Changes to Health Parameters of White-Tailed Deer during a Drought in the US Virgin Islands

Abstract

Resident white-tailed deer (Odocoileus virginianus) on St. John, US Virgin Islands offer a unique case study for understanding a population under pressure from climate change. During a 2015–2016 regional drought, deer health parameters including body condition, coat condition, tick prevalence, musculature, and stress hormones were tracked over three field seasons representing the onset, peak, and recovery phases of the drought. All health indicators showed significant change over the course of the drought, and post-hoc tests suggest some indicators (body condition, musculature, and ticks) were more sensitive during drought onset. High levels of cortisol during the peak period indicated substantial stress to the population, which normalized during recovery. The strongest correlations were between overall health/body condition and musculature and overall health/body condition and coat condition. The weakest correlations were between ticks and the remaining three variables. These results support the hypothesis that various measures of deer health are related. The frequency and intensity of droughts and environmental stressors are predicted to increase in the future due to climate change, which will further challenge this island deer population.

1. Introduction

Historically, the Caribbean region has been characterized by relatively predictable seasonal rainfall patterns and moderate fluctuations in annual temperatures. However, this stability is changing rapidly with climate change, and is projected to be highly variable as levels of greenhouse gases in the atmosphere continue to increase [1, 2]. The most recent climate change data predicts that climate change will bring extreme heat waves at greater frequency, droughts that will occur 2–3 times more often, stronger hurricanes with the trend toward hotter temperatures, and wet and dry extremes in local areas [2, 3]. The Caribbean is considered particularly vulnerable to the effects of climate change as weather events that occur in the region increase in both frequency and duration (www.drought.gov). As a result, this area will experience increasing weather variability and extremes, which will manifest as hotter temperatures, shifting rainfall patterns, more frequent water shortages following decreased annual rainfall, and longer dry seasons. The Virgin Islands have already experienced a series of droughts in recent years, the most severe of which occurred in 2015–2016.

Changing landscape conditions, including more frequent drought, can exert stressors on wildlife populations that have the potential to be detrimental to the health and fitness of individuals [4, 5] and populations over time [6]. For wildlife populations, drought presents multiple simultaneous environmental challenges such as high temperatures, low food availability, and low water availability [1]. Water shortages associated with drought can bring about reduced plant primary productivity and seed survivability, which causes food reduction and changes in water quantity and quality. As a result, suboptimal consumption of protein, vitamins, minerals, and other essential nutrients can lead to malnutrition. Subsequently, malnutrition, prolonged dehydration from heat stress, and parasitism can lead to the depletion of fat reserves, anemia, and poor body condition, and has the potential to result in immunocompromised individuals [7, 8]. These individuals are even more at risk for malnutrition, parasitism, or starvation [9]. Wildlife adapt to drought using physiological and behavioral adaptations, but the stress from prolonged drought can eventually overwhelm their resiliency [10].

Several studies have evaluated deer survival during drought events and demonstrated a clear connection between a weather event and population effects. During drought years, deer often overgrazed available flora and harmed plant species due to intense herbivory [11]. In addition, deer consumed fewer plants, and plants of lower forage quality, and often did not meet their nutritional requirements, which had the potential to limit lactation [12]. Reduced lactation can decrease deer numbers either because of neonate starvation or result in smaller and weaker fawns [13]. Bucks responded to reduced food quantity and quality by displaying smaller body size and antler growth, particularly in young males that were still growing [12]. Therefore, a single year of drought might have lifetime consequences for a cohort of both female and male deer [14]. Overall, the effects of climate change are multidimensional, and exacerbated by the stress of prolonged drought, and can be largely deleterious to the health of wildlife.

