Effect of captivity on morphology: negligible changes in external morphology mask significant changes in internal morphology

Captive breeding programmes are increasingly relied upon for threatened species management. Changes in morphology can occur in captivity, often with unknown consequences for reintroductions. Few studies have examined the morphological changes that occur in captive animals compared with wild animals. Further, the effect of multiple generations being maintained in captivity, and the potential effects of captivity on sexual dimorphism remain poorly understood. We compared external and internal morphology of captive and wild animals using house mouse (Mus musculus) as a model species. In addition, we looked at morphology across two captive generations, and compared morphology between sexes. We found no statistically significant differences in external morphology, but after one generation in captivity there was evidence for a shift in the internal morphology of captive-reared mice; captive-reared mice (two generations bred) had lighter combined kidney and spleen masses compared with wild-caught mice. Sexual dimorphism was maintained in captivity. Our findings demonstrate that captive breeding can alter internal morphology. Given that these morphological changes may impact organismal functioning and viability following release, further investigation is warranted. If the morphological change is shown to be maladaptive, these changes would have significant implications for captive-source populations that are used for reintroduction, including reduced survivorship.

SKCJ, 0000-0002-0905-0624 Captive breeding programmes are increasingly relied upon for threatened species management. Changes in morphology can occur in captivity, often with unknown consequences for reintroductions. Few studies have examined the morphological changes that occur in captive animals compared with wild animals. Further, the effect of multiple generations being maintained in captivity, and the potential effects of captivity on sexual dimorphism remain poorly understood. We compared external and internal morphology of captive and wild animals using house mouse (Mus musculus) as a model species. In addition, we looked at morphology across two captive generations, and compared morphology between sexes. We found no statistically significant differences in external morphology, but after one generation in captivity there was evidence for a shift in the internal morphology of captivereared mice; captive-reared mice (two generations bred) had lighter combined kidney and spleen masses compared with wild-caught mice. Sexual dimorphism was maintained in captivity. Our findings demonstrate that captive breeding can alter internal morphology. Given that these morphological changes may impact organismal functioning and viability following release, further investigation is warranted. If the morphological change is shown to be maladaptive, these then explicitly compare or manipulate environmental factors in captivity to provide robust inferences about the mechanisms for morphological change in captivity. The broad aim of this study was to provide a holistic assessment of the effects of captivity on the internal and external morphology of house mouse (Mus musculus) as a model species for small mammals. The specific aims were: (i) to compare the external and internal morphological traits between captive-reared and wild-caught individuals; (ii) to examine the effect of captivity on external and internal morphology across captive generations; and (iii) to compare the internal and external morphology of each sex from the captive and wild environments.

Study species
The house mouse is a small rodent distributed globally; the wild-derived strain was used in this study. This species is easily maintained in captivity and has a short generation time, which permits multiple generation studies to be conducted over relatively short periods [3]. Further, M. musculus provides a good model for investigating the effects of captivity on small mammals, because it shares a number of life-history traits in common with other small mammals. These include short generation time, high reproductive value, large litter sizes, iteroparity, sexual dimorphism and early age at maturity [37][38][39][40][41][42]. For these reasons, M. musculus is being increasingly used as a model to address questions related to small mammal captive breeding and reintroduction [3,43].

Housing and feeding
All individuals (wild-caught and captive-reared) were maintained separately in opaque plastic cages with a metal top (32 × 18 × 12 cm; MB1 Mouse Box, Wiretainers Pty Ltd, Melbourne, Victoria, Australia). We used wood shavings as cage substrate and all cages were provided with bedding material (shredded paper) and a 6 × 4 cm cardboard tube for cover. Water and food (Vella Stock Feeds Brand Rat and Mouse Nut; The Vella Group, Glendenning, New South Wales, Australia) were available ad libitum, determined as 20 g of food per 100 g of body mass supplied daily [44]. Room temperature was maintained at 22 ± 2°C on a 12 hour reverse light : dark cycle, lights on at 9:00am AEST, with full spectrum UV light provided. Humidity was not controlled; however it was monitored daily and recorded as 75 ± 10%. Animals were monitored daily, with cages cleaned once a week by removing the occupant and placing it in a round escape-proof container (54 × 52 cm; Spacepac Industries Pty Ltd, Wollongong, NSW, Australia) before placement in a new cage. Prior to relocation to the University of Wollongong, captive-reared F 4 mice were housed at UNSW in a temperature (19-25°C) and light-controlled room (12 : 12 h reverse light cycle, lights on at 9:00am AEST). Humidity was not controlled but was approximately 30% (A. Gibson 2014, personal communication). Males were housed separately at weaning but female siblings were housed together in groups of up to three individuals. All animals had been provided with food and water ad libitum. Mice were monitored daily and thoroughly checked three times a week for body condition, injuries and behaviour.

