CASE STUDY: Behavioral, Anatomical, and Physiological Changes in Late Gestation Mares

Introduction

Most scientific studies have attempted to find predictors of parturition in periparturient mares; however, there is a void in the information describing the changes in behavior and certain physiological parameters as parturition approaches. A concise compilation of reference ranges for a variety of changes in the late gestation mare is not readily available but would allow veterinarians, researchers, and mare owners to recognize abnormal or "at risk" pregnancies. Average gestation length in the mare is 340 d (Evans, 1989), but this can vary considerably (Pugh and Schumacher, 1990; Marteniuk et al., 1998; Davies Morel et al., 2002). Eighty-six percent of mares foal during the night or early morning hours when darkness or low light conditions exist (Rossdale and Short, 1967; Hafez, 1980). Some mares, specifically maiden mares (Berger, 1986), may seek seclusion (Barty, 1974) and an increase in periparturient activity is more apparent in semiferal mares (McDonnell, 2000); however, behavior of mares late in gestation does not appear to change significantly (Waring, 1983; Shaw et al., 1988) until just prior to initiation of stage 2 labor (Bazer and First, 1983). The most consistent behavioral change in the periparturient mare is often described as "nervousness" and includes increased walking, pacing, pawing, laying down, and getting up (Barty, 1974; Waring, 1983; Crowell-Davis and Houpt, 1986; Shaw et al., 1988; Evans, 1989); however, timing of these signs in relation to parturition has been variable among mares, and not all mares show all of the aforementioned signs. Some groups have attempted to quantify changes in activity level through visual observation (Crowell-Davis and Houpt, 1986; Shaw et al., 1988) but were unable to detect any significant change in activity until the 30 min prior to parturition. These findings may indicate a lack of sensitivity to detect subtle changes in activity that mares may undergo as parturition approaches. The present study was designed to provide reference ranges for specific behavioral, anatomical, and physiological changes observed in late gestation Quarter Horse mares.

Materials and Methods

Subjects and Management. This project was conducted with prior approval of The Pennsylvania State University's Institutional Animal Care and Use Committee (IACUC #01R139). Ten multiparous Quarter Horse mares used in this study ranged in age from 7 to 19 yr (12.2 ± 1.4), had from 3 to 12 parities (6.8 ± 1,1), and had previous gestation lengths from 330 to 345 d (336.8 ± 1.7). All mares resided in the same herd for at least 7 yr, had been housed within the same group of mares for at least 4 yr, and had foaled at least 3 times under the same management regimen. None of the mares used in this study had delivered foals with perinatal abnormalities or had a history of foaling difficulty.

Management protocol for these mares was consistent throughout the data collection period. Mares were housed in 3.66-m × 5.49-m box stalls between 1600 and 0900 h and were group-housed in a 6-ha pasture between 0900 and 1600 h daily. Box stalls had solid back and side walls, and the front wall was solid 1.22 m from the ground with vertical bars to 2.44 m high. Following parturition, mares were maintained on the same turn-out schedule and same pasture with the remaining pregnant mares during the day. As the weather warmed sufficiently (approximately April 10), mares with foals remained on pasture day and night, and only pregnant mares were housed in stalls at night. At all times there were at least 2 mares housed in the barn during the night-time hours. Mares either had another mare housed in the stall next to them or in a stall directly across the hallway.

Mares were fed approximately 2.27 kg of a 12% protein grain mixture in their stall prior to going to pasture in the morning and again after returning to their stalls in the evening. Grain mixture was prepared on site and consisted of 75% crimped oats, 15% cracked corn, 5% linseed meal, 2.5% soybean meal, 1% wheat bran, 0.75% trace mineralized salt, and 0.75% dicalcium phosphate. Mares were fed twice daily at approximately 0815 h, between 1700 and 1800 h. On data collection days, feeding was withheld until measurements were made and all monitoring equipment was in place. Alfalfa hay was provided ad libitum when mares were on pasture; approximately 5.45 kg of alfalfa hay was offered concurrently with grain feeding during time spent in stalls. Water was provided ad libitum with the use of automatic heated waterers.

