The Relationship between Shell Morphology and Crevice Size Affecting Retreat Selection of the Keeled Box Turtle (Cuora mouhotii)

Fanrong XIAO,Zhen HONG,Jichao WANG,Taggert BUTTERFIELD and Haitao SHI＊

1 Key Laboratory for Ecology of Tropical Islands of Ministry of Education; Key Laboratory of Tropical Animal and Plant Ecology of Hainan Province; College of Life Sciences,Hainan Normal University,Haikou 571158,Hainan,China

2 Escuela Nacional de Estudios Superiores Unidad Morelia,Universidad Nacional Autónoma de México,Morelia 58190,México

Abstract Rock crevices are important retreat sites for many reptiles.Unlike lizards and snakes,crevicedwelling turtles have rigid shells which severely limit the size and shape of crevices they are able to use.Few studies have addressed the relationship between turtle’s shell morphology,behavior,and crevice size.Here,we investigate the relationship between morphology and crevice size selection in the Keeled box turtle (Cuora mouhotii).We radio tracked turtles in the field to document crevice dimensions and turtle behavior associated with crevices,and we conducted a controlled experiment in the lab to understand how a turtle’s morphology influences its decision to select crevices with different dimensions.Both laboratory and field experiments demonstrate that turtles select deep and moderately wide crevices that are low enough to barely accommodate the carapace.Crevices used by turtles tend to be about 2 times as wide,1.2 times as high,and 3 times longer than the turtle’s shell.Moreover,turtles tended to prefer crevices with horizontal openings in the field and position themselves head first in the crevice.We conclude that morphology,especially carapace height,influences the Keeled box turtles’ decision to select specific sizes of crevice.We argue that the tendency for turtles to select crevices that come into contact with their carapace contributes to successful defense against predators,as it would be more difficult for a predator to remove a turtle when the crevice is nearly touching the turtles shell.

Keywords antipredation,burrow,microhabitat,rock crevice,shelter

1.Introduction

The distribution and abundance of organisms is closely related to the availability of suitable microhabitats in the environment,especially retreat sites (Bustard,1969; Price,1978; Milne and Bull,2000; Zhanget al.,2006).Retreat sites can provide refuge from predators and access to a stable microclimate (Pringleet al.,2003; Croaket al.,2008).Therefore,availability of retreat sites may have a strong impact on individual fitness as well as population viability and it has been shown that organisms nonrandomly select retreat sites in the environment that can meet their immediate needs.However,retreat sites are not unlimited,and the morphology of an organism influences the range of retreat sites that it can utilize.Previous studies have focused on how the availability of microhabitats in the environment influences retreat site selection (Milne and Bull,2000; Pringleet al.,2003; Croaket al.,2008; Coxet al.,2018),but few have investigated how an organisms morphology affects the retreat sites they use.

Rock crevices as retreat sites play a central role in the natural history of many reptile species (Pringleet al.,2003; Croaket al.,2008).However,unlike snakes,lizards and geckos,crevicedwelling turtles face a significant challenge because they have a rigid shell that limits their ability to utilize crevices smaller than their shell.For this reason,rock crevices should be big enough to allow turtles to enter and exit,but small enough to prevent predators from removing the turtle.Therefore,turtles’shell shape likely determines the size of crevices that they can use.For instance,Jackson (1988) found that musk turtles(Sternotherusspp.) preferred crevices that were low enough to come into contact the carapace,demonstrating that a turtle’s shell height is correlated to the crevice height.Nevertheless,musk turtles are also aquatic and their compressed shell could reflect an adaptation to be more hydrodynamic in water and not to occupy a specific sized crevice (Claudeet al.,2003; Bonnetet al.,2010; Bensonet al.,2011).In terrestrial turtles on the other hand,especially rock-dwelling turtles such as the pancake tortoise (Malacochersus tornieri) or Namaqualand speckled padloper (Chersobius signatus),it is possible that compressed shells are an adaptation to utilize a certain size of rock crevice crevice but this remains to be seen (Irelandet al.,1972; Malonza,2003; Loehret al.,2006).Previous studies,such as Shiet al.(1995)and Nasengbateet al.(2013) have investigated the size of soil burrows that turtles select,but studies have not tested if there is a relationship between burrow size and shell shape.

