Abstract Although numerous aquatic and terrestrial species adorn their surface with items secured from their surroundings, termed decorating, the physical and environmental factors that drive this behavior are often unclear. One of the best-known examples of this phenomenon are the decorator crabs (Majoidea), with almost 75% of species known to decorate. Here, we examined patterns of decorating in a coral reef-associated majoid, Camposcia retusa, to identify what factors determine patterns of, and investment in, decorating. Observations of natural decoration patterns indicate this species primarily decorate with sponges, fleshy and filamentous algae, and detritus. Decorations were primarily distributed on the carapace and hind walking legs which may reflect exoskeleton morphology. However, decoration cover did not decline with size, as is observed in some other majoid species, suggesting the factors driving decoration investment remain consistent throughout growth stages. From behavioral experiments, we determined that nonuniform decorating is a result of active selection, with crabs preferentially decorating their hind legs and carapace, with only small items placed on the carapace. Decorating in C. retusa appears to function primarily as an antipredator response, with crabs decorating at higher rates when access to shelter is limited. This study is the first to quantify the decorating habits of C. retusa and suggests that behaviorally-mediated decorating has a primary antipredator function. This study also highlights the value of manipulative behavioral experiments as a tool for assessing the behavioral mechanisms that drive decorating in animals. INTRODUCTION In an ecological context, decorating refers to any behavioral and/or morphological adaptations that facilitate the accumulation and retention of environmental material onto an organism’s exterior surface (Ruxton and Stevens 2015). Although decorating is observed in terrestrial and aquatic species from ~25% of all major metazoan phyla (Berke et al. 2006), the underlying drivers are frequently unclear (Ruxton and Stevens 2015). Decorating is generally theorized to have a primary antipredator function, with decorations either disrupting an organism’s visual appearance causing non- or misrecognition, that is, camouflage (Jackson and Pollard 2007), or preventing successful capture post-recognition (McClintock and Janssen 1990; Brandt and Mahsberg 2002). However, decorating may also play additional roles depending on an organism’s environment. For instance, decorations could alter an organism’s physical properties preventing dislodgement or provide a physical barrier preventing structural damage via abrasion or UV light (Dumont et al. 2007; Limm and Power 2011). The widespread evolution of decorating suggests it has clear benefits; however, investment in decorating is not without potential costs. Although direct evidence is limited, highly decorated individuals can experience greater rates of energetic expenditure (Briffa and Elwood 2005) and impaired mobility (Otto 2000) due to carrying additional weight. The diversion of energetic resources towards decorating can affect the physiological condition of individuals, impacting growth and potentially fitness (Mondy et al. 2011, 2012). The physical demands associated with carrying items likely explains why decorating occurs more frequently in aquatic species, with buoyancy compensating for additional weight. Indeed, aquatic species often display extravagant patterns of decoration when compared to their terrestrial counterparts (Ruxton and Stevens 2015). Of these aquatic species, one of the best-known and most studied examples are the decorator crabs (Majoidea). Of the 900 species from within this superfamily, at least 75% are estimated to decorate to some degree (Hultgren and Stachowicz 2009). Decorating in these crabs has both a morphological and behavioral basis making them the ideal model for examining the physical and environmental factors that promote and constrain this mechanism (Hultgren and Stachowicz 2011). Decorations are attached to thin, Velcro-like setae that extend from the exoskeleton, with those used primarily for decorating having a hooked structure that ensures item retention (Szebeni and Hartnoll 2005). Although the abundance and distribution of these setae constrains decoration potential, decoration patterns and composition are largely the result of behavioral selectivity (Thanh et al. 2005; Stachowicz and Hay 1999). These crabs are often assumed to be generalists, decorating with items relative to abundance (e.g., Fürböck and Patzner 2005; Martinelli et al. 2006); however, many in fact exhibit material preferences (Hultgren and Stachowicz 2011). For instance, some species will select materials that are easier to manipulate (Fürböck and Patzner 2005; Hultgren et al. 2006), suggesting decorating efficiency plays a role in material choice, whereas others will select noxious or unpalatable items (Stachowicz and Hay 1999). Many species are also selective with how items are placed on the exoskeleton, displaying consistent patterns in the order of body parts that they will decorate. For example, many majoids will preferentially decorate their rostrum and front claws (cheilipeds), presumable to conceal these body parts that are constantly in motion (Hultgren and Stachowicz 2009). However, patterns of selectivity vary considerably between species, it likely corresponds with the underlying drivers of decorating itself. Decorating in majoid crabs has a number of proposed functions, including storing surplus food (Woods and McLay 1994), mediating intraspecific interactions (Hazlett and Estabrook 1974), and hiding from or attracting prey (Wicksten 1983). However, empirical evidence for these is scarce with most research instead suggesting a primary antipredation function (Hultgren and Stachowicz 2011). Several studies have shown a correlation between decoration abundance and vulnerability to predators (Stachowicz and Hay 1999; Thanh et al. 2003; Hultgren and Stachowicz 2008) or show crabs modify decorating behavior depending on perceived risk (Thanh et al. 2003; but see Stachowicz and Hay 1999). In addition, many majoids reduce decorating intensity as they grow, presumably as they are less vulnerable to key predators (e.g., gape-limited fishes) once they reach a certain size (Hultgren and Stachowicz 2011). Being decorated could reduce predation risk in several ways, one of the most obvious being via camouflage. Among antipredator mechanisms, camouflage is unique in that it allows organisms to remain effectively concealed by disrupting the sensory pathways predators use to find prey (Ruxton 2009). Decorations could reduce the ability of predators to initially perceive an individual, by blending it into the surrounding visual or chemical environment or by breaking up its visual outline, that is, background pattern matching or disruptive crypsis (Ruxton and Stevens 2015). Decorating could also camouflage an individual following perception, by causing misidentification as a nonedible item, that is, masquerade (Skelhorn et al. 2010). Alternatively, decorating could reduce predation risk by making predators less likely to initiate an attack or decreasing the chances of successful capture. For instance, those species of majoid that preferentially decorate with noxious or distasteful algae, sponges, and other invertebrates may make themselves repellent and more difficult for predators to handle (Cutress et al. 1970; Stachowicz and Hay 1999; Hultgren and Stachowicz 2011). Although most evidence suggests majoid crabs decorate to avoid predation, this conclusion is hindered by a lack of controlled empirical studies. To this end, we examined how external factors influence decorating using a little-studied tropical majoid crab, Camposcia retusa, as a model. Specifically, we first examined 1) whether the quantity and type of materials used to decorate varied across the exoskeleton. We then used a novel behavioral method to determine 2) whether nonhomogeneous decorating is the result of active behavioral selection and if C. retusa exhibits consistent decoration patterns between individuals and growth stages. Finally, to examine whether decorating in C. retusa is an antipredator response we then 3) tested if changes to environmental characteristics that could impact predation risk, that is, relative shelter availability, affected decoration rates. METHODS Study species and study location The spider decorator crab, C. retusa (family: Inachidae), is distributed throughout the Indo-West-Pacific. Adult crabs were collected from reefs in the Bohol Sea, Philippines by a commercial supplier (Live Aquaria Ltd.) and maintained in individual 7.6 L aquaria within an 800 L semi-recirculating sea seawater system at the University of Delaware during June and July 2016. Crabs were fed small pieces of shrimp daily; however, no food was provided on days when behavioral trials were conducted. Are decorations distributed uniformly over the exoskeleton, with regards to abundance or materials used? To determine whether decorations are nonrandomly distributed over the crab’s exoskeleton, a survey was conducted soon after collection. A total of 20 crabs were surveyed with each individual first placed into a white plastic container containing a 5-cm scale bar and water to 5-cm depth. A high-resolution photograph of the dorsal surface of the crab was taken and then it was released back to its aquarium. These photos were then analyzed using Adobe Photoshop (version CC 2015.5). First, the comparative size of each crab was