Seasonal dynamics of parasitism and stress physiology in wild giant pandas

Parasitic infection as a form of chronic stress affects the cortisol metabolite levels, and the cortisol levels were consistent with the seasonal variation in parasite infection. Sex, age, reproductive status and diet composition were directly associated with helminth parasite richness of wild giant pandas. Age and diet composition had significant influences on the faecal cortisol levels.


Introduction
Nutrition, parasitism and stress hormones have reciprocal effects (Beldomenico and Begon, 2015). It is well known that food shortages or poor-quality diets not only result in malnourishment (Trites and Donnelly, 2003) but also stimulate the host to produce more stress hormones (Muller and Wrangham, 2004;Behie and Pavelka, 2013;Chinn et al., 2018). For example, a poor-quality diet corresponded with an increase in cortisol in colobus monkeys (Procolobus rufomitratus) (Chapman et al., 2007). Most of the previous studies also indicated that poor-quality food could reduce the parasite resistance of the host (Coop and Holmes, 1996;Coop and Kyriazakis, 1999;Agostini et al., 2017). Meanwhile, stress hormone levels are largely affected by parasite prevalence or infection intensity (Seguel et al., 2019). For instance, the faecal cortisol levels were directly associated with the intestinal parasite richness in wild male chimpanzees (Pan troglodytes) (Muehlenbein and Watts, 2010).
In addition to the food availability, the parasite infection of wildlife is also related to body condition (Sánchez et al., 2018). Animals in good body conditions might experience an increase in the ability to resist infection by parasites or better tolerate high parasite loads (Roy and Kirchner, 2000). Conversely, poor body condition can result in impaired immunological development, maintenance and function, which could increase susceptibility to infection or lead to high parasite loads and a long infection duration (Beldomenico et al., 2008;Beldomenico and Begon, 2010). Meanwhile, faecal cortisol levels are also affected not only by nutrition and parasite infection but also by a variety of other factors, such as sex, age class, body condition, habitat change and diet composition (Girard-Buttoz et al., 2009;Pokharel et al., 2017;Ranglack et al., 2017;Tecot et al., 2019). For instance, stress hormones in the plain zebra (Equus quagga) in Etosha National Park largely peak in the dry season, when parasite infection intensities are lowest and are most strongly correlated with host mid-gestation rather than with parasite infection intensity (Cizauskas et al., 2015). Thus, it is important to concurrently measure multi-factors, such as the sex, age and reproductive season along with their seasonal food fluctuations, influence on the parasitism and stress physiology to better understand their complex interactions.
As an endangered species, the giant panda (Ailuropoda melanoleuca) may suffer from a large number of potential stressors, such as habitat reduction, climate change and human disturbance (Zhu et al., 2010;Li et al., 2015;Wei et al., 2015a). However, seasonal diet quality change and potential food competition with sympatric species as potential stressors have been ignored. Previous studies indicated that giant pandas faced nutritional stress in late winter and early spring (from December to next March) when they consumed a less optimal diet during this period (Schaller et al., 1985;Li et al., 2017;Wei et al., 2019). The golden takin (Budorcas taxicolor bedfordi) has a broad diet, which includes bamboo leaves and shoots and thus caused competition with the giant panda from June to August in Foping National Nature Reserve (FNNR) . Recent study showed that strong food competition may exist between the giant panda and a sympatric species, the wild boar (Sus scrofa), during the bamboo shoot-eating period (mainly from May to July) (Nie et al., 2019a). In addition to food-related stresses, the giant panda is also likely to face some threats from disease due to extraintestinal migration (visceral larva migrans) by an ascarid nematode (Zhang et al., 2008). The parasite fauna of the wild giant panda consists of five species (Baylisascaris schroederi, Ogmocotyle pygargi, Toxascaris seleactis, Ancylostoma ailuropodae n. sp. and Strongyloides sp. (larvae)), dominated by B. schroederi (Zhang et al., 2011). Parasites can cause chronic stress in giant pandas, and the probability of death in wild pandas caused by parasites increased significantly from 1971 to 2005 (Zhang et al., 2008).
In this study, we explored the influence of the sex, age, reproductive season and seasonal food availability on the parasitism and stress physiology in wild giant pandas. We used faecal cortisol as a measure of physiological stress in pandas and examined the number of parasite eggs in faecal samples to assess the prevalence and intensity of gastrointestinal helminth infections. Considering all factors influence the parasitism and stress physiology comprehensively, our purpose is to assess the potential influences of environmental factors and individual conditions on the health and survival of wild giant pandas and provide useful information for the conservation and management of such an endangered species.