The Keeled box turtle (Cuora mouhotii) is distributed throughout southern China,including Hainan Province (Zhao and Adler,1993),and is typically found in moist evergreen forest (Wanget al.,2011).Previous work has already shown that the Keeled box turtle has a relatively compressed shell and prefers microhabitats with rock crevices in the field (Xiaoet al.,2017).However,it is not clear what features of rock crevices influence their selection,how shell shape determines selection,and what behavior they use when occupying crevices.We predict that 1) the observation that Keeled box turtles have a more compressed shell compared to other turtles in the habitat(e.g.Flower back box turtleCuora galbiniforns; Xiaoet al.,2017)is an adaptation to occupy rock crevices,and 2) that the Keeled box turtle prefer crevices that come into contact with the top of their carapace to limit a predators ability to remove them from crevices.Moreover,because the posterior edge of the Keeled box turtle’s carapace is serrated (Xiaoet al.,2017),we hypothesize 3) that the majority of turtles that we observe in rock crevices in the field will be oriented in the crevice head first,with the serrated carapace facing outward as a potential defense from predators.To test these hypotheses,we documented the Keeled box turtles’ behavior in rock crevices,and the sizes of crevices(width,depth,height) that they used in the field.In addition,we did a laboratory experiment in which we manipulated the dimensions of artificial crevices and documented the preference towards different crevices with different dimensions.

2.Materials and Methods

2.1.Ethics statementThis study was approved by the Animal Research Ethics Committee of Hainan Provincial Education Centre for Ecology and Environment,Hainan Normal University (HNECEE-2014-002) and was carried out in strict accordance with the institutional guidelines.Fieldwork was carried out with permission from the Diaoluoshan National Nature Reserve.No turtles were sacrificed for this study or incurred injury.

2.2.Size of rock crevice and behavior of turtles in the fieldWe tracked 7 adult Keeled box turtles using radio telemetry from April 2015 to February 2016 in the Diaoluoshan Natural Reserve in Hainan Province,People’s Republic of China.This reserve is comprised of approximately 18 389 ha of intact rainforest.January is the coldest month,with an average temperature of 15.4oC,and the hottest month is July,with an average temperature of 28oC.Rainfall in the reserve is plentiful,with 1 870-2 760 mm of precipitation falling annually,mostly during the rainy season (May to October).Radio transmitters(diameter ×height=21 mm × 8 mm) were fitted to the last pleural scute of the posterior end of the turtles’ carapace in a location that did not inhibit turtles from entering shelters.Turtles were monitored on a regular basis during the study period,and because they use many different types of shelters we only include shelters that were occupied for more three days into our analysis.For each shelter we recorded the type of shelter (1=rock crevice,2=tree hole,3=fallen log,4=grass clump,5=bamboo clump,6=deciduous leaves) and measured the maximum width (cm; same throughout),maximum depth,minimum height (inner crevice height at which the turtle was located),and maximum height (maximum height at crevice opening).We also recorded whether or not the opening of the crevice was horizontal or non-horizontal and how turtles were positioned when found in rock crevices.For example,we noted if the turtle was positioned in the crevice tail first or head first,and if the top of the carapace was in contact with the ceiling of the rock crevice.

2.3.Selection of artificial crevice size in the laboratoryTo determine if the dimensions of the Keeled box turtle’s carapace influences the crevices it uses,we conducted a laboratory experiment in which we exposed turtles to crevices with different depths,widths,and heights.We did these experiments in a rectangular enclosure (size:145 cm ×65 cm;Figure 1) from April to July 2016 and turtles were exposed to ambient temperatures during the experiment (19-22oC).This enclosure was located indoors,but was exposed to natural light throughout the day through several windows.To remove any potential obstacles that might influence a turtles decision to select a shelter,the habitat within the enclosure consisted of only bare ground and artificial shelters (no vegetation or plants were included within enclosure; Figure 1).All artificial shelters had the same lighting conditions over the course of the day,which were consistent throughout the experiment.Turtles were fed before,but not during the experiments.