determined as the maximum width across the carapace. For analysis of decorations, crabs were divided into 7 sections based on their anatomy; front chelipeds (CHP), first set of walking legs (WL1), second set of walking legs (WL2), third set of walking legs (WL3), fourth set of walking legs (WL4), forward half of the carapace, including the rostrum (CPT), and rear half of the carapace (CPB). For chelipeds and walking legs, left and right sides were grouped. For each crab, the presence or absence of decoration and the type of decoration material used was recorded at 5 randomly selected points on each section. Where decoration type could not be determined from photo a second inspection of the random point was conducted on the live crab. Decorations were grouped into the following categories; filamentous macroalgae, fleshy macroalgae, sponges (Porifera), and miscellaneous detritus. Differences in decoration abundance between sections was first determined using a logistic regression. This regression used a quasibinomial distribution with a logit-link function (function glm) for the presence-absence of total decoration (expressed as % by section), with Tukey’s post hoc comparisons (package multcomp). To determine if the composition of decoration materials differed between sections, multivariate analysis of variance (Manova) was used. This tested the differences in decoration amount, type, and location (section) and allowed the simultaneous comparison of decoration categories as multiple individual responses, in correlation to their location on the exoskeleton. Separate univariate statistics (generalized least squares, package nlme) were run for each decoration type to determine what sections had different amounts and composition of decorative materials. Finally, the presence or absence of a correlation between crab size and the % of exoskeleton that was decorated was determined using a Spearman’s rank-order correlation. Crab sizes varied between 13.96 and 23.49 mm max carapace width. Are nonuniform patterns of decoration abundance due to crab behavior? To examine if differences in decoration abundance or material distribution are mediated by crab behavior a manipulative decorating experiment was conducted. Trials were run within the 7.6 L aquaria. For each trial, a selection of red polyester pompoms (Creatology, Taiwan) were added to each tank at 7 AM. Five pompoms each of 3 diameters were added; small (5 mm), medium (10 mm), and large (15 mm). These pompoms were submerged in salt water for 24 h prior to use with only those pompoms that sunk subsequently used in trials. In addition to size, weight differences between pompom sizes was determined by weighing 20 individual pompoms of each size. Pompom wet weight did differ significantly; small (0.06 g ± 0.001), medium (0.89 g ± 0.05), large (2.0 g ± 0.05). Pompoms were scattered randomly around each container. Following 12 h, the number of pompoms attached to the crab’s exoskeleton, their location, and size of each pompom were recorded. Crab sizes varied between 15.07 and 24.54 mm max carapace width. All crabs used in experimental trials had a similar level of existing decoration cover (~75% of exoskeleton). Differences in the number of pompoms attached to each section of the exoskeleton were examined using a nested Anova (mixed effects model, nlme package; Pinheiro et al. 2017), using square-root transformed data, and post hoc comparison of means (multcomp package; Horthorn et al. 2008). The random effect in the model (crab ID) was also tested for significance. Differences in the size of pompoms attached to each section of the exoskeleton were examined using Manova, where the number of pompoms of each size was tested against each section where the pompoms were placed. Manova was selected as it can analyze multiple dependent variables, in this case, the abundance of various sizes of pompoms attached to the exoskeleton. Due to the small number of count data, data did not conform with the assumptions of multivariate normality and data was positively skewed. However, as Manova is generally robust to these assumptions (Erceg-Hurn and Mirosevich 2008), we chose to use multiple models (i.e., zero-inflated Poisson) to ensure that results aligned. We first checked for the presence of zeros to justify using zero-inflated models (Martin et al. 2005). We followed this with general linear models for the number of small, medium and large-sized pompoms attached using a Poisson distribution. Although these are univariate models, which do not account for the added complexity of the multivariate model, they are better at estimating the size and effect of the response without greater risk of interpreting the results with additional error. Transformations did not alleviate the violation of multivariate normality, as