Captive-reared F 4 generation
For this study, captive-reared F 4 individuals were collected late January 2014 and transported to the Ecological Research Centre at the University of Wollongong, Wollongong, NSW. Mice were weighed (grams) on digital scales (Mettler-Toledo PJ3600, Mettler-Toledo Ltd, Port Melbourne, Victoria, Australia) upon entry into the individual housing (see Housing and feeding).
Once acclimated to the individual housing, captive-reared F 4 individuals were used to breed the F 5 generation. At the conclusion of the breeding period, captive-reared F 4 individuals were then reacclimated to the individual housing for a minimum period of 12 days before quantifying external and internal morphological traits (see External and internal morphological traits).

Captive-reared F 5 generation
Pedigree mapping was used to ensure that individuals from the founder generation were paired so that captive-reared F 5 mice did not share related parents or grandparents. Monogamous breeding pairs were held together for one week. Each breeding pair was housed in the same caging used for all wild-caught and captive-reared individuals in this study (see Housing and feeding).
Once mated, the captive-reared F 4 mothers were minimally disturbed, but were closely monitored on a daily basis around the expected due date to check for young. Offspring were housed with their mother until they were weaned at 25 days of age; this was kept uniform across all litters to reduce differences in maternal investment post-pregnancy. At 25 days of age, the captive-reared F 4 mother was removed from the breeding cage, and the litter housed for 2 days under ad libitum conditions; this was done with the view to reduce post-weaning stress on the litter. After 2 days, the offspring were then housed individually in the same caging used for all wild-caught and captive-reared individuals in this study (see Housing and feeding). The sex of each offspring (henceforth, captive-reared F 5 ) was determined as the mouse was placed in its individual housing (13 males and 14 females). Captive-reared F 5 mice were individually housed until they reached sexual maturity before quantifying external and internal morphological traits (see External and internal morphological traits).

Wild-caught population
Eight adult male and fifteen adult female M. musculus were captured in October-November 2014, at the same agricultural site in the western Sydney area (34°4 36.48 S, 150°34 15.6 E) as the source population of the original wild-caught founder generation (see Captive-reared F 4 generation). Elliott traps (30 × 10 × 8 cm; Sherman Traps Inc., Florida, USA) were set inside and outside sheds and surrounding vegetation. These were checked and emptied daily in the early morning approximately 8.00 am AEST. Elliott traps were baited with honey and peanut butter rolled oat balls.
Once captured, animals were transported to the Ecological Research Centre at the University of Wollongong, Wollongong (34°24 24 S 150°52 46 E) and housed in the same caging as the captivereared generations (see Housing and feeding). Wild-caught individuals were acclimated to the individual housing for a minimum period of 12 days before quantifying external and internal morphological traits (see External and internal morphological traits).

External and internal morphological traits
Animals were euthanased using CO 2 asphyxiation. Immediately following euthanasia, external morphological trait measurement and macroscopic dissection of organs were conducted to study morphometric differences between wild-caught and captive-reared F 4 and F 5 generations. External traits were: body mass (grams), skull length, body length, tail length and foot length (right hind leg; millimetres). Internal traits were: mass of brain, liver, combined kidneys, heart, lungs, testes/ovaries, spleen, stomach, caecum, small and large intestine and the lengths of the small and large intestine. Organs were weighed using scales with ± 0.01 g precision (Mettler-Toledo PJ3600, Mettler-Toledo Ltd, Port Melbourne, Victoria, Australia). Where applicable, digestive organs were emptied of their contents and rinsed with a 0.9% saline solution and weighed. The lengths of the small and large intestine were measured using slide callipers with ± 0.05 mm precision.