Study area and sample collection
This study was conducted in FNNR (107 • 8 E, 33 • 8 N) in the Qinling Mountains, Shaanxi, China. The reserve was established for the preservation of giant pandas and contains the population with the highest density in the wild (State Forestry Administration, 2015). Two bamboo species, wood bamboo (WB) (Bashania fargesii) and arrow bamboo (AB) (Fargesia qinlingensis), grow here at mean elevations of 1600 and 2400 m, respectively. As bamboo-eating specialists, the giant pandas in our study showed seasonal foraging migration associated with the phenologies of the two bamboo species (Nie et al., 2019b).
From September 2013 to April 2018, we collected fresh faeces (0-3 h) of giant pandas in the Sanguanmiao Protection Area in FNNR. First, fresh faecal samples were collected by tracking four GPS-collared pandas, two males named Xiyue and Chaoyang, and two females named ZhenZhen and Niuniu, and the son of Niuniu named Longbao, two or three times per week from September 2013 to May 2014. Then, a total of 30 transects were established in our study area with the purpose of fresh faeces collection (Fig. 1), and 10 field staff members were divided into five groups to conduct the sampling work on six consecutive days each month from November 2017 to April 2018. To obtain a sufficient number of samples, we collected all fresh faeces that we encountered during the field work. Each sample was divided into three parts for the individual identification of pandas, parasite analysis and cortisol measures. The samples were stored on ice in 50-ml sterile centrifuge tubes in the field and usually frozen at −30 • C within 3 h at the field station. We then transferred the samples with dry ice to the lab at the Institute of Zoology, CAS, until laboratory analysis. Thus, there is no freeze-thaw phenomenon in the transportation of the samples. During all stages of sample collection and processing, we used latex gloves to reduce pollution of the samples. Bamboo samples were also collected in the same area from September 2013 to August 2014. Samples were collected according to the ages and included one-year-and multi-yearold leaves, culms and shoots (Nie et al., 2015). All bamboo samples were oven dried at 55 • C and ground to powder with a common multifunctional laboratory mill (KMS-800Y) in the laboratory and then stored at 4 • C for subsequent analysis.

Sex and age determination and individual identification
DNA was extracted from faeces according to Zhang et al. (2006) using standard controls. Fifteen microsatellite primers, Ame-μ5, μ10, μ11, μ13, μ15, μ22, μ24, μ26, μ27, Aime1, AY79, AY95, AY161213, AY217 and GP7 Shen et al., 2005;Wu et al., 2009;Hu et al., 2017), were used to amplify the DNA extracts from the faecal samples. We amplified each extract three times, if the genotype could not be determined, we performed two additional amplifications. Two species-specific sexing primer pairs, SRY and ZF, were designed for sex determination (Zhan et al., 2007), which was conducted three times for each DNA extract. Template DNAs from a male, a female and a negative control were amplified in each PCR. Individual and sex identification PCR amplification procedures are described in detail in Hu et al. (2010).
The bamboo fragment sizes in the faeces were used to distinguish the age class of giant pandas (Schaller et al., 1985). Here, we determined the age class of giant pandas through the bamboo leaves fragments sizes of the faeces. For details, we used a vernier caliper to measure the length and width of bamboo leaves fragments randomly, and no <40 fragments were measured in each faecal sample. The ages of five giant pandas with GPS collars (including a new collared one DianDain) and three offspring (HuZi, HouBao and LongBao) are relatively clear (Table 1). We measured the length and width of bamboo leaves fragments in faeces of these individuals and made the values as the background information for the age estimations of other individuals. Based on this information, all individuals were categorized into one of four age classes: cubs (0-1.5 years), sub-adults (1.5-5.5 years), adults (5.5-20 years) and old (>20 years) (Hu 1987).