transforming data did not affect the spread of zero values (Martin et al. 2005). Finally, the Poisson distribution was selected after comparing separate Poisson, quasi-Poisson, negative binomial [package MASS, function glm.nb()] (Venables and Ripley 2002) and zero-inflated Poisson [package pscl, function zeroinfl()] (Zeileis et al. 2008) models. For both medium and large pompom sizes, the Poisson model performed best by 2 Akaike information criterion (AICc) units (package MuMIn). For the small pompom size, response the zero-inflated model performed best according to AICc, but for the sake of comparison only the results for the Poisson models were presented. However, these models are very similar to the others tested for the measures of the main effects. The presence or absence of a correlation between crab size and the % of exoskeleton that was decorated was again determined using a Spearman’s rank-order correlation. Is behaviorally mediated decorating influenced by shelter availability? To determine whether decorating in C. retusa is conducted to mitigate predation risk we conducted a second manipulative decorating experiment. In this experiment, shelter was either provided or absent. If crabs decorate less when shelter is available this would indicate that decorating is primarily done to induce an antipredator benefit. This experiment was also conducted in the 7.6 L aquaria. For shelter trials, shelter was provided in the form of a 5-cm diameter PVC elbow whereas this was not added for no-shelter trials. For each trial, 5 medium (10 mm) pompoms were added to the aquaria. Each hour for the following 12 h, the number of pompoms attached to the exoskeleton and their location was recorded. Differences in the number and rate of decorating between treatments was determined using a 2-pronged approach. First, a repeated measures Anova (RM-Anova) was used to determine how the number of attached pompoms changed as a function of shelter and time, controlling for individual variation with an additional error term. We then tested the main effect of presence/absence of shelter using linear mixed effect models, using a randomized block design for greater power with crab as the blocking variable (package nlme). There was some unequal variance between the no shelter and with shelter measurements, which was in part alleviated by a square-root transformation. In addition, although there was some deviation from normality there was no evidence of block within-block interactions with shelter and time (Tukey’s nonadditivity test). We compared multiple mixed-effects models with shelter and time as the main fixed effects and subject-level crabs as a random effect. These were 1) random intercept models where time was either a categorical or continuous factor (where the magnitude of the effect the same for each crab), 2) a linear mixed-effect model incorporating up to a third-order polynomial for time and a random-slope, and 3) a random-intercept model (where the magnitude of the effect of shelter did not need to be the same for each crab). Models were evaluated based on AIC, variance components were extracted using the function VarrCorr, and R2 values extracted using the function r.squaredGLMM.lme() (from the package MuMIn). All data analysis was completed using R version 3.3.2 and unless otherwise stated used the base package “stats.” Model assumptions were evaluated graphically. RESULTS Are decorations distributed uniformly over the exoskeleton with regards to abundance or materials used? Camposcia retusa exhibited nonuniform decorating with regards to both exoskeleton coverage and the types of decorations attached (Figure 1). Between the exoskeleton sections surveyed, CHP had the lowest mean % decoration cover, followed by (in increasing order) WL1, WL2, WL3/WL4/CPT (equal), with CPB the most decorated. Among these, CHP and WL1 were significantly different from both each other and all other sections (P < 0.05). However, CPB, CPT, WL2, WL3, WL4 were not significantly different from each other. There was no relationship between crab size and the % of total exoskeleton that was decorated (rs (18) = −0.02, P = 0.93). Figure 1 View largeDownload slide (a) Results of crab survey showing natural levels of variation in decoration coverage and materials used. Each bar represents one section of the exoskeleton: chelipeds (CHP), first set of walking legs (WL1), second set of walking legs (WL2), third set of walking legs (WL3), fourth set of walking legs (WL4), forward section of the carapace (CPT), and rear section of the carapace (CPB). Bar height represents mean % decoration cover (±SE), whereas different shades represent the relative composition of different decorating materials. Total number of crabs n = 20. Figure 1 View largeDownload slide (a) Results of crab survey showing