiency. It should be recognized that failures in

estrus detection efficiency can result from either the failure of cows to have a regular estrous cycle

or from the failure of management to identify and service the estrus once it occurs.

- The likelihood that a service or insemination will result in pregnancy. This factor is typically

referred to as “Conception Rate” but a number of indices have been devised to measure this

propensity for cows to become pregnant on each occasion that they are inseminated.

Reproductive Outcomes and Their Measurement

CALVING INTERVAL

When reporting “optimal” reproductive outcomes, measurements are based on a 12 to 13-month

calving interval, which has been found to realistically be the most suitable for maximizing profit and milk

4
production [9]. A calving interval is the period of time between successive parturitions (usually reported in

months on the herd level). Calving interval is most directly affected by three reproductive outcomes: estrus

detection, days to first service, and voluntary waiting period (VWP), with estrus detection being the most

important [8, 9]. As calving interval increases, days in milk increases and lifetime milk yield decreases [9].

Some studies have shown that estrus synchronization offers one of the best alternatives for decreasing

calving intervals [8, 9]. By optimizing estrus detection for first service inseminations and by reducing herd

variability within days to first service, first service conception rates can be improved and calving intervals can

subsequently be shortened [10], possibly depending on the synchronization program used [9].

Herd reproductive performance is a direct product largely of herd management [9, 11]. Estrus

detection efficiency and accuracy are two of the most important parameters influencing total reproductive

efficiency within a given herd, since they are the major determinants of pregnancy rate. Almost all other

reproductive outcomes (services per pregnancy, days open, days in milk, percent of herd culled for reproductive

purposes, days to first service, conception rate, and calving interval) are influenced in some way by estrus

detection.

CALVING TO CONCEPTION INTERVAL

Calving to conception interval (CCI) also referred to as “Days Open” in some record keeping schemes

is the period of time between parturition and conception and is nearly always inversely related to estrus

detection. This reproductive parameter is influenced by the estrus detection rate, conception rate, VWP, and

culling [5]. Calving to conception interval plus gestation length, in turn, results in the calving interval [5, 12].

Realistically, calving to conception interval should be no more than 113 days to achieve an optimal calving

interval of 13 months [9, 12]. Calving to conception interval can be adversely affected by lameness. One study

found that lame cows had calving to conception interval’s 40 days longer than healthy cows [13].

As days open increase:

1) fewer calves are produced per year leading to lower calf sales and fewer replacement heifers,

2) breeding costs increase due to higher services per pregnancy,

3) milk sold per day decreases due to longer average DIM (>150 days),

4) and veterinary costs increase due to more repeat breeders [12, 14].

5
In fact, the postpartum voluntary waiting period (VWP), estrus detection efficiency, conception rate, and

culling are the determinants of days open within dairy herds [15]. Days open are also reported to increase with

reproductive pathology, systemic illness, and lameness [11].

ESTRUS DETECTION

One of the major factors potentially influencing pregnancy rates within the D. R. is poor estrus

detection. Pregnancy rates are influenced by conception rate and estrus detection efficiency [5, 16]. Estrus

detection and conception rates are in turn influenced by a host of environmental and management factors [8].

Estrus detection can be studied as two separate measures, estrus detection efficiency and estrus detection

accuracy, which have two different management implications. Detection efficiency is a measure of the

breeding eligible cows that are predicted to come into estrus over a given time period that are actually observed

in estrus. Detection accuracy is the proportion of cows thought to be exhibiting estrus activity over a given

period of time that are actually in estrus [15]. Estrus detection accuracy is one of many factors that influence

the likelihood of conception at each estrous that is serviced [5, 15]. So the success in both detection efficiency

and accuracy largely determines herd pregnancy rates. Both parameters affect herd economics. In fact, in

1994, it was estimated that the U. S. lost over $300 million to inadequate estrus detection efficiency and

accuracy [17].

Detection efficiency measurements can be determined for a given herd several different ways. One

calculation involves counting the number of days between all recorded standing estruses for the breeding herd

over a given period of time and dividing by the average estrus length of 21 days. This calculation yields the

proportion of estrous cycles recorded. Another calculation involves listing all the cows eligible for

insemination on a given day and then comparing this with cows found in estrous over the next 24-day period

(assuming average estrous cycle length ranges from 18-24 days). At the end of the 24-day period, the number

of cows detected in estrus are divided by the total number of cows on the list, giving the detection efficiency

proportion for that period [15]. This method is not ideal for a study such as the one conducted in the D. R.

because it relies too heavily on owner/employee compliance. Also, such a list may bias detection efficiency

measures, as workers may be tempted to watch cows not checked off the list more carefully to improve the

herd’s test statistics. Yet another method of estimating estrus detection efficiency is by calculating the ratio of

single (18-24 day) to double (38-46 day) interestrual intervals. A 6:1 ratio of single to double interestrual

periods is considered an acceptable measure for detection efficiency. The Dairy Records Management Service,

6
a large U. S. dairy records processing organization, uses the equation: [number of breedings and estruses in a

test period for eligible cows/(estrous cycle days in the test period for eligible cows/21)] x 100, where “eligible

cows” are those over the voluntary waiting period (VWP) who are eligible during any part of the given test

period. This equation is similar to the one recommended by the Dairy Reproduction Core Parameters

Committee and the American Association of Bovine Practitioners (AABP), the breeding interval (BI) method.

This method is only applicable for cows after the first insemination of a given lactation and makes use of the

following equation and requires complete records: BI = (mean days open for pregnant cows – mean days to first

breeding)/ (breedings per pregnancy – 1). Many other methods exist for estimating estrus detection efficiency.

Choosing the most appropriate method involves determining the type and quality of data available for

calculation [15].

Poor estrus detection efficiency is considered as one of the most costly problems of AI programs or

programs based on hand-mating natural service [5, 14, 15, 17, 18]. Estrus detection programs based solely on

visual observation have been shown to be inadequate [17-20]. As herd size increases, time for estrus

observation for each cow decreases [9, 17]. On average, cows have been found to exhibit standing activity for

less than 1% of the total time they are in estrus [17]. Therefore, significant amounts of time must be spent in

visual observation to adequately observe estrus activity within the herd [17].

Mechanical aids to supplement visual detection of estrus have been shown to improve herd reproductive

efficiency [14, 17, 18]. To optimize estrus detection efficiency and accuracy, mechanical aids should optimally

have the following attributes [17]:

1) they should monitor the cow 24 hours/day,

2) they should last for the entire productive life of the animal,

3) they should accurately and automatically identify which cows are in estrus,

4) they should minimize labor requirements (not as important in a developing country where labor is

relatively inexpensive and plentiful), and

5) they should be able to accurately (95%) and precisely identify the physiological or behavioral

signs of estrus most closely associated with ovulation.

Various technologies currently exist or are being developed to aid in estrus detection. HeatWatch

transponders, pedometry, and implantable sensors to measure changes in uterine secretion electrical

conductivity [17, 21] all meet many of the requirements listed above [17]. Unfortunately, such technologies are

7
too expensive to use on a wide-scale basis in a developing country. However, Xu et al [18] found that tail head

painting with consistent visual observation (2 times daily for at least 20 minutes each time in the morning and

afternoon) yielded results comparable to using HeatWatch transponders.

One example of how advanced technologies can aid in estrus detection is the development of

pedometry. Cows in estrus have been shown to increase their physical activity [16, 17, 20]. In fact, dairy cows

in estrus kept in loose housing display almost four times more physical activity than do their non-estral

herdmates. Therefore, pedometers can potentially be a useful aid in estrus detection [16, 17]. Estrus detection

efficiency may be as low as 60% and as high as 100% using pedometers. Estrus detection accuracy, when used

alone, however, can be quite variable from 22-100%, especially in herds with lameness problems [17].

Estrus synchronization, using commercially available pharmaceutical hormone preparations, has been

shown to increase estrus detection efficiency by increasing the number of cows exhibiting estrus activity at a

given time, as well as by allowing some prediction of the time that estrus will occur [8, 9, 20]. Mounts for

herds with multiple cows exhibiting synchronized estrus activity were shown to be longer [20] and more

frequent [6, 20]. Timed insemination (or systematic insemination), using hormones to synchronize estrous

cycles, can moderately decrease the need for visual detection of estrus and can shift reproductive management

from observing sporadic individual cows to a methodical herd approach, thereby having the potential to have a

greater effect on improving herd reproductive efficiency [8, 9]. The benefits of timed insemination on

improving reproductive efficiency within a tropical climate will be discussed in further detail later.

Estrus detection accuracy is an important measure of the number of inseminations that were performed

on cows truly in estrus. As stated earlier, inaccuracy in estrus detection results in lower conception rates [5,

15]. Nebel et al. [22] showed that 25% of the cows presented for insemination in one study had elevated serum

progesterone levels on ELISA testing, indicating that they were not truly in estrus. Others studies have found

similar results [17]. Heersche and Nebel [15] recommends as part of the herd evaluation investigating estrus

detection accuracy on any farm suffering from low conception rates. Detection accuracy can be estimated by

examining the distribution of return intervals to estrus (18-24 days considered normal interval), conception

rates, palpation of cows presented for insemination, and by blood or milk progesterone levels [15]. Interestrual

intervals and conception rates can also be adversely affected by a number of other factors, particularly climate,

and should be interpreted cautiously especially within a tropical environment [23-26].

8
The highest detection accuracy is obtained in a system where cows are inseminated based on standing

estrus and low blood or milk progesterone [15]. The following table (Table 2) from a study by Reimers, Smith,

and Newman [7] demonstrates the effect of visualizing standing estrus on improving the detection accuracy

within a herd over visualize of secondary signs alone. The average error in estrus detection accuracy was found

to be 5.1% within the sample population (n = 476 herds) of this study, with some herds having as high as 60%

inaccuracy in estrus detection. Standing to be mounted and mounting behavior were found to be the most

accurate signs of estrus. As is evident from the data, producers must put their primary emphasis on the

observation of standing estrus, and not secondary signs, to optimize their detection accuracy.

Table 2. Effect of Standing to be Mounted in Combination With Other Signs of Estrus


on Error of Estrus Detection Based on Milk Progesterone Concentrations.

Progesterone $ 1 ng/mL ($ 1 ng/mL assumed nonestral)


Secondary sign Standing not recorded (%) Standing also recorded (%)

Rough tail head 7.8 1.5


Riding other cows 4.4 1.9
Unusually active 7.6 2.1
Mucus on vulva 7.9 2.3
Bawling 6.7 2.5
Fully triggered Kamar device 17.7 3.9
No milk letdown 10.5 5.7
Partly triggered Kamar device 25.8 8.3
Reimers et al. [7].

Improving estrus detection can have major economic benefits [17]. One such benefit is the reduction

in the number of reproductive culls due to failure to conceive (“problem” cows or “repeat breeders”) [5, 14].

Problem cows have been defined as those cows over 120 days in milk which have not been observed in estrus.

Repeat breeders are those cows with 3 or more inseminations which have not been diagnosed pregnant [4, 12].

Bailie [5] (Table 3) reported data that demonstrates the effect increasing estrus detection efficiency has on

decreasing reproductive culls due to failure to conceive. Older cows (second and third lactation) produce more

milk per lactation than do first lactation cows, so by reducing culling rates fewer replacements have to be added

to the herd each year to maintain herd size and thus overall herd milk yield will often increase [12].

Table 3. Relationships Between Estrus Detection Efficiency and Culling Rate


Based on Failure to Conceive [5].
Estrus Detection Rate Culling Rate for Failure to Conceive
(%) (%)
50 8.2
60 4.3
70 2.1
80 1.2

9
Improving estrus detection efficiency and accuracy also increases individual milk yield over time [5,

14]. Lifetime milk yield increases because as estrus detection efficiency increases, pregnancy rates also increase

which means that, on average, cows within the herd complete more lactations during their lifetime. With each

lactation a cow reaches peak milk and so lifetime milk yield is generally greater, assuming the cow remains

healthy [12, 14].

Successful estrus detection efficiency and accuracy are the foundation upon which successful artificial

insemination and hand-mated bull-bred programs must be built [9]. These measures determine the level of

overall reproductive efficiency on such farms, as they affect the outcome of almost all other reproductive

parameters [5].

CONCEPTION RATE

Conception rate, or the proportion of cows inseminated which actually become pregnant, is often

reported as “percent successful services”. As previously mentioned, estrus detection accuracy may significantly

influence conception rate. It should be noted that even though estrus detection accuracy plays a major role, it is

not the only thing that affects conception rate. Physiological stress from increased milk production or heat

stress, disease (i.e. retained placenta, metritis, mastitis, cystic ovaries, etc.), and a host of other factors can also

influence conception rate [4, 27]. Detection efficiency also influences conception rate to some degree because

producers must be able to consistently identify the beginning of estrus to know when to inseminate for

maximum fertility [10, 14, 16, 19]. Please see Table 3 for a summary of the effects of detection efficiency on

culling rate for failure to conceive [5].

Among diseases causing decreases in conception rate, lameness is a major contributor [13, 28].

Hernandez et. al [13] reported that cows with claw lesions were only 52% as likely to become pregnant as a

healthy cow. Lameness also was found to affect calving to conception intervals and services per pregnancy,

another indicator of conception rate. Studies indicate that lameness may actually affect the maintenance of

pregnancy and may even cause infertility but these theories need further investigation [13, 29].

Services per pregnancy is the measure of the number of inseminations for pregnancy within the

average cow of a given herd. Assuming a VWP of 45-60 days, optimal services per pregnancy to achieve a 13-

month calving interval is around 1.8, based on a 55% successful [12]. Services per pregnancy is adversely

10
affected by many factors including decreased estrus detection accuracy, improper A. I. technique, reproductive

pathology, lameness, and heat stress [13].

DAYS TO FIRST SERVICE

Days to first service tends to be inversely related to estrus detection in the early postpartum period [5]

and is the number of days from parturition to the first insemination of a given lactation. It is also influenced by

VWP. The following table from a study by Bailie [5] demonstrates the effect improving estrus detection

efficiency has on days to first service.

Table 4. Relationship Between Days to First Service


and Estrus Detection Efficiency [5].
Mean interval to first service (days)* Estrus detection efficiency (%)

36 50

31 60
VWP** +
27 70

24 80

* Assumed all cows were cycling normally and a conception rate of 60%.
** Assumed VWP of 50 days.

Notice that days to first service is not within the optimal range (~75 days) [30] to achieve a 13 month calving

interval [9] until estrus detection efficiency reaches 80% [5, 12]. Lameness may affect days to first service due

to lowered estrus detection efficiency in affected cows.

Climate

The physiological effects of temperature and humidity on reproductive outcomes must be discussed to

fully understand the challenges facing dairy producers within the D. R. The Temperature-Humidity Index

(THI) within the D. R. (Santiago) has been shown to be over the physiologically stressful range (> 72 THI), on

average, for all 12 months of the year [31]. Temperatures within such a tropical maritime climate can stay

between 25-35oC for the entire year and dairy cattle have been noted to experience heat stress at air

temperatures of as low as 27 oC [32] with a humidity of 80.9% [24, 33]. Heat stress involves the sum of forces

external to the homeothermic animal which acts to displace the normal resting body temperature [34].

Holsteins are especially susceptible to heat stress [31]. Although older studies may indicate that cows come

into estrus proportionally more often at night [5, 20, 35], newer studies have offered contradictory information

that, on average, cows within a given dairy herd come into estrus equally throughout a 24-hour period [18, 19].

11
Other studies, however, have also indicated that cows suffering from heat stress are less likely to be observed

exhibiting estrous activity and inseminated in a timely manner from the actual onset of the estrous cycle [24-26,

33]. Studies have shown that temperature has a curvilinear relationship with estrus intensity: with frequency of

mounts per unit of time increasing up to 25oC and then decreasing by about 30 oC [14, 27]. This result is

thought to be a due to a lower serum estradiol concentration during proestrus attributable to a combination of

factors which may be caused by a depressed preovulatory surge of luteinizing hormone (LH)[24]. Lactating

cows have been shown to have lower serum proestrus estradiol levels than cows experiencing heat stress [24,

26, 36, 37].

Since elevated serum estradiol levels are responsible for the behavioral changes seen during estrus

[38], cows with diminished levels of the hormone may fail to show estrous activity or may not mount as often

or for as long. Higher serum adrenocorticotropin concentrations have also been found in heat stressed cattle,

which can also block estrous behavior. The length of estrus is also thought to be shorter in dairy cattle suffering

from heat stress [26, 39].

Heat stress is now believed to alter the normal follicular dynamics pattern. One theory of this

disruption is through its influence on follicular estradiol levels [40]. Follicular estradiol levels have been shown

to be decreased during periods of heat stress, causing disruption of normal folliculogenesis of the first-wave

dominant follicle. Thus, secondary follicles are not suppressed properly and the second-wave dominant follicle

appears early [41]. The second-wave dominant follicle does not appear to function normally, having a

decreased size in heifers and producing lower levels of estradiol-17ß in both cows and heifers [42]. Possibly

due to impaired oogenesis, heat stressed dairy cattle have the potential to ovulate older, less fertile oocytes [42-

44].

It is also thought that oocytes, like sperm, have a lower capacity to adapt to heat stress than normal

cells. For example, normal cells increase production of heat shock protein 70 in response to heat stress;

whereas, bovine germ cells have not been shown to have this ability [45, 46].

Estrus detection efficiency can be increased with the use of aids such as tail chalking, Kamar,

pedometers, HeatWatch transponders, androgenized females, or surgically altered males [12, 14, 17, 47].

Little research has been done to test these devices in a heat-stressed environment but some indications are that

they improve estrus detection efficiency within herds also using a prostaglandins for synchronization [14, 24].

12
Timed insemination using hormones is showing promising results by shortening interestrus cycles,

without affecting estrus cycle length itself [14]. By using timed insemination, estrus detection efficiency

increases dramatically because there is no need for visual observation of standing estrus before the first

insemination [30]. Days open [8, 25] and days to first service appear to be shorter, although services per

conception were higher in one study [25]. Conception rates were slightly lower in synchronized cows in some

studies, depending on the protocol used [6, 30, 48] but overall reproductive efficiency was still enhanced due to

higher estrus detection efficiency. A study by Moreira et. al [49] demonstrated improved conception when

timed insemination was used concurrently with bovine somatotropin (bST). Although bST improved

pregnancy rates for timed insemination, by itself bST is reported to decrease estrus detection efficiency by

suppressing or diminishing the behavioral signs of estrus [49]. Timed breeding may improve overall group

reproductive performance, depending on other management decisions.

Although timed insemination is promising, it can do nothing to protect the fertilized egg from being

lost due to heat stress. Cows are not only more difficult to observe in estrus during periods of heat stress, they

are also less likely to become pregnant once inseminated [23, 25, 31]. It is believed that higher early embryonic

mortality results from lower serum progesterone levels, shunting of blood away from the uterus to the periphery

for cooling, and higher uterine temperatures [26, 33, 50]. In one timed insemination study conducted by de la

Sota, first time conception rate was only about 13.9 +/- 2.6% during periods of heat stress (assuming 100%

estrus detection efficiency) [25]. One study by Ullah et. al [33] demonstrated promising results of the ability of

GnRH administration at estrus to increase embryonic survival during periods of heat stress, presumably by

increasing serum progesterone. More work needs to be done to confirm this finding.

Heat stress on the day a cow is bred is not thought to be as important as the environmental temperature

the day after a cow is bred [23]. Optimal environmental temperature the day after insemination with frozen

semen has been shown to be between 100 and 230C [23]. In fact, heat stress up to 7 days post-fertilization has

been shown to increase embryonic mortality [51]. As embryonic development proceeds, it appears that

embryos become more resistant to heat stress such that there is no noticeable effect on lost pregnancies 7 days

post-fertilization [52].

Problems with heat stress and reduction in pregnancy rate is mainly a problem of lactating cows. As

cows produce milk, heat is generated due to the increased rumen heat associated with higher feed intake. The

more milk a cow produces, the more heat she generates and the more susceptible she is to heat stress [32, 53].

13
In one study, season significantly influenced pregnancy rate along with other factors, such as days-in-milk,

herd, milk production, etc. [26].

In summary, during periods of heat stress, chances of detecting dairy cattle in estrus diminish,

especially on a farm that solely uses visual observation for estrus detection. Fertility in cows during estrus also

decreases, which causes conception rates to drop. Low estrus detection efficiency and conception rates have a

disastrous effect on herd pregnancy rates.

Impaired reproductive capacity associated with heat stress does not seem to be a characteristic of all

breeds. Bos indicus breeds appear to be more resistant than Bos taurus breeds to heat stress. Bos indicus breeds

have more normal and more fertile oocytes at high temperatures than do Bos taurus breeds [54, 55]. The

following table (Table 5) demonstrates breed differences found in one study [55]:

Table 5. Breed Differences in Effect of Season on Production of Embryos Via in vitro


Maturation, Fertilization and Development in Louisiana [55].

Breed Variable Cool season Warm season


Holstein No. of oocytes 67 28

Oocytes classified as normal (%) 80.0 +/- 19.1 24.6 +/- 6.3**

Fertilization rate (%) 59.8 +/- 11.7 52.3 +/- 10.6

Oocytes developing to blastocysts (%) 29.0 +/- 14.8 0**

Brahman No. of oocytes 83 89

Oocytes classified as normal (%) 83.3 +/- 17.4 77.0 +/- 6.3

Fertilization rate (%) 83.1 +/- 10.7 79.3 +/- 10.6

Oocytes developing to blastocysts (%) 52.3 +/- 13.5 41.3 +/- 7.2
**
P<0.01

Natural service is not recommended during periods of heat stress as bulls have been shown to be less

fertile during such times [24]. Although artificial insemination (A. I.) is preferable in such situations, frozen

semen must be handled appropriately to decrease semen loss to inappropriate thawing procedures, sudden

temperature changes after thawing, and improper placement of the semen within the uterus. Technicians must

be appropriately trained for conception rates to be at their best [14]. As of January 2001, there has been no

detailed investigation on the effects of heat stress on the survivability and fertilizing capacity of sperm cells in

utero.

14
Proper artificial insemination techniques may increase sperm cell access to the oocyte but they do little

to nothing in increasing fertility within the female. The literature appears contradictory as to whether decreased

female fertility is a result of a decreased oocyte fertilization rate, an impairment of normal embryonic

development, or a combination of both factors [24].

To decrease the devastating effects of heat stress on pregnancy rates, the producer must do at least one

of three things:

1) Make cows cooler by manipulation of their environment (i.e. providing shaded structures and

using cooling aids such as fans and/or sprinkler systems),

2) Genetically select cows which are phenotypically more resistant to physiological changes due to

heat stress, or

3) Alter detrimental physiological processes before damage to either germ cells or fertilized oocytes

occurs [24].

Cooling systems can increase pregnancy rate, even though pregnancy rates of cows in artificially

cooled environments seldom equal those of cows exposed to cooler ambient air temperatures [34]. Some studies

indicate that even cooling cows for several days around insemination can increase reproductive efficiency.

Cooling cows for just 8 days after PGF2α administration and insemination resulted in a 10% increase in the

pregnancy rates between treated and control cows [56].

While selecting for thermally-adapted breeds of Bos indicus cattle may not be desirable for most

producers from a management standpoint (lower milk production, ill temperament of certain breeds), non-

adapted breeds can be genetically selected to help counteract the effects of heat stress on fertility [24, 31].

Selection for coat length [57] and color [58, 59] and for rectal temperatures [50] are several parameters that can

help non-adapted breeds perform better in a warm climate. In fact, heritabilities for such traits are often high, as

is the case for rectal temperatures (0.25-0.65) [60]. Among European breeds, Holsteins’ rectal temperature has

been shown to increase more dramatically than Jerseys’ during periods of heat stress [50]. Unfortunately,

thermotolerance and milk production have a negative correlation to one another and are thought to be

genetically linked [61]. Also, A. I. sires are not commonly tested for performance by environment interactions

[34, 61]. Others propose that in tropical and subtropical countries, where management and feed are low, crosses

between the Bos indicus and Bos taurus species provide the best overall milk production. However, in

15
countries where high quality feed is available and management conditions are fairly good the best success is

attained by using Bos taurus breeds of cattle [34].

Some studies have suggested the increased use of embryo transfer within tropical developing countries

to decrease generation intervals and more accurately predict heritabilities for the production traits and

reproductive performance of popular A. I. sires within a tropical setting [61]. It may be possible in the future,

through transgenics or conventional selection, to transfer heat shock resistance from thermally-adapted to non-

adapted breeds. More work must be done, however, to identify these genes and to incorporate them into the

genotype of non-adapted breeds while preserving as much milk production as possible [24].

One alternative in genetic selection that has proved successful in several studies is the use of F1 Bos

taurus-Bos indicus crosses. F1 individuals have been shown to have shorter calving intervals, are younger at

first calving, and have higher first lactational milk yields than either breed alone in a tropical environment.

Increasing Bos taurus in subsequent crosses did little in increasing reproductive efficiency in subsequent F2

crosses [61].

Nutritional supplementation of heat stressed cattle is a relatively new field. A few reports have

suggested that feeding supplemental niacin [62] or Aspergillus oryzae extracts [63] can decrease the body

temperatures of cattle undergoing heat stress.

Nutrition

Fertility can be profoundly affected by nutrition. Reproductive outcomes may be closely related to the

ration on a given farm. In fact, upon exclusion of male infertility and severe uterine pathology, studies have

shown that 80% of the variance in fertility is due to environmental factors and 50% of that estimate is

influenced by nutrition. Nutritional factors influence fertility in a number of ways, including increasing an

animal’s predisposition for infectious disease and lameness and by creating a negative energy balance and loss

of body condition to name a few. Feed quality and composition are not only important from an energy balance

standpoint, but also because some substances (i.e. mycotoxins) found in poor quality feed can actually adversely

affect reproduction directly [64].