The goal of this study was to assess the resilience of a population of white-tailed deer on St. John, US Virgin Islands as they responded to a severe drought in 2015–2016. Health observations were collected through three successive field seasons representing drought onset, peak, and recovery. This work is the first to document the physiological changes observed during drought for a population of isolated residential island deer in the Caribbean region, and as a result, this work was largely exploratory. We hypothesized the lack of food and water resources associated with prolonged drought would have a negative effect on deer on St. John and we predicted that there would be an adverse change to their physiological condition as a result. However, there is little prior research alluding to chronic, pre-existing stress factors or the time points at which the changes would be demonstrated in the deer, including if and when they would resolve. We hypothesized that there would be strong associations between drought and musculature, coat condition, and body condition, due to limited island resources, and that these parameters might decrease due to direct and indirect effects of the drought. Similarly, we hypothesized that there might be an association between the values for cortisol, T-3 levels, and tick presence, and they might increase with drought-associated stress. However, we did not know the strength of the relationship between these different parameters, or which parameters would show differences within the three evaluation periods of this study. Our goal was to provide quantitative evidence for these changes for a protected island population at three distinct time periods of a drought. The monitoring of stress and nutrition in wildlife populations can provide researchers with valuable insight into the baseline stability and physiological impacts of environmental change on wildlife populations [6]. The deer of St. John represent a unique and intriguing case study for understanding a population facing highly altered future conditions due to climate change.

2. Methods

2.1 Field data collection

The island of St. John is located in the Caribbean Sea between 18°18′ and 18°22′N latitude and 64°40′ and 64°48′W longitude (Figure 1). It is 11 km long and 5 km across at its widest point [15]. St. John is part of the US Virgin Islands which includes St. John, St. Thomas, St. Croix, and Water Island. Virgin Islands National Park (VINP) lies on the island of St. John and comprises 60% of the landmass of the island and protects one of the largest tracts of secondary dry forest in the eastern Caribbean [16]. Dryland plant communities on St. John include forests, shrublands, coastal hedges, and a rare cactus community [17, 18]. The forests and forest structure are largely shaped by hurricanes and drought [16]. The climate is relatively dry, with an average temperature of 27°C with 75% relative humidity. Water is in limited supply because of high temperatures, high evaporation rates, and run-off from steep slopes [15, 19]. There are no permanent streams or rivers on the island and only a few intermittent streams [20]. Precipitation is highest from May through November with a range of 890–1400 mm/yr [15]. The driest months are February and March [20]. The region is prone to cyclical patterns of drought and excessive moisture and structural damage from seasonal hurricanes.

Figure 1.

A map of the study location on the island of St. John, U.S. Virgin Islands in the Caribbean.

This information builds upon a series of articles describing the natural history parameters of the deer of St. John. Despite their isolation, St. John deer demonstrated low inbreeding and average heterozygosity [21], were positive for antibodies for bluetongue and epizootic hemorrhagic disease but no clinical signs of the disease were observed during field observations [22]. A third study identified the ticks found on the St. John deer as a species of cattle tick and tropical horse tick [23].

Data were collected over three field seasons on St. John, USVI.; first during the onset of a drought (July 2015, henceforth “onset”); second at the peak of the drought (March 2016, henceforth “peak”); and third during recovery from the drought (July 2016; henceforth “recovery”) (Figure 2). The drought occurred regionally throughout the Caribbean and was prevalent in 2015 on St. John [23].

Figure 2.

The 12-month rolling average precipitation (calculated using the previous 12 months) on St. John Island for the years 2015–2016, with annotations for the three data collection time points and field notes regarding observed conditions in a study of white-tailed deer (Odocoileus virginianus) health indicators during the onset, peak, and recovery phases of a 2015–2016 drought. Initially (December–April 2015) the rolling average is on par with average precipitation for the previous 10 years. Precipitation decreases beginning in April of 2015. Data collected in July 2015 (“Onset”) occur when the rolling average is low but reserve supplies are still available. By October 2015 the rolling average precipitation drops below the lowest annual precipitation in the previous 10 years. Data collected in March 2016 (“Peak” of the drought) occurred when the monthly average precipitation for the past 12 months has remained lower than usual for several months. Precipitation returned shortly after, bringing the rolling average up to more favorable conditions (“Recovery” phase of the drought).