natural levels of variation in decoration coverage and materials used. Each bar represents one section of the exoskeleton: chelipeds (CHP), first set of walking legs (WL1), second set of walking legs (WL2), third set of walking legs (WL3), fourth set of walking legs (WL4), forward section of the carapace (CPT), and rear section of the carapace (CPB). Bar height represents mean % decoration cover (±SE), whereas different shades represent the relative composition of different decorating materials. Total number of crabs n = 20. The various materials used to decorate were not distributed evenly over the exoskeleton (df = 6, Pillai’s = 0.89, approx F(24,532) = 6.33, P < 0.001), driven by the variable use of sponge and detritus (univariate, sponge: df = 6, SS = 110.0, MS = 18.34, F = 7.89, P < 0.001; detritus: df = 6, SS = 25.2, MS = 4.20, F = 4.38, P < 0.001). However, both algae types were distributed equally (univariate, filamentous: df = 6, SS = 11.5, MS = 1.92, F = 2.16, P = 0.05; fleshy: df = 6, SS = 2.90, MS = 0.483, F = 1.57, P = 0.16). With regards to the use of sponges, the highest mean % was observed over the carapace (55% CPB and 39% CPT, respectively) and WL2-4 (47–50%). Sponges were least abundant on the CHP (19%) and WL1 (27%). Meanwhile, detritus was present in low amounts (8–15%) on CHP, CPB and CPT but was not found at all on WL1-4. Filamentous algae was a significant proportion of WL3 and WL4 (t values = 2.52, P < 0.05 for both) although it was not significantly different across all sections (P > 0.05) and contributed a small % of the total decorations (0–15% of decorations per section). Finally, fleshy algae was a significant decoration on WL3 (t value = 2.28, P < 0.05) it was the least abundant of all the decoration types overall (0–8%), and was therefore not significantly different across the 7 body sections. Are nonuniform patterns of decoration abundance due to crab behavior? The nonuniform pattern of decorating in C. retusa appears to be largely a result of active choice by the crabs themselves, with the total number of pompoms placed on each body section varying significantly [df (6, 114), F-ratio = 159.9/36.07, P < 0.0001] (Figure 2). Highest rates of decoration were observed on the back 2 sets of walking legs with post hoc comparisons indicating that, whereas WL3 and WL4 means were significantly different from each other, they were both higher than and all other body sections (P < 0.05). The lowest levels of decorating were observed on the chelipeds, rear carapace, and first 2 sets of walking legs with CHL, CPB, WL1, and WL2 means not significantly different. Likewise, CPT and WL2 were not significantly different. In this model, the random effect of individual crabs was not found to be significant when comparing models of fixed-random or just fixed effects. AIC values were within 2 units indicating the similarity of both models. Figure 2 View largeDownload slide Results of the second manipulative experiment where crabs were offered artificial decorations (pompoms) of 3 sizes: 5-mm, 10-mm, and 15-mm (5 per size). Each bar represents one section of the exoskeleton: chelipeds (CHP), first set of walking legs (WL1), second set of walking legs (WL2), third set of walking legs (WL3), fourth set of walking legs (WL4), forward section of the carapace (CPT), and rear section of the carapace (CPB). Bar height represents mean number of decorations added (±SE), whereas different shades represent the relative composition of different decoration sizes. Total number of crabs n = 20. Figure 2 View largeDownload slide Results of the second manipulative experiment where crabs were offered artificial decorations (pompoms) of 3 sizes: 5-mm, 10-mm, and 15-mm (5 per size). Each bar represents one section of the exoskeleton: chelipeds (CHP), first set of walking legs (WL1), second set of walking legs (WL2), third set of walking legs (WL3), fourth set of walking legs (WL4), forward section of the carapace (CPT), and rear section of the carapace (CPB). Bar height represents mean number of decorations added (±SE), whereas different shades represent the relative composition of different decoration sizes. Total number of crabs n = 20. Manova indicated that the size of pompoms used to decorate was also significantly different between body sections, for each of the 3 pompom sizes (df 6, 133, Pillai = 0.90, P < 0.001) (Figure 2). However, there were no significant differences seen when sizes were analyzed separately using Poisson-distribution models, likely due to the small means and sample size. From the raw data, the use of small pompoms was more evenly spread across the CPT and WL2-4, whereas medium and large pompoms were only used on both WL3 and WL4. Checking for zero-inflation against a Poisson distribution, 10–12% more zeros were observed than was