Nutrient analysis
Through our long-time tracking and foraging behaviour observation on giant pandas, we quantified the diet composition according to the types of bamboos they ingested, for instance, the proportion of one-year and multi-year bamboo leaves in the diet that we recorded. We also calculated the proportion of different bamboo components (e.g. shoots, culms and leaves) in the dung of pandas to confirm the observational result. The macronutrient composition of foods eaten by giant pandas were mainly crude protein, crude fiber, crude fat and carbohydrate. Unlike crude protein and crude  fiber, the total crude fat and carbohydrate concentrations were low and did not show significant difference in leaves and shoots of WB and AB (Nie et al., 2019b). For the bamboo diet of pandas, protein is thought to be a very important nutrient with high value, which plays key roles in the pandas' life history (Nie et al., 2015;Nie et al., 2019b), and the crude protein:fiber ratio can be used to show the food quality of giant panda (Wei et al., 2014b). A highquality diet usually has high protein and low fibre (e.g. young shoots and leaves), thus we use the crude protein:fiber ratio as an index to measure the nutrition intake of giant pandas. We collected faecal samples of giant pandas all years and calculated the proportion of different bamboo components to quantify the diet composition in each month. The quality of the diet in each month was determined by multiplying the proportion of different bamboo components by the protein-to-fibre ratio of each item and summing all products. The total nitrogen content was assessed using Kjeldahl procedures with a Kjeltec™ 8400 analyser. The measurement of total nitrogen provides an estimate of crude protein (protein = N * 6.25). Acid detergent fibre was measured using the methods of Van Soest et al. (1991) with a Fibertec™ 2010 Automatic Fiber Determination System.

Parasites infection analysis
Each fresh faecal sample was examined for parasite eggs using the floatation/microscope technique as described by Zhang et al. (2011). Eggs and larvae of parasites were counted and identified based on egg colour, shape, content and size. Prevalence of parasites was determined by dividing the number of positive samples/individuals by the total number of samples/individuals examined. Parasite infection intensity, with uninfected samples being excluded, was defined as the number of conspecific parasites eggs per gram of wet faeces.

Extraction and mensuration of hormones
Glucocorticoids are an indicator of stress (Möstl and Palme, 2002), and the concentration of hormone metabolites (cortisol) can be measured in animal faeces or other body fluids (Wasser et al., 2000;Young et al., 2004;Di Francesco et al., 2017). For details on the hormone analysis, see Nie et al. (2012 a, b), with slight modification in regard to sample pretreatment. We weighed 5-10 g of the fresh faecal samples and then mixed the subsamples and placed them in Petri dishes, and dried the samples for >1 week in a freeze dryer. The freeze-dried faecal samples were crushed with a grinder to ensure that the samples were mixed evenly. We weighed 1 g of the ground faecal samples into a 15-ml centrifuge tube, added 4 ml of an 8:1 mixture of analytically pure methanol and distilled water and mixed and vibrated the tube for 3 min. To remove the lipids, which may impact steroidantibody binding in the radioimmunoassay, we added 2 ml of analytically pure petroleum ether into the centrifugal tube and vortexed it for 1 min. We rotated the tube in a centrifuge at 10 000 rpm for 15 min at −4 • C before transferring 2 ml of the methanol layer (the supernatant) to another tube. The supernatant was centrifuged at a speed of 10 000 rpm for 3 min at −4 • C, and then 1 ml of solution was added to a new 1.5 ml centrifuge tube, which was cryopreserved at −20 • C before analysis.