Multivariate analysis
To examine the effects of rearing environment on the external and internal morphology of mice, we used permutational analysis of variance (PERMANOVA) with 9999 permutations in Primer 7 (PRIMER-E Ltd, Plymouth, UK) [45] and PERMANOVA+ B version [46]. Permutational analyses were selected in favour of parametric analyses for these data because they are suitable for small and unequal sample sizes when comparing treatments [47][48][49] and for examining changes in morphology across multiple generations [50].
To control for the effects of body size on morphological traits, we calculated the residuals of a leastsquares regression of each morphological trait on body size using body mass or body length where lengths were measured. We then normalized the morphological trait data so that all morphological traits would take values within the same limits (−2 to +2 to cover all entries).
To test whether morphological traits differed between rearing environment and sex, a two-factor PERMANOVA was used on the external and internal morphological traits. In this analysis, the factors were rearing environment (three levels orthogonal and fixed; wild-caught; captive-reared F 4 and captivereared F 5 ) and sex (two levels orthogonal and fixed; female and male), with acclimation period (number of days acclimated) included as a covariate. An interaction term between rearing environment and sex was also included to account for any interactive effects of rearing environment and sex on morphology. All analyses used Euclidean similarity measures. Where significant in the main PERMANOVA test, PERMANOVA pairwise comparisons were conducted to compare external and internal morphology between rearing environments and between males and females. Pairwise comparisons were conducted in PERMANOVA+ B version [46]. Similarity percentage (SIMPER) analysis was used to identify the morphological traits that were primarily responsible for the compositional differences in external and internal morphology between captive-reared F 5 , captive-reared F 4 and wild-caught animals and between males and females. Only traits that contributed greater than 10% to compositional changes were used in univariate analyses, as these traits were likely to be primarily responsible for the compositional differences. One individual was excluded from external and internal morphological trait SIMPER analysis due to missing morphometric values.

Univariate analyses
To examine the effects of sex on external morphology of mice, four external morphological traits that contributed greater than 10% to compositional changes in external morphology between sexes in SIMPER were analysed using analysis of variance (ANOVA, table 3; electronic supplementary material). To correct p-values for multiple testing on external morphological traits, a Bonferroni adjusted alpha level (α = 0.0125) was used. To control for the effects of body size on external morphological traits, we used the residuals of a least-squares regression of each morphological trait in analyses. Two individuals were unable to be sampled and subsequently removed from the analysis (one captive-reared F 4 male due to an unexpected death; one captive-reared F 4 female with a previously damaged tail). Residuals from ANOVAs were inspected to verify normality and homogeneity of variances. For all morphological data, Tukey's HSD pairwise comparison tests were used for post hoc comparisons between treatments. Where normality was unable to be met, Kruskal-Wallis tests were used, with post hoc comparisons made using Wilcoxon tests.
To examine the effects of rearing environment and sex on the internal morphology in mice, internal morphological traits that contributed greater than 10% to compositional changes in internal morphology between rearing environments and sex in SIMPER were analysed using analysis of covariance (ANCOVA, table 3; electronic supplementary material). To correct p-values for multiple testing on internal morphological traits, a Bonferroni adjusted alpha level was used (α = 0.00625). For internal morphology, the effects of rearing environment and sex were the fixed effects, and acclimation period (number of days acclimated) was the covariate. An interaction term between rearing environment and sex was also included. To control for the effects of body size on internal morphological traits, we calculated the residuals of a least-squares regression of each morphological trait on body mass (or body length where length was measured). Where individuals were unable to be sampled for specific internal morphological traits, the degrees of freedom for these respective analyses were adjusted to account for these exclusions. Residuals from ANCOVAs were visually inspected to verify normality and homogeneity of variances. As there was no interaction between rearing environment and sex on any internal morphological traits, ANOVAs were then conducted to estimate the effect of rearing environment or sex on internal morphological traits (brain, liver, combined kidneys, spleen, small intestine length, large intestine, large intestine length, caecum) showing significance in the ANCOVA. Where the assumptions of normality and/or homogeneity of variance were not met, Kruskal-Wallis tests were used, with post hoc comparisons made using Wilcoxon tests. All morphological data were analysed in the JMP 11.2.0 statistical package.