Deer musculature, coat condition, body condition, and tick levels were recorded using two data collection methods: remote assessment and direct assessment of tranquilized deer. Remote assessments of nutritional conditions offer a noninvasive alternative when capture of deer is not possible, and can be used to evaluate changes at a population level [24]. Remote assessments were conducted at all three time points (onset, peak, and recovery of the drought), including collecting data on deer seen either on transects or opportunistically by trained technicians working in pairs and using binoculars at a distance of less than 25 m. Deer on the island are highly acclimatized to humans and did not move when observed by technicians, particularly near popular tourist trails or beaches. This allowed the research team to observe deer directly for accurate data collection. Technicians for each field season were trained to determine body condition, musculature, coat condition and tick levels by completing both pre-visit trainings and in-field calibration sessions to reduce the amount of inter-observer variability and to retain a high fidelity of rating. A body scoring system that uses anatomical landmarks provided an immediate evaluation that is non-invasive [25]. Data were collected in real time using binoculars, iPads, and a custom app for data storage that was specifically designed for this project. For examples of the different parameters measured, see Figure 3 and Table 1. Pregnant deer and fawns were purposefully excluded.

Figure 3.

(A) A white-tailed deer (Odocoileus virginianus) showing excellent coat condition, musculature, and body condition, and an absence of ticks (B) showing good coat condition, musculature, and body condition, and no ticks (C) showing fair coat condition, musculature, and body condition, with some ticks and (D) showing a poor body condition, with a high degree of tick infestation in the ears. These conditions were observed in a study of deer health indicators during the onset, peak, and recovery phases of a 2015–2016 drought on St. John Island.

	Body composition	Coat condition	Musculature
Excellent	Hip, rib and spine bones not visible Rump rounded and possible extra fat seen	Shiny and smooth hair with luster No blemishes or scars	Adequate/robust development Toned and well-defined muscles Abundant fat reserves
Good	Hip, rib and spine bones starting to be visible Rump not as rounded	Mostly free of scars Less shiny and smooth and losing luster Some blemishes on hair	Healthy muscle present Not well defined Adequate fat reserves
Fair	Hip, rib and spine bones visible Rump starting to flatten/ concave.	Some scars No luster or shine Small patches of missing hair	Atrophy of muscle seen No definition Depleted fat
Poor	Hip, rib and spine bones prominent Rump concave and obvious absence of muscle	Dullness to coat Large clumps of missing or patchy hair Large amounts of scars	Severe atrophy Underdeveloped No fat reserves

Table 1.

White-tailed deer (Odocoileus virginianus) were scored as excellent, good, fair, and poor for body condition, coat condition, and musculature on St. John Island during the onset, peak, and recovery phases of a 2015–2016 regional drought. Ratings loosely follow [26].

The second method of data collection was conducted via direct assessment when deer were tranquilized. This method was only used during the recovery phase of the drought [22]. Deer data were collected throughout St. John [21] and were geographically representative of the deer population in all areas of the island. Pregnant or nursing does and deer that were one year old were not tranquilized. In total, twenty-three adult deer were tranquilized using butorphanol, azaperone, and medetomidine (BAM, Wildlife Pharmaceuticals, Windsor, Colorado, USA). Relative body size was used to determine the administered dose according to the manufacturer’s guidelines. Most deer were small to medium in size (41.1 ± 13.2 kg), receiving 1.0–1.5 mL intra-muscularly in the hind quarter by pneumatic dart gun (Pneu-Dart, Williamsport, Pennsylvania, USA). Vitals monitored included heart rate, respiratory rate, mucous membrane color, body temperature, time to recumbency, and recovery. Body temperatures were stabilized with a wet cooling blanket (Equi Cool-Down, Jacksonville, Florida, USA). After examination, the anesthesia was reversed with 2–3 mL of atipamezole (25 mg/mL) and 0.5 mL of naltrexone (50 mg/mL, Wildlife Pharmaceuticals). This work was conducted under Scientific Research and Collection Permit VIIS-2016-SCI-0026 for the Virgin Islands National Park to S. Nelson and IACUC (1602.01-15Mar2016) from the University of Colorado at Boulder and the National Park Service to S. Nelson.