Although some studies disagree [11, 13], other studies indicate that as milk yields increase,

reproductive efficiency generally decreases [9] due to additional physiological demands on animals, although

the exact mechanism involved remains unclear. Hernandez et. al [13] found, however, that low yielding cows

had longer calving-to-conception intervals, probably due to inadequate dry matter intake and subsequent

16
negative energy balance. Whatever the impact of milk yield on reproductive efficiency, investigators seem to

agree that its major effect on reproductive performance is due to improper nutritional management. Therefore,

proper ration management is critical in optimizing reproductive performance in herds where increasing milk

production is the primary goal and/or in herds where dry matter intake may be insufficient, like many of those

within the D. R.

Proper feeding for optimal fertility within dairy cattle is guided by several important principles. First,

rations must be balanced properly for all stages of lactation as well as the dry period. Secondly, both excess

nutrients and deficient nutrients can cause a problem with reproductive performance. Energy deficiencies can

cause a cascade of physiological effects that can severely decrease reproductive efficiency within dairy herds.

Energy deficiencies in the face of high levels of protein are thought by some to cause excess levels of

circulating ammonia to flow through the liver, resulting in inactive ovaries, endometritis, and early embryonic

death [64].

The following schematic, taken from Lotthamer 1989 [64], provides a summary of the interactions

between nutrition (specifically energy balance) and reproductive efficiency:

17
Figure 1. Effects of Energy Imbalance and Acidosis on Health and Reproductive Performance in
Dairy Cows Both Before and After Calving [64].

Dry excess of deficiency of deficiency of


ENERGY ENERGY CRUDE FIBER

Period
(fatty liver, liver damage) ketosis acidosis

lack of acetic acid

Subclinical ketosis Liver damage estrogen deficiency

Parturition Milk Production retained placenta


Low intake metabolic insufficiency
Lactation endometriosis

ENERGY DEFICIENCY
Adrenal ketosis
Stress fatty liver Acidosis
C. N. S. damage
Hypothalmus
Pituitary immunosuppression

FSH and LH mastitis

Ovarian dysfunctions atonia uteri


Silent heat
Delayed ovulation delayed involution general
Follicular cysts infections
dystrophy
endometritis
Laminitis
Decreased estrus detection efficiency and conception rates

It has been suggested that temperature and humidity play one of the most significant roles in dry matter

intake within a tropical environment, especially for lactating Holstein dairy cows [31]. Heat stress can trigger

the central nervous system and hypothalamus to decrease the amount of forage consumed. Perhaps

surprisingly, heat stress does not seem to concurrently decrease concentrate intake, which can lead to lameness

if forage to grain ratios are inappropriate. Feed quality and quantity are thought to be the most important non-

environmental factor influencing dry matter intake in most settings, and subsequently milk yield and

reproductive performance [31]. Better feed quality in the lactating ration has been shown to increase herd

conception rate [64]. Unfortunately, there seems to be no set of definitive directives about proper ration

management within the tropics [31, 64].

18
Chapter 3

Materials and Methods

Initial Observational Visit

An initial four-week visit was made to the Dominican Republic, during late December 2000 – early

January 2001, for the purpose of evaluating the quality and general quantity of data retrievable from average

dairy record keeping systems and to get an idea about general reproductive and herd management. These

observations were later used to design a questionnaire which will be discussed in further detail later.

During the initial visit, 22 farms were evaluated within four of the five regions of the country [2] (see

Appendix A for regional breakdown of the farms visited). Farm visits were not evenly distributed between

regions and were chosen based on the expertise and previous work experience of Dr. Leonardo Tineo, a

Dominican veterinarian and professor with over 30 years experience, and on the recommendation of local

producers. Farm size, owner membership in local cattlemen’s associations, and previous cooperation on the

farm were some of the determining factors of farm choice. The visits consisted (of an interview of about 17)

questions assessing the characteristics of farm reproductive management, type of record keeping, and history of

reproductive problems (i.e. postpartum metritis, high incidence of repeat breeders, etc.) (see Appendix B for

initial interview questions). A chart was also left at all the farms visited, except those with the PCDART record

keeping system, and was to be filled out by the herd manager and/or herd owner (see Appendix B). The chart

contained the information needed to calculate reproductive statistics (palpation pregnancy rate, estrus detection

accuracy and efficiency, etc.) each month between the initial observational visit and the second visit scheduled

for about 6 months later. Only one of these forms was returned on the subsequent visit to the country, rendering

it of little use in data collection.

Survey design and observational data collection

Based on observations made during the initial visit, a survey was designed to assess reproductive

management (see Appendix B for survey form). Since the second visit was to be largely observational in nature,

survey questions were fairly broad and designed to assess typical farms within a given region and to gather

information for inferences about reproductive, nutritional, and general farm management. Since little has been

written about the typical dairy within the D. R., this type of survey was necessary that important information

not be missed. Also, it was decided that it would be extremely difficult to statistically compare farms with a

19
small number cattle (10-15) and larger farms having several hundred cattle in the milking herd. Based on the

expertise of Dr. Tineo regarding typical D. R. farm size, it was decided that farms should be selected in the

medium to large size range (> 40 cows in milk preferably).

During the second visit, which lasted 10 weeks, 43 farms (6461 cows) were evaluated throughout four

of the five regions of the country. These were selected as follows: first, a list of farms meeting set research

criteria (discussed below) was made from the major milk-producing provinces within the five major regions of

the country[2]. Individual farms were selected mainly with the expertise of two Dominican veterinarians, Dr.

Leonardo Tineo and Dr. Ramon Martinez, a veterinarian working for a local milk processing plant (Leche Rica)

in genetic extension and semen sales. Farms were selected based on their size (preferably 40 adult cows or

more), availability of data, and demonstrated motivation by the owner in improving the herd (i.e. membership in

a local cattlemen’s association, owner’s willingness to adopt new technology, history of thorough record

keeping, etc.)

Farm visits were tentatively scheduled by region. Since most of the farms fitting our survey criteria

were located to the north and west of the capital, we did not visit the eastern region of the country. Each week,

farm owners were contacted and visits arranged, preferably when the owner would be available for interview.

Visit times were scheduled with an attempt to arrive prior to milking when lactating cows could be inspected

individually as they came through the parlor. This practice was not always possible, hence it was determined

that at least 70% of the milking herd should be available for body condition and lameness scoring for the

estimate to be considered representative of the lactating herd.

Herd evaluation was broken into three major components: owner interview, farm and data evaluation,

and adult cow evaluation. The farm and data evaluation were combined and consisted of an 11-question survey

and 2 charts, which were to be completed by the investigator (see “Operator Questionnaire”, “Records Survey”,

and “Milking Herd Evaluation” in Appendix B). This questionnaire was divided into three main parts: general

information (8 questions), record information (3 questions), and a milking herd evaluation chart.

The “general information” category assessed the perceived level of owner cooperation to get an idea

of owner attitudes toward the project and to gauge the level of herd information to which the investigator was

given access. Average body condition and lameness scores of all adult cattle (lactating and dry) were evaluated

to assess nutritional management and to assess the prevalence of lameness in the herd, since lameness has been

shown to affect reproduction [13] and because specialists working in the dairy industry perceive a problem with

20
lameness in adult cows. Body condition and lameness scores were taken from 70-100% of the individual

animals within each milking herd, depending on availability of milking cows on the day of the visit. The

average body condition score of the dry cows was taken by looking at the whole dry cow group and comparing

the average body condition score of these cows to that of the milking herd to assess dry cow nutritional

management [65]. Lameness was graded on a scale from 1 to 4: with “1” being normal gait and posture while

standing and walking; “2” being normal posture while standing with abnormal gait (possibly intermittent

between normal and abnormal) and a curved back when walking; “3” being an arched back while standing and

walking and an abnormal gait; and “4” being resisting touching foot to the ground while standing with

reluctance to move [66]. Cows were typically body condition scored during milking and lameness scored on

concrete or hard-packed soil as they exited the milking parlor. If grading did not begin until the milking was

underway or ended before the milking was done, individuals at the beginning and ending respectively of the

milking line were body condition and lameness scored to avoid a convenience sample. Occasionally, however,

cows had to be graded in muddy lots or while on pasture. In this event, animals were graded from all portions

of the field trying to avoid skewing outcomes with a potential convenience sample.

Breed was determined by the investigator for each milking cow during body condition and lameness

scoring. Breed was collected individually from the records for any cow not evaluated during the body

condition and lameness scoring.

To ensure that breeds were evenly distributed between the lactating and dry groups, after individually

assessing the lactating cows, the dry cow group was observed to see if breeds were proportioned evenly

between the two groups. Several farms have been observed to have a high proportion of non-pregnant cows

within the dry group: cows that did not get pregnant throughout their lactation and were dried off and then

inseminated with the hope that they would get pregnant once they were free from the physiological stress of

lactation. Producers also complain of poorer conception rates during the hotter months (March to November).

Given this situation, by comparing breed distributions within both the dry and lactating groups especially during

the summer months, it was hoped that any appreciable breed differences in fertility would become evident.

Presence and type of cow cooling devices were recorded for further investigation of any possible links

between cooling systems and fertility. Cooling systems were recorded according to the following categories:

“fans,” “sprinkler systems,” “shaded barns,” or “other”. The “other” category contained a description of the

shade structure or cooling device.

21
The “record information” category assessed the number of cows in lactation, type of records used, and

contained a table to record reproductive events based on reproductive pregnancy exam findings over the past

year, following the format of the chart left during the initial visit. Due to low compliance, this chart was not

useful since the needed information could only be gleaned from 2 farms.

Reproductive information was collected by hand on individual cows during the second visit. The

“milking herd evaluation” table was used as a means of individually recording data for adult cows for the

following information* (see Appendix B):

-Identification (to link individual cows with body condition and lameness scores)

-Breed

-Body condition score (lactating cows only)

-Lameness score (lactating cows only)

-Date of most recent parturition

-Whether or not the birth was normal (to determine incidence of dystocia and abortion)

-Whether or not twins were born

-Presence or absence of postpartum metritis

-Date of previous parturition

-Date of first recorded breeding for the current lactation

-Date of last recorded breeding for the current lactation

-Number of services since the last calving (separate categories for dry and lactating cows)

-Pregnancy status

-Number of services for last successful calving for non-pregnant cows

-Date of last conception (as determined by the breeding records and rectal palpation)

-Whether conception was by natural service or artificial insemination (pregnant cows only)

-Days in milk (days since last calving for lactating cows).

The following equations were then used to calculate outcome (dependent) variables:

Service Rate = Number of inseminations / (days in milk – VWP)

22
Services per Pregnancy = Number of inseminations from calving to conception for current lactation
(pregnant cows)
OR
Number of inseminations from calving to conception for the previous lactation
(non-pregnant cows)

Projected Calving Interval = (Date of conception* – date of parturition for this lactation) + 283

Days to First Service = Date of first insemination for this lactation – date of parturition for this
Lactation

Historical Calving Interval = Calving date for this lactation – Calving date for the previous lactation

Days Postpartum = Date of farm visit by investigator – Calving date

Calving to Conception Interval = Date of conception* – Calving date

* Usually determined as last insemination before diagnosed pregnant on palpation. If the insemination date was not
available it was estimated from palpation records.

The owner interview consisted of a 43-question survey. The survey was divided into three major

categories: general farm information, information on the farm’s reproductive program, and information on the

farm’s nutritional program. The “general information” category of the owner interview consisted of 7 questions

and assessed such areas as farm income as proportion of the owner’s overall income, frequency of owner visits

to the farm, average milk production, farm size, etc. The purpose of this section was to gauge owner interest in

and motivation to improve the farm and to determine the scale of the operation. The section also contained a

chart designed to describe the typical day for a milking cow on the farm, to assess number of times cows were

milked daily and a preliminary idea of what activities cows were engaged in when they were observed for

estrus.

The “reproductive program” category consisted of 29 questions and assessed the farm’s estrus

detection program, insemination program, pregnancy evaluation schedule, etc. The first nine questions of the

survey were designed to find out if the farm used mechanical aids to help with estrus detection, how many

people were responsible for estrus detection on the farm, and who was responsible for and how were observed

estruses recorded. The latter was to give the investigator an idea of how “error prone” estrus records might be

(i.e. it was assumed that the longer it took for an estrus to be recorded the more likely it was to be recorded

incorrectly). Survey questions 10 and 11 were designed to assess the presence/absence and appropriateness of

the farms’ VWP and to see if the farm rigidly adhered to it when initiating breeding. Question 12 assessed the

23
number of hours the milking herd spent on concrete, since it was assumed that the more hours that were spent

on concrete the less likely cows were to exhibit standing estrus behavior and possibly the higher the incidence

of lameness within the herd. Questions 13-15 investigated the farm use of reproductive hormones for

synchronization. Questions 16-21 assessed if the farm used bulls for breeding, the extent of their use, and the

aspects of bull reproductive management. Questions 22-25 investigated the farm’s artificial insemination

program including number and training of inseminators, number of times per day cows are inseminated on the

farm, and one question on artificial insemination technique. Question 26 was applicable to both artificial

insemination and hand-mating herds and asked about time between first observed estrous activity and

insemination. Finally, questions 27-29 assessed the farm’s palpation program, which provided, among other

helpful information, an assessment of the accuracy of pregnancy status data within the records.

The “nutritional program” category consisted of 7 questions and assessed the procedure for ration

formulation and feeding, types of feed used in the ration, and feed bunk management. Due to the profound

effect nutrition can have on reproductive efficiency and the prevalence of laminitis observed incidentally during

the initial visit, a short section assessing nutritional management was a necessary part of the survey.

Once the on-farm evaluation was complete, survey responses and any written observational data were

entered into a spreadsheet to serve as a “back-up” should hard copies of any survey data be lost. Record

information was also kept in a computer spreadsheet. Each farm was coded with a number, which would later

serve as its identification code. Farms were also coded as “reliable” or “not reliable” based on the perceived

quality of the information (i.e. amount of information present, consistency with which information was recorded

as evidenced by gaps within the data). All individual cows with perceived gaps in their breeding and/or calving

records were also coded based on the type of error present in case it was determined that they should be

excluded from later analyses.

Statistical preliminary analysis

Once observations had been made and catalogued, it was determined that all data would be

summarized on the herd level. This summary step was taken for two reasons: to ensure that data from larger

farms did not skew survey data which was taken on the herd level and to simplify data manipulation and model

construction. Herd averages were calculated for body condition scores, lameness scores, and all potential

dependent variables (see Table 6). These averages along with the survey data were incorporated into one SAS

data set.

24
Although the data gathered in this study is best appreciated in a descriptive context, multiple regression

techniques were useful in selecting the variables which described the most variation within the data. This

method was chosen due to the number of possible independent variables involved in the study.

Based on the quality and quantity of information available, 7 dependent variables and 18 independent

variables were initially chosen for investigation. These variables are summarized in the following tables:

Table 6. Initial Candidate Dependent (Outcome) Variables

Dependent Variables Description*


Service Rate Proportion of inseminations over the total number of estrus cycles
expected for a given period of time, expressed as the likelihood of
insemination per day

Days to First Service Days from parturition until the first insemination

Services per Pregnancy Number of services required for pregnancy to occur

Days Postpartum Number of days since last calving for both lactating and dry cows

Historical Calving Interval Number of days between last parturition and the previous parturition

Calving to Conception Interval Period of time (days) between last parturition and insemination resulting
in pregnancy (pregnant cows only)

Projected Calving Interval Number of days between last parturition and next expected parturition
(pregnant cows only)
*For equations used in calculation of dependent variables, see Appendix C.

25
Table 7. Initial Independent (Regressor) Variables

Independent Variables Type of Description


Variable
Region Categorical Region of the country where the farm was
located

Average milk yield Continuous Average daily milk yield per cow of the farm in
liters

Cows/herd Continuous Number of cows in the milking herd (dry cows


not included)

Cooling Categorical Type of cooling system used (none, shade,


evaporative/enhanced)

Hours on concrete Continuous Number of hours cows within the milking herd
spend on concrete per day

Amount Concentrate fed Continuous Average Amount of Concentrate fed per cow
per day (kg)

Method of Estrus Detection Categorical Method of Estrus Detection used (none—bull


stays in the herd, visual, visual with mechanical
aids)

Number of people involved in Continuous Number of people involved in estrus detection


estrus detection

Time for estrus detection Continuous Length of Time spent each time cows are
observed for estrus (minutes)

Hormone Use Categorical Are hormones used for estrus synchronization


on the farm? (yes/no)

A. I. versus N. S. Categorical Is artificial insemination used to predominantly


breed the cows? (yes/no)

Body condition score (BCS) Continuous Average body condition score of the lactating
herd (1-5)

Lameness score Continuous Average lameness score of the herd (1-4)

Reproductive Health Statistics Categorical Presence of postpartum metritis (yes/no)

Breed Categorical Holstein/Jersey/Brown Swiss/Mixed

Type of Records Categorical Type of record keeping system used on the farm
(notebooks, individual cow cards, computer
program)

Days in milk (DIM) Continuous Number of days spent in current


lactation(lactating cows only)

Nighttime estrus detection Categorical Does the farm check for estrus between 6 p.m.
and 6 a.m.? (yes/no)

26
“Days in Milk” was added to the potential list of dependent variables, instead of acting as an

independent variable, because an investigation of the factors influencing the parameter was desired. However,

the measure is cross-sectional in nature and must be interpreted loosely. More confidence could be placed in

validation of this measure if multiple visits had been made to the same farm over the course of several months

to a year, to adjust for seasonal differences in calving and management.

“Reproductive Health Statistics” was eventually determined to be an unsatisfactory measure of

reproductive problems as some farms did not keep records of cows presenting with postpartum metritis and the

category did not take into account other reproductive health events such as twinning and dystocia. Therefore,

“Reproductive Health Statistics” was combined with twinning and abnormal births to become “Reproductive

Event”, a continuous variable, and was calculated as the proportion of cows within the herd that had

experienced any reproductive anomaly including dystocia, twinning, and/or postpartum metritis. Cows were

assigned a “1” or a “0” based on the presence or absence respectively of abnormalities in any or all of these

three categories. Then the proportion of the herd experiencing a reproductive event was calculated and reported

as an average percentage for the herd.

“Breed” was interpreted as a continuous variable rather than a categorical representing it as the

percentage of Holsteins within a given herd. This decision was made because all other variables were

summarized on the herd level and it was very cumbersome to summarize breed as a categorical variable on the

herd level. Also, there was insufficient information to interpret the category “mixed” breed as there was no way

to tell from the data which crosses made up a given individual (small versus large-framed animals, thermally-

adapted versus non-adapted breeds). Also, it was felt that smaller native, tropical Criollo mixed breeds would

perform better in the D. R.’s tropical environment than Holstein crosses; therefore, combining all crosses into

one category was not desirable for determining breed by environment interactions [31]. Since Holsteins made

up about 60% of the animals for which data was collected in the study followed by Brown Swiss (17%), it was

felt that the single best herd estimate for breed interactions with reproductive outcomes would be the proportion

of Holsteins within a given herd.

Another important issue is the question of feed standardization. Since the dry matter intake for larger

and smaller breeds differ, there was discussion about whether the “average amount of concentrate fed” should

be standardized to reflect those breed differences. Upon observation of the data, however, it was found that

27
Jerseys only make up about 2% of the total sampled population with only two farms having more than 5%

Jerseys (9.6% and 48.82%) and only one of these farms had a recorded “average amount for concentrate fed”.

Also, as previously stated, there was no way to distinguish small and large breed crosses from the data so it was

virtually impossible to tell which farms had a significant proportion of large and small mixed breeds.

Therefore, feed was not standardized by breed for this analysis. Since breed had been added to the list of

regressors, an interaction term between “breed” and “amount of concentrate” could be added later if thought

necessary.

Using SAS1 [67], frequency charts were made for all categorical independent variables looking for

gaps within the data distribution. “Cooling” was found to have only one farm listed under “none”. Upon

reviewing notes taken while on the farm it was decided that this farm could be re-categorized “shade” as it had a

few trees in pasture and a small shaded barn in the milking holding area, changing “cooling” from a three to a

two-category variable: shade only or enhanced mechanical cooling. All remaining three-category variables

were converted into “dummy” variables such as can be easily handled by SAS. Cross tabulations were made in

SAS of all independent categorical variables. “Method of Estrus Detection” was found to only have 2 herds

within the “natural service” category. Since natural service, depending on bulls to catch and inseminate cows in

estrus, and the other two categories, which relied on some sort of human observation of estrus, are so different,

it was decided to include these two herds in the analysis. It was hoped that the bull-bred herds might give some

insight into the role visual estrus detection plays in reproductive outcomes in this tropical environment. These

two observations had to later be removed from the study since the farms were limiting the number of total

observations in other areas of record keeping during model building due to absence of breeding records, etc.

For the voluntary waiting period variable, farms reporting not having a VWP were assigned a herd average of

40 days since, based on field experience, it was felt that it would take the average herd about 40 days to

properly involute and begin cycling normally across all individuals [68].

After the data had been examined for gaps in distribution, collinearity and influence diagnostics were

performed. All the independent and dependent variables were plotted against all other variables looking for

collinearity and/or points outside the data set which might inaccurately influence the model. In the plots of the

dependent variables, several points below the biological minimum were seen so the following criteria were set

1
SAS, version 8.1, SAS Institute Incorporated, Cary, NC, 27513

28
to help eliminate erroneous data for several of the reproductive outcomes with lower values than would

normally be physiologically possible for individual animals:

-Projected Calving Interval must have a minimum of 295 days,

-Historical Calving Interval must have a minimum of 295 days,

-Days to First Service must have a minimum of 25 days,

-and Calving to Conception Interval must have a minimum of 25 days [68].

A criteria was also set that all cows must have no more than 500 days in milk, since it was felt that it

was biologically not likely for a cow to be lactating for over 500 days [68]. If a cow had a reproductive

outcome recorded which was lower than these, it was assumed to be an error and the outcome was subsequently

removed from the data set. In general, however, the problem of high influence points lying above the rest of the

other farms was not so easily remedied. All 7 dependent variables had “suspect” or high influence points lying

well above the range for the rest of the data. Each high influence point was examined for possible errors. The

high influence points for “Services per Pregnancy” and “Days to First Service” were found to be accurate and

were left in the study. “Days in milk”, “Days Postpartum”, “Projected Calving Interval”, “Historical Calving

Interval”, and “Calving to Conception Interval”, however, were all skewed averages resulting from one farm

with a few extreme outlying individuals. Further inquiry suggested that these individuals were cull cows that

had not been removed from the records and so their record information was subsequently deleted from the data

set.

Within the dot plots of the independent variables, high influence points were found for “average

amount of milk produced on the farm”, “number of employees involved in estrus detection”, and “reproductive

event”. These points were examined individually and additional information was sought. (Since no published

averages exist for most of these values, information was gleaned from Dr. Tineo and other veterinarians with

work experience in the D. R. [68].) The data points were then deleted or kept depending on whether the

information was felt to be inaccurate or real respectively.

In addition to dot plots, Pearson Correlation Coefficients and p-values were calculated to look for

collinearity between variables (see Table 8 for the Pearson Correlation Coefficients and p-values for the

dependent variables). Since several of the variables were highly collinear and several were not as interpretable

as would be desirable due to their cross-sectional nature (i.e. days postpartum and days in milk), upon

29
examination of the scatterplots, correlation coefficients, and p-values, the list of 8 dependent variables was

reduced to the following four:

1) Projected calving interval—which gives an indication of overall reproductive performance [4],

2) Services per Pregnancy—which gives an indication of fertility or the likelihood of conception for

each breeding,

3) Service rate—which gives an indication of estrus detection efficiency, and

4) Days to first service—which also gives an indication of estrus detection efficiency and postpartum

management (i.e. how efficiently postpartum cows return to normal cycling activity and are

judged fit for insemination).

Table 8. Pearson Correlation Coefficients and p-values


for Potential Dependent Variables.

Service Rate Days to First Services/ Days


Service Pregnancy Postpartum
Service Rate Corr coef. 1.00 -0.24 0.27 -0.19
p-value 0.14 0.09 0.22
Days to First Corr coef. 1.00 -0.25 0.49
Service p-value 0.12 0.002
Services/ Corr coef. 1.00 0.18
Pregnancy p-value 0.28

Historical Calving to Projected Days in Milk


Calving Conception Calving
Interval Interval Interval
Service rate Corr coef. -0.06 0.003 0.10 -0.11
p-value 0.73 0.98 0.55 0.48
Days to First Corr coef. 0.44 0.49 0.48 0.39
Service p-value 0.01 0.001 0.002 0.01
Services/ Corr coef. 0.03 0.29 0.45 -0.76
Pregnancy p-value 0.88 0.07 0.004 0.64
Days Corr coef. 0.69 0.83 0.88 0.81
Postpartum p-value <.0001 <0.0001 <.0001 <.001
Historical Corr coef. 1.00 0.92 0.78 0.35
Calving p-value <.0001 <.0001 0.05
Interval
Calving to Corr coef. 1.00 0.93 0.38
Conception p-value <.0001 0.02
Interval
Projected Corr coef. 1.00 0.41
Calving p-value 0.01
Interval

Next, regression analysis, using the “proc reg” function in SAS, was begun on “service rate”, one of

the four dependent variables, to look for potential problems before model building was initiated in earnest.