The rectangular enclosure was divided into four sections and we placed a different sized artificial shelter in three of the four sections (Figure 1).No artificial shelter was included in the fourth section (bare ground) as a control and turtles could move freely between each section.Artificial crevices were assembled using bricks,and depending on the experiment,were designed to exceed the mean length,width,or height of a turtle by 0.2,1,and 2 times,respectively.In the crevice width experiment,for example,we wanted to know if the width of crevice influences a turtle’s decision to select a crevice.To do this,we put three crevices in their own section of the arena,and each crevice exceed the mean carapace width (11.86 cm) of the Keeled box turtle by 0.2 (crevice width=14.24 cm),1 (crevice width=23.73 cm),and 2 (crevice width=35.59 cm) times(Figure 1).The crevice dimensions that were not manipulated in each experiment were set to exceed a turtle’s mean dimensions by 0.5 times.The average measurements of the Keeled box turtle that were used to determine crevice dimensions were based on the measurements of five individuals and include length=17.12 cm,width=11.86 cm,and height=6.85 cm.See Table 1 for dimensions and names of each crevice in each experiment.

During each experiment,five Keeled box turtles were released individually into the arena and their behavior was recorded for 84 hours using 24-hour video monitoring.These turtles were the same individuals that were tracked in the field.We only used the last 72 hours (three days) of video recordings to collect data because some turtles appeared stressed immediately after being released into the enclosure.We reviewed the recordings for each experiment and calculated the amount of time that turtles spent inside each artificial crevice,or on bare ground of the enclosure.At the end of experiments,the elapsed time that each turtle in the three artificial crevices and bare ground was used for subsequent data analysis.

2.4.Statistical analysisFor the field experiment,Chi-squared test was used to assess whether the Keeled box turtle differed in its preference of type of shelter.We tested for normality in width,depth,and height of crevices used by the Keeled box turtle using a Kolmogorov-Smirnov test.A Chi-square test was used to test turtles preference in crevice openings (horizontal vs.non-horizontal),orientation in the shelter (tail vs.head first),and whether or not the carapace was in contact with the roof of the shelter (contact vs.no contact).For the laboratory experiment,we used generalized linear mixed model (GLMM)to test whether the amount of time (hours) that the Keeled box turtles spent in a particular artificial crevice could be explained by the dimensions of the artificial crevice.The time spent inside each section (hours) was tested as a function of crevice type (including bare ground) which was defined as the fixed effect.Whereas,turtle individual number and size (i.e.carapace length,width or height,and each of them corresponds to its own experiment) were defined as random effects.A separate GLMM was done for each experiment.Pair wise comparison of time spent in different sections was tested in the GLMM using pair wiset-tests.All data were expressed as mean ± SE.For all statistical analyses,an alpha level ofα=0.05 was selected as criterion for statistical significance.All statistical analyses were carried out in SPSS 19.0.

Figure 1 The rectangular enclosure (145 cm × 65 cm) with four sections (three artificial crevices and bare ground) in the laboratory.This picture shows the layout of three artificial crevices with different widths (but with same depth and height) in the rectangular enclosure.All the remaining areas of the enclosure were classified as bare ground except for the artificial crevices.

Table 1 Dimensions of artificial crevices in the three experiments in which we manipulated the width,depth,or height of the crevice.These dimensions were selected to exceed the mean dimension of the Keeled box turtle (Cuora mouhotii) 0.2,1,and 2 times,respectively.The two artificial crevice dimensions that were not being manipulated were set to exceed the same measurement of a turtle by 0.5 times.Mean carapace width,length,and height of the Keeled box turtle was 11.86 cm,17.12 cm,and 6.85 cm,respectively (n=5).

3.Results

The seven turtles equipped with radio transmitters had the mean (± SD) carapace length,width,and height of 16.28 ± 0.71 cm (range 14.07-18.30 cm),11.45 ± 0.19 cm (range 10.19-11.83 cm),and 6.63 ± 0.13 cm (range 6.00-7.10 cm),respectively.From April 2015 to February 2016,turtles were observed inactive in shelters for more than three days on 41 occasions.Of these shelters,23 were rock crevices (56.10%),6 were tree holes (14.46%),5 were bamboo clumps (12.20%),4 were fallen logs (9.76%),2 were grass clumps (4.88 of time) and 1 was deciduous leaves (2.44%).Showing that turtles prefer rock crevices when compared to other shelter types (Chi-square test,x2=48.415,P＜ 0.0001).The maximum crevice width (40.74 ± 5.14 cm; range 13-110 cm),maximum crevice depth (57.55 ± 5.31 cm; range 15-110 cm) and the minimum crevice height (9.30 ± 1.40 cm; range 7-12 cm)were normally distributed (Kolmogorov-Smirnov test,Z=0.918,P=0.368;Z=0.728,P=0.665;Z=1.02,P=0.249; Figures 2A,B,D).Whereas,the distribution of maximum crevice height(15.43 ± 6.29 cm; range 10-36 cm) was right-skewed (skewness=1.982) and not normally distributed (Kolmogorov-Smirnov test,Z=1.371,P=0.047; Figure 2C).These data comprise of 204 independent observations of turtles using 23 different rock crevices,indicating that the Keeled box turtle frequents rock crevices that average 40 cm wide,57 cm deep,9 cm minimum height,and 15 cm maximum height.Chi-square test results show that the Keeled box turtles prefer crevices with horizontal openings (95.65%;n=23,x2=19.17,P＜ 0.0001; Figures 3A,C),prefer to be oriented head first in the shelter (Table 2; Figure 3D),and to be touching the roof of the rock crevice with their carapace (Table 2; Figures 3B,D).

