Skip to main content
Dryad

Influence of multiple predators decreases body condition and fecundity of European hares

Cite this dataset

Weterings, Martijn et al. (2022). Influence of multiple predators decreases body condition and fecundity of European hares [Dataset]. Dryad. https://doi.org/10.5061/dryad.2jm63xsqp

Abstract

We assessed the hypothesised negative correlation between the influence of multiple predators and body condition and fecundity of the European hare, from 13 areas in the Netherlands.

Year-round abundance of predators was estimated by hunters. We quantified predator influence as the sum of their field metabolic rates, as this sum reflects the daily food requirements of multiple individuals. We determined the ratio between body mass and hindfoot length of hares as an index of body condition, and the weight of their adrenal gland as a measure of chronic exposure to stress, and we counted the number of placental scars to estimate the fecundity of hares.

As hypothesised, we found that the sum of field metabolic rate of predators was negatively correlated with body condition and the number of placental scars, whereas it was positively related to the weight of the adrenal glands. In contrast to the sum of the field metabolic rate, the total number of predators did not affect the investigated risk responses.

The sum of the field metabolic rate can be a useful proxy for the influence of multiple predators and takes into account predator abundance, type, body weight, and food requirements of multiple predators.

With our findings, our paper contributes to a better understanding of the risk effects of multiple predators on prey fitness. Additionally, we identify a potential contributor to the decline of European hare populations.

Methods

Hare harvest and density estimation

In November and December 2013 we collected 73 hares (37 females, 35 males, 1 unknown) that were shot on 14 hunts (X  ± SD = 5.6 ± 2.8 hares/hunt) within subareas in the hunting leases. Hares were hunted by hunters on foot and at fixed positions during drives. Drives consisted of a dense line (a person every 5-10 m) of hunters and beaters with or without dogs. We accompanied the hunters during the drives on clearly demarcated subareas and counted the number of hares flushed (i.e., total count) and harvested to estimate hare density and the percentage of hares shot in a hunting lease. After the hunts, we took a random subset of the total number of hares shot in a hunting lease, although in two occasions hunters removed some of the hares before we could take a sample. Hares were stored at low temperatures (< 7°C) and dissected within 1 to 4 days (X  ± SD = 1.8 ± 0.8 days) after the hunt.

Body condition

We determined the ratio between body mass and hindfoot length of each animal (i.e., BM/HFL) as an index of body condition because this index has been shown to be highly correlated with total bone-marrow fat in other lagomorphs (i.e., snowshoe hares, Lepus americanus; Murray 2002) (see Appendix II). Additionally, we conducted a general health assessment of hares sampled before and during dissection, by assessing the presence of parasites, as well as lesions and other abnormalities that could affect body condition (Appendix III).

Age

We determined the weight of the eye lenses to distinguish different age classes (Peig & Green 2010). Eye lenses were removed and stored in 10% formalin solution. After 29.6 days ± 9.1 (SD) since first storage, we air-dried the eye-lenses at 80°C for 6 days and then weighed each lens to the nearest 0.1 gram. We assigned each hare to an age class based on eye lens weight (Broekhuizen and Maaskamp 1979) and the presence of an ulna coalescence (Stroh 1931). Individuals with lens weight > 270 mg and ulna absent were indicated as adult (> 1 year), while individuals with an ulna present were indicated as sub-adult (≤ 1 year old).

Fecundity

Female hares can have up to 5 litters each year, with a mean litter size between 2 and 3 leverets (Marboutin et al. 2003). For harvested female hares, the uteri were removed and frozen at -18˚C after our dissection. We later (205.9  days ± 10.4) thawed uteri and counted the total annual number of placental scares to provide an index of the number of pregnancies as an estimator of fecundity. As uterine walls of European hare regenerate during anoestrus, placental scar counts represent an index of fertilised eggs that implant during the preceding breeding season (February-August 2013). The average annual fecundity of European hares was found to be similar across regions (about10-11 placental scars; Hackländer et al. 2011). Placental scars were counted and stained by following the protocol by Hackländer et al. (2001). The number of scars were independently assessed, discussed, and verified by Weterings and Hackländer using a 7-30x magnification zoom stereoscopic binocular.

Weight of adrenal glands

During the lifetime of many species, the weight of the adrenal glands increases as a result of a prolonged period of exposure to stress (Harder & Kirkpatrick 1994). We carefully removed and weighted the adrenal glands without adhering tissue as an additional estimator of stress due to chronic exposure to the potential predation risk imposed by multiple predators.

Predator assessment

Because of the difficulty in estimating the year-round abundance of 23 different predator species, each with their specific census methods and biases, we made use of estimates provided by hunters (see validation of hunter estimates in Appendix V). Experienced hunters (X  ± SD = 31 ± 14 years of hunting experience, Appendix I Table A1) that assessed the number and type of predators in their hunting leases weekly (X  ± SD = 8 ± 10 h/week, Appendix Table A1, hunter effort) were interviewed to provide estimates of the year-round presence and abundance of 23 potential predator species of hares active on their hunting lease during the last year (Appendix VI). Potential predator species were chosen based on the literature (Tapper & Yalden 2010) and discussions with hunters. Hares (especially when they are young) can be predated by multiple predators, such as foxes, birds of prey and members of the mustelid family. Predation of young hares may negatively affect the condition of adult female hares via physiological pathways (Travers et al. 2010; Zanette et al. 2014).

sFMR and hunting risk calculations

The influence of predators on prey species was expressed as the sum of the field metabolic rate (sFMR) of all potential avian and mammalian predators of hares present in a hunting lease during the year before the collection of the harvested hares. We assigned each predator to a specific predator type (i.e., all birds, Pelecaniformes, mammal omnivores, and mammal carnivores) based on Nagy et al. (1999) (Appendix VI). We then calculated the average of the lower and higher limit of the body weight for each predator species (BWavg; birds: Del Hoyo et al. 1992; 1994; 1996; 1999; 2009; mammals: Lange et al. 2003). The average body weight per predator species was then used in the allometric relationships of Nagy et al. (1999) to calculate the field metabolic rate (FMRBWavg) for each predator species (per equation 2). Finally, for birds, we calculated the proportion of the year each species was resident in the Netherlands, as many birds migrate towards southern latitudes in winter (Vogelbescherming 2017).