30
Only 16 observations were recorded for “service rate,” indicating that 26 out of the 42 farms were missing

observations for the dependent variable itself or one or more of the independent variables, causing the farm to

completely drop out of the analysis. Additionally variance inflation factors (VIF) were high in the presence of a

high R2 (R2 = 0.77) indicating that two or more of the independent variables were highly correlated with one

another and were explaining some of the same variation within the model. Charts were made to calculate the

number of observations for each dependent and independent continuous variable to determine which variable(s)

could be eliminated to increase the number of observations (farms) included in model building. The following

table summarizes the results:

Table 9. Number of Observations Recorded for Dependent and Independent Continuous Variables.

Variable Name Independent or Dependent Number of observations


Variable recorded
Service rate Dependent 41

Services per pregnancy Dependent 40

Projected calving interval Dependent 39

Days to first service Dependent 39

Nighttime estrus detection Independent 40

Average amount of milk Independent 41


produced by the farm

Amount of concentrate fed Independent 41

Time observed for estrus Independent 26


detection
Average body condition score Independent 42

Average lameness score Independent 41

Reproductive event Independent 41

Due to the low number of numerical observations in “time observed for estrus detection” (26), this variable was

dropped from subsequent analyses. The more farms that could be included in a given analysis, the higher the

degrees of freedom and the more statistical power imparted to the analysis.

Next, eigenvalues, condition indices, and variance proportions were calculated as a further diagnostic

for collinearity between the regressors. Eigenvalues near zero with condition indices of greater than 30

indicated collinear groups of regressors. Variance proportions were used to determine exactly which regressors

31
were explaining the same variation within the model. It was initially determined that “average amount of milk

produced per cow by each farm” (avmilk) and “hormone usage” (hormone) were very strongly correlated.

“Average milk” was removed for the next calculation of eigenvalues and the variance proportion dropped from

91.3% in the first calculation to 11.8% in the second, indicating that one of the collinear variables had truly

been eliminated. Therefore, it was determined that both variables should not remain in the same model at the

same time since they both describe the same variation, introducing instability into any resulting models.

Once this problem was resolved, variable selection and model building began for each of the four

dependent variables selected as described above. (For specifics on model building for each dependent variable,

please see chapters 5-8 pertaining to the respective variable.) In general, using SAS, Mallow’s CP analysis was

used to determine the maximum number of regressors each model could contain with minimal bias. A

Mallow’s CP graph plots the CP (a test statistic) value against the number of predictors within the model with

the highest R2 for a given number of regressors. As the simulated model approaches the true but unknown

model, the expected value of CP equals the number of predictors (E(CP) = p), a linear relationship. However,

since errors exist within regressor coefficients, the simulated model plot appears parabolic. Therefore, a straight

line is drawn through the parabola, such that CP = p, to find the point at which the parabola first crosses the

line. This point is the point of minimum bias for the parameter estimates, which also corresponds to the

maximum number of predictors (regressors) allowed for minimal bias [69]. (For an example of a Mallow’s CP

plot, please see Figure 2.)

32
Figure 2. Mallow’s CP Plot for Projected Calving Interval.

Max R2 was used in variable selection. Due to the Dominicans’ interest in the potential effects breed

has on reproductive efficiency, breed (“percent Holstein”) was forced into every model so that the effects of

breed on each outcome variable could be observed. Breed interactions were created but not forced into the

models, unless otherwise specified, for average milk production (avmilk), body condition score (BCS), amount

of concentrate fed (howmuchg), and cooling system used (cool).

When running max R2, a criterion was maintained such that each additional variable added at least 5

points to R2, while keeping the model statistically significant at a level of p< 0.10. The p-values for the

coefficients were not maintained at a p< 0.10 significance level because although variables selected during the

model building process had low p-values, these p-values changed once other component variables had to be

forced into the model. Interpretation of models, however, was generally limited to those variables with

coefficients having p-values < 0.15.

After variable selection was completed, the model was examined for high influence points using the

“proc reg” function in SAS with the “/influence” option statement. In specific, DFBeta values were primarily

examined. This procedure is recommended for a small to moderate sample size, such as the one in this project

[70]. DFBeta values give the number of standard errors that a given coefficient changes if the ith observation

33
were set aside [70]. First, DFBeta values of >2 or <-2 [70] were identified and the farm responsible was

deleted. Secondly, DFBeta values were calculated again to ensure that no other farms contained high influence

observations. Once all farms with high influence observations had been deleted, variable selection using max

R2 was repeated on all possible independent variables to ensure that variable selection remained the same. Final

models were based on these selection procedures with high influence farms missing. By approaching variable

selection in this manner, chances of a high influence observation affecting the results of model building are

reduced drastically.

34
Chapter 4

Descriptive analysis

During the 10-week course of this study, 43 farms were surveyed throughout 4 of the 5 regions of the

Dominican Republic. Although extreme cases existed in all regions, each region had general differences and

similarities in management intensity, climate, and reproductive management practices. The differences and

similarities between regions have not been described in detail before. This chapter will characterize each region

by providing summary statistics for specific dependent and independent variables of interest. (See Appendix E

for additional descriptive information by region.)

To fully appreciate the descriptive summary, however, data collection should be characterized more

thoroughly. Survey data evaluating management intensity, reproductive management practices, and nutritional

management was gathered by surveying farm personnel, preferably the owner. During the survey, 32 (74.4%) of

those questioned were farm owners, 8 (18.6%) were herd managers, and 3 (7.0%) were members of the owner’s

immediate family (spouse, older son). The farms surveyed were geographically distributed as follows: East and

West Cibao region 16 farms (37.2%), Santo Domingo region 19 farms (44.2%), and the Southwest region 8

farms (18.6%).

In addition to the survey data collected from the 43 farms, individual cow data was gathered for

roughly 6460 lactating and dry adult cows of various breeds: 3286 (50.9%) Holstein, 1087 (16.8%) Brown

Swiss, 951(17.5%) mixed breed, and 117 (1.81%) Jersey, with 1019(15.8%) cows having no breed information

available within the records. Individual cow data were taken from 12 (27.9%) farms that used notebooks, 17

(39.5%) farms that used individual cow cards, and 14 (32.6%) farms that used computer-based record keeping

systems. Notebook data was found to be less complete, in general, than other forms of record keeping often

only containing one year of information, depending on the organization of the individual cow information.

Distribution of record keeping systems by region will be discussed in the next section.

Reproductive Outcomes by Region

Reproductive management, nutritional management, and environmental management all influence

reproductive efficiency [4, 31]. Furthermore, it is assumed by participants in the Dominican dairy industry that

there are significant differences in reproductive outcomes between the various regions. An objective of this

study was thus to compare differences in reproductive outcomes by region. The following table summarizes

35
seven reproductive outcomes which were calculated based on individual cow data that was collected from

records secured on each farm. Indices that were calculated include: service rate ( svcrate), projected calving

interval (projci), days to first service (dysfstsvc), services per pregnancy (svcprpreg), days postpartum (dayspp),

historical calving interval (histci), and calving to conception interval (clvconin).

Table 10. Breakdown of Reproductive Outcomes by Region

Region No. Variable Median Mean Std. Minimum Maximum


Obs. Devia-
tion
Cibao 16 Svcrate (insem. / 100 d) 1.07 1.10 0.80 0 3.27
Projci (d) 421 456 93.2 380 725
Dysfstsvc (d) 89.4 98.1 35.4 57.3 193
Svcprpreg (services) 2.12 2.44 1.03 1.86 6.00
Dayspp (d) 239 277 107 184 602
Histci (d) 427 485 225 375 1157
Clvconin (d) 155 187 124 98.0 584

Santo 19 Svcrate (insem. / 100 d) 1.26 1.38 0.58 0.16 2.51


Dom- Projci (d) 448 435 40.7 348 497
ingo Dysfstsvc (d) 84.6 89.5 20.8 64.8 133
Svcprpreg (services) 2.62 2.44 0.46 1.59 2.98
Dayspp (d) 248 242 32.1 150 286
Histci (d) 435 442 59.5 375 575
Clvconin (d) 163 156 42.6 78.1 258

South- 8 Svcrate (insem. / 100 d) 0.47 0.74 0.58 0.43 2.11


west Projci (d) 402 412 47.6 360 478
Dysfstsvc (d) 112 109 26.3 70.2 142
Svcprpreg (services) 1.44 1.90 1.13 1.00 4.38
Dayspp (d) 232 229 35.0 168 276
Histci (d) 498 492 84.0 376 573
Clvconin (d) 146 156 72.7 77.3 297

Total 43 Svcrate (insem. / 100 d) 0.93 1.07 0.65 0.00 3.27


Projci (d) 424 434 60.5 348 725
Dysfstsvc (d) 95.3 98.7 27.5 57.3 193
Svcprpreg (services) 2.06 2.26 0.87 1.00 6.00
Dayspp (d) 239 250 58.1 150 602
Histci (d) 454 473 123 375 1157
Clvconin (d) 155 166 79.9 77.3 584

Since all reproductive outcomes were calculated from recorded information collected on each cow, the

type and quality of the record keeping systems are highly pertinent to the present discussion (see Table 13 to

review record keeping systems by region).

36
East and West Cibao Region

Comparing the medians and means, the data for all seven reproductive outcomes appear to be fairly

evenly distributed around the mean (mean lying close to the median). An examination of the standard

deviations reveals that several of the values (i.e. historical calving interval, calving to conception interval, days

postpartum, and days to first service) have rather large standard deviations, indicating a wide distribution of the

data.

The Cibao region has an average estrus detection efficiency (calculated by multiplying service rate ×

21 d) of 23.1% ± 16.8%, based on the mean and standard deviation of the service rate with a minimum and

maximum of 0% and 68.7% respectively. (The minimum for herd average of 0% service rate probably resulted

from at least one farm using notebooks records that did not record inseminations but was included in the service

rate calculation.)

Overall the farms sampled within the Cibao region have an average about a 15-month projected

calving interval (± 3.1 months) or about two months longer than optimal [9]. The mean historical calving

interval of 488 d or 16.0 months (median 428 d) may be slightly longer due to one farm with a very long 1157 d

(38 month) historical calving interval, however, this farm had several gaps in records for previous years so

some cows included in the calculations may have actually calved again or been culled.

Mean days to first service for the region is 98.1 d. (This average may be somewhat influenced by one

farm with an average days to first service of 193 d. This number is thought to be real as the farm has

historically had a problem with estrus detection and has a service rate of 0.006 [or 12.9% estrus detection

efficiency].)

Services per pregnancy ranged from excellent to poor (1.86 – 6) [12]. The farm reporting 6 services

per pregnancy had very incomplete records with whole lactations missing. However, the mean services per

pregnancy (2.49) for the region is not drastically different when the mean is calculated without the maximum

farm included (2.20).

The mean days postpartum for the Cibao region is 277± 107 d. Individual farm values ranged from

184 d to 602 d. The data appears slightly skewed having a median of 239 d. The mean days postpartum for the

region is higher than for the sampled farms within the country as a whole (239 d) and much higher than the U.S.

optimum of 198 d [12]. The maximum value for the Cibao region (602) is reported for a farm whose records

37
information may have contained serious errors. This farm also reported an average days in milk of 435 d but

insufficient data was available to calculate further reproductive averages. The next highest herd value for the

region was 439 d. Therefore, it is possible that days postpartum are closer to the total sample mean of 239 d

across all regions than the data mean suggests.

The mean calving to conception interval for the Cibao region is 187 ± 124 d. Individual farm values

range from 98.0 d to 584 d. The mean calving to conception interval is higher for the Cibao region than for the

country as a whole (166 d) and also higher than the U. S. optimum of 113 d [68]. The maximum for calving to

conception interval (584 d) is reported for a farm that had severe gaps in its records. Therefore, this number is

likely not a correct value for the farm and is a result of some culled cows that have not been removed from the

records and are skewing the herd average. The next highest calving to conception interval is 282 d. Eleven of

the 14 farms surveyed had calving to conception intervals lower than 200 d suggesting that the regional mean is

probably much lower than first calculated.

Santo Domingo Region

By examining the summary data for the seven reproductive outcomes, herd averages appear to be fairly

equally distributed around the mean. Standard deviations are typically small indicating a generally small spread

of data points. The region has the lowest average days to first service, calving to conception interval, and

historical calving interval, while having the highest service rate between the regions. However, the region also

has the highest services per pregnancy. This could be a result of poorer reproductive management or better

insemination records.

Mean service rate for the region is 1.38 " 0.58 or 29.0 ± 12.2% estrus detection efficiency. Values

ranged from 0.16 (3.36%) to 2.51 (52.7%). Since service rate is very dependent on the completeness of the

individual cow records, it is important to note that two of the lowest service rates were found on farms with

notebook record keeping systems. Service rate for one farm out of the 19 sampled in the region could not be

calculated due to insufficient data. This farm also has a notebook record keeping system. Therefore, service

rate must be interpreted in relation to other reproductive parameters (i.e. projected calving interval) to validate

whether or not the calculated value is most likely real. Most average service rates (10 of 18) fell between the

1.1 (23.1%) to 1.5 (31.5%) range.

38
The mean projected calving interval for the Santo Domingo region is 435 ± 40.7 d. Values ranged from

348 d to 497 d. The projected calving interval of the Santo Domingo region is about average for the country as a

whole. The farm reporting the minimum projected calving interval (348 d) also reported the minimum service

rate (0.16 or 3.23%). This indicates that the insemination record may not be complete since one would expect

that as service rate increases (more estrous cycles are observed and eligible cows are inseminated), projected

calving interval would decrease. The farm reporting the maximum projected calving interval (497 d) also

reports almost 130 average days to first service and 2.4 average services per pregnancy per cow. Optimal U. S.

projected calving interval is around 395 d [12].

The average days to first service for the Santo Domingo region is 89.5 ± 20.8 d. Values ranged from

64.8 to 133 d. The mean for the Santo Domingo region is lower than for the country as a whole (98.7 d),

however is higher than the U.S. optimum of 75 d. The minimum days to first service within the region (64.8 d)

is reported on a farm using solely visual observation for estrus detection and using artificial insemination for all

cows, except repeat breeders. However, this farm only has 48 cows in milk allowing greater individual

attention for estrus detection. The farm reporting the highest days to first service (133 d) also reports a mean

projected calving interval of 471 d and an mean 2.8 services per pregnancy. This indicates that although

records were kept in a notebook, the farm average was probably an accurate picture of the poor reproductive

efficiency on this particular farm.

Mean services per pregnancy for the Santo Domingo region is 2.44 ± 0.46. Values ranged from 1.59 to

2.98. The mean for the region is slightly higher than that for the country as a whole (2.26) and is 0.60 services

per pregnancy higher than the U.S. average (1.80). The farm reporting the minimum services per pregnancy

(1.69) also reported the minimum projected calving interval and minimum service rate which, as previously

stated, is a further indication that the insemination records on this farm are incomplete since service rate should

be high for such a farm. The maximum services per pregnancy (2.98) is reported on a farm using solely

artificial insemination which also reports a 75 d to first service, a calving to conception interval of 180 d, a

projected calving interval of 459 d, and an estrus detection efficiency of 25.8%. Record keeping being thorough

on this farm, these averages indicate that cows are observed and inseminated fairly early during lactation but it

takes a long time for them to get pregnant. About half the cows in this herd (64 of 160 in milk) were graded

“lame” (lameness score 2-4).

39
The mean days postpartum for the Santo Domingo region is 242 ± 32.1 d. Values ranged from 150 to

286 d. The mean for the region appears slightly lower than the total sample average for the country as a whole

(250 d) but is higher than the U. S. optimum of 198 d [12]. The farm reporting the maximum days postpartum

(286 d) is the same farm reporting maximum days to first service (130 d) and projected calving interval (497 d).

This farm is comprised of a roughly 70% Holstein milking herd using artificial insemination, except for

problem cows, with inseminations being done by off-farm personnel. The farm reporting the minimum days

postpartum (150 d) also reported the minimum service rate and days to first service and was discussed above. It

is unclear if the number for days postpartum is therefore real due to the questionable quality of the records. The

milking herd on this farm is comprised of 84.5% Holsteins.

The mean historical calving interval for the Santo Domingo region is 442 ± 59.5 d. Values ranged

from 375 to 575 d. The mean for the region appears lower than for the sample population within the country as

a whole (473) but is higher than the U. S. optimum of 395 d [12]. The farm reporting the minimum historical

calving interval (375 d) also reports the minimum projected calving interval (393 d), a days to first service of

85 d, and a services per pregnancy of 1.78. However, this farm also reports a 14.1% estrus detection efficiency

indicating a discrepancy in the insemination record since estrus detection efficiency should be higher for such a

low projected and historical calving interval. This farm used individual cow cards for record keeping and

consisted of a totally non-Holstein milking herd.

The farm reporting the maximum historical calving interval (575 d) also reported a 456-day projected

calving interval, a121-days to first service, 1.59 services per pregnancy, a 257 calving to conception interval,

and 25.6% estrus detection efficiency. The farm’s milking herd, composed of 1.18% Holsteins, had an mean

lameness score of 1.64 with 55 of its 151 cows being graded “lame” (lameness score 2-4) and was an artificial

insemination herd, except for problem cows.

The mean calving to conception interval for the Santo Domingo region is 156 " 42.6 d. Values ranged

from 78.1 d to 258 d. The mean for the region appears slightly lower than that for the country as a whole

(166 d), although the median for the region (163 d) is higher than that of the country (155 d), but is higher than

the U. S. optimum of 113 d. The maximum calving to conception interval is reported for the same farm that

reported the maximum historical calving interval described above. The minimum calving to conception interval

40
is reported for the farm that also reported the minimum service rate, days postpartum, and days to first service

and was also discussed above.

Southwest Region

When the summary data for the seven reproductive outcomes for this region are compared with herd

averages for the Santo Domingo region, they appear to have a comparable distribution, with all standard

deviations being equal to or slightly higher than those calculated for the Santo Domingo region. Also, by

examining the medians and means it appears that data is fairly equally distributed around the mean. This

indicates that data from the farms in Azua and Barahona do not seem to skew the mean calculated for the San

Juan province. While data appears fairly consistent, 5 out of the 8 herds surveyed reported natural service for

the entire herd, all 5 having bulls housed with the milking herd. Therefore, all reproductive parameters for the

region that require observation of estrus activity or insemination (days to first service, services per pregnancy,

service rate, calving to conception interval and projected calving interval) must be interpreted with this in mind.

In general only the service resulting in pregnancy was recorded in the herd where bulls ran with the cows at all

times.

Summary statistics for days to first service in the Southwest shows that the mean for the region is

109 ± 26.3 d (optimum U. S. average = 75 d). The minimum of 70.2d value was calculated for a farm placing

strong emphasis on maximizing reproductive efficiency. (As evidenced by assigning 2 employees specifically to

the task of reproductive management and by using mechanical aids for estrus detection.) The next to lowest

days to first service (96.5 d) was recorded for a farm in the San Juan province which uses natural service

keeping the bull with the cows at all times. In this situation, estrus detection is assumed to be high, depending

on the health status and libido of the bull and assuming that all cows are cyclic by soon after calving. However,

the two highest values for days to first service, 142 d and 132 d, were also recorded for farms which keep

breeding bulls with the milking herd. Therefore, the reliability of days to first service on such farms is

dependent on the ability of the person responsible for observing and recording mounting behavior by the bull. In

fact, the reason for higher days to first service in this region could be due in part to a records error, since 5 of

the 8 farms surveyed within this region leave breeding bulls with the milking herd and the first insemination

may not be consistently observed.

41
The mean days postpartum for the Southwest region is 229 ± 35.0 d. Values ranged from 168 to 276 d.

Optimal days postpartum, expected to be associated with a 13-month calving interval should be approximately

198 d (395 days/ 2). Longer days postpartum may be observed in the Southwest region as a result of low estrus

detection efficiency and accuracy, nutritional or lactation stress, or increased incidence of pathology (i.e.

lameness) within the herd. Low service rates indicate problems with estrus detection efficiency within the

region and higher than optimal services per pregnancy for some of the farms indicate problems with cow and/or

bull fertility.

The maximum value for average days postpartum (276 d) is recorded for the farm in Barahona and

may be due to low conception rates seen in many cows transferred from another domestic farm which suffer

from chronic lameness and which had a historically high services per pregnancy before they arrived at the new

facility. The minimum value for average days postpartum (168 d) is recorded for a bull-bred herd within the

San Juan province. This value is likely a result of high estrus detection efficiency and good conception rates by

the bull.

The mean projected calving interval for the Southwest region is 412 ± 47.6 d. Values ranged from

360 d to 478 d. As previously stated, the optimal U. S. projected calving interval to maximize profits and milk

production is around 395 d (13 months) [12].

The maximum projected calving interval (478 d) was recorded for the farm in Barahona. The next

highest value (475 d) was recorded for a farm in the San Juan province. These two farms also reported the

highest services per pregnancy (4.4 and 2.4 services/pregnancy respectively). It appears that the overall

reproductive efficiency on these farms could be improved, as measured by decreased projected calving interval,

if solutions to poor conception rates can be found. Conception rates could be low due to high incidence of

lameness, nutritional imbalances, environmental challenges (especially heat stress), lactation stress, semen

quality and handling problems, or bull infertility (not likely as both herds use mainly artificial insemination) [4,

12, 31, 64].

The mean historical calving interval for the Southwest region is 492 ± 84.0 d. Values ranged from

376 to 573 d. As previously stated, optimal calving intervals within the U.S. are 395 d, making the historical

calving interval within this region almost 100 d longer than U.S. targets. Historical calving intervals could not

be calculated for 3 of the farms in the region due to inadequate records information on previous lactations.

42
Two farms in the San Juan province reported historical calving intervals over 500 d. These farms also

reported 126 and 131 d to first service. Service rates for these farms were 0.008 (or 18.7% estrus detection

efficiency) and 0.004 (or 9.30% estrus detection efficiency). The farm reporting the minimum historical calving

interval (375 d) also had the shortest days to first service (96.5 d) and minimum days postpartum (167 d). This

herd was a bull-bred herd from the San Juan province.

The mean services per pregnancy for the Southwest region is 1.90 ± 1.13 services/ pregnancy/cow.

Values ranged from 1.00 to 4.38 services/pregnancy/cow. It must be remembered that these are services per

pregnancy for pregnant cows only and the numbers would certainly be higher if all cows were included. The

mean services per pregnancy in this region appear to be lower than for the country as a whole (2.26) but this

could be due to 5 of the 8 farms surveyed leaving breeding bulls with the milking herd at all times resulting in

not all inseminations being recorded. Barahona reported the maximum services per pregnancy (4.3), thought to

be mainly attributable to some domestic cows that were transferred to the facility with histories of high services

per pregnancy (10+).

The mean service rate for the Southwest region is 0.74 ± 0.58 services per 100 d or 15.5% ± 12.2%

estrus detection efficiency. Values ranged from 0.43 to 2.11 services per 100 d or 9.03% to 44.3% estrus

detection efficiency respectively. Service rate for the region appears lower than for the country as a whole (1.07

services per 100 d) and is much lower than the U.S. optimum of 3.33 services per 100 d (70 % estrus detection

efficiency). Service rate is highest in the farm in Barahona (2.11 services per 100 d or 44.3% estrus detection

efficiency) probably due to advanced estrus detection methods with the next highest service rate being reported

for Azua (0.8 services per 100 d or a calculated 16.8% estrus detection efficiency). Estrus detection efficiency

is difficult to interpret in Azua since breeding bulls are left with the milking herd. Five of the 8 farms surveyed

in the region leave breeding bulls with the milking herd. Therefore, although service rate appears lower in the

San Juan region, this measure may not be a good representation of the actual situation since estrus detection is

actually being done by the inseminating bull.

The mean calving to conception interval for the Southwest region is 156 ± 72.7 d. Values ranged from

77.3 to 297 d. The mean calving to conception interval for the region is slightly lower than for the country as a

whole (166 d). The highest calving to conception interval (296 d) is recorded for a San Juan province herd

which uses artificial insemination to a large degree. This farm also reports a 568-day historical calving interval,

43
a 476-day projected calving interval, 126 days to first service, and 2.4 services per pregnancy. The minimum

calving to conception interval (77 d) is recorded for a San Juan herd, using natural service. Calving to

conception interval, in general, within the bull-bred herds of San Juan is relatively low with values of 77, 94,

and 96 d reported, although one farm reported an interval of 149 d.

Summary

Reproductive outcomes are only as reliable as the data from which they are calculated. Therefore, it is

crucial to carefully evaluate data, keeping in mind that every region has extremes in record keeping

management. It was found to be very difficult to secure complete records from which to calculate reproductive

outcomes. It is also apparent that producers do not regularly use these values for making herd management

adjustments and therefore do not appreciate the importance of the reliable information that can be extracted

from complete and reliable records.

Although several farms report values comparable to U.S. optimal values, means for the country are still

not optimal for levels of reproductive performance that would support efficient milk production. Table 26

indicates the differences between published U.S. optimums and means that the present study has calculated for

the Dominican Republic.

Table 11. Comparison of D. R. Means with U. S. Optimum Values and


Averages for the State of Virginia for Select Reproductive Outcomes.