Figure 2 The frequency distribution of maximum crevice width(A),crevice depth (B),crevice height (C),and minimum crevice height (D) in the field.

In the lab,we analyzed 72 hours of recordings that correspond to 15 recording events (three experiments for each turtle).In each experiment,the GLMM revealed significant effect of treatment (crevice size) on time spent in crevices (width experiment:F3,16=3.569,P=0.047; depth experiment:F3,16=4.062,P=0.029; height experiment:F3,16=3.458,P=0.041).There was no inter-individual (width experiment:Z=1.34,P=0.18;depth experiment:Z=1.528,P=0.126; height experiment:Z=1.766,P=0.77) or effect of body size (width experiment:Z=0.559,P=0.576; depth experiment:Z=1.146,P=0.252; height experiment:Z=1.375,P=0.169) in any of the experiments.In the width experiment,pair wise comparisons of mean time spent in crevices or bare ground were significantly different between all sections,with most time being spent in mediumwidth crevices (Figure 4A).In the depth experiment,pair wise comparisons of mean time spent in each crevice or bare ground reveal significance differences between all sections except bare ground,with most time being spent in deeper crevices (Figure 4B).In the height experiment,pair wise comparisons of mean time spent in each crevice or bare ground were significant between all sections except bare ground,with most time was spent in low crevices (Figure 4C).In sum,these results suggest that the Keeled box turtle prefers crevices that are about 2 times as wide,1.2 times as high,and 3 times as deep as the turtle.

4.Discussion

Our study revealed that the Keeled box turtle prefer crevices that are low enough to barely accommodate the carapace both in the field and the laboratory.To the best of our knowledge,this study is the first to quantify the relationship between the crevice height and carapace height in a terrestrial turtle.Although some studies focus on how animal morphology affects the choice of crevice height (Jackson,1988),few studies have shown that amphibians or reptiles choose low rock crevices to prevent predators from approaching them (Cooperet al.,2000; Thompson and Withers,2005; Smithet al.,2017).The civets (e.g.Paguma larvataandParadoxurus hermaphroditus)are the predators of the Keeled box turtle in the wild,thereby,choosing a lower crevice helps turtles prevent being predated by these mammals.Our observation that a more compressed carapace allows turtles to fit in lower cervices suggests that the flatted carapace of the Keeled box turtle has evolved a more compressed carapace to fit into rock crevices.

Table 2 Behavior of the Keeled box turtles (Cuora mouhotii) while hidden in the rock crevices in the field.

Figure 3 Two examples of a Keeled box turtle (Cuora mouhotii) in a typical rock crevice in the field.Location of the turtle from a distance can be seen on the left (A,C),inside white circles,whereas a photo of the same turtle inside the crevice (indicated with an arrow) on the right (B,D).All photos were taken by F.XIAO in 2015.

The Keeled box turtle preferred crevices with a medium width,indicating that there is a tradeoff between the space turtles needs to move and to avoid predation.Narrow crevices are effective against predators,but cannot be affectively used turtles if they cannot maneuver within the crevice (Schlesinger and Shine,1994; Milne and Bull,2000; Williamset al.,2016).Therefore,the medium width of the crevice not only ensures that the turtle has space to turn around and perform other movements,but combined with its small height,prevents predators from taking advantage of the larger space to enter or reach into the crevice.