Stress hormones were analyzed from fresh deer fecal material. Feces were collected opportunistically throughout St. John when we could directly attribute the fecal deposit to an individual deer, which allowed us to describe the age, sex, and health condition of the deer [27, 28]. Fecal samples were collected using sterile gloves and placed into labeled plastic bags, stored with ice packs, and given a unique identifying number [29]. The samples were frozen and sent to the Wasser lab at the University of Washington for stress hormone analysis. Samples were analyzed for both cortisol (ng/g) and T3 levels (ng/g). All samples were freeze-dried and homogenized, and then 0.1 g was extracted using 15 ml of 70% ethanol [30].

2.2 Statistical analysis

Deer musculature, coat condition, body condition, and tick levels were determined to be excellent, good, fair, or poor (Figure 3 and Table 1). The initial four categories of excellent, good, fair, and poor were collapsed into binary outcomes: Good/Excellent vs. Fair/Poor to simplify interpretation and bolster sample sizes. Borderline cases were omitted (e.g., deer rated as “Good/Fair”). After collapsing, all four variables satisfied conventional conditions for sample size for Chi-Square tests of independence to assess the relationship between deer health and time point (onset, peak, and recovery phase, respectively). In post-hoc analyses, for each of these measures three pairwise two-proportion two-tailed z-tests were conducted to discern differences between time points: drought onset vs. recovery, peak vs. recovery, and onset vs. peak. No corrections for multiple testing were used so as to flag potential effects for future study.

To evaluate relationships between the four health indicators collected via visual assessment (musculature, body condition, coat condition, and ticks; both remote and direct), the phi coefficient was calculated for each of the pairwise relationships. Marginal distributions indicate that it is impossible for phi to reach a magnitude of 1; to avoid inflating the perceived relationships, no adjustment was made. The resultant correlations are conservative.

Two deer that had been tranquilized were also observed remotely in July 2016 following the drought. The remote observations were removed from the primary analyses, but used to check consistency between data collected remotely and data collected by tranquilizing. Observations of body condition and musculature were rated the same, however, ticks were observed more readily within the tranquilized deer when compared when they were viewed remotely. Thus, to promote consistency when comparing across time periods, observations of tranquilized deer were not included in the analysis for ticks. Data from tranquilized deer were included for the other variables because inconsistencies were not meaningful after the collapse into binary groups.

To quantify the stress from drought using fecal cortisol levels, two F-tests were conducted (one for average cortisol and one for average T3 levels). There were no extreme outliers and conditions for the test were otherwise adequately met. In post-hoc analyses three pairwise two-sample t-tests were conducted to discern differences between time points. All analyses were conducted using R statistical software (version 3.6.0).

3. Results

In total, 515 deer were observed remotely and 23 deer were tranquilized over three study seasons. In July 2015, 265 deer were observed, in March 2016, 189 deer were observed, and in July 2016, 61 deer were observed remotely, and 23 were tranquilized. At each time point, the samples were slightly more heavily represented by female deer; approximately 60–70% were female. The remote samples were heavily represented by mature deer, and ranged from approximately 75–90% mature deer at each time period.

3.1 Deer health indicators associated with drought conditions

The data confirm the hypothesis that there is an association between deer health and phase of drought conditions. For all six measures, a relationship between the stage of drought was statistically discernible (significant), which suggests that the onset, peak, and recovery phases were not all equally difficult for the deer. More specifically, a relationship was found between time points relative to drought and average cortisol levels (F_{(2, 63)} ≈ 18, p < .00001) as well as average T3 levels (F_{(2, 63)} ≈ 5.8, p ≈ 0.005). Similarly, statistically discernible (significant) associations were found between time point relative to drought and proportion of deer with fair/poor rankings on each of the four measures observed visually (tick level, coat condition, body health, and musculature, with respective p-values of .013, .01, .005, and .001, respectively) (Figures 4, 5 and Table 2). In summary, all six measures revealed an association between drought conditions and health of the deer, which does not support a hypothesis that the onset, peak, and recovery phases are all equally difficult on the deer.