Reproductive Outcome U. S. optimum Virginia DHI D. R. Mean D. R. Standard


12/2001 deviation
(436 herds)
Service rate (services / 100 d) 3.34 1.81 1.07 0.65
Projected calving interval (d) 395 438 434 60.5
Days to first service (d) 75 97 98.7 27.5
Services per pregnancy services) 1.8 2.1 2.26 0.87
Days postpartum (lactating and dry) (d) 198 191 DIM 250 58.1
Historical calving interval (d) 395 438 473 123
Calving to conception interval (d) 113 159 166 79.9

It should be noted that although the average projected calving intervals for Virginia and the D. R.

appear similar, their calculations are slightly different. Virginia’s projected calving interval is calculated using

not only the pregnant cows but the non-pregnant cows which have received an insemination but which have not

been palpated. So, in essence, the Virginia estimate may actually be higher than the reported value, depending

on the conception rate for inseminated cows [68]. The averages for the D. R. were calculated using only the

44
pregnant cows in each herd partially due to the prevalence of incomplete insemination records. Therefore, this

estimate may actually be lower in reality depending on the reproductive performance of the inseminated cows.

The Cibao region reported the highest projected calving interval and calving to conception interval

among the regions sampled. The data from several of the farms in this region contained severe gaps,

sometimes missing whole lactations, so means may be higher than in reality.

The Santo Domingo region reported the highest mean service rate and services per pregnancy, while

reporting the lowest days to first service and calving to conception interval. The records from which data was

extracted for this region were in general more complete than recorded data from other regions of the country.

This region has means closer to the optimal values than the overall mean for all reproductive outcomes except

services per pregnancy and projected calving interval.

In general the Southwest region reported the lowest mean service rate and the highest days to first

service but the lowest projected calving interval, services per pregnancy, days postpartum, and historical

calving interval when compared to the other regions surveyed throughout the country. The region has means

closer to the optimal values than the total country means for all reproductive outcomes except days to first

service and historical calving interval. However, since 5 out of the 8 farms sampled allowed breeding bulls to

remain with the non-pregnant milking cows, it is possible that any reproductive outcome requiring detailed

insemination records for accuracy, especially service rate, days to first service, and services per pregnancy is

questionable in accuracy.

Herd descriptors by region

The following table summarizes the following information: average body condition score of the

lactating herd (avbcslac), the number of lame cows within the milking herd (nolame), the average lameness

score of the milking herd (avlame), the proportion of Holsteins within the herd (percenthol), the number of

cows within the milking herd (nucomh), and the average amount of milk produced (avmilk).

45
Table 12. Continuous Herd Descriptives by Region.

Region No. Obs. Variable Median Mean Std. Dev. Min. Max.
Cibao 16 Avbcslac 3.03 3.03 0.25 2.70 3.67
Avlame 1.40 1.40 0.20 1.10 1.80
Percenthol 76.4 66.1 25.5 13.3 95.7
Nucomh 94.0 98.1 48.4 33.0 186
Avmilk 16.9 16.4 3.50 10.9 22.8

Santo Domingo 19 Avbcslac 3.07 3.04 0.23 2.65 3.39


Avlame 1.60 1.60 0.30 1.10 2.00
Percenthol 64.4 52.7 36.1 0.00 100
Nucomh 122 146 97.0 33.0 400
Avmilk 15.9 16.0 4.10 10.0 28.5

Southwest 8 Avbcslac 2.49 2.58 0.42 1.89 3.20


Avlame 1.00 1.10 0.20 1.00 1.40
Percenthol 48.9 53.2 39.8 12.9 100
Nucomh 32.5 76.3 108 20.0 336
Avmilk 12.2 13.8 7.60 7.00 31.4

Total 43 Avbcslac 2.86 2.88 0.30 1.89 3.20


Avlame 1.33 1.37 0.23 1.00 2.00
Percenthol 63.2 57.3 33.8 0.00 100
Nucomh 82.8 107 64.0 20.0 400
Avmilk 15.0 15.4 5.07 7.00 31.4

The following figure summarizes the types of record keeping systems sampled by region:

Figure 3. Types of Record Keeping Systems by


Region

20 17
Number of Farms

14
15 12
9
10 7
5 6
4 4 3 3
5 2
0
Cibao Santo Domingo Southwest Total
Region
Notebooks Individual Cow Cards Computer

East and West Cibao Region

The East and West Cibao regions were combined, in part, to increase sample size and decrease degrees

of freedom for the study. Any climatic differences are small and comparable to those found within the other

two regions sampled. Temperature and humidity is generally slightly higher than in the other two geographic

regions.

46
There is a tendency for the data from the Cibao region to be less widely distributed than data from

other regions. Milking herd sizes for herds surveyed in the Cibao were generally medium-sized and produced

on average 16.36 liters of milk with average milk production fairly evenly distributed about the mean. Herds

are mainly comprised of Holsteins, with 11 out of the 16 farms surveyed within the region having >50%

Holsteins within their milking herds.

Average body condition scores for lactating cows in the herds surveyed in the region are 3.03,

however, 5 out of the 16 farms have average scores under 2.90 which could impair reproductive efficiency,

especially during the early stages of lactation when negative energy balances can cause problems in getting

cows pregnant [4]. Herds seem to have a similar incidence of lameness apparent by to the fairly uniform, yet

narrow distribution of the data around the mean of 1.4. Average lameness is quite high in some herds with 5

out of the 16 surveyed having average lameness scores of 1.5 or more. Several producers complained about

laminitis problems within their herds.

Record keeping within the region was variable in intensity and sophistication. This region was the

only region with farms which used ComputoGanadero, a computer-based records system that operates

completely in Spanish. This system was very useful in securing data for this study but also had significant

limitations. For example, days to first insemination could not be adequately calculated from the records since

the program only preserves the date of the most recent insemination.

Although many of the farms gave considerable attention to record keeping, 3 out of the 16 farms had

records from which large quantities of data were missing, sometimes information for entire lactations was

missing. Two of these farms in this regions use PCDART®, a computer-based system, and one used individual

cow cards.

Santo Domingo Region

The largest survey sample was drawn from the Santo Domingo region due to the number of farms

accessible in the region that met the study criteria. The region varies somewhat in climate between provinces

with some areas being near the ocean and others being inland. Areas such as the Baní province and the Santo

Domingo district are both near the ocean but differ slightly in humidity and altitude. Three farms were

surveyed in the Santo Domingo district, 11 farms in the Monte Plata district, and 5 farms in the Baní district.

The average milk production per farm within the region was mainly less than 20 liters of milk per cow per day,

with 16 of the 19 farms reporting milk/cow/day between 10 and 20 liters. The number of cows in the milking

47
herd varies widely but is mainly below 200 cows as indicated by the following frequency distribution of cows

within the herd: 301-400 cows (2 farms), 201-300 cows (2 farms), 101-200 cows (8 farms), 33-100 cows (7

farms). The proportion of Holsteins within the herd varies widely as well but is fairly uniformly distributed

around the mean of 52%.

Record keeping systems were mainly kept in notebooks or on individual cow cards, although 4 farms

had computer-based systems (3 had PCDART®, 1 had DairyFlex®). None of the farms sampled within the

Baní province had computer based systems. Two out of the three farms sampled within the Santo Domingo

district used computer-based systems. The data from one of these two farms was not kept on the premises but

stored on a computer at a local milk processing plant. Records data was fairly complete, in general, with

producers putting a demonstrated emphasis on maintaining individual cow records, although a few exceptions

did exist.

Average body condition scores were quite evenly distributed around the mean of 3.04 and respectively.

Average body condition scores for lactating cows were between 2.9 and 3.3. Average body condition scores for

herds within the Santo Domingo region were lower than 3.0.

Lameness is definitely a problem within the region with a mean lameness score of 1.6. Ten of the 19

farms sampled were found to have average lameness scores greater than or equal to 1.5. Most farms

complained of problems with laminitis. Five of the 19 farms surveyed in the region averaged lameness scores

of 2 or greater for over 45% of their milking herds. At least two of these farms reported historical problems

with laminitis after implementing certain ration changes.

Southwest Region

The Southwest region presented a significant challenge in data collection. First, due to limits of time

and availability of farms meeting the criteria for research in size and record quality, only 8 farms were surveyed

from this region. Extremes within the survey population are more likely to skew averages because of the small

sample size. Secondly, six out of the eight farms were located within the province of San Juan, which is

inland. The survey population from this region was characterized by small farms (20-32 cows in milk)

producing from 7-14 liters of milk per cow per day. Several of these farms were pasture-based nutritionally,

perhaps partially explaining the low milk production reported. Farms were also generally characterized by low

proportions of Holsteins within the herd (4 farms having <20% and only 2 farms having >70%). Farms with

<20% Holsteins making up the herd were generally composed of mainly Brown Swiss and/or mixed breed

48
cattle, some of which were native Criollo crosses. This is the only region visited where some of the farms

within the survey population milked cows by hand.

Record keeping within this region was generally confined to notebooks and individual cow cards

making quantity of the data available sometime limited, especially with notebook systems. However,

information from two farms was gleaned from computer-based systems. These databases were not maintained

on-farm but at a central government MEGALECHE office within the local town where extension agents would

regularly (about once every 2-3 weeks) visit the farm and gather the information kept by the owner. (On the

date of the survey, individual cow records had been updated within the last 3-4 days). Information was stored

in AfiFarm®, an Israeli database system designed specifically for aiding in dairy farm management. This

computer-based records management program had just begun within the past year in the D. R. to provide

producers with regular herd production and reproductive averages. However, many producers are still

unconvinced of the program’s practicality, since they are not accustomed to having so much information and do

not fully realize the benefits of regular herd summaries on molding herd management decisions.

Two farms within the survey population for this region, however, were located in the Azua and

Barahona province, both of which border the Caribbean Sea on the south and southeast respectively. These

farms have management styles much different from the rest of the herds sampled within the Southwest region.

Both farms have a fairly substantial capital base, one being a government run farm and the other a farm owned

by Leche Rica, one of the major milk processing companies within the country. The government-run farm had

been plagued by poor pregnancy rates and declining milk production until a new herd manager took over about

two years ago. He implemented changes to improve pregnancy rates to boost milk production in the short term,

such as leaving breeding bulls in with the cows, until long-term solutions could be determined. Pregnancy rates

and milk production are now improving but the farm is still suffering from past mistakes, resulting in much

lower milk production than at the Leche Rica facility (14 versus 31 liters respectively). The processing

company owned facility, Lecheria San Antonio, is a state-of-the-art facility composed completely of Holsteins,

many of which were imported as pregnant heifers from the United States, and has a goal to become the largest

dairy cattle operation in the country eventually milking up to 2000 cows. This facility began production in

2000 and still suffers from some lameness problems and poor reproductive efficiency in cows sent to the farm

from its former location in another region of the country that was thought to be less suited for dairy cattle

49
production due to high heat and humidity. Record keeping systems on both farms are computer-based,

Lecheria’s database being maintained by two computer-based systems, PCDART® and AfiFarm®.

Differences in general herd management between the four small farms of the San Juan province and

the two farms in Azua and Barahona provinces are quite clear. It might be helpful to classify these regions as

“Southwest A” and “Southwest B” for descriptive purposes. This step was not done for the analysis of data

since the averages were not skewed drastically by the data from the Azua or Barahona farms and two farms

were not enough to perform an adequate statistical assessment of a region. This “similarity” in averages for

herd parameters will probably not continue to exist once the larger farms begin to benefit from better

management, facilities, and environment.

Average body condition scores do not differ drastically within the region as a whole with a regional

average BCS of 2.58 and the San Juan farms with 2.40. An important note, however, is that average body

condition scores are quite low for the San Juan province (all averages <3.00 for milking cows). Lack of body

condition may be associated with decreases in both reproductive efficiency and milk production within the

region.

Average lameness scores also do not differ drastically within the region as a whole, with a regional

average of 1.14 and the San Juan farms having an average of 1.05. Lameness may also be a factor in reduced

reproductive efficiency and unless further lameness is prevented these farms may continue to experience

decreases in efficiency [13].

Overall, the incidence of lameness was high all over the country, with the exception of the farms

sampled in the San Juan province. On average, one-third of the milking herd had lameness scores of 2.

Reproductive Management Descriptors by Region

To investigate associations with variation in reproductive outcomes several management parameters

thought to impact reproductive outcomes were investigated. Those management practices studied include type

of cooling system used, average hours spent on concrete per day, method of estrus detection, number of people

involved in estrus detection, reproductive hormone use, artificial insemination (A. I.) versus natural service (N.

S.), and whether or not the farm personnel detect for estrus between 6 p.m. and 6 a.m. In this section, these

variables are summarized by region under one of three categories: Cow Cooling (Environmental Control),

Estrus detection and Synchronization, and Breeding Program.

Cow Cooling

50
Attempts are made in this tropical environment to make cows more comfortable through cooling.

Studies have shown that heat stress decreases reproductive efficiency [23, 24]. Cooling systems are a potential

approach to abating the detrimental effects of heat stress. Few farms surveyed had sufficient shade to

completely cover all watering troughs and feeders for all cows: only about 14.6% (6 of 41). Partial shading was

found in 85.4% of the farms surveyed and ranged from mild lack of shade (i.e. uncovered watering trough) to

severe lack of shade (i.e. only a few trees in the pasture). The following figure summarizes the types of cooling

systems found by region:

Figure 4. Distribution of Cooling Systems by Region

40 37
Number of Farms
Reporting Use

30

20 16 16 14 16
9 9 11
10 6 6 7 5 6
4 2 3 4
1 1 1
0
Cibao Santo Domingo Southwest Total*
Region

Fans Sprinkler systems Shaded barn Shade netting Trees

*More than one possible answer per farm. 43 farms surveyed.

East and West Cibao Regions

The farms surveyed in East and West Cibao regions were mainly characterized by shade-only

structures. Only 25% of the farms within the Cibao region were classified as being completely shaded. At least

four of the farms had free-stall barns where cows were housed for much if not all of the day. However, during

survey visits, it was found that the free stalls often had little to no bedding, decreasing the frequency with which

they were used by cows. Additionally, on one farm, free stalls were exposed to the sun during the hottest part

of the day. Sprinkler systems also did not appear to be functioning on some of the farms visited. Proper cow

cooling is especially important in this region as temperature and humidity, on average, are higher than in other

regions sampled

51
Santo Domingo Region

In general, farms within the Santo Domingo region utilize multiple methods in attempting to keep

cows cool, although all 19 farms surveyed were classified as “partially” shaded. Only seven of the 19 farms

surveyed within this region reported that they depended solely on shade structures (i.e. shade netting, trees, or a

shaded barn) and only 3 of those did not house cows in a barn. Most farms had shade structures around feeding

areas and, to a lessor extent, the watering areas. Very few herds had access to a free stall barn.

Southwest Region

In terms of cooling systems, the farms sampled within the province of San Juan are managed very

differently than the two farms located in Azua and Barahona. The general trend within the farms surveyed in

the San Juan province is toward shade structures only (natural or man-made). All four farms made use of trees

for cooling, and 3 out of the 6 farms sampled had shaded barns, although these were mostly just used during

milking time. The facilities in San Juan were, in general, unsophisticated. In at least two of these farms the

cows were milked by hand. However, neither the farm in Azua or Barahona utilized trees for shading. These

facilities were “high overhead” systems in comparison to the 4 farms within the San Juan province. Cows in

both facilities were completely shaded under a barn (loose housing and free stall respectively). In addition to a

state-of-the-art free stall barn, the facilities in Barahona contained a sprinkler system for lactating cows and the

holding facility before milking contained RainBird sprinklers and fans. Reproductive outcomes for the

Southwest region need to be interpreted, in part, with these differences in mind.

Estrus detection and Synchronization

As previously stated, estrus detection based on visual observation alone has been reported to be inadequate to

maintain optimal herd reproductive performance [17-20]. Visual observation is nonetheless the only method

employed by 31 of the 43 farms (72%) surveyed within the D. R. The following figure summarizes methods of

estrus detection by region within the D. R.:

52
Figure 5. Methods of Estrus Detection by Region

50
40
Number of Farms

40

30

16 18
20

10 4 3 6 5 5
0 0 0 0 2 0 0 1 1 0 1 1 1 1 1 2
0
Cibao Santo Domingo Southwest Total*
Region
Visual observation Kamar Chalk (tailhead)
Timed breeding Pedometer Bull

*More than one possible response per farm. (16 farms surveyed in Cibao, 19 surveyed in Santo Domingo, and 8
surveyed in the Southwest region.)

Since most of the farms surveyed (72%) rely solely on visual observation characterizing how the

detection is carried out is of interest. It is important to assess exactly when and by whom visual estrus

detection is done. Estrus detection done during regular farm activities, such as milking and feeding, may not be

as efficient since mounts may be missed. In terms of personnel, having more people involved in estrus

detection may actually be less successful since often no one takes full responsibility. When evaluating a dairy

herd for estrous activity, it is frequently desirable for 1-2 people to observe the herd several times throughout

the day and evening for a specified period of time (~30 minutes), being exclusively engaged in estrus detection

[4]. The following figures characterize farms by region in terms of when and by whom estrus detection is

performed:

53
Figure 6. Estrus Detection Efforts in Relation to
Milking or Feeding Activities by Region

35 32
Number of Farms
30
25
20
13 14
15
8
10 5
3 4
5 1
0
Cibao Santo Domingo Southwest Total*
Region

Not Related Related


*3 Observations missing

Figure 7. Summary of the Number of Employees Involved in Estrus


Detection by Region

12
No. of Employees

10
8
6
4
2
0
Cibao Santo Domingo Southwest All Regions
Region
Median Mean Std. Dev. Min. Max.

54
The literature reports that cows suffering from heat stress are less likely to be observed exhibiting

estrous activity [24-26, 33]. Studying the time of day that estrus detection is done may also be illustrative.

Nighttime estrus detection, defined here as occurring between 6 a.m. and 6 p.m., may increase the likelihood of

finding cows in estrus since radiant energy from the sun is no longer a problem and temperatures are usually

somewhat lower at night. The following figure summarizes nighttime estrus detection by region:

Figure 8. Nighttime Estrus Detection by Region


(6 p.m. to 6 a.m.)

45
40
Number of Farms

35 12
30
25 No
20 Yes
15 4 7 28
10 1
5 12 11 5
0
Cibao Santo Southwest Total
Domingo
Region

*3 Observations missing

To optimize estrus detection efficiency, estrus synchronization has been recommended, especially for

first service inseminations [8, 9]. Characterization of estrus detection should an evaluation of reproductive

hormone use by region (Table 19).

55
Table 13. Reproductive Hormone Use by Region.

Response Cibao Santo Southwest Total


Domingo
n No hormone 2 0 6 8
% use 12.5 0.0 75.0 18.6

n Hormones 14 19 2 35
% used 87.5 100.0 25.0 81.4

Specific Patterns of Reproductive Hormone Use by Region*

n Prostaglandins 13 13 1 27
% occasionally 81.3 68.4 12.5 62.8

n Prostaglandins 1 6 1 8
% regularly 6.3 31.6 12.5 18.6

n GnRH 8 14 0 22
% occasionally 50.0 73.7 0 51.2

n GnRH 2 1 2 5
% regularly 12.5 5.3 25.0 11.6

n Progesterone 1 1 0 2
% regularly 6.3 5.3 0 4.7

n Estradiol 0 2 0 2
% valerate and 0 10.5 0 4.7
Norgestomet
program
*More than one possible response per farm.

Finally, the following figure summarizes the average hours a typical lactating cow spends on concrete

within a given region. Concrete can exacerbate lameness and may become slippery causing cows to be

reluctant to mount. Cows have been reported to display less estrous activity when being held on concrete.

56
Figure 8. Summary of Hours the Milking Herd Spends on
Concrete by Region

30
24 24
25 22
20
Hours

14
15 12 12.4 12 12.9 12
10 6.2 6
4 4.9 4 4
5
0
Cibao Santo Domingo Southwest
Region
Median Mean Std. Dev. Min. Max.

East and West Cibao Region

Similar to other regions, visual observation is the principle method of estrus detection within the East

and West Cibao region. All 16 farms sampled reported primarily use of visual observation to detect for estrus.

In addition, over 80% of the farms surveyed reported observing for estrus during milking or feeding. Seventy-

five percent reported estrus detection between 6 a.m. and 6 p.m. Five farms reported use of mechanical aids for

estrus detection, with two farms reporting the use of both Kamar and chalk. However, 11 out of the 16 farms

sampled reported 12 or more hours on concrete (2 farms reporting 24 hours) which, depending on the

prevalence of lameness, might counteract any benefits gained by mechanical aids.

Even though the average number of employees involved in estrus detection is lower than for the Santo

Domingo region, five out of the 16 farms surveyed (31.3%) reported 5 or more employees being responsible for

estrus detection, indicating that estrus detection is “everyone’s job” as opposed to a specific responsible

employee. The Santo Domingo region had a similar proportion (31.6%) of farms assigning “everyone” to estrus

detection.

Although some regular use of reproductive hormones is reported, use of such hormones is generally

reserved for occasional use on problem cows, with two farms not using hormones at all. This region seems to

be slowly moving in the direction of using more reproductive hormones as part of their overall reproductive

management program.

57
Santo Domingo Region

Of 19 farms surveyed in the Santo Domingo region, 18 used visual observation as their primary

method of estrus detection with the remaining farm housing a bull with the herd. Only two of these farms used

an aid to estrus detection (chalk). The average number of employees involved in estrus detection for the

region is 3.2 and six of the 19 farms surveyed had 5 or more employees involved in estrus detection. This

indicates farms in which no one specifically has been assigned to detect estrus and it is “everybody’s job”.

Over 60% of the farms in the Santo Domingo region observe estrous activity between 6 a.m. and 6 p.m.

Finally, average hours on concrete were 12.9 for the region with only 5 out of the 19 sampled reporting less

than 10 hours on concrete.

Producers within the Santo Domingo region seem to be the most progressive in adopting

synchronization programs. Of all the regions, regular reproductive hormone use was the greatest for this region

with 31.6% of producers reporting regular prostaglandin usage and 5.3% reporting regular GnRH usage. In

addition to these synchronization programs, 3 farms reported using alternative programs (1 progesterone, and 2

estradiol valerate and Norgestomet implants). In general, even though occasional use is more prevalent in the

region, it appears that the trend is to adopt regular use of reproductive hormones.

Southwest Region

Of the six farms surveyed within the San Juan province, estrus detection is done by visual observation

in four of six farms. The remaining 2 farms in this province keep bulls with the herd so no estrus detection is

done. For those farms detecting for estrus, no mechanical aids are used to help in the process, except for one

farm that uses a teaser bull to aid in estrus detection. Three of the four farms detect for estrus during milking or

feeding, generally having no more than 1-2 employees in charge of estrus detection, although one farm reported

3 employees responsible for estrus detection. Three of the four farms regularly observing for estrus observe

between 6 a.m. and 6 p.m.

By far the most notable observation within the San Juan province was that none of the farms used

reproductive hormones for synchronization, even for problem cows. Veterinarians providing service to these

farms do not recommend hormone injections. Estrus synchronization has not been well accepted within the

region, possibly due to lack of information. On average, herds are not examined for pregnancy on a regular

basis, probably because of the small number of cows to be examined.

58
Most of the herds (5 out of 6) sampled within the San Juan region only have cows on concrete 12 hours

or less daily, with all six farms allowing cows access to pasture. In general, reproductive management within

the San Juan province appears fairly rudimentary in comparison with other regions of the country. The general

conclusion, based on observations made within the area and conversations with various veterinarians and

government agricultural officials working in the province, is that the area needs more technical support to

introduce producers to newer developments in reproductive management so they can make more effective

decisions.

The two farms sampled in Azua and Barahona were much different than those in the San Juan province

in their approach to reproductive management. The most notable differences between the farms is the method

of estrus detection and reproductive hormone usage. The farm in Azua, much like other farms within the

region, only uses visual observation for estrus detection performed exclusively by 1-2 employees, the farm in

Barahona uses a combination of methods including visual observation, Kamar patches, pedometry, and estrus

expectancy lists. Both farms also use reproductive hormones for synchronization, both using GnRH on a

regular basis and one using prostaglandins on a regular basis.

Another difference between the two farms in Azua and Barahona and the farms in the San Juan

province is the length of time cows spend on concrete. The farms in Azua and Barahona are intensively

managed and cows are not allowed access to pasture. This leads to cows remaining on concrete for longer

periods than for herds in the other farms in the region, 24 and 18 hours respectively.

Breeding Program

Aside from estrus detection, timing and manner of insemination is a key part of any farm’s

reproductive management program. The timing and manner of insemination influences conception rates and,

depending on whether artificial insemination or natural service is used, involves such factors as appropriateness

of timing of insemination, semen handling and placement, and bull fertility and libido [4].

Since many factors can influence conception rates it is important, when evaluating causes of possible

reproductive inefficiency, to examine specific management practices, such as timing of insemination [22],

which can influence herd pregnancy rates and overall profitability. The following table summarizes the intervals

between initial observation of estrous activity and insemination by region:

59
Figure 10. Interval between First Observed Estrous Activity and
Insemination by Region

30 26
Number of Farms

25
20
15 11 12

10
4 5
5 2 3 3 3 3
1 0 0 0 1 0 1 0 0 1
0
Cibao Santo Domingo Southwest Total
Region

Within 6 hours 6-8 hours


10-12 hours More than 12 hours later
All cows serviced at same time (1-2 times/day)

*5 Observations missing

Appropriate timing of insemination will maximize the opportunity of conception by placing semen in

the appropriate location in the female tract at the most fertile time. The well-established practice is to breed by

the “a.m.-p.m. rule” which has been shown to provide the best results, depending on the methods of estrus

detection used [16, 35]. However, more recent studies have suggested comparable success with one time a day

breeding [35]. Since the length of estrus is also thought to be shorter in dairy cattle suffering from heat stress

[24, 26], intervals between first observation of estrus and insemination are not so straightforward to interpret. It

is imperative that appropriate intervals between first estrus activity and insemination for tropical zones be

studied and established.