Statistical analyses
For statistical model building, faecal samples were separated into two seasonal groups, the mating season (March-April) and non-mating season (other months), based on the reproductive season of giant pandas (Nie et al., 2012b;Zhou et al., 2019). Generalised linear mixed-effects models (GLMM) had an advantage in handling non-linear data, which could avoid transformations through the use of a link function, and also non-linear effects between dependent and independent variables. We used GLMM to predict the infection intensity of the parasites and cortisol metabolite concentration in wild giant pandas based on sex (female/ male), age (old, adult, sub-adult/cub), reproductive season (mating season/non-mating season) and diet type (bamboo leaves, shoots/culms). Model selection was based on Akaike information criterion corrected (AICc) for small samples, hence models with lowest AICc values were selected as the most parsimonious. Analysis were carried out in R version 3.5.1, using the statistical package lmerTest. Pearson correlation was used to calculate the correlation coefficient (r) of the nutrition (protein/fiber), infection intensity of the parasites and faecal cortisol metabolites. Statistical differences between these factors were determined using the Wilcoxon rank sum test and the Kruskal-Wallis test. All tests were two-tailed tests, with P < 0.05 indicating significance.
The statistical values are all expressed as the means ± SEM. ArcGIS 10.5 software (Environmental Systems Research Institute Inc., Redlands, USA.) was used to draw a map.

Results
A total of 304 faecal samples were collected during this study, and 206 of them were identified as originating from 29 giant pandas (13 females and 16 males) (

Seasonal dynamics and correlation of nutrition, parasite infection intensity and cortisol levels
The diet nutrients showed obvious seasonal changes ( Fig. 2A). Pandas preferred one-year-old WB leaves (protein/fibre %, 22.59 ± 3.90) than multi-year-old WB leaves during the first several months after they move back to the low elevations, usually from September to November (Table 2). As the consumption of one-year WB leaves gradually increased, multiyear WB leaves increased gradually in their daily diet items (protein/fibre %, 17.87 ± 3.08) from December to April of the next year (Table 2). We found that pandas would also feed on a small part of WB culms (protein/fibre %, 2.59 ± 0.35) and AB culms (protein/fibre %, 2.09 ± 0.41) in March (12.3% WB culms and 4.1% AB culms) and April (12.5% WB culms and 7.5% AB culms). In May, the WB shoots were the main food of giant pandas when they were available, and the percent of protein/fibre in the WB shoots (69.23 ± 9.93) was significantly higher than that in the WB leaves. When the WB shoots matured, the percent of protein/fibre decreased, and pandas migrated to the high-elevation site and switched to AB shoots (protein/fibre %, 47.51 ± 7.51) and one-yearold AB leaves (protein/fibre %, 29.92 ± 2.38) from June to August.
The seasonal variation of cortisol levels in faeces was similar with the dynamic of parasite infection intensity, which is highly consistent with the seasonal variation of nutrition intake in pandas ( Fig. 2A). The intensity of the parasite infection varied in different months and seasons, and the highest value occurred in May (41.47 ± 12.11 eggs/g of wet faeces). From September to November, the parasite infection intensity was very low, among which, it reached the lowest level in October (0.48 ± 0.21 eggs/g of wet faeces) ( Fig. 2A). The correlation coefficient between the parasite infection intensity and the nutrition intake of giant pandas was r = 0.799 (Fig. 2B). The cortisol levels in the faeces ranged from 33.2 to 1026.8 ng/g, varying in different months and seasons, which rose in April and peaked in May (619.34 ± 70.55 ng/g dry faeces), and went back to low level in September ( Fig. 2A). The correlation coefficient between cortisol levels and nutrition intake of giant pandas was r = 0.900 (Fig. 2B). Meanwhile, we also found a significant correlation between parasite infection intensity and cortisol levels (r = 0.671) (Fig. 2B).