Figure 4.

Percentages of deer with coat condition issue (i.e., categorized as poor or fair condition) at each of the three time points (drought onset, peak, and recovery). This was the only indicator that presented higher prevalence of issues during the peak of the drought, while having lower prevalence both at the drought onset and recovery phases. Chi-square test indicates an association between time and prevalence of coat condition issues (𝜒² (2 degrees of freedom) = 8.9, p = .01).

Figure 5.

Indicators that presented higher prevalence of issues both at the onset and peak of the drought, while having lower prevalence after drought in a study of white-tailed deer (Odocoileus virginianus) during a 2015–2016 drought on St. John Island. (a) Percentages of deer with body decomposition (i.e., categorized as poor or fair condition) at each of the three time points (onset, peak, and recovery of the drought). Chi-square test indicates an association between time and prevalence of body decomposition (𝜒² (2) = 15.1, p = .005). (b) Percentages of deer with muscle atrophy observed at each of three time points (onset, peak, and recovery of the drought). Chi-square test indicates an association between time and prevalence of muscle atrophy (𝜒² (2) = 17.9, p = .0001). (c) Percentages of deer with ticks observed at each of three time points (onset, peak, and recovery of the drought). Chi-square test indicates an association between time and prevalence of ticks (𝜒² (2) = 8.8, p = .013).

	P-value for Two-tailed Pairwise Test
Health indicator	Onset vs. peak	Peak vs. recovery	Onset vs. recovery
1. Coat condition	.007	.08	1
2. T-3	.2	.02	.02
3. Body condition/overall health	.9	.0005	.0002
4. Musculature	.3	.001	.00004
5. Ticks	.6	.006	.01
6. Cortisol	.02	.0006	.0005

Table 2.

Post-hoc results (two-tailed p-values) of white-tailed deer (Odocoileus virginianus) health indicator comparisons across time points relative to drought (onset, peak, recovery) on St. John Island during a 2015–2016 regional drought. To flag potential effects, no correction for multiple testing was used.

Post-hoc analyses were conducted to explore various hypotheses about patterns of how each of the six health measures appeared to respond to the different stages of drought (Table 2). There are a few similarities; perhaps not surprisingly, all six measures had high levels of poor health at the peak of the drought. Another similarity is that all six measures revealed relatively improved health at the third time point (during the recovery phase after the drought). This suggests the deer population showed resilience in the recovery phase.

On the other hand, there were differences in the extent to which the health measures appeared to be sensitive to the onset stage of the drought: some measures indicated poorer health even in the onset, whereas other measures were less sensitive. Three patterns of behavior were observed and named according to relative health at each of the onset, peak, and recovery periods of the drought. The patterns are named: “Low-High-Low,” “High-High-Low,” and “Medium-High-Low” (where “High” indicates high levels of poorer health).

Coat condition displayed the Low-High-Low pattern. This measure had discernibly higher (statistically significant) prevalence of poorer health only at the peak of the drought (Figure 4 and Table 2). In the drought onset and recovery, percentages were significantly lower than the peak. No statistically discernible (significant) differences were observed when comparing onset vs. recovery phases of the drought (p > .5). This supports a hypothesis that coat condition is not particularly sensitive to the onset stage of the drought.

In contrast, the T-3 levels, body condition, musculature, and tick indicators displayed the “High-High-Low” pattern. These three measures showed higher prevalence of poorer health both during the drought onset and peak (Figures 5, 6 and Table 2). To be clear: for these indicators there were no statistically discernible (significant) differences in deer health when comparing the first two time points: onset versus peak of drought (p = .2, .3, .6, and .9, respectively, for T-3, musculature, ticks, and body condition measures). The prevalence of poor health on these indicators was only discernibly different (significant) at the third time point, after the drought. This supports the hypothesis that T-3 levels, body condition, musculature, and tick prevalence are sensitive to drought even in the onset stage.