Bull use and management are also important factors in determining reproductive efficiency. When

bulls are exclusively used for insemination on a given farm, they make up 50% of the mating herd and their

reproductive health must not be ignored. Whenever natural service is used in a mating program, it is

recommended that bulls get regular vaccinations and be tested for reproductive diseases and semen quantity and

quality. Hoof care and nutritional management are also important in farms using natural service for

insemination [4]. The following tables summarize bull use and frequency of breeding soundness exams by

region:

60
Figure 11. Bull Use by Region

50
4
Number of Farms

40

30 Bulls not used


20 2 Bulls used
2 39
0
10 17
14
8
0
Cibao Santo Southwest Total
Domingo
Region

Figure 12. Groups of Cows Inseminated with Natural Service by


Region
30 28
Number of Farms

25
20
15 12 13
9
10
5
5 2 2 2 3 2
0 0
0
Cibao Santo Domingo Southwest Total*
Region
The entire herd Heifers and problem cows Problem cows/ repeat breeders

*4 Observations missing.

61
Figure 13. Farms Conducting Bull Breeding Soundness
Examinations by Region

45
40
Number of Farms

35 8
30
25 Yes
20
0 5 No
15 32
10 3
14 15
5 5
0
Cibao Santo Domingo Southwest Total*

Region

*3 Observations missing

One important note is that “breeding soundness examination” appeared to be a relatively new concept

to many of the producers surveyed. Full-scale breeding soundness examinations (vaccination against and

testing for reproductive diseases, testicle palpation and scrotal circumference, rectal exam of seminal vesicles,

and sperm motility and morphology examination) seem to rarely if ever done in the D. R. Therefore, during the

survey, it was applied loosely as any reproductive preventative health measure taken regarding the bull.

East and West Cibao Regions

Most farms surveyed in the Cibao region (68.8%) inseminate eligible cows using the “a.m.-p.m. rule”.

This interval may not be appropriate given the extreme heat of the climate so that the 31.2% of the farms

surveyed that inseminate cows 8 hours or less after they are observed in estrus may be more appropriate. Only

one farm reported having a veterinarian or off-farm technician inseminate eligible cows. This indicates that

someone working on the farm whose schedule is not subject to other farm visits usually handles insemination.

Of the farms surveyed within the Cibao region, 87.5% reported using natural service in some portion

of their breeding program. Only 2 farms reported using natural service for the entire herd, with the remaining

12 farms using natural service for problem cows that had been artificially inseminated repeatedly without

resultant pregnancy. Only 1 of the 2 farms maintain bulls with cows (74:1 cow to bull ratio) and the bulls on

this farm are rotated to help maximize fertility. On the farms using natural service, the average age of bulls is

around 30 months. Finally, none of the farms sampled reported conducting breeding soundness examinations

on their bulls. Therefore, reproductive pathology within bulls is likely not detected until pregnancy rates drop.

62
Santo Domingo Region

Two thirds of farms in the Santo Domingo region inseminate according to the “a.m.-p.m.” rule. Of the

farms surveyed within the region, 11% inseminate cows 8 hours or less after observation of estrus and 21% of

the farms surveyed inseminate eligible cows 12 hours or more after estrus is observed. Five of these farms

utilized solely off-farm personnel to inseminate cows, which may be a factor in the longer intervals between

observation and insemination than seen in the other regions.

Similar to the Cibao region, 89.5% of the farms surveyed in the Santo Domingo region reported use of

natural service within their herds. Most of the herds surveyed (76%) only use natural service for cows that had

been artificially inseminated multiple times and had not gotten pregnant. Four farms maintain bulls within

groups of non-pregnant cows and/or heifers with cow to bull ratios ranging from 25-100 cows per bull. The

farms maintaining larger groups of cows with bulls usually rotate bulls regularly. Average age of herd bulls is

around 39 months with about half of the herd bulls being 36 months or older. Finally, “breeding soundness”

programs were only in place for 5 of the farms surveyed with 3 of these farms reporting bull examinations

within the last 3 months but it is apparent that these examinations were not conducting according to high

standards involving semen evaluation.

Southwest Region

Only 4 out of the 8 surveyed farms in this region artificially inseminate cattle. In addition, farms

surveyed that used artificial insemination in the San Juan province sporadically inseminate eligible cows 10-12

hours after observation of estrus. The farm surveyed in Barahona inseminates multiple eligible cows at fixed

times (6-8 hours). The only farms reporting off-farm inseminators are two within the San Juan province.

All 8 farms within the Southwest region use natural service to some extent. Within this category, there

is no dramatic difference between the San Juan province and the farms sampled from Azua and Barahona.

However, as a whole, the Southwest region is different from other regions in the proportion of herds sampled in

which bulls are maintained with the lactating cows. Average bull to cow ratio is lower at 20 cows to 1 bull,

depending on the proportion of the lactating herd not pregnant, with the average age of bulls used being around

28 months old.

Even though 37.5% (3/8) of the farms surveyed reported conducting breeding soundness examinations, none of

the farms conducted full breeding soundness examinations on all bulls in the herd. “Breeding soundness”

programs on most of these farms consisted only of vaccination against Leptospirosis with periodic

63
Brucellosis testing. Bull testicles had been palpated and measured on at least one of the farms but similar to

other regions, semen morphology, quantity, and motility is rarely if ever evaluated.

Nutritional Management by Region

Since the present study was not designed specifically to deal with nutritional management a detailed

discussion will not be given. Nevertheless, summarizations of several variables of interest related to nutritional

management are summarized in the following figures by region.

Figure 14. Average Amount of Grain Fed to


Lactating Cows per Day

18
Amount of Grain (Kg)

16
14
12
10
8
6
4
2
0
Cibao Santo Southwest Total
Domingo
Region
Median Mean Std. Dev. Min. Max.

Table 14. Procedure Determining the Amount of Feed per Cow per Day.

Response Cibao Santo Domingo Southwest Total


n All cows receive the same ration 5 9 2 16
%
31.25 50.00 25.00 38.1

n Cows are fed according to level of 11 9 5 25


production
% 68.8 50.0 62.5 59.5

n Quantity depends ultimately on 0 0 1 1


availability of feed
% 0.0 0.0 12.50 2.4

Total* 16 18 8 42
*1 Observation missing

64
Figure 15. Daily Access to Pasture by Region

30 27

Number of Farms 25
20
15
15 12
9
10 6 7 6
5 2
0
Cibao Santo Domingo Southwest Total*
Region

No access to pasture Access to pasture


*1 Observation missing

Figure 16. Farms Depending on Pasture for a Significant


Proportion of Forage Dry Matter Intake for Lactating Cows

45
40
35 15
Number of Farms

30
25
20
6
15
7 27
10 2
5 8 13 6
0
Cibao Santo Domingo Southwest Total*
Region

Farms not depending on pasture Farms depending on pasture

*1 Observation missing.

On average, dairy producers in the D. R. feed lactating cows between 6-8 Kg of grain per day.

Producers within the Cibao region feed the most concentrate at 7.9 Kg/cow/day. When determining the

quantity each cow receives, 59.5% of the producers surveyed said that they based feeding on individual milk

production. The Cibao again had the highest proportion of producers (68.75%) allotting concentrate on the

basis of milk production.

Access to pasture varied by region, but overall, 35.7% of the producers surveyed reported that their

cows had some access to pasture, with 35.7% also saying that pasture provided a significant proportion of their

herd’s dry matter intake. The San Juan region within the Southwest utilized pasture the most, with 75% of the

farms sampled depending on pasture for a significant proportion of the dry matter intake for their herds.

65
Conclusions

The Cibao region provided the most variable data. Even though the region reported more use of

computers for record keeping than other regions, record quality and the quantity of needed information for

calculation of reproductive outcomes was variable, depending on the emphasis put on record keeping and the

type of system used.

The Cibao region is generally characterized by medium sized farms (average 98.1 lactating cows/herd)

and has the highest average herd composition of Holsteins (66.1%) than any other region sampled within the

country. Average body condition score is good (3.0) for the region with an average lameness score (1.383)

being intermediate between the other regions surveyed. Nutritionally, the region is roughly half pasture-based

and half-non-pasture-based with 46.7% of all producers depending on pasture for a significant proportion of the

dry matter intake for their cows.

Although 87.5% of producers use natural service to inseminate cows, reproductive management is

generally based on artificial insemination with only 14.3% of those surveyed depending totally on natural

service for insemination of their herds. Reproductive hormones are used in 87.5% of the herds sampled within

the region, with only 6.3% reporting regular prostaglandin use but 12.5% reporting regular GnRH use. All

farms relied on visual observation to detect estrus with 4 farms using Kamar devices and 3 using chalk to aid

in estrus detection.

The Santo Domingo region, though fairly broad geographically in distribution, provided some of the

most consistent data. The region contained, on average, larger farms with an average number of cows in the

milking herd of 145.6 cows. The cows within the region were also, on average, had good body condition scores

(average = 3.0) although lameness scores were also higher (1.552). Nutritionally, 63.18% of the producers

surveyed report that their lactating herd has access to pasture, although only 31.58% say that they actually

depend on pasture for a significant proportion of their herd’s dry matter intake.

Reproductive management within the Santo Domingo region generally involves artificial insemination

with only 11.76% of the producers relying on natural service to inseminate their whole herd. Also, 100% of all

producers surveyed use reproductive hormones to assist in getting cows pregnant. This region had the highest

reported regular use of prostaglandins (31.58%). Estrus detection is generally done by visual observation

(94.74%) with 2 farms using chalk and 1 farm using a teaser bulls. Reproductive outcomes for the region were

66
closer to optimal for service rate, days to first service, and calving to conception date and farther from optimum

in services per pregnancy, which could be due to higher lameness scores (see Chapter 6).

Although an overgeneralization, when considering the differences between the San Juan province and

the Azua and Baní provinces, the Southwest region is characterized by small to medium-sized farms that are

generally pasture-based nutritionally. Average BCS scores are too low (2.4) in the San Juan region of the

Southwest possibly due to depending too heavily on poor quality tropical grass pastures for dry matter intake.

More work needs to be done to increase the availability and quality of forages and to identify any other factors

influencing low body condition scores within cows of the San Juan region.

Lameness is lower than in other regions, on average, possibly due to the greater amount of time spent

on pasture and/or the lower average amount of grain fed within the region. Although the average lameness score

appears fairly good for the region (1.1), the farms sampled within the Azua and Barahona provinces both have

average lameness scores of 1.4. This indicates that more emphasis needs to be put on preventing new lameness

cases and culling chronically lame cows as soon as it is economically feasible on these farms, due to the

detrimental effects lameness can potentially have on reproductive efficiency and milk production [13, 71].

The Southwest provided an anomaly in not only data collection but management styles. This region

contained both the highest technology and the lowest technology farms of the country. All farms sampled

reported using natural service for some portion of their breeding programs with 62.5% depending on such bulls

to inseminate cows. Reproductive hormone usage was only reported in 25% of the herds surveyed in this

region. None of the farms in the San Juan province reported using hormones to help achieve pregnancies within

their lactating cows. Estrus detection is mainly by visual observation (75%) with 2-3 farms within the region

using mechanical aids or teaser animals for estrus detection. Five out of the 8 farms leave breeding bulls in

with the lactating herd eliminating the need for estrus detection, although inseminations still need to be

observed and recorded.

In general, the D. R. is following a basic pattern of development in their dairy management and

reproductive programs. Regions in closer proximity and/or with better means of communication with the

capital city, Santo Domingo, seem to have a higher level of sophistication in management than those regions

farther geographically from the capital or with fewer means of communications. Since the capital city usually

has the best infrastructure and data managing capabilities within the developing country, it is, therefore, an

important point of entry for new information through literature or in bringing professionals from other

67
countries. Depending on the infrastructure throughout the rest of the D. R., new information will either be

spread efficiently or sequestered. Like in many developing countries, information usually appears to reach

large producers first since they often have access to more resources for extension. However, for the Dominican

dairy industry to become self-sufficient, special attention needs to be placed in educating both large and small

producers in all areas of the country on the value of using calculated reproductive outcomes to make wise

management decisions.

68
Chapter 5

Days to First Service

Abstract

A model was constructed using regression techniques that selected the independent variables

(regressors) explaining the most variability within the data for the variable “days to first service”. The resulting

model contained eight regressors: breed, cooling system, both geographical region categories, average amount

of milk produced, number of employees involved in estrus detection, and the interactions of breed by cooling

system and breed by average amount of milk produced. Most of the variance within the model was explained

by the region categories. Relative to the Cibao region, the effects of the Santo Domingo region were associated

with fewer days to first service and the effects of the Southwest region were associated with higher days to first

service. An increase in days to first service was also associated with the number of employees involved in

estrus detection and the average amount of milk produced.

Introduction

Days to first service is inversely related to estrus detection in the early postpartum period [5] and is the

number of days from parturition to the first insemination in a given lactation. Optimal days to first service to

achieve a 13-month calving interval is around 75 d. As days to first service increase, days open and

subsequently calving interval and average days in milk also tend to increase; thereby, decreasing lifetime milk

production and profits.

Postpartum estrus detection and postpartum management (i.e. how efficiently postpartum cows return

to having a normal reproductive tract and to normal cycling activity) can be assessed by examining days to first

service. By identifying the regressors that explain the largest proportion of the variation within the data, the

variables that are associated with postpartum estrus detection and postpartum reproductive management can

also be identified. Average days to first service range from 89 to 108 in the farms surveyed (see Table 9,

chapter 4). Decreasing these averages could help set the foundation for improving reproductive efficiency

within the D. R.

69
Variable Selection and Description

Based on Mallow’s CP analysis, the optimal size of the model for days to first service should be

around seven regressors to minimize bias. Using a significance level of p< 0.10 for the model, max R2 was used

in SAS (“proc reg” with option statement “selection=maxr”) to facilitate selection of variables, following the

criteria outlined in the “Methods and Materials” section. After forcing in breed (percenthol), variables which

increased R2 by at least 5 % were added successively. The interaction variables for breed with both average

amount of milk produced and cooling system were added to the model during this selection process. Therefore,

both the interaction terms and the variables for average amount of milk produced and cooling system were

forced into the model. One of the two categories was added to the model during the selection process. Since

the region category consists of two “dummy” variables, both region categories had to be forced into the model.

Once all the necessary variables were forced into the model, the following table summarizes the independent

variables (regressors) that were part of the final model:

Table 15. Names and Descriptions of Variables influencing Days to First Service.

Variable Description
Percenthol** Proportion of Holsteins within a given herd

cool** Type of cooling system used (0=shade only,


1=evaporative/enhanced)

b_cool** Interaction between percenthol and cool (selected)

regcat1* ** Region category 1 (Santo Domingo region)


(selected)

regcat2* ** Region category 2 (Southwest region)

avmilk** Average amount of milk produced by the farm


(liters)

b_avmilk** Interaction between percenthol and avmilk


(selected)

Nuemphd Number of employees involved in estrus detection


(selected)
* When regcat1 and regcat2 both equal 0, the Cibao region is represented
**Indicates variables that were forced into the final model.

The backwards elimination procedure was run resulting in variable selection that proceeded in the same order as

the max R2 procedure (the same variables had to be forced into the model). The model was then examined for

the presence of high influence observations by calculating DFBeta values (“proc reg” in SAS with option

statement “/influence”). One farm was found to have high influence observations (Farm 1) and was deleted

70
from the selection process for the days to first service variable. Max R 2 was then repeated on all potential

regressors. Variable selection proceeded in the same manner as the first selection process, including having to

force in the same variables, with the resulting model having a 0.55 R2 compared to a 0.43 R2 from the first.

Results

Upon excluding high influence observations, the following model best explained the variance in days

to first service within the data, given minimal bias as calculated by Mallow’s CP and given that each regressor

explained at least an additional 5% of the variance independently within the model at a significance level of

p< 0.10 for the model as a whole:

dysfstsvc = 87.3 + 0.29 percenthol + 61.3 cool - 0.27 (percenthol)(cool) – 30.2 regcat1 + 17.7 regcat2

– 0.04(percenthol)(avmilk) + 0.17 avmilk + 8.66 nuemphd

This model was found to explain 54.8% of the variation within the data and was found to be statistically

significant at P = 0.0036. The following table lists the variables within the final model with their parameter

estimates, p-values, and 95% confidence intervals (β ± 1.96*SE):

Table 16. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Days to First Service.

Variable Parameter Estimate p-value 95% Confidence


Interval (days)
percenthol 0.29 0.60 -0.77 to 1.35
cool 61.3 0.12 -13.3 to 135.8
(percenthol)(cool) -0.27 0.60 -1.26 to 0.72
regcat1 -30.2 0.005 - 49.4 to - 11
regcat2 17.7 0.14 -5.33 to 40.7
(percenthol)(avmilk) -0.04 0.30 -0.11 to 0.03
avmilk 0.17 0.95 -5.48 to 5.80
nuemphd 8.66 0.004 3.28 to 14.04

Discussion

The following table (Table 17) summarizes the major findings from the model. It also contains the

tenth and ninetieth percentile values reported in the D. R. for any continuous variables and the “0” and “1”

categories for any categorical variables. Finally, it summarizes the differences according to the model in days

to first service between the tenth percentile or “0” category and the ninetieth percentile or “1” category to show

how dramatically days in first service can change between farms.

71
Table 17. Extreme Predicted Values from Days to First Service Model Building.

Variable 8.66 -30.2 17.7 61.3 cool


nuemphd regcat1 regcat2
10th percentile 1 “0” “0” 0cool
reported value or category category shade
“0” category Cibao Cibao
Days to first 8.66 0 0 0
service (days) for
10th percentile
reported value or
“0” category
90th percentile 5 “1” “1” 1cool
reported value or category category mechanical
“1” category Santo Southwest
Domingo
Days to first 43.3 -30.2 17.7 61.3
service (days) for
90th percentile
reported value or
“1” category
Difference in 34.6 -30.2 17.7 61.3
Days to First
Service (days)

According to the model, the independent variable that is most important associated with decreasing

days to first service is the region category for the Santo Domingo region, decreasing days to first service by

about 30 days (P = 0.005). An increase of about 18 days to first service, however, appears to be associated with

the Southwest region, (P = 0.14). The Cibao region appears to have an intermediary association with days to

first service between the Santo Domingo region and the Southwest. Regional differences are difficult to

interpret by themselves and could be a result of different management practices, different ambient temperature

and humidity, etc. One management practice in which the regions differ is in estrus synchronization. Although

hormone usage was not selected during model building, hormones are rarely used as part of a routine estrus

synchronization program within most of the country. Regular prostaglandin was only reported on 8 of the 43

farms sampled and regular GnRH use was only reported for 14% of the 43 farms sampled. Some studies have

shown that estrus synchronization offers one of the best techniques for optimizing estrus detection for first

service inseminations and reducing herd variability within days to first service, depending on the

synchronization program used [8, 9]. Within the Santo Domingo region, 100% of the farms sampled reported

72
either regular or occasional hormone use, followed by 87.5% in the Cibao region, and 25% in the Southwest

region. The frequency of hormone use between regions follows the pattern established in the model.

Given a larger sample size it might become apparent that the effects seen within the model are due to

hormone usage and not other regional differences. This hormone use pattern follows a trend often seen in

developing countries. In general, regions in closer proximity and/or with better means of communication with

the capital city and research universities have a higher level of sophistication in management than those regions

farther from the capital. Since the capital, Santo Domingo in the D. R., usually has the best infrastructure and

data managing capabilities in the country, it is, therefore, the point of entry for new information. Depending on

the infrastructure throughout the rest of the country, new information will either be spread efficiently or

sequestered [3]. Information usually reaches large producers first since they usually have access to more

resources for extension [3]. The San Juan province within the Southwest region provides one such example.

The San Juan province makes up 6 of the 8 farms sampled in the Southwest. Until recently, few information

sources, including veterinarians, have been available to inform producers of the use, benefits, and potential

benefits of reproductive hormone usage.

The most important independent variable associated with increasing days to first service is the number

of employees involved in estrus detection. According to the model, every additional employee assigned to

estrus detection, is associated with almost a 9-day increase in days to first service. In fact, days to first service

could increase by almost 35 days, as the number of employees increases from 1 to 5 (see Table B). On the

surface this finding seems contradictory, however, this result may be due to decreasing individual responsibility

as numbers of employees increase. Thus farms where “everybody” is responsible for estrus detection seem to

foster the “If it’s everybody’s job, then it’s nobody’s job” mentality. Producers within the D. R. should be

encouraged to select 1-2 reliable people who will be totally responsible for estrus detection. In this manner,

days to first service will not be unnecessarily elevated by too many people being responsible for detecting

estrus. Also, if estrus detection efficiency appears lower than desired, the source of the problem with visual

observation will be more quickly recognized when there are fewer people involved in the process.

73
Chapter 6

Services per Pregnancy

Abstract

A statistical model was constructed using various regression techniques to select the independent

variables (regressors) explaining the most variability within the data set as they related to “services per

pregnancy” while minimizing bias. The resulting model contained four regressors: average lameness score,

breed, and both geographic region categories. Most of the variance within the model was explained by average

lameness, supporting findings in earlier research which suggest that lameness decreases conception rate and

thus increases services per pregnancy.

Introduction

Services per pregnancy is usually reported as the number of inseminations per pregnancy for the

average cow within a given herd. Services per pregnancy is one of the most commonly used measurements to

assess herd fertility, the likelihood that any given service will result in conception. For U. S. conditions, to

maximize milk production and profits, the optimal services per pregnancy to achieve a 13-month calving

interval is around 1.8, assuming a VWP of 45-60 days [12]. Services per pregnancy can be adversely affected

by a large number of factors including decreased estrus detection accuracy, heat stress, improper A. I.

technique, reproductive pathology, and lameness [13].

Variable Selection and Description

Based on Mallow’s CP analysis (see Figure 2), the optimal size of the model should be around 4

regressors to minimize bias. Using a significance level for the model of p< 0.10, max R2 was used in SAS

(“proc reg” with option statement “selection=maxr”) to aid in variable selection, following the criteria outlined

in the “Methods and Materials” section. Forcing breed (percenthol) into the model, average lameness (avlame)

within the herd was added to the model, as was the geographical region variable for the Southwest (regcat2).

Since the region variable is a “dummy” variable matching another component, regcat1 (the Santo Domingo

region), the second “dummy” variable was forced into the model and the analysis run again, leaving the

variables listed in Table 28 in the final model:

74
Table 18. Names and Descriptions of Variables Influencing Services per Pregnancy.

Variable Description
percenthol** Proportion of Holsteins within a
given herd

avlame Average lameness score of a given


herd

regcat1* ** Region category 1 (Santo Domingo


region)

regcat2* ** Region category 2 (Southwest


region)
* When regcat1 and regcat2 both equal 0, the Cibao region is represented
**Indicates variables that were forced into the final model.

Backwards elimination was run and the same results were obtained when breed and region categories 1 and 2

were forced into the model. The model was then examined for the presence of high influence observations by

calculating DFBeta values (“proc reg” in SAS with option statement “/influence”). One farm was found to have

high influence observations (Farm 4) and was deleted from the selection process for the services per pregnancy

variable. Max R2 was then repeated on all potential regressors. Variable selection proceeded in the same

manner as the first selection process, including having to force in the same variables, with the resulting model

having a 0.44 R2 compared to 0.34 from the first model.

Results

Upon excluding high influence observations, the following model best explains the variance in services

per pregnancy within the data, given minimal bias as calculated by Mallow’s CP and given that each regressor

explains at least an additional 5 % of the variance independently within the model at a significance level of p<

0.10 for the model as a whole:

svcprpreg = 0.77 + 0.008 percenthol + 0.27 regcat1 - 0.04 regcat2 + 0.65 avlame

This model was found to explain 44.5% of the variation within the data and was found to be statistically

significant at p= 0.0007. The following table lists the variables within the final model with their parameter

estimates, p-values, and 95% confidence intervals (β± 1.96*SE):

75
Table 19. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Services per Pregnancy.

Variable Parameter Estimate p-value 95% Confidence


Interval (services)
percenthol 0.008 0.01 0.002 to 0.014
regcat1 0.27 0.22 -0.16 to 0.70
regcat2 -0.04 0.86 -0.50 to 0.42
avlame 0.65 0.15 -0.22 to 1.51

Discussion

The following table (Table 20) summarizes the major findings from the model (coefficients with p-

values < 0.15 in the final model). It also contains the tenth and ninetieth percentile values reported in the D. R.

for any continuous variables. Finally, it summarizes the differences according to the model in days to first

service between the tenth percentile and the ninetieth percentile to show how dramatically days in first service

can change between farms.

Table 20. Extreme Predictive Values from Services per Pregnancy Model Building.

Variable 0.008 percenthol 0.65 avlame

10th percentile 7.00% 1.02


reported value
Services/ pregnancy 0.06 0.66
for 10th percentile
reported value
90th percentile 98.0% 1.80
reported value
Services/ pregnancy 0.78 1.17
for 90th percentile
reported value
Difference in 0.73 0.51
Services/ pregnancy

One of the only statistically significant coefficients was for percentage of Holsteins within the herd.

The percentage of Holsteins within the herd appears to have little effect on services per pregnancy (see Table

20). It appears that as the percentage of Holsteins within a given herd increases from the tenth to the ninetieth

percentile observation in the data, the number of services per pregnancy increase almost negligibly.