expected for the small-sized pompoms, and 7% more zeros were observed than expected for medium-sized pompoms; however, estimates of large-sized pompoms were not overly represented in the data. Is behaviorally mediated decorating influenced by shelter availability? Crabs with no shelter decorated more than crabs with shelter F(1, 375) = 9.77, P < 0.01 (Figure 3). Crabs with no shelter used approximately 17% more decorations; however, there was a high amount of variation between individuals. The random-slope, random-intercept model, accounting for between-individual variation for the effect of shelter on decoration was by far the best model with the greatest statistical power. It ranked significantly higher than the next ranking model (the polynomial) by comparing loglikelihood ratios. R2 conditional (how much the model explained) was high at 85%, R2 marginal (the percent explained by fixed factors) was just 20%. According to the best-fitting model, the proportion of the variance was approximately 43% (individual), 46% (with/without shelter) and 11% (residual), respectively with the cross-level effects. In other models that only incorporated random effects, variance due to individual crabs was high (58–60%) whereas the proportion of the variance due to the fixed effects of shelter and time was lower (40–42%). Figure 3 View largeDownload slide Results of manipulative experiment where crabs were offered five 10-mm pompoms in the presence or absence of shelter, with the subsequent number of pompoms attached recorded each hour for a total of 12 h. Total number of crabs n = 16, with each individual tested both with and without shelter in a random order. Figure 3 View largeDownload slide Results of manipulative experiment where crabs were offered five 10-mm pompoms in the presence or absence of shelter, with the subsequent number of pompoms attached recorded each hour for a total of 12 h. Total number of crabs n = 16, with each individual tested both with and without shelter in a random order. DISCUSSION Camposcia retusa exhibits a nonuniform, yet consistent, pattern of decorating with regards to both the materials used and their placement over the exoskeleton. Decoration patterns appear to be driven by crab behavior, with decorating intensity varying depending on shelter availability. Crabs collected from the wild had significantly less material on the chelipeds and first set of walking legs compared to the remaining ventral surface of the exoskeleton. Lower rates of decorating may be due to the dominant setal morphology on these appendages, or due to differences in appendage function. For example, microscopic examination of the exoskeleton suggests that cheliped setal composition is primarily composed of the straight form used as tactile sensory appendages rather than the hooked forms associated with decorating (Figure 4) (Berke and Woodin 2009). In addition, as the chelipeds are principally used for feeding, sorting, communication, and other non-locomotory tasks (Mariappan et al. 2000), attaching substantial amounts of material to these appendages could impede these essential functions and reduce the foraging efficiency of individuals. The tendency for C. retusa to have highly decorated walking legs differs from many majoids; however, this is consistent with what is seen in some closely-related species from within its family (Inachidae), for instance Podochela hemphilli (Hultgren and Stachowicz 2011), suggesting this decoration pattern has an evolutionary basis. Camposcia retusa also differed from many majoids in having a fairly uniform amount of material over the whole carapace. In contrast, many species decorate disproportionately on the front section, ostensibly to conceal the rostrum and antennae (Hultgren and Stachowicz 2009). In many majoids, a decline in decoration abundance is seen with increasing size (Hultgren and Stachowicz 2011). However, this was not observed in C. retusa, indicating the drive to decorate remains high at all ontogenetic stages. If decorating in C. retusa is primarily an antipredator defense, this suggests crabs remain vulnerable to local predators throughout their lifespan. Figure 4 View largeDownload slide Detail of C. retusa exoskeleton showing variation in setal morphology. (a) Detail of setae located on the chelipeds (CHP) (i) and rostrum at the anterior of the carapace (CPT) (ii). Note the similar straight structure often attributed to setae used primarily as sensory appendages. (b) Detail of setae located on the rear walking legs (WL4) (iii). Note the hooked structure often attributed to setae used primarily for attaching decorations. (c) Detail of setae located on the carapace (CPB) (iv). Note the primarily hooked structure. (d) Detail of hooked setae used for decorating. Scale bars indicate either 5 mm (a–c) or 1 