Figure 6.

Levels of fecal T3 (a) and cortisol (b) in white-tailed deer (Odocoileus virginianus) populations in a study of deer health indicators during the onset, peak, and recovery phases of a 2015–2016 drought on St. John Island. Sample sizes collected indicated below each period.

Cortisol was the only measure to display a “Medium-High-Low” pattern. This measure had statistically discernible (significant) differences when comparing each of the three time points with each other (Figure 6 and Table 2), with a mild sensitivity during the onset stage, increased severity at the peak of the drought, and of course the ability to recover in the third stage.

3.2 Relationships between health measures

Positive relationships were also present between each of the visually observed measures of deer health (Table 3). For example, a deer with fair/poor musculature was more likely to also have fair/poor coat condition (φ = .34). Conversely, a deer with good/excellent musculature was more likely to also have excellent/good coat condition. Overall health/body condition was strongly correlated to musculature (φ = .64) and coat condition (φ = .48). Coat condition and tick infestation were moderately correlated (φ = .31), as were coat condition and musculature (φ = .34). Smaller correlations were observed between tick infestation and musculature (φ = .2), and between tick infestation and health/body condition (φ = .26). The fact that all the pairwise correlation coefficients between health/body condition, musculature, coat condition, and ticks are positive supports the hypothesis that that these four measures of deer health are related, and/or largely dependent on similar underlying reasons. For this work, that reason was thought to be the multifaceted effects of stress associated with prolonged drought on the island.

	1	2	3	4
1. Coat condition	1.00
2. Body condition/overall health	0.48	1.00
3. Musculature	0.34	0.64	1.00
4. Ticks	0.31	0.26	0.2	1.00

Table 3.

Correlation (as measured by Phi coefficient) indicating strength of association between health indicators (coat condition, body condition, musculature, and ticks) for white-tailed deer (Odocoileus virginianus) scored as Excellent/Good vs. Fair/Poor on St. John Island during the onset, peak, and recovery phases of a 2015–2016 regional drought. To avoid inflation, no adjustment for binary outcomes was used (e.g., adjusted phi, etc.). Thus, the correlations given here are conservative.

4. Discussion

Positive pairwise correlations were present between all four measures of individual deer: tick prevalence, musculature, coat condition, and body condition. That is, deer that had one health issue were more likely to also have another. This could simply reflect that some measures may be related (e.g., coat condition and overall body condition have some common sources of overlap). However, other measures (e.g., tick prevalence and musculature) do not naturally overlap. Therefore, our results suggest health factors are indeed associated perhaps because the underlying stress of deer represents a common etiology. The population of deer on St. John are largely disease-free [22], and demonstrate relatively high allelic diversity for an isolated island population [21] which may act as a mitigating factor for stress. However, this study demonstrates a strong response during peak drought for all of the variables.

Our observations suggest that deer on St. John were highly nutritionally deficient, particularly within the drought peak in July 2015, when deer faced severely limited feeding within a denuded landscape. During the peak drought period, deer appeared to be eating atypical forage, often called famine foods, in the absence of typical and available forage. Famine foods are eaten by animals when all known food sources have been exhausted. They often contain low levels of protein, are low in calories provided [31], and can be very energetically expensive to metabolize. Observations during the drought peak included a fawn eating dead leaves and adult deer spending considerable time and energy digging up tubers and roots. Dependable sources of water on St. John were also severely reduced in both quantity and quality, and deer were often seen near anthropogenic water sources such as pipes, foot baths, and shower areas. Their body condition during the height of the drought indicated both severe water and nutritional stress [23]. Below we analyze multiple indicators of deer health to show individual responses to drought onset, peak, and recovery.