Parasite prevalence and infection intensity in giant pandas
The overall prevalence of helminth infections was 56.79% (163/287), with 18.43 ± 2.81 eggs/g of wet faeces. Three different eggs of helminth species were identified in the faecal samples. Prevalence of these helminths was not homogenous and B. schroederi was dominant, which occupied 96.93%

Conservation Physiology • Volume 8 2020
Research article Model selection showed that the interaction between reproductive season and diet was the best model (Table 3). Parasite infection intensity was the highest during the WB shoot-feeding period (27.95 ± 6.93 eggs/g wet faeces) than other time (X 2 = 30.009, P < 0.0001), which in mating season (14.55.95 ± 2.89 eggs/g wet faeces) was higher than nonmating season (7.86 ± 2.02 eggs/g wet faeces) (W = 6708.5, P < 0.0001) (Fig. 3C, 3D). Although age class was not included in the best model, we also found the age class had statistically significant effect on parasites infection intensity in wild giant pandas (X 2 = 18.47, P = 0.00035) (Fig. 3B). However, sex had no effect on the parasite infection intensity in wild giant pandas (W = 4269, P = 0.2991), although males appeared to have a slightly higher average infection intensity than females (Fig. 3A).

Levels of metabolites of cortisol in faeces
The best model showed that age class, reproductive season and diet affected cortisol metabolite levels in wild giant pandas (Table 3). The sex of giant pandas had no significant influence on the faecal cortisol levels (females, 307.27 ± 24.13 ng/g dry faeces; and males, 305.56 ± 20.85 ng/g dry faeces; W = 3175, P = 0.4825) (Fig. 4A). The faecal cortisol levels of cubs (570.63 ± 68.89 ng/g dry faeces) and sub-adults (450.40 ± 76.08 ng/g dry faeces) were higher than adult and old pandas (X 2 = 20.35, P = 0.00014) (Fig. 4B). Although reproductive season was included in the best model, we failed to find any statistically significant effect of this variable on faecal cortisol levels in wild pandas (mating season, 302.13 ± 21.18 ng/g dry faeces; and non-mating season, 313.63 ± 19.53 ng/g dry faeces; W = 4898.5, P = 0.9209) (Fig. 4C). Peak cortisol levels were observed in the bamboo shoot-eating period (AB shoots, 610.8 ± 217.89 ng/g dry faeces; and WB shoots, 556.40 ± 50.81 ng/g dry faeces), which were significantly higher than the leaves and culms period (W = 6708.5, P < 0.0001) (Fig. 4D).