Stall lighting was provided by incandescent lights that yielded 4 foot-candles of light at floor level of the stall. The total daily amount of natural daylight to which each mare was exposed varied because of the increasing day length as foaling season progressed. To provide a consistent duration of light exposure, stall lights remained on until 2100 h.

Data Collection. Data was collected on each mare on alternate days until parturition. To decrease the likelihood of missing data, some mares were started prior to d 320 of gestation based on their history of gestation lengths of <340 d and known breeding dates. Data collected included calcium ion concentration in mammary secretions, rectal body temperature, mare behavior, activity level, heart rate, mammary gland parenchymal tissue thickness, and teat diameter. In addition, once a mare entered the data collection period, ultrasound images of the croup were taken twice weekly, and once 1 to 2 d post-foaling.

Calcium Ion Concentration in Mammary secretions. Calcium ion concentration in mammary secretions was measured using Predict-A-Foal (Animal Health Care Products, Williston, VT) test strips per manufacturer's label instructions. Mammary secretions were obtained from mares prior to the evening meal by gently squeezing the base of the nipple. Briefly, secretions were collected in a plastic collection cup and diluted by adding 0.6 mL of secretions into a test tube containing 3 mL of deionized distilled water. The diluted sample was inverted gently to mix, and a test strip was inserted in the vial so that all of the colored squares were submersed in solution. The strip was removed from the test tube, and a reading was taken after 1 min to determine the number of squares that changed color from green to red. After using the Predict-A-Foal test strips on 203 mares of various breeds, Brook (1990) reported the likelihood of a mare foaling within 12 h in relation to the number of squares that changed color, as follows: O squares = 1%, 1 square = 1%, 2 squares = 10%, 3 squares = 40%, 4 squares = 80%, and 5 squares = 90%. The results of the Predict-AFoal test strips were used to determine post hoc whether the mares progressed in a normal manner toward parturition relative to calcium ion concentration in mammary secretions.

Rectal Temperature. Rectal temperature was taken twice daily using a standard commercial digital thermometer. Temperature was taken prior to each mare being turned out to pasture in the morning and again upon return from pasture in the evening. Rectal ternperatures were not used in an attempt to predict the time of parturition, but rather to assist in the posthoc determination of any clinical disease.

Mare Behavior An audio-enabled video camera was positioned in a corner above each stall, and mares were videotaped for a 6-h period from approximately 1800 to 2400 h. The location and angle of the camera were selected to allow for viewing maximal area of the stall. Because of the layout of the stall, there were areas in which the view of the mare was partially or completely obscured from view on the videotape. Because of the size of the mares, the area of the stall in which no part of the mare could be seen was negligible. The only time that a mare could not be seen was when she stood against the wall directly under the camera. During times when part of the mare or the entire mare could not be seen, the audio portion of the tape, along with other visual cues, was used to help interpret mare behavior.

State behaviors occur for a relatively long time and are generally recorded in duration-an example is posture. In contrast, event behaviors occur over a very short time and are recorded as occurring at a particular time and most commonly reported by frequency over time-one example would be drinking. For this experiment, a bout was defined as each individual occurrence of a state behavior. Videotapes were viewed in fast forward mode, and specific behaviors were recorded using a check sheet. The check sheet was blocked by minute, and placement of an "X" in the box further delimited that block into 15-s intervals. Therefore, all state behaviors were timed to the nearest 15s, and all event behaviors had at least a 15-s interbout interval. Real time was displayed on screen as the videotapes were viewed; therefore, tapes could be slowed to observe specific behaviors to improve accuracy of observation and quantification.

There were 2 broad classes of behaviors. The first class was posture of the mare and included the following mutually exclusive behaviors: lateral recumbency, sternal recumbency, standing, and walking. The only time that one of the posture behaviors was not recorded was during a time of significant interruption. Periods of interruption were defined as those times when a mare was prevented from behaving spontaneously; this occurred at least once during every recording period when a person needed to enter the stall and collect data on the mare. The second class recorded was mare activity. Event behaviors were superimposed on standing time and included eating hay, eating grain, drinking, pawing, tail rubbing, defecating, and urinating.