Field metabolic rate (FMR) per predator species for each hunting lease (KJ day-1 ha-1) (based on Nagy et al. 1999):

 

[equation 2]  

= FMR based on average body weight (KJ day-1),

P  = proportion of the year being resident (birds only),

A  = size of the hunting lease (ha).

Hunting risk

We also investigated the effect of the risk of being killed by hunting on prey body condition and fecundity, to be able to assess its relative effect compared to the influence of predators, as prey responses to hunting can be stronger than responses to predators (Proffitt et al. 2009). The risk of hunting mortality was expressed as the percentage of hares shot from the total number of hares counted in a hunting lease during the hunting drives. Hunts were restricted to the period between 15 October and 31 December, with a frequency between 1 and 5 hunts per season (n = 8 hunting leases). We assumed that the risk of hunting mortality did not change between years, based on our communications with the local hunting groups. We thus assessed the risk of hunting mortality of the hunting period before the collection of the harvested hares.

 

Data analysis

Model investigated

First, we investigated the correlation between the sum of the predator field metabolic rate (sFMR) and the risk of hunting mortality as predictor variables and the body condition index as response variable using a linear mixed model (LMM) in R (package lme4 version 1.1-12; Bates et al. 2015; n = 66). Additionally, we investigated an alternative LMM with the total number of predators as a predictor variable and the body condition index as a response variable to investigate whether predator abundance better explains body condition compared to sFMR (see Appendix IV for an overview of the global models fitted). We included the sex of the hares, their age class, and the days since the start of the data collection as fixed effects, because female hares fatten up within several weeks at the end of the year to prepare for the next breeding season (Valencak et al. 2009). Besides, body condition varies during the season (Van Vuuren & Coblentz 1985) and scales differently between sexes (Murray, 2002). We included hunting lease as a random factor, with subareas nested within hunting lease. We excluded one adult female that had a very low body weight (2416 gr) compared to the rest of the adult females (X  ± SD = 3642 ± 318 gram).

Second, we investigated the correlations between sFMR and the risk of hunting mortality as predictor variables and the average weight of the adrenal gland as response variable using a LMM (n = 66). We included the age class and sex of hares as fixed effects, as adrenal glands of mammals are assumed to increase in size by chronic exposure to stress during their lifetime (Harder & Kirkpatrick 1994). Additionally, we expected a sex-specific stress response and perception of risk, as females have to fatten up to prepare for their first litter in winter (Valencak et al. 2009) and therefore probably respond differently to predation risk compared to males. Again, we used subareas nested within hunting lease as a random factor. We excluded one adult female that had a very high average weight of the adrenal glands (0.61 g) compared to the rest of the adult females (X  ± SD = 0.31 ± 0.076 g). Similarly to body condition, we also ran a model with the total number of predators as a predictor variable.

Third, we investigated the correlations between the sFMR, the risk of hunting mortality, body condition, and the weight of the adrenal gland as predictor variables and the number of placental scars as a response variable. Subareas nested within hunting lease were used as a random factor. Correlations were investigated by fitting generalized linear mixed models in R, with a binomial error structure (B(n=19, p) and logit link (n = 18) given that we modeled the success or failure of a fertilised egg implant in the uterus (i.e., placental scar present or absent) for each of the maximum number of possible implant locations (i.e., 19; Hackländer et al. 2001; Smith et al. 2010) in the uterus. We did not use a Poisson distribution, as this distribution did not approximate our distribution (i.e., the number of trials (n) multiplied by the probability of success (p) was much higher than 5 (NIST-SEMATECH, 2013)). The following females were excluded from the analysis of fecundity: females with inactive uteri (i.e., uteri that were too small for reproduction after visual inspection; n = 13; 1 adult, 12 sub-adults), females with active uteri that did not reproduce (i.e., these females are possibly sterile, especially in northwest European areas, see Smith et al. 2010; n = 3; 1 adult, 2 sub-adults), as well as females of which the uterus contained tumors or other abnormalities (n = 3; 2 adults, 1 sub-adult). Again, we also ran a model with the total number of predators (n=18) as a predictor variable instead of sFMR.

We used standardized regression coefficients to assess the effect size of the predictor variables on the three response variables. Continuous predictor variables were standardized and scaled by dividing their mean by two standard deviations (Gelman 2008). sFMR and the total number of predators were log10 transformed to normalize a right-skewed distribution. Multicollinearity of continuous predictor variables was not an issue because the Variance Inflation Factor (VIF) of all continuous predictor variables remained below 1.5 for all models. We tested the linearity between the predictors and the response variables using a Generalized Additive Mixed Model (package gamm4 version 0.2-6). The predictors had an effective degree of freedom (edf) close to 1 and were therefore linearly related to the response variables. Model selection was performed by using the ‘drop1’ protocol of Zuur et al. (2009) and the Akaike Information Criteria (AIC). The fit of the models was assessed using plots of model residuals.

Funding

Dutch Research Council, Award: 023.001.222

Wageningen University