This model indicates that increased services per pregnancy may be associated with lameness. With

every additional point that average lameness increases, an average cow requires an additional 0.65 services to

achieve pregnancy. These results agree with a study done by Hernandez et. al [13] in which cows with claw

76
lesions had a significantly higher services per pregnancy (median 5) than healthy cows (median 3).

Unfortunately, even though the t-tests for the coefficient was significant at the p = 0.15 level, the 95%

confidence interval for the coefficient also contained negative values and zero, which does not allow a

definitive statement on the role of lameness in increasing services per pregnancy. However, with P = 0.15 for

the parameter estimate, it is fairly reasonable to assume that the coefficient is not equal to zero and actually may

increase services per pregnancy by as much as 1.5 services (see Table z). Based on the findings in this study

and in others [13, 72], Dominican producers with chronically lame cows who are experiencing low conception

rates should consider preventing new lameness cases while evaluating their herd for lameness and culling

chronically lame cows as soon as it is economically feasible. These cows may only hurt reproductive efficiency

in the long run, decreasing farm profits and milk production.

Both region category variables were not statistically significant in the final model having p-values of

0.22 and 0.86 for the Santo Domingo and Southwest regions respectively. Therefore, making recommendations

based on the parameter estimates is imprecise at best.

77
Chapter 7

Projected Calving Interval

Abstract

A model was constructed using various multiple regression techniques that selected the independent

variables (regressors) explaining the most variability within the data regarding projected calving interval while

minimizing bias. The resulting model contained six regressors: average lameness score, breed, amount of

concentrate fed, cooling system, and interactions of breed by amount of concentrate and breed by cooling

system. Average lameness appears to be the factor associated with the largest and most statistically significant

per-unit increase in the projected calving interval.

Introduction

Calving interval is the period of time between successive parturitions (usually reported as an average

in months on the herd level). When measuring reproductive outcomes, a typical goal is a 12 to 13-month

calving interval, which has been found to be realistically the most suitable for maximizing profit and milk

production [9]. Calving interval is most directly affected by three areas of reproductive management: estrus

detection, the factor having the biggest impact on efficiency, voluntary waiting period (VWP), and conception

rate[8, 9]. As calving interval increases in a herd, days in milk for the average cow increases and lifetime milk

yield decreases [9]. .

Projected calving interval can be defined as the calving to conception interval for pregnant cows plus a

283-day gestation. It is one of the single best measurements for examining overall reproductive efficiency

within a given herd. Whereas historical calving intervals examine reproductive management for the previous

lactation, projected calving intervals examine more current reproductive management and are affected by

almost all other reproductive outcomes (days to first service, estrus detection efficiency and accuracy,

conception rates, and average days open). Therefore, by examining the variables which explain most of the

variation within projected calving interval data, we can make recommendations to producers within the D. R.

which could potentially improve their overall reproductive efficiency.

Variable Selection and Description

Based on Mallow’s CP analysis, the optimal sized model can only contain 6-8 regressors to minimize

bias. Using a significance level of p < 0.10, max R2 was used in SAS (“proc reg” with option statement

78
“selection=maxr”) to facilitate selection of variables, following the criteria outlined in the “Methods and

Materials” section. After forcing in breed (percenthol) into the model, variables were added successively which

increased R2 by at least 5%. The interaction variables for breed with both amount of concentrate fed and

cooling system were added to the model during this selection process. Therefore, both the interaction terms and

the variables for the amount of concentrate fed and cooling system were forced into the model. Once all the

necessary variables were forced into the model, the following table summarizes the independent variables

(regressors) that were included:

Table 21. Names and Descriptions of Variables Influencing Projected Calving Interval.

Variable Description
percenthol** Proportion of Holsteins within a given herd

avlame Average lameness score of a given herd


(scale of 1-4)

howmuchg** Amount of concentrate fed per cow (kg)

cool** Type of cooling system used (0=shade only,


1=evaporative/enhanced)

b grain** Interaction between percenthol and howmuchg

b cool ** Interaction between percenthol and cool


**
Indicates variables that were forced into the final model.

The backwards elimination procedure was run resulting in variable selection that proceeded in the same order as

the max R2 procedure (the same variables had to be forced into the model). The model was then examined for

the presence of high influence observations by calculating DFBeta values (“proc reg” in SAS with option

statement “/influence”). One farm was found to have high influence observations (Farm 4) and was deleted

from the selection process for the days to first service variable. Max R 2 was then repeated on all potential

regressors. Variable selection proceeded in the same manner as the first selection process, including having to

force in the same variables, with the resulting model having a 0.68 R2 compared to a 0.62 R2 from the first.

Results

Upon excluding high influence observations, the following model best explained the variance in

projected calving interval within the data, given minimal bias as calculated by Mallow’s CP and given that each

regressor explained at least an additional 5% of the variance independently within the model at a significance

level of p< 0.10 for the model as a whole:

79
projci (in days) = 321 + 0.69 percenthol – 0.06(percenthol)(howmuchg) – 0.49 howmuchg + 5.29 cool

+ 0.48(percenthol)(cool) + 61.1 avlame

This model was found to explain 68.4% of the variation within the data and was found to be statistically

significant at P < 0.0001. The following table lists the variables within the final model with their parameter

estimates, p-values, and 95% confidence intervals (β ± 1.96*SE):

Table 22. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Projected Calving Interval.

Variable Parameter Estimate p-value 95% Confidence


Interval (days)
percenthol 0.69 0.09 -0.09 to 1.47
cool 5.29 0.83 -42.4 to 53.0
(percenthol)(cool) 0.48 0.17 -0.19 to 1.15
howmuchg -0.49 0.91 -8.68 to 7.71
(percenthol)(howmuchg) -0.06 0.34 -0.18 to 0.06
avlame 61.1 0.002 25.2 to 97.0

Discussion

The following table (Table 23) summarizes the major findings from the model. It also contains the

tenth and ninetieth percentile values reported in the D. R. for any continuous variables. Finally, it summarizes

the differences according to the model in projected calving interval between the tenth percentile and the

ninetieth percentile to show how dramatically days in first service can change between farms.

Table 23. Extreme Predicted Values from Projected Calving Interval Model Building.

Variable 0.69 percenthol 61.1 avlame

10th percentile reported value 7.0% 1.02


Projected CI (days) for 10 th 4.83 62.3
percentile reported value
90th percentile reported value 98.0% 1.80
Projected CI (days) for 90 th 67.6 110.0
percentile reported value
Difference in Projected CI 62.8 47.7
(days)

Lameness appears to be the factor associated with the largest and most statistically significant per-unit

increase in the projected calving interval. During model building, lameness alone explained 11% of the

variation seen in projected calving interval data, more than any other selected regressor. According to the

model constructed, as average lameness increases by 1 point (for example, from score 1 = normal to score 2 =

80
mildly lame), projected calving interval increases from 25-97 days, based on a 95% confidence interval (see

Table 22)! Therefore, it is strongly suggested that producers take all steps possible to prevent and consider

culling chronically lame cows as soon as it is economically feasible to improve their overall reproductive

efficiency.

Dramatic increases in the projected calving interval are also associated with the percentage of

Holsteins in the herd. As the percentage of Holsteins in a given herd increases from 7% to 98%, projected

calving interval increases by 63 days, according to the model! A 100% Holstein herd would be expected to

have 69 additional days open compared to a non-Holstein herd. However, the association between projected

calving interval and the percentage of Holsteins within the herd is not so easily explained for two reasons: 1) the

95% confidence interval for percent Holstein contains negative values in addition to positive values and 2)

interactions were also part of the final model.

Since the coefficient was statistically significant at P = 0.09, it is unlikely that the coefficient is equal

to zero or is a negative number. In fact, the 90% confidence interval for the coefficient lies between 0.03 and

1.35 (β ± 1.65*SE). So, we are 90% confident, according to the final model, that increasing percentage of

Holsteins in herds is associated with increased projected calving intervals.

Unfortunately, however, neither the interactions or their component variables are statistically

significant enough in the final model for a definitive conclusion on how increasing the amount of concentrate

fed in herds with different percentages of Holsteins and the use of different cooling systems in herds with

various percentages of Holsteins are associated with changes in the projected calving interval. A larger data set

from a designed study in the future might be useful in interpreting the relationship between these variables and

their effects on overall reproductive performance. In the mean time, Dominican producers, especially with

higher percentages of Holsteins in their herds, should focus on preventing new incidences of lameness and on

culling chronically lame cows as soon as it is economically feasible.

81
Chapter 8

Service Rate

Abstract

A model was constructed using various multiple regression techniques that selected the independent

variables (regressors) explaining the most variability within the data regarding service rate while minimizing

bias. The resulting model contained five regressors: breed, average amount of milk produced, interaction of

breed by average amount of milk produced, and the two categories of reproductive records variables. Service

rate does not appear to be dramatically changed by any single variable within the model, although breed and

level of milk production effects appear to have the most important implications by improving reproductive

efficiency.

Introduction

Service rate, or the likelihood of an insemination per eligible day for cows, is an indicator of estrus

detection efficiency. Since the average estrous cycle of a dairy cow is around 21 days (18-24 days) [19], all

non-pregnant cows that have passed their voluntary waiting period should display estrus activity, be observed in

estrus, and inseminated about every 21 days. By determining what percentage of these cycles were actually

serviced, estrus detection efficiency can be evaluated. As applied in this study, there is no way to distinguish

whether the failure to have the service per cycle is due to failure of the cow to come into estrus or failure of

personnel to observe estrus that is manifest, undoubtedly there are elements of each phenomenon. Most farms

keep records of inseminations which makes service rate an ideal measure of this important reproductive

outcome. To maintain a 13-month calving interval, service rate should be around 0.033 services/day or 0.69

services/21-day cycle (estrus detection efficiency = service rate (services/day)( 21 days (100%) to obtain an

estrus detection efficiency of around 70%, assuming all non-pregnant cows are cycling normally.

Poor estrus detection efficiency is considered to be one of the most costly problems of AI programs [5,

14, 15, 17, 18]. As already discussed, several factors can result in poor estrus detection efficiency. For

example, estrus detection based on visual observation alone has been found to be inadequate [17-20]. This was

observed in 31 out of the 43 farms (72%) surveyed within the D. R. Also as herd size increases, time for estrus

observation for each cow tends to decrease [9, 17]. Other studies have also indicated that cows suffering from

heat stress are less likely to be observed exhibiting estrous activity and inseminated in a timely manner in

82
relation to the actual onset of the estrous cycle [24-26, 33]. Increasing herd sizes, estrus detection based solely

on visual observation, negative energy balance, and heat stress are just a few problems facing Dominican dairy

producers which could potentially result in low estrus detection efficiency. Mechanical aids to supplement

visual detection of estrus [14, 17, 18] have been shown to improve estrus detection efficiency. Estrus

synchronization has also been recommended as a method of increasing first insemination estrus detection

efficiency in postpartum cows, depending on the synchronization program used [8, 9].

Variable Selection and Description

Based on Mallow’s CP analysis, the optimal size of a model analyzing service rate should include four

regressors to minimize bias. Using a significance level of p < 0.10, max R2 was used in SAS (“proc reg” with

option statement “selection=maxr”) to facilitate variable selection, following the criteria outlined in the

“Methods and Materials” section (see Chapter 3). Forcing in percentage of Holsteins within the herd

(percenthol), variables were added successively which increased R 2 by at least 5%. To ensure that percentage

of Holsteins within the herd and amount of milk produced were not linearly dependent, percent Holstein and

average milk were examined for collinearity. The collinear relationship (R2 = 0.06) was not thought to be

strong enough to influence the model allowing both variables to remain in the model.

The interaction variables for breed with average amount of milk produced were added to the model

during this selection process. As a result, the variable for average amount of milk produced was forced into the

model. Once all the qualifying variables were forced into the model, the following table summarizes the

independent variables (regressors) that were included:

Table 24. Names and Descriptions of Variables influencing Service Rate.

Variable Description
Percenthol** proportion of Holsteins within a given herd

Avmilk** average amount of milk produced (liters)

b_avmilk** percenthol/avmilk interaction variable

Reccat1* ** reproductive record category 1 (individual cow


cards)

Reccat2* ** reproductive record category 2 (computer


program)
*When both record category 1 and 2 equal 0, notebook record keeping
systems are represented.
**Indicates variables that were forced into the final model.

83
The Backwards elimination was run resulting in variable selection that proceeded in the same order as the max

R2 procedure (the same variables had to be forced into the model). The model was then examined for the

presence of high influence observations by calculating DFBeta values (“proc reg” in SAS with option statement

“/influence”). No high influence observations were identified for the service rate variable.

Results

The following model best explains the variance in service rate within the data, given minimal bias as

dictated by Mallow’s CP and given that each regressor explains at least an additional 5% of the variance

independently within the model at a significance level of p< 0.10:

Svcrate (per 100 days) = 0.13 – 0.003 percenthol + 0.03 avmilk + 0.0005(percenthol)(avmilk) +0.44 reccat1

+ 0.05 reccat2

This model was found to explain 34.5% of the variation within the data and was found to be statistically

significant at P = 0.01. The following table lists the variables within the final model with their parameter

estimates, p-values, and 95% confidence intervals (β ± 1.96*SE):

Table 25. Independent Variables and Their Parameter Estimates, p-values, and 95% Confidence
Intervals for the Final Model of the Response Variable Service Rate.

Variable Parameter Estimate p-value 95% Confidence


Interval (services/100
days)
Percenthol -0.003 0.79 -1.813 to 1.807
Avmilk 0.03 0.59 -0.09 to 0.15
(percenthol)(avmilk) 0.0005 0.50 -0.001 to 0.002
reccat1 0.44 0.07 -0.03 to 0.91
reccat2 0.05 0.86 -0.46 to 0.56

Discussion

None of the variables selected appear to have a dramatic effect on service rate and only one variable

was statistically significant in the final model at P = 0.07(see Table ). However, several important points should

be mentioned. During the variable selection process, “ avmilk” was consistently added to the model, suggesting

that this parameter is probably one of the more resilient regressors in explaining the variability within the data.

However, average amount of milk produced may have been consistently added due to the forcing of breed into

the model. One should be extremely careful to examine breed, average milk, and the interaction between the

two variables when trying to understand the variation within service rate. Unfortunately, none of these

84
variables had coefficient with p- values < 0.15 in the final model, making a discussion of the exact associations

between the variables and service rate imprecise.

The following table (Table 26) summarizes the major findings from the model. It the “0” and “1”

categories for the categorical variables selected during model building. Finally, it summarizes the differences

according to the model in service rate between the “0” category and the “1” category to show how dramatically

days in first service can change between farms.

Table 26. Extreme Predicted Values from Service Rate Model Building.

Variable 0.44 reccat1 0.46 reccat2

“0” category Notebook Notebook


Service rate (per 100 days) for 0 0
“0” category
“1” category Cow Cards Computer
Service rate (per 100 days) for 0.44 0.46
“1” category
Difference in Service rate per 0.44 0.46
100 days (change in estrus (9.2%) (9.7%)
detection efficiency)

The largest and only statistically significant parameter estimate was the reproductive record category

(reccat1) for individual cow cards. As far as recording inseminations, the most complete service records were

found, in general, in cow card systems, followed by computer systems, and then notebooks. These results

follow the implications of the model.

Cow card systems were often more complete than computer-based systems because they were

frequently updated more regularly, kept on the farm (as opposed to computer-based systems which sometimes

were kept in the nearby town), and they were less problematic. Notebooks, on the other hand, only provided

partial insemination records for about a year at a time and individual cow information can be easily overlooked,

especially if cow identification numbers are not recorded clearly every time. Whatever record keeping system

Dominican dairy producers use (preferably individual cow cards or computer-based systems), it is imperative

that there be complete and accurate insemination records to be able to adequately assess reproductive outcomes

within the herd. Record keeping systems also serve as the basis for hormone usage systems that demonstrated

advantages in other models in this study.

85
Service rate might best be assessed in a more controlled study where record keeping is very thorough.

Also, performing variable selection without forcing breed may result in the development of a model that

explains more of the variation associated with service rate, although it might not be as useful in a practical sense

to Dominican producers.

86
Chapter 9

Summary and Conclusions

For the Dominican Republic to meet the ever-increasing domestic demand for milk, dairy producers

must improve their ability to meet the challenges of efficient reproductive management. Many obstacles to this

goal exist. The purpose of this project was to identify some of these obstacles and to provide a current

evaluation of reproductive efficiency and major factors believed to impact the same for the dairy industry within

different regions of the D. R.

During the course of the study, 43 farms were visited over a 10-week period and evaluated using a

survey and the collection of individual cow reproductive data from owner records. Once observations and cow

information had been made and catalogued, all data was summarized on the herd level and analyzed

descriptively. In addition to descriptive analysis, multiple regression techniques were used to select

independent variables which explain as much of the variance as possible for each of four reproductive

outcomes: days to first service, services per pregnancy, projected calving interval, and service rate.

Projected calving interval is a comprehensive index that gives an indication of overall reproductive

performance. Average lameness appears to be the factor associated with the largest and most statistically

significant per-unit increase in the projected calving interval, impeding overall reproductive performance.

Average lameness also explained the largest variation (11%) in projected calving interval. Due to this strong

association, Dominican dairy producers should strive to prevent new lameness and cull chronically lame cows

as soon as it is economically feasible to do so. Reducing the incidence of lameness could, in itself, dramatically

improve reproductive efficiency within the D. R.

Findings also indicate that increases in the percentage of Holsteins within the herd are also associated

with increased projected calving interval. This could be due to the decreased thermotolerance of the breed since

cooling systems also affect projected calving interval and an interaction between the breed and cooling system

was seen in the final model.

Days to first service gives an indication of estrus detection efficiency and postpartum management (i.e.

how efficiently postpartum cows return to normal cycling activity and are judged fit for insemination). Optimal

days to first service by U. S. standards is 75 days. Increases in the number of employees involved in estrus

detection, both geographical region categories, and the type of cooling system used were associated with higher

87
days to first service. Every additional employee assigned to estrus detection was associated with almost a nine-

day increase in days to first service. Producers within the D. R. should be encouraged to select 1-2 reliable

people to be given total responsibility for estrus detection. Estrus detection should be done several times

throughout the day and evening for a specified period of time (~30 minutes), with those responsible being

exclusively engaged in estrus detection [4]. Milking or feeding tasks distract both employees and cows and

decrease estrus detection efficiency.

The Santo Domingo region was associated with lower days to first service possibly due to widespread

reproductive hormone use within the region or more consistent record keeping which resulted in days to first

service appearing lower than in other regions which did not keep records as completely. In the Santo Domingo

region 100% of the farms sampled reported either regular or occasional hormone use compared with 87.5% in

the Cibao region, and only 25% in the Southwest region. These hormone-use patterns follow the same trend

established by the model. Some studies have shown that estrus synchronization offers one of the best

alternatives for optimizing time to insemination for first service inseminations. Synchronization programs may

also reduce herd variability in days to first service, depending on the synchronization program used [8, 9].

Researchers recommend that a structured reproductive program that is properly managed be in place before

considering estrus synchronization on a herd basis [6]. Thorough records are necessary in these programs so

that pregnancies are not lost and cows are inseminated at the proper times. Sporadic reproductive hormone use

for problem cows under veterinary supervision should be advocated when these conditions do not exist. As

record keeping systems improve, an appropriate synchronization program could be recommended.

Aids to estrus detection were not included in the model for days to first service perhaps because they

are used by an insufficient number of farms in this study. A number of other authors report that estrus detection

efficiency can be improved by using aids. Estrus detection based on visual observation alone has been reported

to be inadequate [17-20]. Visual observation is the only method employed by 31 of the 43 farms (72%)

surveyed in this study. Mechanical aids to estrus detection (tailhead chalk, Kamar patches, etc.) may help

increase estrus detection efficiency in herds currently only relying on visual observation. Simple aids can help

improve poor estrus detection efficiency. HeatWatch® transponders and tailhead painting with visual

observation have been shown to provide comparable results in estrus detection efficiency [18]. Producers

within the D. R. might not be able to afford expensive computer-based estrus detection systems but they can

improve their estrus detection efficiency by using tailpainting as a simple aid to visual observation.

88
Of all the farms sampled, partial shading was found in 85.4% and ranged from mild lack of shade (i.e.

uncovered watering trough) to severe lack of shade (i.e. only a few trees in the pasture). Producers should be

encouraged to see that all feeding and watering areas are shaded and that cows have a place to rest out of the

sun.

Services per pregnancy gives an indication of fertility or the likelihood of conception for each

breeding. Both average herd lameness score and percentage of Holsteins in the herd are associated with higher

services per pregnancy but neither are associated with increases >1 service per pregnancy.

One of the major obstacles facing Dominican dairy producers in attaining greater reproductive

efficiency is lameness in their herds. One-third of the average milking herd had lameness scores of 2 or greater

(mild to severe lameness) across all regions sampled within the D. R. In this study, average lameness within the

herd was associated with increases in explaining the variation in both services per pregnancy and projected

calving interval. As average herd lameness increases by 1 point (based on a 1 to 4 scale), services per

pregnancy and projected calving interval increases by 0.65 services/pregnancy and 61.1 days respectively.

These results agree with a report by Hernandez et. al. [13] in which lameness was shown to impair reproductive

efficiency by increasing average services per pregnancy within the herd. In addition to impaired reproductive

efficiency, lameness can also cause decreased milk yield. Although average lameness is highest in the Santo

Domingo region (mean score of 1.6), the prevalence of lameness is evident throughout all regions except for the

San Juan province within the Southwest.

Finally, although the percentage of Holsteins within the herd appears be associated with an increase in

the services per pregnancy. As the percentage of Holsteins within a given herd increases from the tenth to the

ninetieth percentile observation in the data, however, the number of services per pregnancy increase almost

negligibly.

Service rate was used in this study to give an estimate of estrus detection efficiency. None of the

variables within the model dramatically affected service rate. Service rate appears to increase slightly with

increasing milk production, especially in herds with lower proportions of Holsteins. This may result from

associations with increasing management intensity and higher numbers of thermally adapted breeds.

Many challenges face the Dominican dairy producer but the D. R. but increased management and

adoption of technology will allow dairy producers to meet those challenges. This study should by no means be

the final investigation on the factors influencing reproductive inefficiency. Since this study was mainly

89
observational, specific findings of interest should be researched using designed studies to get a broader

understanding of causes and effects. Also convenience sampling when sampling farms may have influenced

outcomes, making them more representative of intensely managed farms.

For example, one investigation that is badly needed within the D. R. will establish proper nutritional

management and forage to grain ratios that will aid producers in the use local tropical grasses and other forages.

Proper nutritional management is needed to provide for higher levels of milk production but prevent the high

incidence of lameness currently seen in the D. R.

Also a controlled investigation of breed interactions in the tropical environment of the D. R. would

help to determine breed characteristics that are best suited to allow for reproductive efficiency [24]. Another

helpful area of study involves the investigation of the changes that cattle undergo physiologically, specifically

hormonally, during periods of heat stress. The action of LH, serum estradiol-17β, and other reproductive

hormones during heat stress and its effect on fertility needs to be investigated more thoroughly [33]. If these

changes are more fully understood, solutions to the year-round heat stress of a tropical environment should be

more forth coming. For example, GnRH is known to increase estradiol concentrations and influence the size of

the ovulatory follicle [10]. Appropriate use of this hormone in a breeding program could offer one solution to

the suppression of estrous cycles caused by heat stress. However, research needs to be done to determine if

GnRH increases estradiol concentrations sufficiently in heat stressed cows to override the potential effects of

high temperature and humidity.

In addition to research, educational services should be further developed which translate known

technology to producers on relevant issues such as appropriate use of reproductive hormones and bull

management. Due to the effects of heat stress on male fertility [24], farms using dairy bulls as the sole source

of semen must be educated on proper bull management and the necessity of regular reproductive exams,

including examining semen for motility, morphology, and concentration.

The findings presented in this study, combined with studies conducted in other tropical environments,

and educational services to convey the results of these investigations must all work together to aid the producer

in making wise management decisions. Wise management decisions will eventually lead the Dominican dairy

producer to improved reproductive efficiency and the country to dairy self-sufficiency.

90
Literature Cited

1. Government, C.I.A.o.t.U.S., The World Factbook 2001. 2001, Central Intelligence Agency of the
United States Government.

2. Tejada Cabrera, F.A., Diagnóstico del Sector Lácteo en República Dominicana. 1998, Santo Domingo:
Colores, S. A. 126.

3. Norton, G.W. and J. Alwang, Introduction to Economics of Agricultural Development, in McGraw-Hill


Series in Agricultural Economics, P.J. Barry, Editor. 1993, McGraw-Hill, Inc.: New York. p. 20.

4. Radostits, O.M. and D.C. Blood, Dairy Cattle--Maintenance of Reproductive Efficiency, in Herd
Health: A Textbook of Health and Production Management of Agricultural Animals. 1985, W. B.
Saunders Company: Philadelphia. p. 66-89.

5. Bailie, J.H., Management and Economic Effects of Different Levels of Oestrus Detection in the Dairy
Herd. The Veterinary Record, 1982. 110(10): p. 218-221.

6. Jobst, S.M., R.L. Nebel, and e. al., Evaluation of Reproductive Performance in Lactating Dairy Cows
with Prostaglandin F2α, Gonadotropin-Releasing Hormone, and Timed Artificial Insemination.
Journal of Dairy Science, 2000. 83: p. 2366-2372.

7. Reimers, T.J., R.D. Smith, and S.K. Newman, Management Factors Affecting Reproductive
Performance of Dairy Cows in the Northeastern United States. Journal of Dairy Science, 1985. 68: p.
963.