Data labeled as standing time were the total amount of time spent standing and included times when mares were standing and eating hay, drinking water, eating grain, or performing some other behavior. An additional calculated endpoint was time spent standing idle. Standing idle time was calculated by subtracting the total of time spent eating hay, drinking water, eating grain, and any other behavior performed while standing from the total time spent standing. Data labeled as time standing idle represent the time that a mare spent standing without performing any other behavior.

Activity Level. Number of steps as an indication of activity level was quantified using a digital pedometer. A Bodytronics Q One Step (Bodytronics, Fayetteville, GA) counter was attached between the knee and fetlock on the left front leg of each mare. The leg was wrapped with quilted cotton, and the pedometer was secured just above the outside of the fetlock using a cotton polo wrap. The pedometer records abrupt motion, not actual steps; therefore, if a mare lifted the leg that had the pedometer attached and placed it back on the ground abruptly without taking a step the pedometer would record the action. This could lead to false measures of steps if mares did a lot of pawing. On days of data collection, the pedometer was attached prior to the mare being put on pasture. When the mare returned from pasture the pedometer was removed. Number of steps was recorded along with total hours and minutes of turn-out. The pedometer was cleared and replaced on the mare in the same location and manner as described previously. The pedometer was removed after the 6-h collection period while in the stall, and the number of steps and exact duration were recorded.

Maternal Heart Rate. Heart rate of each mare was recorded during the 6-h evening data collection period. Heart rates were obtained using a Polar heart rate monitor (Polar Electro Europe BV, OuIu, Finland) and downloaded into the Polar Precision Performance® software package (Polar vs. 3.0 Software, Polar Electro). In addition, heart rate was recorded for each mare during one 6-h evening session between 5 and 8 mo of gestation to obtain a baseline average for all mares. Baseline measures were taken during the same week in November to minimize potential environmental bias.

All mares were acclimated to the equipment prior to the beginning of the study to minimize effects of the equipment on heart rates. Heart rate monitors were secured on each mare using a standard training surcingle. The positive lead of the monitor was placed just below the withers on the left side of the mare, and the negative lead was placed just behind the left elbow. Water-soluble ultrasound jelly was used under the leads to ensure contact between the skin and electrodes (hair was not clipped). Heart rate recordings were obtained at 1-min intervals. Data were stored in the instrument and then downloaded to a computer for subsequent analysis. To standardize data for each mare, only the first 360 min of heart rate data per night were used during analyses. If the electrodes briefly lost contact with the mare and heart rate was detected as zero, those values were removed from the analyses.

Ultrasonographic Measurements of the Mammary Gland. Mammary gland parenchymal tissue thickness was measured on the right mammary gland of each mare using an Aloka Micron 500 (Aloka USA, Wallingford, CT) ultrasound machine fitted with a 5-MHz linear transducer. Water-soluble ultrasound gel was used as a couplant to ensure acoustical contact. The internal calibrating program of the ultrasound machine was used to measure thickness of the gland. Parenchymal tissue thickness was measured from the innermost layer of skin to the median suspensory ligament. Subcutaneous edema was identified by difference in echogenicity from parenchyma and excluded from the measurement.

Mammary gland parenchymal tissue was measured in 3 locations on the gland. The dorsal measurement was taken as closely to the abdominal wall as was possible. The ventral measurement was taken where the body of the mammary gland met the dorsal aspect of the teat, and the middle measurement was taken midway between the dorsal and ventral sites. Ultrasound images and measurements were saved using either a Sony Videographic Printer (Sony Corporation, Tokyo, Japan) or a Sony Mavicap (Sony Corporation) for subsequent analyses (Figure 1).

Ultrasonographic Teat Measurements. Teat measurements were obtained using an Aloka Micron 500 equipped with a 7.5-MHz linear transducer and polyvinylchloride gel standoff. Water-soluble ultrasound gel was used as an acoustic couplant between the standoff and the teat as well as between the standoff and the transducer.