4.1 Indicators of deer health

4.1.1 Body condition and coat condition

Results for body condition followed the High-High-Low pattern, demonstrating a higher prevalence of poor health during both the drought onset and peak. This indicates that the deer were in a state of stress at the drought’s onset and were without the reserves needed from the lack of food and water availability during the drought peak. Body condition scoring represents an amalgam of multiple health parameters representing external markers of internal health. Many of the factors that were evaluated with body condition, including coat condition and musculature, are complementary data for assessing the health and nutritional state of an animal. These factors are influenced by both current and past food availability, and can therefore indicate health in both the present and recent past. Prolonged nutritional challenges such as drought can cause individuals to exhaust stored fat reserves which can result in a deterioration in body condition [6]. In contrast, coat condition demonstrated a Low-High-Low pattern during the drought. This suggests that coat condition is not as sensitive to the onset stage of the drought as body condition, and coat condition is able to return to original condition during drought onset within the drought recovery period. Coat condition represents a sub-category of the body composition assessment, and one that can further indicate nutritional health. Essential vitamins, as well as fats and oils in the diet, are needed to provide coat and hair luster [32, 33].

4.1.2 Muscle and muscle atrophy

Results for musculature followed the High-High-Low pattern, demonstrating a greater prevalence of poor health both during the drought onset and peak. The response of musculature is similar to body condition in that the pattern indicates that individual deer were stressed at drought onset, indicating a deficiency in both food availability and in high quality foods that contain adequate protein reserves to support muscle development and growth. Muscle is an energetically expensive tissue to build and maintain [34]. Muscle presence and appearance can be an indicator of overall health and an excellent external indicator of both health and protein reserves. However, with chronic and insufficient protein consumption, muscle catabolism can occur, resulting in muscle atrophy [33]. In addition, because more than 80% of protein in the animal body is dedicated to maintaining proper functioning of the immune system, disease state can indirectly result in muscle loss and atrophy. As a result, there can be a strong correlation between muscle appearance and the disease state of a deer. If muscle atrophy is present, this indicates that the animal body is using muscle as an energy source in the absence of any remaining fat reserves. More energy is released when a unit of fat is metabolized as compared to a unit of protein. Therefore, when a deer has metabolized most of its fat and is using muscle for energy, it will lose weight very quickly as subcutaneous fat is already depleted [35] and will appear gaunt and emaciated in appearance as a result [32]. In contrast, excellent musculature indicates that the deer is eating sufficient protein to both maintain muscle and to support the high protein demands of the immune system [36]. These changes in subcutaneous fat can be evident via visual assessment [24]. The results of this study add to evidence that both musculature and fat reservoirs are sensitive to drought even in the early onset stages within island populations.

4.1.3 Ticks

The dynamics of infection often depend on the host’s vulnerability, as poor body condition is likely to predispose individuals to infectious and parasitic diseases [6, 7]. Tick prevalence in this study followed a High-High-Low pattern, indicating that ticks were already both prevalent during both the onset and peak of the drought. The additive effect of this continued high level of parasitism has negative consequences for individual deer health for blood loss and the potential for disease development. A common parasite found on the deer of St. John included ticks. The two tick species found on St. John include the southern cattle tick (Rhinocephalus (Boophilus) microplus (Canestrini) and the tropical horse tick, Dermacentor (Anocenter) nitens Neumann [23]. High tick densities could result in associated health problems, including pruritis, alopecia, anemia, and low weight gain. In addition, ticks can deplete the iron resources of the deer through each blood meal taken, which can result in the development of iron-deficiency anemia and deprive tissues of necessary oxygen [37, 38]. There can be a considerable energy requirement to replace daily blood loss which could result in further accelerated nutritional decline and weight loss. Additionally, the relative energetic cost associated with compensating for blood loss is higher for animals in poor condition who can experience more pronounced energy and protein deficits compared to healthier animals [37].

4.1.4 Stress hormones

Stress hormones help the body handle adverse conditions, and levels can vary by situation and species [39]. Cortisol is released by the body when there is either an acute or chronic stressor. Cortisol can suppress the conversion of T4 thyroid hormone into T3, and lower circulating T3 levels can be an indicator that cortisol levels are high. Together, both T3 and cortisol levels create a complementary data set that demonstrates stress levels in individual animals [28].