This study found that the Keeled box turtle selected medium-deep crevices in the field,but preferred deep crevices in the laboratory.This may be due to the fact that the mediumdeep crevice (57.55 cm) in the field was deeper than the deepest crevice in the laboratory (51.35 cm),which results in the tendency of turtles to choose crevices with the maximum depth available in the laboratory.We have observed that the Keeled box turtle always hides as deep as possible into the rock crevices in the field.The medium-deep rock crevices in the wild are sufficient for protection.We suggest that the Keeled box turtles exhibit a preference for a deeper crevice to reduce the risk of being dragged out by the predators,and to avoid climatic extremes,similarly to other reptiles (Milne and Bull,2000; Pikeet al.,2010).In fact,the microclimate of rock crevices in general is an important factor that influences organisms decision to select specific crevices (Mile and Bull,2000; Pringleet al.,2003; Coxet al.,2018).Therefore,future work should aim to understand how temperature and humidity inside and outside of rock crevices influences a Keeled box turtle’s decision to select specific crevices.

Figure 4 Bar graphs indicating the amount of time that the Keeled box turtles (Cuora mouhotii) spent in each section of the rectangular enclosure in each experiment.A,B and C correspond to the experiment in which we manipulated crevice width,depth,and height,respectively.Asterisks denote the difference (＊P ＜ 0.05)based on pair wise comparisons (all df=16) in GLMM.

Field observations have shown that Keeled box turtles often make their carapace touch to the roof of the crevice with the help of their limbs.This is consistent with a previous study on musk turtles (Jackson,1988),and one explanation for this behavior is that a turtle can better resist a predator trying to force it out of a crevice by creating friction between the carapace and the roof of the crevice when it extends its limbs(Jackson,1988).Our recent research confirmed that the Keeled box turtles can produce more friction with their carapaces than the Flower back box turtles (Cuora galbinifrons) which do not utilize rock crevices (unpublished data).We also found that the crevices used by the Keeled box turtle had mainly horizontal openings in the wild,which increases the contact area between the carapace and the ceiling of crevice,thus increasing the friction.Many animals prefer this type of crevice because horizontal openings below big rocks are not easily found by predators (Schlesinger and Shine,1994).Another explanation is that the preference for the rock crevices which can touch carapace increases the likelihood that turtles will use the lowest crevices,because these rock crevices can reduce the available space for predators (Jackson,1988).Our field and laboratory studies confirm that the Keeled box turtle preferred low rock crevices,showing that this kind of behavior and morphology was correlated with the height of crevice.Therefore,the Keeled box turtle’s carapace height was the most important factor in determining the size of the rock crevice when compared with the carapace length and width.

Many aquatic turtles or semi-aquatic turtles,such as musk turtles,prefer mud or vegetation for shelter (Belleau,2008;Markleet al.,2018).Thus,Jackson’s (1988) study on musk turtles was a simplification of size selection of rock crevice in a bottom-walking species.Unlike musk turtles,many evidences (Lyet al.,2013; Xiaoet al.,2017),including this study,have demonstrated that the Keeled box turtle prefers to take refuge in rock crevices shelters in the field.Therefore,our study strongly implies that shell morphology plays an important role in a crevice-dwelling turtle’s decision to select a shelter.

In addition,the Keeled box turtle’s carapace has a serrated posterior edge (Xiaoet al.,2017) and we found that they most often take refuge inside rock crevices with their posterior end of the carapace facing out (head first) in the field.Such a behavior might also help deter predators from attempting to remove turtles from rock crevices.The Black-breasted leaf turtle (Geoemyda spengleri) also has a carapace with prominent serrated edges and prefers taking refuge inside rock crevices in the field,but whether it also adopts this kind of behavior to defend against predators requires further research.

In summary,we found that Keeled box turtles prefer crevices that have a similar height to their shells,and that turtles most frequently take refuge in the deepest parts of rock crevices so that their carapaces are in close contact with the ceilings of the crevices.There also seems to be a tradeoff in crevice width,in which Keeled box turtles select crevices that are wide enough to move around,but not too wide to allow the predators to get in.We conclude that the tendency for Keeled box turtles to select crevices with a similar height as their carapace,and to enter the crevices head first contributes to their success in avoiding predators.It is likely that the compressed carapace in Keeled box turtles and the serrated posterior edge of their carapace are adaptations to life in rock crevices.This seems even more evident when considered that the Keeled box turtle is sympatric with the flower back box turtle,which does not have similar shell characteristics or behavior (it does not use rock crevices).

AcknowledgementsWe are grateful to the Diaoluoshan National Nature Reserve for access to the field site and providing convenience for our works.Funding was provided by the National Natural Science Foundation of China(No.31772486) and the Hainan Provincial Natural Science Foundation of China (No.319MS047).