Teat measurements were obtained at the dorsal aspect of the right teat for all mares. Distance was recorded in 2 directions: caudal to cranial and abaxial to axial. Teat images were printed or stored on disk using the same equipment described for the mammary gland measurements (Figure 2).

Ultrasonographic Measurement of the Croup. Ultrasonographic assessment of the croup was made twice weekly while mares were in the data collection phase of the study and one time postfoaling. All measurements were taken using a PIE Scanner 200 (Classic Ultrasound Equipment, Tequesta, FL) ultrasound machine with an 18-cm, 3.5-MHz linear transducer. The probe was placed just cranial and lateral to the tail head and on an angle toward the point of the hip (Figure 3). The probe was positioned such that the poverty line (line of separation between the biceps femoris and semitendinosus) was in the center of the screen for viewing. Mares were scanned on the right hip; however, one mare was scanned on her left hip because of chronic atrophy of the muscles over her right hip.

Two images were obtained for each mare on each day, and averages were used for data analyses. Internal measurement software was used to determine subcutaneous fat measurements, muscle depth of the semitendinosus, and viewable muscle area of the biceps femoris from the saved ultrasound images (Figure 4). Subcutaneous fat thickness was measured in 3 locations: at the poverty line between the biceps femoris and semitendinosus, 2.54 cm caudal to the poverty line, and 2.54 cm cranial to the poverty line. Total viewable muscle area for the biceps femoris and muscle depth of the semitendinosus, 2.54 cm cranial of the poverty line, were measured.

Ultrasonographic Measurement of Ventral Edema. Ventral edema was measured on mares twice weekly, on the same day that the ultrasonographic croup measurements were taken, when ventral edema was apparent by visual inspection. Edema was measured using the PIE Scanner 200 ultrasound machine with an 18-cm, 3.5-MHz linear transducer and a curved polyvinylchloride gel standoff. The ultrasonographic images were taken at the most ventral location of the abdomen and were saved to a laptop computer for subsequent interpretation and analysis.

Statistical Analysis. Data were analyzed using 2 sample f-tests in Statistix (Analytical Software, Tallahassee, FL) comparing each observation day to all other days. For all variables studied, except heart rate, daily measurements for each mare were analyzed. For heart rate data, daily averages for each mare were analyzed. All data, with the exception of croup ultrasonographic measurements, were analyzed from d -1 (day closest to parturition) through d -15 (either d 14 or 15 prior to parturition). Data for d -17 and -19 prior to parturition were only collected on 2 mares and 1 mare, respectively; therefore, these data were omitted from further analyses.

Because ultrasonographic measurements of ventral edema and the croup region were only recorded twice weekly, there are inconsistencies among mares related to the number of days prior to parturition on which the measurements were obtained. For analysis of data, collection days were merged into the following periods: d -18 through -20, d -15 through -17, d -12 through -14, d -9 through -11, d -6 through -8, d -3 through -5, and d O through -2 with the day closest to parturition serving as the indicator for reporting. All postparturient measurements were grouped together and reported as post. Average values for each of the above-assigned days for each mare were then analyzed using 2 sample f-tests with Statistix.

Results and Discussion

One mare was removed from the study, as her foal showed signs of dysmaturity, including partial failure of passive transfer and incompetent nursing behavior. Gestation lengths for the remaining 9 mares ranged from 328 to 342 d (332.9 ± 1.4 d) and resulted in normal, healthy foals. The 9 foals exhibited normal neonatal behavior, i.e., stood and nursed within an acceptable time as described by Shideler and Voss (1984). Data collection began 20 d prior to the expected parturition date. Data are reported according to day of collection relative to parturition (d 0; range, d -15 to -1). Because most of the mares studied foaled earlier than the anticipated 340 d of gestation, total days in the study for the 9 mares varied. Therefore, data were collected for 9 mares from d 0 to -9, for 7 mares from d 0 to -11, for 6 mares from d 0 to -13, and for 5 mares from d 0 to -15.