Cortisol was the only measure from this study to display a “Medium-High-Low” pattern. This measure had statistically discernible (significant) differences when comparing each of the three time points with each other. Our data for all three phases of the drought show that the deer were already stressed in July 2015 at the start of the drought, and to a degree that shows widespread stress across the population. This was further exacerbated at the height of the drought, in March 2016, when the deer were in acute stress. The stress hormone data ranged from 24.3 to 110.1 ng/g for T3 and 26.5 to 258.5 ng/g for cortisol.

The cortisol values for St. John deer at the onset and recovery of the drought were within published values for animals within ongoing stressful situations, but cortisol values for deer during the peak of the drought were more consistent with wildlife undergoing acute trauma. For example, values ranged from 23.9 to 114.9 ug/g for musk deer in crowded conditions [40], and deer undergoing high parasite burdens had cortisol values that reached 93–144 ng/g [39] and could result in diminished body condition for deer over time [25]. These published values for deer with ongoing stress are consistent with the values found in this study for St. John deer during drought onset and recovery. However, the highest values found in our study at the peak of the drought (258.5 ng/g) were instead consistent with values that more reflected acute and traumatic events for animals. For example, koalas who had been in vehicle collisions (202 ng/g), were burn victims (200 ng/g), or were in an area during land clearance (669 ng/g) showed cortisol values similar to the deer of St. John during the peak of the drought [41]. High fecal cortisol levels can often be used as a predictor of mortality. Ring-tailed lemurs that died following traumatic events showed average fecal values of 51.1 ng/g [42]. The deer of St. John both endured sustained cortisol levels above that value for the duration of the drought (2015–2016), and also at values five times that amount at peak drought. This highlights the noteworthy resiliency of this population to continue to live with this ongoing stressor present on island for over 2 years.

Stress hormones are also important regulators of energy balance. When food availability diminishes to the point of starvation, cortisol remains chronically elevated, body condition declines, and fat stores are used for energy [43, 44]. Prolonged elevation of glucocorticoids can result in the suppression of reproduction, growth, immune function, and responses to pathogens and parasites [45].

5. Conclusions

This study was able to describe the resilience of an isolated population of white-tailed deer on St. John, US Virgin Islands during a severe drought in 2015–2016. Health observations collected through three successive field seasons representing drought onset, peak, and recovery of the drought enabled novel work documenting physiological changes of deer under the stress of climate change For all six health measures in this study (musculature, coat condition, body condition, the presence of ticks, cortisol, and T-3 levels), there is an association between deer health and time relative to drought (onset, peak, or recovery phase). All measures indicate a rapid return to health following the drought peak.

This study provides a baseline foundation for future research needed to inform the extent to which isolated populations of herbivores cope with increasing climate variability. Further work is needed to explore health differences between males and females during various stages of drought. Future studies would also benefit from using tagging methods for a study design that enables direct observations to individual deer and clarifies how the population maintains resilience in extreme weather. In addition, future work may include a closer analysis of stress hormone levels related to reproduction and survival of individuals. Periodic re-evaluation of the St. John deer population and health will add additional data to population densities that can be supported by the island ecosystem. This work provides a critical baseline to document physiological changes to an isolated Caribbean deer population, and the results of this work can be extended regionally throughout the Caribbean and to analogous species.

Acknowledgments

Thank you to the hard-working field crews who made this work happen, including D. Masters, C. Walls, B. Daughton, D. Duncan, J. DePompolo, B. Wilkins, P. Ewing, M. Johnson, and M. Malone. We thank T. Kelley and A.S. McKinley from the Virgin Islands National Park for their help. C. Anderson and K. Hendricksen contributed to the statistical analysis. Thank you to the Sam Wasser lab for stress hormone analysis. This project was funded by Friends of the Virgin Islands National Park to S. Nelson and the IPHY department at the University of Colorado at Boulder to S. Nelson. This project was completed under permit VIIS-2015-SCI-0027 for 2015 for the Virgin Islands National Park to S. Nelson. The findings and conclusions in this article are those of the author and do not necessarily represent the views of the US Fish and Wildlife Service.