mm (d). Figure 4 View largeDownload slide Detail of C. retusa exoskeleton showing variation in setal morphology. (a) Detail of setae located on the chelipeds (CHP) (i) and rostrum at the anterior of the carapace (CPT) (ii). Note the similar straight structure often attributed to setae used primarily as sensory appendages. (b) Detail of setae located on the rear walking legs (WL4) (iii). Note the hooked structure often attributed to setae used primarily for attaching decorations. (c) Detail of setae located on the carapace (CPB) (iv). Note the primarily hooked structure. (d) Detail of hooked setae used for decorating. Scale bars indicate either 5 mm (a–c) or 1 mm (d). A variety of materials, all common components of benthic reef environments, were found attached to the exoskeleton of wild crabs, including sponges, fleshy macroalgae, filamentous macroalgae, as well as miscellaneous unidentifiable detritus. Each of these categories contained a variety of species; however, these could not be identified. Of these, sponges were the primary material used across the exoskeleton suggesting crabs exhibit a decoration preference, assuming all materials were equally abundant. Decoration preferences for sponges have been observed in a number of majoids, including several inachids (reviewed in Hultgren and Stachowicz 2011). Assuming decorating in C. retusa is an antipredator response, sponges and algae would make effective decorations; as many are chemically defended they could mask an individual’s odor or make it distasteful to predators (Stachowicz and Hay 1999; Hultgren and Stachowicz 2011). The highly variable, 3-dimensional growth form of these sessile organisms would also effectively break up an individual’s outline, increasing the chances of nonrecognition or misidentification by predators (including, potentially, other crabs). However, as sponges and algae can grow rapidly it is not known if individuals select these materials at a specific size and to what degree crabs control their growth once attached. In addition to sponges and algae, detritus and other incidental materials could also increase an individual’s overall similarity to the surrounding sensory environment, improving the effectiveness of any background matching camouflage. The results of our first manipulative experiment provide further evidence that the nonrandom decoration patterns observed in wild crabs are due to selective placement of materials. When given a selection of artificial decorations of 3 standardized sizes, crabs decorated most heavily on the rear legs, with no decorations placed on the front sets of walking legs or chelipeds, in congruence with natural decoration patterns. In addition, crabs were selective with regards to decoration size; although all decorations were added to the rear walking legs only the smallest decorations were added to the carapace. Material selection based on physical attributes may have several underlying drivers. For instance, crabs may differentiate between materials based on relative weight, with heavier decorations placed on the bigger, presumably stronger, rear walking legs to reduce the energetic costs associated with carrying them. Placing larger decorations onto these dominant walking legs, and prioritizing their decoration in general, could also provide the greatest camouflage benefit by breaking up a crab’s overall outline most effectively, given that these appendages make up a substantial proportion of a crab’s overall size. The prioritization of decorating specific body parts also appears to reflect how this species interacts with its environment. For instance, when shelter is available crabs will generally remain stationary at the opening with only the rear walking legs exposed (E. C. Muñoz Ruiz Pers. Obs.). Presumably this position maximizes their ability to observe their environment while reducing investment in vigilance, but may further explain why the rear appendages are decorated most heavily. Similarly, the need to access shelter may explain why crabs avoid placing large decorations onto their carapace as these items will increase the maximum depth of the crab and have the unwanted effect of restricting the minimum diameter of shelter crabs can access. Placing larger decorations onto the legs may not induce the same restrictions due to their high degree of articulation and comparatively thin size. No decorations were placed on the front chelipeds by any crabs during the experiment which further suggests this species does not prioritize decorating on these appendages. Although patterns of decorating are consistent between individuals, investment in decorating appears to be plastic and reflects surrounding environmental conditions. Greater rates of decorating were seen when access to shelter was limited. This suggests decorating