None of the mares showed any sign of clinical illness, required any medication, showed evidence of premature lactation by visual inspection, or showed any evidence of vaginal discharge. All mares maintained rectal temperatures consistent with the lack of clinical signs of disease. Morning rectal temperatures of mares ranged from 36.8° to 38.2°C; evening rectal temperatures ranged from 36.7 to 38.2°C. All mares had calcium carbonate concentrations consistent with normal progression toward parturition and onset of lactogenesis, according to the literature provided by the manufacturer. On the last collection day prior to foaling, 8 mares had mammary secretions with calcium carbonate concentrations, indicating a 90% possibility of foaling within 12 h; one mare indicated an 80% possibility of foaling within 12 h.

All data were analyzed using 2 sample t-tests. After looking at all of the data, it was apparent that the basis for significant comparison should be relative to d -1. For this reason, results of the analyses of most data are presented compared with d -1.

Behavior Observations and Pedometer Results. Average total time that mares were observed via video camera per night ranged from 358 to 360 min. Difficulty with video recording resulted in the data from 4 video recordings being removed from data analyses. Additionally, one mare foaled at 70 min from the start of the video on her data collection night. The data from this observation, as well as all other data collected on this mare for this day, were removed from the analyses.

Behavioral changes seen in the group varied among mares, except for a significant increase in frequency of change in behavioral state. Standing, walking, and lying down behaviors for mares while stalled (6 h) and steps per hour while on pasture or stalled (6 h) for 15 d prior to parturition (d O) are given in Table 1. Total time spent standing idle did not change over the study period. The frequency of standing bouts was significantly greater on d -1 compared with d -3, -5, -7, -9, -11, and -13 of the study (P < 0.05). Average duration of standing bouts was significantly shorter on d -1 compared with d -3 (P < 0.05). Total walking time and walking bout frequency were significantly greater on d -1 compared with d -3, -5, -7, -9, -11, and -13 (P < 0.05). Average duration of walking bouts did not change significantly (P > 0.05). Frequency of standing bouts and frequency of walking bouts were linked by design; therefore, similar trends were seen, as illustrated in Figure 5.

Total time eating hay or grain and drinking along with average duration for each day for all mares is shown in Table 2. Total time spent eating hay, eating grain, or drinking water did not change throughout the study. Frequency of hay and grain eating bouts were significantly greater on d -1 compared with d -7 and -9 (P < 0.05); frequency of drinking bouts did not change. Duration of hay eating bouts was significantly less on d -1 compared with d -7 (P < 0.05). Duration of grain eating bouts was less on d -1 compared with d -9 and -11 (P < 0.05). The longest grain eating bout was less on d -1 compared with d -3, -7, -9, and -11 (all P < 0.05). Duration of drinking bouts was less on d -1 compared with d -3, -7, -9, and -13 (all P < 0.05).

Mares were highly variable in time spent lying down. Two of the mares did not lie down on any of the days while being videotaped. One mare only lay down on 1 d (d -11). Of the mares that lay down, only one mare lay down on d -1 (Table 1).

Previous reports on preparturient mare behavior demonstrated no substantive change until just prior to parturition (Crowell-Davis and Houpt, 1986; Shaw et al., 1988). The findings of this study agree with those reports. The mares in this study did not show any change in behavioral traits until parturition was within 1 to 2 d. Of the behaviors recorded, frequency of standing bouts and walking bouts significantly increased on d -1. When a mare was standing, she most commonly needed to walk to change her behavior; therefore, as standing bouts increased so did the bouts of walking. The increased frequency of walking and standing bouts could therefore be summarized as more frequent changes of state as parturition approaches. The only other mutually exclusive behavior in that class was to lie down. Lying down was relatively infrequent and variable among and within mares.