8. Heuwieser, W., P.A. Oltenacu, and et al., Evaluation of Different Protocols for Prostaglandin
Synchronization to Improve Reproductive Performance in Dairy Herds with Low Estrus Detection
Efficiency. Journal of Dairy Science, 1997. 80: p. 2766-2774.

9. Nebel, R.L. and S.M. Jobst, Evaluation of Systematic Breeding Programs of Lactating Dairy Cows: A
Review. Journal of Dairy Science, 1998. 81: p. 1169-1174.

10. LeBlanc, S.J., K.E. Leslie, and e. al., Measures of Estrus Detection and Pregnancy in Dairy Cows
After Administration of Gonadotropin-Releasing Hormone Within an Estrus Synchronization Program
Based on Prostaglandin F2α. Journal of Dairy Science, 1998. 81: p. 375-381.

11. Eicker, S.W., Y.T. Grohn, and J.A. Hertl, The Association Between Cumulative Milk Yield, Days Open,
and Days to First Breeding in New York Holstein Cows. Journal of Dairy Science, 1996. 79: p. 235-
241.

12. Barnes, M., Reproduction and Lactation. 2001, Principles and Practices in Bovine Reproduction:
Blacksburg.

91
13. Hernandez, J., J.K. Shearer, and D.W. Webb, Effect of lameness on the calving-to-conception interval
in dairy cows. Journal of the American Veterinary Medical Association, 2001. 218(10): p. 1611-1614.

14. Walker, W.L., R.L. Nebel, and M.L. McGilliard, Time of Ovulation Relative to Mounting Activity in
Dairy Cattle. Journal of Dairy Science, 1996. 79: p. 1555-1561.

15. Heersche Jr., G. and R.L. Nebel, Measuring Efficiency and Accuracy of Detection of Estrus. Journal of
Dairy Science, 1994. 77(9): p. 2754-2761.

16. Maatje, K., S.H. Loeffler, and B. Engel, Predicting Optimal Time of Insemination in Cows that Show
Visual Signs of Estrus by Estimating Onset of Estrus with Pedometers. Journal of Dairy Science, 1997.
80(6): p. 1098-1105.

17. Senger, P.L., The Estrus Detection Problem: New Concepts, Technologies, and Possibilities. Journal
of Dairy Science, 1994. 77: p. 2745-2753.

18. Xu, Z.Z., et al., Estrus Detection Using Radiotelemetry or Visual Observation and Tail Painting for
Dairy Cows on Pasture. Journal of Dairy Science, 1998. 81: p. 2890-2896.

19. Nebel, R.L., What We Know About Heat Detection. 2001, Principles and Practices in Bovine
Reproduction class: Blacksburg.

20. Hurnik, J.F., G.J. King, and H.A. Robertson, Estrous and Related Behaviour in Postpartum Holstein
Cows. Applied Animal Ethology, 1975. 2: p. 55-68.

21. Schams, D., et al., The Oestrous Cycle of the Cow: Hormonal Parameters and Time Relationships
Concerning Oestrus, Ovulation, and Electrical Resistance of the Vaginal Mucus. Acta
Endocrinologica, 1977. 86: p. 180-192.

22. Nebel, R.L., et al., Comparison of On-farm and Laboratory Milk Progesterone Assays for Identifying
Errors in Detection of Estrus and Diagnosis of Pregnancy. Journal of Dairy Science, 1987. 70: p.
1471.

23. Gwazdauskas, F.C., J.A. Lineweaver, and W.E. Vinson, Rates of Conception by Artificial Insemination
of Dairy Cattle. Journal of Dairy Science, 1981. 64: p. 358-362.

24. Hansen, P.J., et al., Adverse Impact of Heat Stress on Embryo Production: Causes and Strategies for
Mitigation. Theriogenology, 2001. 55(1): p. 91-103.

25. de la Sota, R.L. and J.M. Burke, Evaluation of Timed Insemination During Summer Heat Stress in
Lactating Dairy Cattle. Theriogenology, 1998. 49: p. 761-770.

92
26. Thompson, J.A., et al., Management of Summer Infertility in Texas Holstein Dairy Cattle.
Theriogenology, 1996. 46(3): p. 547-558.

27. Gwazdauskas, F.C., J.A. Lineweaver, and M.L. McGilliard, Environmental and Management Factors
Affecting Estrous Activity in Dairy Cattle. Journal of Dairy Science, 1983. 66: p. 1510.

28. Lee, L.A., J.D. Ferguson, and D.T. Galligan, Effect of Disease on Days Open Assessed by Survival
Analysis. Journal of Dairy Science, 1989. 72: p. 1020.

29. Bakerma, H.W., et al., The Effects of Lameness on Reproductive Performance, Milk Production and
Culling in Dutch Dairy Farms. Preventative Veterinary Medicine 1994, 1994. 20: p. 49-259.

30. Pursley, J.R., M.C. Wiltbank, and e. al., Pregnancy Rates Per Artificial Insemination for Cows and
Heifers Inseminated at a Synchronized Ovulation or Synchronized Estrus. Journal of Dairy Science,
1997. 80: p. 295-300.

31. Johnson, H.D. The Lactating Cow in the Various Ecosystems: Environmental Effects on its
Productivity. in FAO Expert Consultation. 1989. Bangkok, Thailand: Food and Agriculture
Organization of the United Nations.

32. Berman, A., et al., Upper Critical Temperatures and Forced Ventilation Effects for High-Yielding
Dairy Cows in a Subtropical Environment. Journal of Dairy Science, 1985. 68: p. 1488-1495.

33. Ullah, G., et al., Effect of Gonadotropin-Releasing Hormone at Estrus on Subsequent Luteal Function
and Fertility in Lactating Holsteins During Heat Stress. Journal of Dairy Science, 1996. 79: p. 1950-
1953.

34. Hansen, P.J. and C.F. Aréchiga, Strategies for Managing Reproduction in the Heat-stressed Dairy
Cow. Journal of Dairy Science, 1999. 82 (Suppl. 2): p. 36-50.

35. Foote, R.H., Time of Artificial Insemination and Fertility in Dairy Cattle. Journal of Dairy Science,
1978. 62: p. 355-358.

36. de la Sota, R.L., et al., Effects of recombinant bovine somatotrophin (Sometribove) on ovarian function
in lactating and non lactating dairy cows. Journal of Dairy Science, 1993. 76: p. 1002-1014.

37. Gwazdauskas, F.C., et al., Hormonal patterns during heat stress following PGF2a-Tham salt induced
luteal regression in heirfers. Theriogenology, 1981. 16: p. 271-285.

38. Senger, P.L., Pathways to Pregnancy and Parturition. 1 ed. 1999, Pullman: Current Conceptions, Inc.
281.

93
39. Abilay, T.A., H.D. Johnson, and M. Madan, Influence of Environmental heat on Peripheral Plasma
Progesterone and Cortisol During the Bovine Estrous Cycle. Journal of Animal Science, 1975. 58: p.
1836-1840.

40. Wolfenson, D., et al., Seasonal and Acute Heat Stress Effects on Steroid Production by Dominant
Follicles in Cows. Animal Reproduction Science, 1997. 47: p. 9-19.

41. Wolfenson, D., et al., Effect of Heat Stress on Follicular Development During the Estrous Cycle in
Lactating Dairy Cattle. Biological Reproduction, 1995. 52(106-1113).

42. Wilson, S.J., et al., Effects of Controlled Heat Stress on Ovarian Function of Dairy Cattle. 2. heifers.
Journal of Dairy Science, 1998. 81: p. 2132-2138.

43. Putney, D.J., et al., Embryonic Development in Superovulated Dairy Cattle Exposed to Elevated
Ambient Temperatures Between the Onset of Estrus and Insemination. Animal Reproduction Science,
1989. 19: p. 37-51.

44. Mihm, M., et al., Effect of dominant follicle Persistence on Follicular Fluid Oestradiol and Inhibin
and on Oocyte Maturation in Heifers. Journal of Reproduction and Fertility, 1999. 1999: p. 293-304.

45. Edwards, J.L. and P.J. Hansen, Elevated Temperature Increases Heat Shock Protein 70 in bovine Two-
Cill Embryos and Compromises Function of Maturing Oocytes. Biological Reproduction, 1996. 55: p.
340-346.

46. Edwards, J.L. and P.J. Hansen, Differential Responses of Bovine Oocytes and Preimplantation
Embryos to Heat Shock. Molecular Reproductive Development, 1997. 46: p. 138-145.

47. Kerr, O.M. and W.J. McCaughey, Tail Painting Technique as an Aid to Oestrus Detection in Cattle.
The Veterinary Record, 1984. 114(25): p. 605-607.

48. Xu, Z.Z. and L.J. Burton, Estrus Synchronization of Lactating Dairy Cows with GnRH, Progesterone,
and Prostaglandin F2α. Journal of Dairy Science, 2000. 83: p. 471-476.

49. Moreira, F., C.A. Risco, and e. al., Use of Bovine Somatotropin in Lactating Dairy Cows Receiving
Timed Artificial Insemination. Journal of Dairy Science, 2000. 83: p. 1237-1247.

50. Gwazdauskas, F.C., C.J. Wilcox, and W.W. Thatcher, Environmental and Managemental Factors
Affecting Conception Rate in a Subtropical Climate. Journal of Dairy Science, 1974. 58(1): p. 88-92.

51. Putney, D.J., M. Drost, and W.W. Thatcher, Embryonic Development in Superovulated Dairy Cattle
Exposed to Elevated Ambient Temperature Between Days 1 to 7 Post Insemination. Theriogenology,
1988. 30: p. 195-209.

94
52. Ealy, A.D., M. Drost, and P.J. Hansen, Developmental Changes in Embryonic Resistance to Adverse
Effects of Maternal Heat Stress in Cows. Journal of Dairy Science, 1993. 76: p. 2899-2905.

53. Cole, J.A. and P.J. Hansen, Effects of Administration of Recombinant Bovine Somatotropin on the
Responses of Lactating and Nonlactating Cows to Heat Stress. Journal of the American Veterinary
Medical Association, 1993. 203: p. 113-117.

54. Rivera, R.M., et al., Short Communication: Seasonal Effects on Development of Bovine Embryow
Produced by in vitro Fertilization in a Hot Environment. Journal of Dairy Science, 2000. 83: p. 305-
307.

55. Rocha, A., et al., High environmental temperature and humidity decrease oocyte quality in Bos taurus
but not in Bos indicus cows. Theriogenology, 1998. 49: p. 657-665.

56. Ealy, A.D., et al., Effectiveness of Short-term Cooling and Vitamin E for Alleviation of Infertility
Induced by Heat Stress in Dairy Cows. Journal of Dairy Science, 1994. 77: p. 3601-3607.

57. Olson, T.A., A.C. Hammond, and C.C. Chase Jr., Evidence for the Existence of a Major Gene
Influencing Hair Length and Heat Tolerance in Senepol Cattle. Journal of Animal Science, 1997. 75:
p. 75 (Suppl. 1): 147 (abstr).

58. King, V.L., et al., Effects of a Hot Climate on the Performance of First Lactation Holstein Cows
Grouped by Coat Color. Journal of Dairy Science, 1988. 71: p. 1093-1096.

59. Rivera, R.M. and P.J. Hansen, Development of Cultured Bovine Embryos After Exposure to Increased
Temperatures in the Physiological Range. Reproduction 2001, 2001. in press.

60. Finch, V.A., Body Temperature in Beef Cattle: Its Control and Relevance to Production in the
Tropics. Journal of Animal Science, 1986. 62: p. 531-542.

61. Cunningham, E.P., The Genetic Improvement of Cattle in Developing Countries. Theriogenology,
1989. 31(1): p. 17-28.

62. Spain, J.N. and D.E. Spiers, Effect of Niacin Supplementation on Milk Production and
Thermoregulatory Responses of Dairy Cows. Journal of Dairy Science, 1997. 80 (Suppl. 1): 153
(abstr.).

63. Huber, J.T., et al., Heat Stress Interactions with Protein, Supplemental Fat, and Fungal Cultures.
Journal of Dairy Science, 1994. 77: p. 2080-2090.

64. Lotthamer, K.H. Influence of Nutrition on Reproductive Performance of the Milking/Gestating Cow in
the Tropics. in FAO Expert Consultation. 1989. Bangkok, Thailand.

95
65. Wildmann, E.E., et al., A Dairy Cow Body Condition Scoring System and Its Relationship to Selected
Production Characteristics. Journal of Dairy Science, 1982. 65: p. 495-501.

66. Whittier, W.D., Lameness Scoring Dairy Cattle on a 1 to 4 Scale. 2001, Angela Billings: Blacksburg.

67. SAS/STAT User's Guide, Version 6. 1990, SAS Institutes: Cary, N. C.

68. Whittier, W.D., Appropriate Parameters for Select Reproductive Outcomes and for Voluntary Waiting
Periods Based on Field Experience. 2001, Angela Billings: Blacksburg.

69. Draper, N.R. and H. Smith, Applied Regression Analysis. Third Edition ed. Wiley Series in Probability
and Statistics. 1998: John Wiley and Sons, Inc.

70. Myers, R.H., Classical and Modern Regression with Applications. Second ed, ed. M. Payne. 1990,
Belmont: Duxbury Press.

71. Rajala-Schultz, P.J., Y.T. Grohn, and C.E. McCulloch, Effects of Milk Fever, Ketosis, and Lameness
on Milk Yield in Dairy Cows. Journal of Dairy Science, 1999. 82: p. 288-294.

72. Collick, D.W., W.R. Ward, and H. Dobson, Associations Between Types of Lameness and Fertility.
The Veterinary Record, 1989. 125: p. 103-106.

73. Kautz, W.P., et al., Pioneer Forage Manual: A Nutritional Guide, C. Holland and W. Kezar, Editors.
1995, Pioneer Hi-Bred International, Inc.: Des Moines. p. 55.

96
Appendix A

Survey Results from First Visit to the Dominican Republic

Farm Name Farm Owner Location Approximate Number of Current Milk Type of record Duration records
of Farm Farm Size Cows in Herd Production keeping system are retained
(Region) (acres) (in Milk)
Finca Rositica Manuel Antuña Banao 469 200-300 Not obtained Computer Permanently?
Cabral (East (PCDART)
Cibao)
Jubaca Dr. Julio Brache Villa Mella Not obtained 200-400 (no Not obtained Computer For several years
(Santo exact number (PCDART)
Domingo) exists)*
Rancho Acedo- Salus Tiano Santo 250 Not obtained 26 lbs./cow/day Computer Permanently
Hacienda la Acebal Domingo (136) (DairyFlex)
Aldea (Santo
Domingo)
Hacienda Jorge Cury Baní (Santo 59 215 (?) 36 lbs./cow/day Notebook  Permanently?
Primavera Domingo) Individual Cow
Cards
Hermanos Two brothers Srs. Baní (Santo 33 340 (110) 14-16 Notebook  Permanently?
Incarnación Incarnación Domingo) lbs./cow/day Individual Cow
Cards
Las CaoBas Juan Precio Baní (Santo Not obtained 250 (100) 36 lbs./cow/day Calendar  For the life of the
Domingo) Individual Cow cow? (except number
Cards (going to of cows in milk which
computer— is discarded annually)
PCDART soon)
Rancho San Nelson “Chacho” Baní (Santo Not obtained 318 (118) 29 lbs./cow/day Notebook  Notebook is discarded
Raphael Landestoy Domingo) Individual Cow annually; owner
Cards keeps cards
permanently
Marta Cecilia Cesáreo Contreras Monte Not obtained ?(425) 26-28 Notebook  Permanently?
Plata lbs./cow/day Individual Cow
(Santo 
Cards PCDART
Domingo)


Hacienda Violeta Luis Felipe Checo Amina 137.5 350 (126) 44 lbs./cow/day General Since the computer is
(West notebook Specific broken, I do not know
Cibao) notebook (computer how much info. is
has been broken for available before 3 mo.
~3 mo.) prior to my visit.
Breeding notebook
kept permanently.


Rancho San Dr. Joselyn Hatillo, 352 568 (168) 40 lbs./cow/day General Several years
Diego Rodriguez Conde Paraje notebook Breed-
San Diego specific notebook
(East
Cibao)
Hacienda Pedro Idelfonso San 35 122 (105) 13.7 L/cow/day Notebook  Record kept until cow
Prikaren Franciso de Individual Cow leaves farm
Marcois Cards (computer to
(East be installed mid-
Cibao) year)
Hacienda Abigail Angel Camello San 156 164+small >15.5 L/cow/day Notebook  Permanently
Franciso de calves (93) Computer
Marcois (program?)
(East
Cibao)
Hacienda Cesa Taveras San 57.5 ? (60) 20 L/cow/day Notebook Permanently
Taveras Franciso de
Marcois
(East
Cibao)

97
Farm Name Farm Owner Location Approximate Number of Current Milk Type of record Duration records
of Farm Farm Size Cows in Herd Production keeping system are retained
(Region) (acres) (in Milk)
Hacienda Elba Santiago Paulino San 125 326 (140) 15.8 L/cow/day Notebook  Several Years
Franciso de Individual Cow
Marcois Cards
(East
Cibao)
Hacienda Jose Frank Banao 136 120 (70) 18 L/cow/day Individual Cow Many Years
Exagasa Guzman (East Cards
Cibao)
? (neighbor to Antonio Vargas Banao 11 + another 60 (24) 7.9 L/cow/day Notebook Kept Permanently
Finca Rositica) (East lot for heifers since 2 years ago
Cibao)
Hacienda Luis Columna Banao 234 100 (70) 14.4 L/cow/day Notebook Several Years
Columna (East
Cibao)
Rancho Belen Crisostomo Banao 46.9 150 (42) 8.6 L/cow/day Notebook Several Years
Risario (East (palpation lists only 1
Cibao) year)
Finca Castillo Virgilio Castillo Bani (Santo 43 143 (43) 5.0 L/cow/day General Notebook  Several Years
Domingo) Individual Cow
Cards
Finca Johanna Opinio Peña Bani (Santo 39 360 (125) 10.4 L/cow/day Notebook (each cow Throughout cow’s
Domingo) has her own page life?
since birth)—owner
planing on putting in
a computer mid year
Establo San Firpo Pementel Bani (Santo 15 160 (52) 15.8 L/cow/day Notebook (each cow Most of the previous
Antonio Domingo) has her own page years are saved—
since birth and some missing
summary pages are
kept each month)


Lecheria San Dr. Julio Brache Barahona 540 ~480 (260) 58 lbs./cow/day Notebook/daily Permanently?
Antonio-Cabral (Southwest) diary PCDART (number of cows in
and AFIFarm milk deleted yearly)

98
Appendix A (continued)
Specific Reproductive and Record Keeping Information

Farm Methods of Freq. AI only Person Freq. Number Frequency Are Frequency Number of Synchronization
Name Heat heats are herd (s) res- breedings of of records freshening cows in milk used? (often,
Detection recorded (yes/ ponsible are breedings pregnancy kept of dates are easily sometimes,never)
and (during no) for AI recorded in past checks preg recorded determined?
(Person(s) vWP) year checks? (yes/no)
Res- easily (yes/no)
ponsible) determine
d?
(yes/no)
Rositica Visual; K- Often Yes Farm Often Yes Once Yes Often Yes Often-on whole
mar; chalk (yes) manager monthly herd?
(3 people) and one by a local
assistant vet
Jubaca Visual; K- Often No Farm Often Yes Once Yes Often No* Sometimes-on
mar (yes) (“clean manager monthly cows not seen in
(manager -up and a by Dr. heat
primarily- bulls veterinary Tineo
all are employee
employees used)
are
responsible
also)
Rancho Visual; Often No Farm Often Yes Once Yes Often Yes Sometimes-on
Acedo- chalk (yes) (bulls manager monthly problem cows
Hac- (watchman used by local
ienda la and when vet; Dr.
Aldea milkers) insemin Tineo
-ator is checks
not problem
present and fresh
) cows
Hac- Visual Often No N/A Historic In theory, Once Yes (have Often No Sometimes-on vet
ienda (herd (yes) (bull ally, yes monthly to look recommendation
Primaver manager) bred often, by local through
a only but now vet individua
since 1 probably l cards)
yr. ago) sometim
es
Herman Visual Often Yes Farm Often Yes (have Once Yes Often Yes Sometimes-on
os (farm (yes) manager to look monthly problem cows
Incarnac manager) through by local
ión individua vet
l cards)
Las Visual (all Often No (AI Two Often Yes (have Twice Yes Often Yes (kept for Uses PGs during
CaoBas employees) (yes) and technician to look monthly one year) the hot season
Bull s through by local only
Bred) individua vet
l cards)
Rancho Visual Often No Farm Often Yes (have Once Yes Often Yes (not kept Sometimes
San (farm (yes) (“clean manager to look monthly on-farm) (conditions not
Raphael manager) up” through by local specified)
bull individ- vet
used) ual cards)
Marta Visual; Often Yes One Often Yes Twice Yes Often Yes Often—whole
Cecilia chalk (one (yes) employee monthly herd
employee at each by local
at each barn vet
barn)

99
Farm Methods of Freq. AI only Person Freq. Number Frequency Are Frequency Number of Synchronization
Name Heat heats are herd (s) res- breedings of of records freshening cows in milk used? (often,
Detection recorded (yes/ ponsible are breedings pregnancy kept of dates are easily sometimes,never)
and (during no) for AI recorded in past checks preg recorded determined?
(Person(s) vWP) year checks? (yes/no)
Res- easily (yes/no)
ponsible) determine
d?
(yes/no)

Hac- Visual (all Sometim No 2 Often Yes Varies Yes Often Yes Sometimes-on vet
ienda employees) es (yes) (“clean technician for AI; with recommendation
Violeta (“some up” s (1 on- no number of
escape”) bull farm, the indicatio animals to
used) other n about check by
from a recordin local vet
neighbori g of bull
ng farm) breeding
s
Rancho Visual Sometim No (AI Farm Often Yes Once Yes Often Yes Sometimes-on vet
San (farm es and manager monthly recommendation
Diego manager (no)— Bull by local
and 1 just cow bred) vet
assistant) being
bred
Hac- Visual Often No Cesa Often Yes for Twice Yes Often Yes Never
ienda (several (yes) (“clean Taveras for AI; AI—Bull monthly
Prikaren employees) up” (neighbor no breedings by local
bull ing indicatio ? vet
used) owner) n about
recordin
g of bull
breeding
s
Hac- Visual Often No Owner Often Yes Once Yes Often Yes Sometimes-on
ienda (farm (yes) (“clean monthly problem cows
Abigail manager) up” by local
bull vet
used)
Hac- Visual Often No Owner Often Yes Once Yes Often Yes Sometimes-on vet
ienda (owner) (yes) (“clean monthly (must recommendation
Taveras up” by local look
bull technician through
used) notebook
)
Hac- Visual (all Often No Owner’s Often Yes (have Varies Yes Often Yes Sometimes-on vet
ienda employees) (yes) (“clean son to look with recommendation
Elba up” through number of
bull individua animals to
used) l cards) check by
local vet
(usually 1-
2
times/mo.)
Hac- Visual; Often Yes? A local Often Yes (have Once Yes Often Yes Sometimes-on vet
ienda chalk (all (yes) technician to look monthly (Pregnant recommendation
Exagasa employees) through by local ,
individua vet postpartu
l cards) m, and
problem
lists kept)
? (neigh- Visual Often No— A local Often Yes Varies Yes Often Yes Never
bor to (farm (yes) AI and technician with
Finca manager) Bull number of
Rositica) used animals to
check by
local
technician

100
Farm Methods of Freq. AI only Person Freq. Number Frequency Are Frequency Number of Synchronization
Name Heat heats are herd (s) res- breedings of of records freshening cows in milk used? (often,
Detection recorded (yes/ ponsible are breedings pregnancy kept of dates are easily sometimes,never)
and (during no) for AI recorded in past checks preg recorded determined?
(Person(s) vWP) year checks? (yes/no)
Res- easily (yes/no)
ponsible) determine
d?
(yes/no)

Hac- Visual (all Often No Local Often Yes Daily by Yes Often Yes Sometimes-on vet
ienda employees) (yes) (“clean vet— local vet recommendation
Columna up” comes
bull daily
used)
Rancho Visual (all Never No N/A Often Yes Varies Yes Never Yes Never
Belen employees (hand with
—farm breed owner’s
manager to a discretion
primarily) bull)
Finca Visual (all Often No N/A Often Yes Twice Yes Often Yes Never
Castillo employees) (no) (bull monthly
bred by local
since 2 vet
yrs
ago)
Finca Visual Often No 1 Often Yes Twice Yes Often Yes Sometimes-(used
Johanna (farm (yes) (bull employee monthly mainly on AI
manager mostly and the by local heifers)
and 1 other —AI neighbor’ vet
employee) only s son
for
heifers)
Establo Visual (all Often No Owner’s Often Yes Once Yes Often Yes Sometimes-on
San employees (yes) (“clean son monthly problem cows
Antonio —cash up” by farm
incentive bull manager
for used)
detection)
Lecheria Visual; K- Often Yes but Two Often Yes Varies Yes Often Yes Sometimes-on vet
San mars; (yes) will be technician when a vet recommendation
Antonio- PCDART bringin s is available and on occasional
Cabral expected g in a groups of heifers
heat lists; “clean
pedometers up”
; (all bull
employees soon
with 3
primarily
responsible
)
*Most of the milking herd is being moved to Lecheria San Antonio-Cabral in Barahona. (See table for more information on this herd.)