Discussion
Stress responses exhibit seasonality in giant pandas, and the seasonal changes in parasite infection were consistent with the stress responses. Parasite infection intensities are the highest in May, when the cortisol levels were also at peak values. Surprisingly, we also found that the lowest protein-to-fibre ratio in the diets, which were occurred during the nutritional limited period in late winter and early spring, neither elevated the parasitic infections (i.e. high prevalence of B. schroederi) nor increased the cortisol levels. This seems to be inconsistent with the nutritional restriction hypothesis, which suggests that food and water shortages that act as stressors can elevate cortisol levels and parasitic infections in wild animals (Chapman et al., 2006;Behie and Pavelka, 2013;Hernandez et al., 2018).
Parasites and hosts coevolve, and the parasites often more rapidly evolve in comparison to their hosts, as they cycle through many generations during a single generation of the host (Anderson and May, 1982). In contrast to other mammals, giant pandas exclusively consume bamboo that has a particularly low nutritional content, and this may be the main reason for the seasonal change in parasite infection. Through the long period of coevolution, parasites may have adjusted many aspects of their life histories to adapt to the lownutrient diet of giant pandas. Our study obtained a consistent result with those of previous studies, and the number of faecal parasite eggs was the highest during the bamboo shooteating period (Peng et al., 2017). We speculated that essential nutrients were needed for the growth and reproduction of parasites and might increase ovulation when pandas (hosts) used a high-quality diet, bamboo shoot (Coop and Kyriazakis, 2001).  The model with lowest AICc values (bold) were selected as the most parsimonious for our data analysis.
mer. In environmentally transmitted stages, the parasites eggs may be discharged with faeces and washed away by rainfall to infect other individuals through the polluted water (see Fig. 2B) (Gatcombe et al., 2010).
Although the results show that sex had no effect on the infection intensity of the parasites in wild giant pandas, we combined the results of individual identification and found that males were more likely to be infected than females. Pandas expand their home range during the bamboo shoot-eating period and mating season, which could increase encounters with the environmental transmission stages of parasites (Ostfeld et al., 2005). Additionally, the larger home range of males not only provides the chance to come into contacts with more conspecifics of the opposite sex but also increases the risk of acquiring contagious parasites (Jardine et al., 2014). However, females have smaller home ranges (Pan et al., 2001), which might reduce the chances of being exposed to parasites or maintaining high parasite loads. Similar to the infection patterns of raccoons (Jardine et al., 2014), cubs of giant pandas harboured the most parasites (the number of the B. schroederi eggs per unit mass of faeces) and might be easily susceptible to parasites because of their immature immune system. It is surprising that four old individuals had relatively lower parasite infection intensity than adults and cubs. We suggested that this might be caused by the seasonal sampling bias on the old individuals, because most of the faeces of the old ones were collected from October to April of the following year, when the parasite ovulation was significantly lower than the bamboo shoot period.
In this study, we found that the seasonal changes in cortisol levels in faeces were similar with the dynamic of parasite infection intensity. A high parasite load, might as a stressor, led to increased cortisol levels in giant pandas, and the cortisol levels might also influence the parasite load. Previous study indicated that low food availability was associated with increased faecal cortisol and parasite levels in primates (Chapman et al., 2006(Chapman et al., , 2007. Similar to parasite infection, we also suggested that the increased faecal cortisol levels were also related to the diets because low food availability could limit the nutrient intake. However, the cortisol levels of the pandas did not change in late winter and early spring,  when the ratio of protein to fibre in the food was at the lowest value, but significantly increased from April to August, when the high protein-to-fibre ratio was found. Although the giant panda occupies different foraging niches in comparison to the sympatric wild boar and golden takin in the most time of the year, they show remarkable dietary competition with each other when consuming a highly preferred food, bamboo shoots, from May to August Nie et al., 2019a). The wild boar, as the main competitor during the bamboo shoot-eating period, can not only consume bamboo shoots thoroughly but also destroy bamboo roots. The increased interspecific competition between the wild boar and giant panda might also cause increased intraspecific competition for the giant panda. Thus, we suggested that interspecific and intraspecific competition for bamboo shoots might be an important reason for the increase in cortisol levels during this period. Meanwhile, the elevated stress related to competition might also make the pandas more vulnerable to parasites.
The young pandas have higher cortisol level than the old ones indicated that they might face more pressures. We suggested that the exploration on their living environments might contribute this. We did not find the significant effects of sex and reproductive season on the cortisol levels of pandas. Our previous study showed that steroid hormone (e.g. testosterone) only increased in the short estrus period (usually one week), and went down to the baseline level immediately after that. As a large solitary species with no threats from predators, giant pandas often maintain a stable and low cortisol levels most of the time. Usually, only a few individuals would be in the reproductive status in the mating season. Therefore, at the whole mating season scale and the population level, the mean cortisol level did not show significant difference with that of other time.

Conclusions
The seasonal dynamic of the cortisol levels and parasite load in faeces co-vary with the seasonal nutrient intake levels of the giant pandas, which concurrently arrived the peaks in the season when nutrition intake was high. Age class and seasonal food availability were the main factors affecting the parasite load and stress physiology in giant pandas. To the best of our knowledge, this is the first study to evaluate the influence of the sex, age, reproductive season and seasonal food availability on the parasitism and stress physiology in wild giant pandas. Our findings help to understand how environmental factors might influence the physiology, behaviour and health of wild pandas. More broadly, our study has implications for the conservation and management of this endangered species, not only for the wild populations but also the captive ones. For example, the effects of the seasonal food quality and availability limitation should be considered into the population monitoring of the wild pandas, especially for cubs and sub-adults. For the captive populations, high-nutrition diets should be controlled, under the premise of ensuring proper nutrition, to avoid high parasite load.