There were no definite trends toward change in specific behaviors during the last 15 d of gestation; however, mares did alter their behavioral state more frequently. Previous studies anecdotally described mares becoming more "nervous" or "restless" as parturition approached (Barty, 1974; Waring, 1983; Crowell-Davis and Houpt, 1986; Shaw et al., 1988; Evans, 1989). Those studies also described the variability in behavioral changes as parturition approached. We suggest that an increased frequency of change in state behavior within the last 2 d prior to parturition is a more precise and objective measure of what is typically described as "nervousness." It is possible that the time during which the mares were observed during this study did not allow for accurate recording of the specific behaviors such as pawing, tail rubbing, etc. Because this group of mares was only observed every other day, the last observation period prior to parturition occurred as much as 30 h prior to parturition. If mares only significantly change their behavior immediately prior to parturition, then nightly observations would be more likely to quantify this change.

Pedometer readings among mares and within mares among days were highly variable while on pasture and while stalled (for all d, P > 0.05). Mares stepped more frequently while on pasture compared with when stalled. The highest average number of steps taken/h was recorded on d -5 both on pasture and while stalled. The number of steps taken/h by mares was highly variable on all days. There was no significant change in steps/h as parturition approached. In the detailed measures of behavior taken in the stall, the frequency of walking bouts and the total time spent walking increased significantly on d -1. Seven of the 9 mares showed an increase in number of steps taken on d -1 compared with d -3 while stalled. Taken collectively, these observations corroborate the increase in frequency of walking bouts and total time spent walking. The increase in total time spent walking could be in part a result of experimental design. Because behaviors were recorded in 15-s intervals, each bout was at least 15 s long. Most times, mares walked for <15 s. For this reason, as walking bout frequency increased so did the total time spent walking. This same phenomenon did not occur in relation to the total time spent standing. Although the frequency of standing bouts increased, the average duration of stand bouts was 6.4 min, and the 15-s recording interval did not have as profound of an effect.

Heart Rate. As gestation length increased, maternal heart rate increased. Average heart rate for all mares on each observation day and between 5 and 8 mo of gestation (baseline) is shown in Figure 6. Average heart rate during late gestation [55 beats/min (bpm) on d -15 to 63 bpm on d -1] was significantly greater on all days of the study (P < 0.05) compared with the average baseline heart rate (45 bpm). This supports previous work with horses and other species (Metcalf and Parer, 1966; Leduc, 1972; Myhrman et al., 1982; Wong et al., 2002).

The increased heart rate would lead to an increase in cardiac output that is necessary to meet the increasing demand for blood flow to the developing fetus in utero and the developing mammary gland. When using all recorded heart rate values, analysis of variance using repeated measures detected significant differences within and among mares and among every day of the study; however, these small increments detected as statistically significant are not clinically significant on a practical level. Analysis of daily averages for each mare using 2 sample t-tests more accurately described changes in heart rate without over- or underrepresenting the variability among mares. Clearly (as shown in Figure 6), there is a trend toward increasing average heart rate in mares over the last 15 d of gestation (55 ± 3 bpm to 63 ± 1 bpm). It is common for an elevated resting heart rate to indicate a metabolic disorder, clinical disease, or some other associated stress in horses. However, a heart rate as high as 63 bpm in a late gestation Quarter Horse mare in the absence of other clinical findings can be considered normal and does not necessarily indicate any abnormality with the mare or the pregnancy.

Mammary Gland. Mammary gland parenchymal tissue thickness (abaxial to axial) increased (P < 0.05) in all 3 locations comparing d -15 with d -1 (Figure 7). The day of the study at which this change in thickness became statistically significant as compared with the parenchymal tissue thickness on d -15 varied for each of the locations. The dorsal section of the gland reached a significant increase (P < 0.05) in thickness at d -5, the middle section at d -9, and the ventral section at d -3. When compared with d -15 parenchymal tissue thickness, the dorsal section increased 34%, the middle section increased 40%, and the ventral section increased 37% by d -1 of the study period.