Effects of rearing environment and sex on morphology
There was a significant interaction between rearing environment and sex on internal morphology (Internal: Pseudo-F: 1.926, p = 0.018; table 1). There was no significant interaction between rearing  The internal morphology significantly differed between individuals from differing rearing environments (Internal-Rearing environment: Pseudo-F = 2.853, p = 0.004; table 1), between sex (Internal-Sex: Pseudo-F = 6.296, p < 0.0001; table 1) and acclimation period (Internal-Acclimation period: Pseudo-F = 8.678, p < 0.0001; table 1). SIMPER analysis revealed four external and nine internal morphological traits were driving the compositional differences in external and internal morphology between captive-reared F 4 , captive-reared F 5 and wild-caught individuals and sex (only morphological traits with greater than 10% contribution were considered; table 3; electronic supplementary material).

Effect of captivity on internal morphology across multiple generations
The sizes of the internal organs significantly differed between captive-reared F 5 and captive-reared F 4 females (t 25 = 1.650, p = 0.007; table 2) and between captive-reared F 5 and wild-caught females (t 25 = 1.805, p = 0.001; table 2). There was no significant difference between captive-reared F 4 and wildcaught females (t 26 = 1.094, p = 0.293; table 2). SIMPER analysis revealed five morphological traits (brain, combined kidney mass, stomach, caecum and ovaries; electronic supplementary material) were driving compositional differences in internal morphology between captive-reared F 5 and captive-reared F 4 females.
The sizes of the internal organs did not significantly differ between captive-reared F 5 and captivereared F 4 males (t 20 = 1.186, p = 0.219; table 2) or between captive-reared F 5 and wild-caught males (t 17 = 1.151, p = 0.223; table 2). Further, there was no significant difference between captive-reared F 4 and wild-caught males (t 15 = 0.996, p = 0.434; table 2). SIMPER analysis revealed four morphological traits (large intestine, large intestine length, lungs and caecum; electronic supplementary material) were driving compositional differences in internal morphology between captive-reared F 5 and captive-reared F 4 males.

Sexual dimorphism in external and internal morphology
The external morphology differed significantly between female and males (t 64 = 1.884, p = 0.009; table 2), SIMPER analysis revealed body mass, body length, skull and tail lengths were driving compositional differences in external morphology between the sexes (electronic supplementary material). Only body mass differed significantly between females and males in external morphological traits (χ 2

Effects of rearing environment and sex on morphological traits
There was no significant interaction between rearing environment and sex on any internal morphological traits (table 3). Acclimation period had a significant effect on liver mass (F 1,68 = 9.899, p = 0.003; table 3). Three internal morphological traits and one external morphological trait contributing greater than 10% to compositional differences in external and internal morphology were significant between females and males following Bonferroni adjustment in the ANCOVA (table 3; electronic supplementary material). The body mass (χ 2 1 = 10.296, p= 0.001), brain (F 1,68 = 11.229, p = 0.001) combined kidney masses (F 1,68 = 47.262, p < 0.0001) were significantly lighter and large intestine length (F 1,68 = 8.644, p = 0.004) significantly shorter in females compared with male house mouse (tables 3 and 4).
There was a significant effect of rearing environment on combined kidney and spleen mass (table 3). The combined kidney and spleen mass were lighter in captive-reared F 5 compared with captive-reared F 4 individuals (Combined kidney: F 2,68 = 6.711, p = 0.002; Spleen: F 2,68 = 5.433, p = 0.006; table 5). Further, there was evidence that multiple generations maintained in captivity results in a shift away from wild phenotypes; combined kidney mass was lighter in captive-reared F 5 compared with wild-caught individuals (table 5).