101
Appendix B

Survey Forms

Survey Questions for Initial Observational Visit

With Dr. Tineo acting as my interpreter, the following basic questions were asked of each farm owner or
manager (depending on who was available for us to speak with):

1) What type of record keeping system does the farm use (i.e. computer program, notebook, individual cow
card, etc)?

2) How long is information kept on the farm (i.e. one year, multiple years, etc)?

3) Who is responsible for heat detection on the farm?

4) Are heats recorded sometimes, often, or never? Are they recorded even if the cow is not going to be
serviced, for example, during the voluntary waiting period?

5) Who is responsible for insemination? Is AI always used or is a bull used as well?

6) Are breeding dates recorded sometimes, often, or never?

7) Can you easily determine the number of breedings for the last year? Are all bull breedings recorded?

8) Who is responsible for pregnancy checks? How often are pregnancy checks done?

9) Are lists with all the cows checked during pregnancy exams and the diagnoses (open vs. pregnant) kept and
maintained?

10) Are freshening dates recorded sometimes, often, or never?

11) At any given time can one determine the number of cows in milk?

12) Does your herd make use of synchronization hormones and if so, under what circumstances?

13) In your opinion, what is the greatest problem you face in your herd reproductively?

14) How much land does the farm cover?

15) What is the current average production for the farm?

16) What do you feed your production animals and dry cows?

17) How many times a day do you feed your cows?

102
Initial Observational Chart
(Left on non-PCDART farms)

Reproductive Performance in Dairy Herds in the Dominican Republic*

Date Number of cows presented Number of cows Number of Number of heats


(Month) for pregnancy exam diagnosed cows within observed within the
pregnant the milking last month
herd

*This chart was translated into Spanish.

103
Study Survey Form

(Used During Second Visit)

Dominican Dairy Reproductive and


Nutritional Management Study

VA-MD Regional College of Veterinary


Medicine and The Dominican Association for
Dairy Self-sufficiency

Herd:

Herd owner:

Responder to
questionnaire:

Position/job of the
Responder: ___________________

Region of country: ___________________

Date:

104
Dominican Dairy Reproductive and Nutritional
Management Study

Operator Questionnaire

A. GENERAL INFORMATION:

A.1. Does the herd manager/owner seem interested


in cooperating in this study?…………………………....A101 Y1___ P2___N3____

A.2. What is the average BCS of the milking herd?…...A102 _________________

A.3. Are the BCS of the milking herd representative


of the dry cows as well?……………………………….. A103 Y1______N2_______
If no, please explain……………………………. A104
__________________________________________________________________
__________________________________________________________________

A.4. How many cows were lame within milking herd?..A105 _________________

A.5. List the breeds and approximate numbers of each


represented within the milking herd on the day of
the visit…………………………………………………..A106
___________________________________________________________________________
___________________________________________________________________________
__________________________________________________________________

A.6. Are these breeds and proportions representative


of the dry cows?…………………………………………A107 Y1______N2_______
If no, please specify……………………………..A108
__________________________________________________________________
__________________________________________________________________

A.7. Are feeding and watering areas shaded?……………A109 C1___P2___NA3____


(completely, partially, or not at all)

A.8. Which, if any, of the following devices are used to aid


in cooling cows? (check all that apply)
Fans?…………………………………………….……….A110 Y1______N2_______
Sprinkler system?………………………………………...A111 Y1______N2_______
Shaded barn?……………………………………………..A112 Y1______N2_______
Other?…………………………………………………….A113 Y1______N2_______

105
Dominican Dairy Reproductive and Nutritional
Management Study

Records Survey

B. RECORD INFORMATION:

B.1. How many cows are in the


milking herd on the day of the visit?……………………….. B114 __________________

B.2. Record the following information from the last 12-14 months of records:

NA= not available


Month 7/01 6/01 5/01 4/01 3/01 2/01 1/01 12/00 11/00 10/00 9/00 8/00 7/00 6/00

No. of cows
presented
for preg
check
No. of cows
diagnosed
pregnant
No. of cows
diagnosed
with cystic
structures on
ovaries
No. of
animals
diagnosed
open and
normal

B.3. What type of records are used on this farm? __________________

106
Milking Herd Evaluation
Farm Name:

Date of visit: Region of Country:

Srvs Srvs # svc from


Last since since Conc- last
BCS Lame (1- Date of First Normal Twins Currently Conc- last last eption Last Prev. successful
(lact. 4) (lact. Date Recorded birth born Postpartum Pregnant eption calv. calv. by NS Recorded Calving calving to
Cow No. Breed Only) Only) Fresh Breeding (Y/N) (Y/N) metritis (Y/N) (Y/N) Date dry Lact or AI DIM Svc Date conception

107
Dominican Dairy Reproductive and Nutritional
Management Study

Herd Owner/Manager Questionnaire*


* This part of the survey was translated into Spanish.

C. General Information

C.1. Is this farm the primary source of income for the owner? Y1______N2_______

C.2. What is/are the primary motivating factor(s) of the farm?


(Please circle all that apply.)
a. The farm is a hobby for the owner.
b. The owner inherited the farm and keeps it going
because farming is a family tradition.
c. The farm is working to improve production or milk
and/or components.
d. The farm is working to produce breeding stock for
sale to other producers.
e. Other
If other, please specify. __________________

_______________________________________________________________

_______________________________________________________________

C.3. How often does the owner visit the farm?


(Please circle the most appropriate answer.)
a. every day
b. several times per week
c. at least 2-3 times per month
d. every 3-6 months
e. once or twice a year

C.4. What is your current average milk production


per cow? ____________________

C.5. Please describe how you determine this


amount/weight? __________________
__________________________________________________________________
__________________________________________________________________

C.6. How much arable land does your farm cover? __________________

108
C.7. Using the following chart, please describe a typical day for the cows on your farm. Begin with recording
all milking times separately and then record the activities of cows between milking. Please indicate during
which of these activities, if any, that cows are observed for heats by circling “Yes” or “No” for each activity.

Start Time Finish Time Heat Detection


Milking Yes/No

Yes/No

Yes/No

Access to Feed Bunk Yes/No

Yes/No

Yes/No

Yes/No

Access to free Yes/No


stalls/barn
Yes/No

Yes/No

Access to pasture Yes/No

Yes/No

Yes/No

Yes/No

If cows are involved in another activity which is not listed, please describe the activity (include time of day the
activity occurs) and indicate whether cows are observed for heat during this activity.

________________________________________________________________________

________________________________________________________________________

________________________________________________________________________

D. Reproductive Program

D.1. Which method of heat detection is currently being


used on the farm? (check all that apply)
Visual observation? Y1______N2_______
K-mar? Y1______N2_______
Chalk? Y1______N2_______
Other? Y1______N2_______
If other, please specify. __________________
__________________________________________________________________

109
D.2. Who is primarily responsible for heat detection on your farm?
Herd owner only? Y1______N2_______
Herd manager only? Y1______N2_______
Total number of employees involved in heat detection? __________________

D.3. Who is primarily responsible for recording


heats on your farm?
(Please circle the most appropriate answer for your farm.)
a. Herd owner only?
b. Herd manager only?
c. 1-3 people? (including a and b)
d. More than 3 people? (including a and b)

D.4. Where are heats recorded? (Check all that apply)


Calendar? Y1______N2_______
Notebook? Y1______N2_______
Individual Cow Card? Y1______N2_______
Computer? Y1______N2_______
Other? Y1______N2_______
If other, please specify. __________________

D.5. On average, how soon after observation are


heats recorded? (Please circle the most appropriate answer
for your farm.)
a. Within 30 minutes?
b. Within the hour?
c. The same day?
d. Heats are recorded when the cow is bred.

D.6. How many times per day are cows exclusively


observed for estrus? __________________

D.7. Please list all times of day that cows are


observed for estrus? __________________________________________

__________________________________________________________________

D.8. Are these times associated with milking or feeding


activities? Y1______N2_______

D.9. For how long is the herd observed for estrus each time? _________________
(i.e. 15 minutes, 1 hour, etc.)

D.10.What is the voluntary waiting period on your farm? _________________

D.11.Do you ever breed cows before the voluntary waiting


period if they are seen in heat? Y1______N2_______

110
D.12.How much time per day do the cows in your milking
herd spend on concrete?
(Please circle the most appropriate category.)
a. None
b. 6 hours or less
c. 6-12 hours
d. 12-18 hours
e. 18-24 hours

D.13.Do you use hormones in helping cows get inseminated? Y1______N2_______


If so, please answer the following table regarding
hormone use within your herd.

Frequency of hormone use Yes No


I use prostaglandins (i.e. Lutalyse) only on an “as needed” basis.
I use prostaglandins as part of a regular routine program.
I use GnRH (i.e. Conceptal) only on an “as needed” basis.
I use GnRH as part of a regular routine program.
If other program used, please specify.
________________________________________________________________________

________________________________________________________________________

________________________________________________________________________

D.14.How many animals in the herd are included in the


hormone treatments?
(Please circle the most appropriate answer for your farm.)
a. Whole herd?
b. Select groups within the herd (i.e. heifers)?
c. Vet recommendation or problem cows only?

D.15.Are hormone treatments used:


Year round? Y1______N2_______
Only during certain times of the year (i.e. hot season)? Y1______N2_______
When? __________________
__________________________________________________________________

D.16.Do you breed any of your cows with a bull? Y1______N2_______

111
D.17.If so, for which of the following groups do you use a bull?
(Please circle the most appropriate answer for your farm.)
a. Whole herd?
b. Heifers only?
c. Milking herd only?
d. Problem cows (i.e. repeat breeders)?

D.18.If your bull is kept with a group of cows for breeding


purposes, how many cows are in his breeding group? _________________

D.19.How old are your bulls? _________________

D.20.Do you have a vet check your herd bulls? Y1______N2_______

D.21.If so, when was the last time your bulls were checked
by a vet?
(Please circle the most appropriate answer for your farm.)
a. Within the last 3 months
b. Within the last 6 months
c. Within the last year
d. Over a year ago

D.22.Within the last 6 months, who performed the inseminations on


your farm? (Please circle the most appropriate answer.)
a. Herd owner?
b. Herd manager?
c. Other?
If other, please specify. ________________________________________________________

________________________________________________________________________________________

Within the last six months, how many people total have been
responsible for inseminating cows on your farm? ________________________

112
D.23.How did the inseminator(s) learn to use AI?
(Please, check the most appropriate answer for each inseminator.)
From a From Self- Other (if Proportion of
training another taught other, inseminations performed
course employee please within the last 6 months
specify) (%)
Inseminator 1

Inseminator 2

Inseminator 3

Inseminator 4

Inseminator 5

D.24.Does the person(s) responsible for inseminating:


Breed multiple cows during the same time period. Y1______N2_______
Breed cows sporadically as they are seen in heat. Y1______N2_______

D.25.If multiple cows are bred during the same time period,
what are the maximum number of straws of semen
that are thawed at one time? (i.e. 1,2,3…) __________________

D.26.On average, how long after a cow is seen in heat is she


bred? (Please circle the most appropriate response.)
a. Within the hour?
b. Within 6 hours?
c. 6-12 hours later?
d. 12-24 hours later?
e. Fixed time, all cows in heat are bred at the same time (1-2 times daily)

D.27.Who is responsible for palpating your cows?


(Please circle the most appropriate answer.)
a. Cows are not routinely palpated.
b. An employee?
c. A technician?
d. A veterinarian?
e. The herd owner?
f. Other?
If other, please specify. ________________________

D.28.For the person doing the most palpation, how long has
this person been palpating your cows? __________________

113
D.29.Does this person ever recommend hormone injections for
problem cows and/or repeat breeders? Y1______N2_______

D.30.How often are your cows palpated?


(Please circle the most appropriate answer for your farm.)
a. Never
b. Varies with the number of cows to be checked?
c. Once monthly?
d. Twice monthly?
e. Other?
If other, please specify.
____________________________________________________________

____________________________________________________________

Nutritional Program

D.31.Within the last 8-12 months, when determining what to feed your cows, have you used
the services of the following professionals/experts/consultants?
(Check all that apply.)
Veterinarian? Y1______N2_______
Nutritional consultant? Y1______N2_______
Feed representative? Y1______N2_______
Neighbor/Local producer? Y1______N2_______
Cattle magazines or Industry Journals? Y1______N2_______
None of the above, I formulate my own ration. Y1______N2_______
Other. Y1______N2_______
If “other”, please specify. __________________

__________________________________________________________________

D.32.How do you determine how much to feed your cows?


(Please circle all that apply.)
a. All cows get the same amount
b. Cows are fed according to their level of production
c. Amount fed depends on feed availability
d. Amount fed changes in the hot season

114
D.33.To which of the following feeds do milking cows
in your herd have access? (Check all that apply.)
Concentrate/grain? Y1______N2_______
If yes, how much per day per cow? __________________
Please specify the type of grain fed __________________

_________________________________________________________________

By-products? Y1______N2_______
If yes, how much per day per cow? __________________
Please list the by-products fed.
__________________________________________________________________

________________________________________________________________________

Cut grass? Y1______N2_______


If yes, how much per day per cow? __________________
What type of cut grass do you feed?
(Please circle the most appropriate choice for your farm)
a. Long grasses (i.e. King grass)
b. Short grasses (i.e. African star)
c. Mix of long and short grasses

Pasture? Y1______N2_______
If yes, how many hours per day are cows
on pasture? _____________hours
Do you depend on pasture as a
significant source of dry matter intake
for your cows? Y1______N2_______

TMR? Y1______N2_______
If yes, how much per cow per day? _________________

D.34.How often do you feed your cows?


1 time per day? Y1______N2_______
2 times per day? Y1______N2_______
3 times per day? Y1______N2_______
Free choice? Y1______N2_______
Other? Y1______N2_______
If other, please specify. __________________________________________

__________________________________________________________________

D.35.Please estimate the percentage of feed NOT consumed


by cows between feedings? ________________%

D.36.How often do you clean the feed bunks? ______________/day

115
D.37.Do you feed grain in the parlor during milking? Y1______N2_______

Thank you for all your cooperation on this survey.

116
Appendix C

Calculation of Dependent Variables (Outcome Variables)

1) Service Rate = Number of inseminations / (days in milk – VWP)

2) Services per Pregnancy = Number of inseminations from calving to conception for current lactation
(pregnant cows)
OR
Number of inseminations from calving to conception for the previous lactation
(non-pregnant cows)

3) Projected Calving Interval (pregnant cows only) = (Date of conception* – date of parturition for this
lactation) + 283

4) Days to First Service = Date of first insemination for this lactation – date of parturition for this
lactation

*
As determined by breeding and palpation records

117
Appendix D

Conversion Tables [73]

Area

1 hectare = 2.47 acres

1 acre = 6.4 tareas*

Fluid

1 pint = 1 pound

1 pint = 0.57 liter

740 mL = 1 bottle *

Weight

1 pound = 0.45 kilogram

*
Personal Communication, Dr. Leonardo Tineo, Tecnico Coordinador, Dominican Association for Dairy Self
Sufficiency, Santo Domingo, D. R.

118
Appendix E

Additional Descriptive Statistics of Interest

Table A. Descriptives for continuous survey data.

Region No. Obs. Variable Median Mean Std. Dev. Min. Max.
Cibao 16 Arablelnd 56.94 56.42 29.55 11.38 98.38
Numtimilk 2.000 2.063 0.250 2.000 3.000
Hdnutimes 2.000 2.636 1.098 1.000 4.500
Vwp 45.00 47.34 7.386 30.00 60.00
Agebull1 26.00 30.54 18.79 9.500 84.00
Agebull2 25.00 29.00 13.43 12.00 48.00
Agebull3 39.00 39.00 29.70 18.00 60.00
Agebull4
Nuresins 1.000 1.267 0.458 1.000 2.000
Maxstraw 1.000 1.077 0.277 1.000 2.000
Nuyrspalp 6.000 7.344 6.172 1.000 20.00
freqfed 2.000 2.571 1.016 1.000 5.000

Santo Domingo 19 Arablelnd 51.63 75.20 71.31 9.474 253.0


Numtimilk 2.000 2.053 0.229 2.000 3.000
Hdnutimes 2.000 2.083 0.996 0 4.000
Vwp 50.00 51.67 9.096 35.00 70.00
Agebull1 36.00 41.50 23.46 12.00 108.0
Agebull2 42.00 39.56 16.15 18.00 60.00
Agebull3 24.00 28.67 11.72 20.00 42.00
Agebull4 32.00 32.00 14.14 22.00 42.00
Nuresins 2.000 1.667 0.767 1.000 4.000
Maxstraw 1.000 1.733 1.033 1.000 4.000
Nuyrspalp 6.000 6.647 4.523 1.000 18.00
Freqfed 2.000 2.567 0.821 2.000 4.500

Southwest 8 Arablelnd 25.24 54.07 66.03 9.300 190.3


Numtimilk 2.000 2.125 0.354 2.000 3.000
Hdnutimes 2.000 1.667 1.506 0 4.000
Vwp 52.50 51.07 10.98 30.00 60.00
Agebull1 22.00 23.50 6.568 18.00 36.00
Agebull2 36 31.50 16.49 8 48
Agebull3 30 30 30 30
Agebull4 42 42 42 42
Nuresins 1.5 1.750 0.957 1 3
Maxstraw 1.5 1.750 0.957 1 3
Nuyrspalp 2.5 3.609 3.468 0.375 10
freqfed 2 2.25 0.886 1 4

119
Table B. Descriptives for categorical survey data.

Are breed types and proportions equally distributed between the lactating and dry cows?

Response Cibao Santo Southwest Total


Domingo
Count No 1 0 0 1
Frequency (%) 8.33 0.00 0.00
Count Yes 11 14 5 30
Frequency (%) 91.67 100.00 100.00
Total* 12 14 5 31
*Frequency Missing = 12

Is the farm the principal source of income for the owner?

Response Cibao Santo Southwest Total


Domingo
Count No 2 8 6 16
Frequency (%) 12.5 42.11 75
Count Yes 14 11 2 27
Frequency (%) 87.5 57.89 25
Total 16 19 8 43

What is the primary goal of the farm?

Response Cibao Santo Southwest Total


Domingo
Count The farm is a 0 1 0 1
Frequency (%) hobby 0 5.26 0
Count The farm is a 3 3 0 6
Frequency (%) family tradition 1.88 1.58 0
Count Increase milk 16 17 7 40
Frequency (%) production 100 89.5 87.5
Count Produce 4 5 3 12
Frequency (%) breeding stock 25 26.3 37.5
Count Other 0 0 1 1
Frequency (%) 0 0 12.5
Total* 23 26 11 60
*Multiple answers could be applicable for a given farm.

120
How often does the owner visit the farm?

Response Cibao Santo Southwest Total


Domingo
Count Every day 12 9 6 27
Frequency (%) 75 47.37 75
Count Several times 4 10 1 15
Frequency (%) a week 25 52.63 12.5
Count At least 2-3 0 0 1 1
Frequency (%) times/month 1 1 12.5
Total 16 19 8 43

On average, how soon after an estrus is detected is it recorded?

Response Cibao Santo Southwest Total


Domingo
Count Within 30 4 3 2 9
Frequency (%) minutes 25 16.67 25
Count Within the 0 2 1 3
Frequency (%) hour 0 11.11 12.5
Count The same day 3 8 1 12
Frequency (%)
Count At 8 5 4 17
Frequency (%) insemination 50 27.78 50
Count Other 1 0 0 1
Frequency (%) 6.25 0 0
Total* 16 18 8 42
*1 observation missing

Are cows inseminated before the end of the voluntary waiting period (VWP)?

Response Cibao Santo Southwest Total


Domingo
Count No 12 5 5 22
Frequency (%) 85.71 29.41 62.50
Count Yes 2 11 2 15
Frequency (%) 14.29 64.71 25.00
Count Farm has no 0 2 1 3
Frequency (%) VWP 0 11.12 12.5
Total* 14 18 8 40
*3 observations missing

121
Of the farms using hormones for synchronization: which groups of cows are
included in hormone treatments?

Response Cibao Santo Southwest Total


Domingo
Count The whole 0 1 0 1
Frequency (%) herd 0 5.26 0
Count Heifers only 3 0 2 5
Frequency (%) 21.43 0 100
Count Problem cows 14 18 1 33
Frequency (%) 100 94.74 50
Total* 17 19 3 38*
*More than one possible response per farm. No answer recorded for 8 of the farms.

Within the past six months, who was primarily responsible for inseminating
the adult cows on your farm?

Response Cibao Santo Southwest Total


Domingo
Count Owner 5 2 0 7
Frequency (%) 31.25 10.5 0
Count Herd manager 7 11 0 18
Frequency (%) 43.8 57.9 0
Count Other 2 0 1 3
Frequency (%) Employee 12.5 0 12.5
Count Veterinarian 0 8 1 7
Frequency (%) 0 26.3 12.5
Count Off-farm 1 3 1 5
Frequency (%) technician 6.25 15.79 12.5
Count Bull bred herd 1 1 4 6
Frequency (%) 6.25 5.26 50.0
Count Neighbor 1 0 0 1
Frequency (%) 6.25 0 0
Count Other 1 0 1 2
Frequency (%) 6.25 0 12.5
Total* 18 25 8 51*
*More than one possible answer per farm.

122
In what manner are cows inseminated?

Response Cibao Santo Southwest Total


Domingo
Count Multiple cows 3 9 1 13
at specified
Frequency (%) times 21.4 50 25
Count Sporadically as 11 9 3 23
Frequency (%) seen in estrus 78.6 50.0 75.0
Total* 14 18 4 36
*7 Observations missing

Who is primarily responsible for the palpation of your cows?

Response Cibao Santo Southwest Total


Domingo
Count Farm owner 0 1 0 1
Frequency (%) 0 5.26 0
Count Technician 0 0 2 2
Frequency (%) 0 0 25.0
Count Veterinarian 16 18 6 40
Frequency (%) 100.0 94.74 75.0
Total 16 19 8 43

Does the person that is primarily responsible for palpating your cows recommend hormone injections for
problem cows or repeat breeders?

Response Cibao Santo Southwest Total


Domingo
Count No 1 0 6 7
Frequency (%) 6.25 0 75.0
Count Yes 15 19 2 36
Frequency (%) 93.75 100 25.0
Total 16 19 8 43

Do you offer grain to your lactating cows during milking?

Response Cibao Santo Southwest Total


Domingo
Count No 7 5 2 14
Frequency (%) 43.75 26.32 25.0
Count Yes 9 14 6 29
Frequency (%) 56.25 73.68 75.0
Total 16 19 8 43

123
With what frequency are your cows palpated?

Response Cibao Santo Southwest Total


Domingo
Count Varies with no. 3 3 3 9
of cows to be
Frequency (%) checked 18.75 15.79 37.5
Count One time per 6 4 2 12
Frequency (%) month 37.5 21.05 25.0
Count Twice 3 5 0 8
Frequency (%) monthly 18.75 26.32 0
Count Never
Frequency (%)
Count Once or more 0 4 1 5
Frequency (%) weekly 0 21.05 12.5
Count Once every 2 2 3 0 5
Frequency (%) months 12.5 15.79 0
Count <1 time every 2 0 2 4
Frequency (%) 2 months 12.5 0 25.0
Total 16 19 8 43

Within the last 8-12 months, when determining what to feed your herd, have you used
the services of the following professionals/experts/consultants?

Response Cibao Santo Southwest Total


Domingo
Count Veterinarian 3 4 4 11
Frequency (%) 18.75 21.05 50.0
Count Nutritional 11 11 2 24
Frequency (%) consultant 68.75 57.89 25.0
Count Feed 1 1 2 4
Frequency (%) representative 6.25 5.26 25.0
Count Neighbor/local 1 0 1 2
Frequency (%) producer 6.25 0 12.5
Count Magazine/ 0 4 1 5
Frequency (%) newspaper 0 21.05 12.5
Count I formulate 5 8 5 18
Frequency (%) my own ration 31.25 42.11 62.5
Count Other 1 2 1 4
Frequency (%) 6.25 10.53 12.5

Are lactating cows given access to a total mixed ration (TMR)?

Response Cibao Santo Southwest Total


Domingo
Count No 11 15 7 33
Frequency (%) 68.75 78.95 87.5
Count Yes 5 4 1 10
Frequency (%) 31.25 21.05 12.50
Total 16 19 8 43

124
Vita

Angela Renea Billings was born in Maeberry, West Virginia on December 22, 1974. Her family soon moved to

Sugar Grove, Virginia four years later. After graduating from Marion Senior High School in 1992, she attended

Emory & Henry College in Emory, Virginia, where she earned a B.S. in Biology in 1996. During 1995-96, she

helped organize and worked on two student-led projects investigating intramammary treatments for mastitis in

dairy cattle. During her undergraduate career, she also logged over 100 hours of observation experience at local

small and large animal veterinary clinics. After graduation from E&H in the spring of 1996, she accepted a

summer job at the Chilhowie Animal Hospital working as a technician’s assistant. That fall, she began

attending veterinary school at the Virginia-Maryland Regional College of Veterinary Medicine where she

graduated with her D.V.M. in May of 2000. During vet school, she decided to pursue a career in veterinary

missions, prompted in part by a six-week missions trip to Bolivia, South America. When the opportunity arose

to improve her Spanish and help the Dominican dairy producers, she applied and was accepted to her current

program. After finishing her Master’s degree, she plans to marry and move to Baltimore, MD, where in

addition to serving as a veterinarian, she plans to be active in a local Hispanic ministry. Her long-range goals

include serving as a veterinary missionary in a developing Hispanic country.

125

Вам также может понравиться