The dorsal location measured on the mammary gland increased in depth less than the other 2 sections. The mammary gland is a tissue of ectodermal origin that grows from the skin toward the body (Akers, 2002). The mammary gland develops within the mammary fat pad and will not grow into the muscle layer of the abdominal wall. Therefore, as parenchymal tissue thickness increases within the mammary fat pad, the skin surrounding the mammary gland distends laterally and ventrally. Because the location of the hind leg, the dorsal section of the gland is limited in the amount of possible lateral distention. Evans (1989) stated that distention of the udder usually begins 2 to 6 wk prior to parturition. This study did not begin until 15 d prior to parturition; therefore, it is likely that mammary development of these mares had begun prior to their entrance into the data collection period. This early mammary development would be most evident in the dorsal portion of the gland prior to the skin distending ventrally. The overall increase in parenchymal tissue thickness during the 2 wk leading up to parturition was 34% (dorsal measurement), 40% (middle measurement), and 37% (ventral measurement). It is important to note that these measurements represent the parenchymal tissue thickness only and do not include subcutaneous edema that might have been present. Therefore, when evaluating mammary gland distention in the late gestation mare, either visually or manually via palpation, it is important to account for the presence of subcutaneous edema to establish an overall assessment of mammary development. Additionally, when evaluating mammary development, it is important to evaluate the entire gland. Most commonly when manually palpating, and even while visually evaluating, only the ventral area of the gland would be assessed. This study suggests that evaluation of mammary gland development needs to be inclusive of all areas of the gland and not just the most ventral portion.

Teat measurements did not show any significant change throughout the study (Table 3, all P > 0.05). There was no significant change in the size of the teats using either the abaxial-to-axial or the cranialto-caudal measurements. The commonly perceived increase in size of the teat as parturition approaches may be valid. It is possible that an increase in diameter of the teat occurred ventral to the location of the measurement reported. Also, teat length was not measured in this study, and if the teats lengthen with the approach of parturition, this lengthening could be perceived as overall increase in size of the teat. Subsequent studies should take teat measurement in several locations and include a measurement of teat length.

Ultrasonographic Croup Measurements and Ventral Edema. Subcutaneous fat thickness of the croup did not change in any of the 3 locations measured (Table 4, all P > 0.05), and there was no significant change (P > 0.05) in either semitendinosus depth or biceps femoris area. Relaxation of the sacrosciatic ligament may cause a change in the tone, volume, or composition of the muscles, or the fat layer of the croup region, or both; however, muscle shape and fat thickness remained constant throughout this study. The technique used in this study does not suggest a more practical method of assessing the relaxation of the sacrosciatic ligament than the one described by Dufty (1971).

All mares developed ventral edema, but the depth did not change (P > 0.05) throughout the study (Table 4). Although there was no significant change in the amount of ventral edema measured as parturition approached, it is important to recognize that otherwise normal and healthy mares developed edema in the ventral region during the late gestation period. Mares were turned out to pasture daily and were housed in box stalls adequately sized to allow ambulation. Therefore, the presence of ventral edema in late gestation mares is a normal phenomenon that does not indicate any problem with the mare or the pregnancy, nor is it necessarily associated with the mare's activity level.

Implications

This study showed an increase in frequency of change in state behavior during the last 1 to 2 d of gestation. This increased frequency of change could more accurately describe what is commonly described as "nervousness" in late gestation mares. Additionally, maternal heart rate and mammary development during the last 15 d of gestation was quantified, and ventral edema occurs consistently without any associated problems for the mare or fetus. The following changes can be observed in mares during the last 15 d of gestation: 1) mares change their behavioral state more frequently during the last 1 to 2 d of gestation; 2) maternal heart rate increases; 3) mammary gland parenchymal tissue thickness increases 34 to 40%; and 4) normal mares exposed to normal exercise schedules develop ventral edema without any clinical problems. The results of this study provide information that can be helpful in determining whether a mare is progressing normally in the late gestation period.

Acknowledgments

Appreciation is expressed to Ed Jedrzejewski for overseeing management of mares and student labor.

© 2006 American Registry of Professional Animal Scientists Provided by ProQuest LLC. All Rights Reserved.