Effects of captivity on morphology
Captive-reared mice showed differences in internal morphology, but not external morphology, when compared with their wild-caught conspecifics. Differences in morphology between captive and wild environments can be expected due to these environments differing in a multitude of biotic and abiotic factors [19]. The absence of significant changes to external morphology can be explained in one of two possible ways. First, differences between captive and natural environments may induce changes     in life-history organization, such as early sexual maturity, instead of changes to somatic growth in external morphological traits. Indeed, this has been observed in hatchery chinook salmon (Oncorhynchus tshawytscha), with egg size decreasing across a 20-year period but with no change in female body mass [51]. Second, external morphological traits may be less plastic, with changes in external morphology occurring more slowly and taking multiple generations to manifest [3,9]. Indeed, in captive black-footed ferrets (Mustela nigripes), skull and dental traits were 5-6% smaller than wild populations (founder population; museum specimens) and 3-10% smaller than wild-caught populations (collected near the founding population), but these differences only became apparent after more than 10 years of captive breeding [18]. In this case, the captive-reared house mouse used in our study may not have been sufficiently removed from the wild-caught founders (individuals were three to five generations removed) for changes in external morphology to become apparent [9]. In this study, the absence of changes in external morphology masked internal morphological changes. Specifically, combined kidney and spleen mass were lighter in captive-reared individuals compared with wild-caught individuals. The finding that some but not all internal morphological traits changed in captivity (i.e. some traits displayed remarkable consistency between wild and captive populations) may have occurred because not all traits strongly impact individual fitness in the captive environment [9]. This could relax selection pressure for trait change. Alternatively, widespread changes in morphology may require multiple generations to manifest because individual traits differ in how quickly they respond to change. To test this, and demonstrate both the speed and directionality of change, and that changes are biologically relevant, it would be necessary to track changes over additional generations.
Changes in organ sizes occurring in captivity could be due to the functional capacity of these organs being in excess of the actual demands, which would make the organs expensive and inefficient to maintain. Subsequently, the size of organs may have altered to deal with such inefficiency [22,29,52]. For example, changes in the mass or length of the digestive organs may have occurred due to an increased digestive efficiency associated with a higher quality captive diet [21,53], and changes in accessory organs such as kidneys and spleen may have occurred due to decreased immunological and disease exposure in the captive environment [54][55][56][57]. However, identifying the specific mechanisms that cause morphological changes can be challenging. This is largely because multiple environmental factors can affect internal morphology, and the effects of these factors are likely to be interactive [28]. Future studies would benefit from explicitly comparing the nutrient and energy content of diets, or even by manipulating these in captivity to provide robust inferences about the mechanisms for morphological change in captivity. the wild-caught phenotype across one generation in captivity, but only in internal morphological traits [9]. Previous studies report change in internal morphological traits can occur after just one generation maintained in captivity [14,19]. The lack of significant differences in other internal and external morphological traits between captive-reared F 5 and captive-reared F 4 house mouse may indicate that these traits do not play a significant role in fitness in captivity [3,9].

Effect of captivity on morphology across multiple generations
Interestingly, the overall sizes of the internal organs of captive-reared F 5 male mouse did not differ significantly from captive-reared F 4 or from the wild-caught males. The reason for this finding is not clear, but sex-based differences in the magnitude of change in response to captivity across generations may explain why male internal morphology was more stable [3,9]. A previous study investigating the effects of selective breeding for high activity in house mouse reported females and males having differing rates of morphological change in response to high activity, indicating that trait plasticity differed between the sexes across generations [55]. Given these findings, changes in internal morphology may take multiple generations to manifest in males [3,9]. If directional changes in internal morphology are shown to occur across multiple generations of mouse maintained in captivity, this is likely to have significant implications for captive-bred animals following release [3]. For example, one study found 83% of offspring post-release were of same-source parentage in released third-generation captive-bred and wild-caught M. musculus, suggesting that morphological traits changing in captivity (such as body size) may be important to mate selection and may infer assortative mating [43]. In this case, the captivebred reintroduced animals appear to have higher reproductive success, but exactly how captivity affects the chances of survival post-release remains to be studied.

Effect of captivity on sexual dimorphism in morphology
Both external and internal morphology were found to differ significantly between females and males, and these sex-based morphological differences occurred in both captive-reared and wild-caught house mouse. While we can expect sex-based differences in morphology as an outcome of sexual selection favouring different trait values in males and females, we expected a loss of sexual dimorphism in captivity due to changes in resource availability and the strengths and targets of sexual selection [3,13]. The maintenance of sexual dimorphism in the present study suggests that sexual selection pressures remained unchanged in the captive environment. Alternatively, changes to or loss of sexual dimorphism may take multiple generations to manifest, and may not have been observed in our study [3,9]. There is emerging evidence that sexual dimorphism can be maintained in captivity; however, most studies have not investigated whether relaxation or reduction in sexual selective pressures occurs in captivity [3,12]. As such, to allow for a greater understanding of the effects of captivity on sexual dimorphism, it would be valuable to test for sex-specific differences in various morphological traits across a diversity of taxonomic groups. In recognition of this possibility, several recent studies have begun to explore whether sexual selection theory can be used to inform management strategies [43,58].

Implications for captive breeding programmes and management
Our finding that negligible changes in external morphology masked significant changes to internal morphology has implications for captive breeding programmes. Changes to internal morphology in captivity are known to impact digestive efficiency [21,53] as well as immune responses and disease resistance [54][55][56][57]. Consequently, rapid changes in internal morphology could have severe and unforeseen effects on the viability of small mammals held in captivity. However, this is dependent on what morphological traits change, and whether those changes are maladaptive for natural environments. If the morphological change is shown to be maladaptive, these changes would have significant implications for captive-source populations that are used for reintroduction. While there is currently no information on the effect of internal changes on the post-release viability of small mammals, there is some evidence for these effects in birds [21]. Future research on small mammals would benefit from investigating the extent to which internal morphological changes occurring in captivity are maladaptive under natural conditions, and whether these impacts can be mitigated by manipulating the captive environment.
Importantly, it is unknown whether the observed changes in morphology were due to phenotypic plasticity or changes in selection pressures experienced in captivity. Further, case studies investigating morphological plasticity in captivity are limited [3]. Plastic responses in morphology to changes in environmental conditions are common, and these plastic responses can be fast, repeatable and reversible  [17,52,[59][60][61]. If morphological traits are shown to be plastic, this presents an opportunity for strategic management of morphological phenotypes [4]. That is, the captive phenotype may be altered to better suit the wild environment; but tailoring methods (such as pre-release exposure) may be required to increase likelihood of survival following release [62]. For example, post-release survival of pheasants (Phasianus colchicus) was higher in those that had been exposed to more natural diets prior to release. One of the mechanisms to explain this increased survivorship was the development of gut morphology (changing intestine and caecum lengths) to suit a natural diet [63]. Alternatively, if morphological change is due to selection pressures in captivity, environmental factors that change the strength or direction of selection pressures could be manipulated to drive selection for favourable morphological phenotypes [3,4,64]. In support of this notion, there is emerging evidence that understanding and controlling selection may be able to mitigate the effects of captivity that influence the success of captive animals released into the wild [64].

Conclusion
This study aimed to investigate whether morphology differed between captive-reared and wild-caught individuals, the effect of captivity on external and internal morphology across generations and whether the sexes responded differently to the captive-rearing environment. The absence of changes to external morphology masked internal morphological changes. Moreover, there was evidence for a shift in internal morphology across one generation in captivity. It was found that morphology significantly differed depending on sex, and that sex-based morphological differences were maintained in the captiverearing environment. Identifying the magnitude and direction of morphological changes in captivity is an important first step towards developing and refining methodologies to minimize unfavourable phenotypic changes in captivity. In turn, this knowledge may be used to improve captive breeding and reintroduction programmes. Overall, these findings draw attention to the need to consider the potential for both internal and external morphological changes when developing methods to improve captive breeding programmes and reintroduction programmes.
Additional supporting information may be found in the electronic supplementary material.