intensity is dependent on ambient risk, as an inability to access shelter would increase the proportion of time a crab is exposed and subsequently increase vulnerability to predation. Lack of shelter may also extend the proportion of time crabs spend actively searching, increasing their chances of encountering potential decorations. However, although this result indicates that decorating is an antipredator response, the exact mechanism or mechanisms that are being employed remain to be determined. Given that individuals decorate with a wide range of items found on the seabed, leading to an overall mottled appearance, it is highly likely that crabs employ a form of camouflage that limits initial perception by predators by blending into their environment, such as background pattern matching, rather than a form that increases the chances of misidentification as a nonedible item following perception, such as masquerade (Skelhorn et al. 2010; Ruxton and Stevens 2015). The reliance on potentially noxious or chemically defended materials, such as sponges and algae, suggests they may also decorate to deter predators post-recognition (Stachowicz and Hay 1999; Hulgren and Stachowicz 2011). Additional experimentation is needed to determine the underlying mechanism for decorating in this species and further examine the degree to which decorating is context dependent. Future work should also examine what sensory cues crabs use to determine relative risk and if they preferentially select materials based on their visual or chemical properties. In addition to providing insight into decorating in C. retusa, this study also demonstrates the value of controlled experiments using artificial decorations as a tool for examining behaviorally mediated decorating in animals. Camposcia retusa exhibits consistent yet nonrandom patterns on decorating, both in the types of materials used and their distribution, suggesting that decorating in this species is constrained by a combination of behavioral and morphological factors. Our data shows that decorating in C. retusa is primarily conducted to mitigate predation risk with investment in decorating dependent on ambient conditions. FUNDING This work was supported by the National Science Foundation Research Experiences for Undergraduates Program. Data accessibility: Analysis reported in this article can be reproduced using the data provided by Brooker et al. (2017). REFERENCES Berke SK, Woodin SA. 2009. Behavioral and morphological aspects of decorating in Oregonia gracilis (Brachyura: Majoidea). Invert Biol . 128: 172– 181. Google Scholar CrossRef Search ADS Berke SK, Miller M, Woodin SA. 2006. Modelling the energy-mortality trade-offs of invertebrate decorating behaviour. Evol Ecol Res . 8: 1409– 1425. Brandt M, Mahsberg D. 2002. Bugs with a backpack: the function of nymphal camouflage in the West African assassin bugs Paredocla and Acanthaspis spp. Anim Behav . 63: 277– 284. Google Scholar CrossRef Search ADS Briffa M, Elwood RW. 2005. Metabolic consequences of shell choice in Pagurus bernhardus: do hermit crabs prefer cryptic or portable shells? Behav Ecol Sociobiol . 59: 143– 148 Google Scholar CrossRef Search ADS Brooker RM, Muñoz Ruiz EC, Sih TL, Dixson DL. 2017. Data from: Shelter availability mediates decorating in the majoid crab, Camposcia retusa. Behav Ecol . http://dx.doi.org/10.5061/dryad.602hq Cutress C, Ross DM, Sutton L. 1970. The association of Calliactis tricolor with its pagurid, calappid, and majid partners in the Caribbean. Can J Zool . 48: 371– 376. Google Scholar CrossRef Search ADS Dumont CP, Drolet D, Deschênes I, Himmelman JH. 2007. Multiple factors explain the covering behaviour in the green sea urchin, Strongylocentrotus droebachiensis. Anim Behav . 73: 979– 986. Google Scholar CrossRef Search ADS Erceg-Hurn DM, Mirosevich VM. 2008. Modern robust statistical methods: an easy way to maximize the accuracy and power of your research. Am Psychol . 63: 591– 601. Google Scholar CrossRef Search ADS PubMed Fürböck S, Patzner RA. 2005. Decoration preferences of Maja crispata Risso 1827 (Brachyura, Majidae). Nat Croat . 14: 175– 184. Hazlett BA, Estabrook GF. 1974. Examination of agonistic behavior by character analysis. I. The spider crab Microphrys bicornutus. Behaviour . 48: 131– 144. Google Scholar CrossRef Search ADS Hothorn T, Bretz F, Westfall P. 2008. Simultaneous inference in general parametric models. Biom J . 50: 346– 363. Google Scholar CrossRef Search ADS PubMed Hultgren KM, Stachowicz JJ. 2008. Alternative camouflage strategies mediate predation risk among closely related co-occurring kelp crabs. Oecologia . 155: 519– 528. Google Scholar CrossRef Search ADS PubMed Hultgren KM, Stachowicz JJ. 2009. Evolution of decoration in majoid crabs: a comparative phylogenetic analysis of the role of body size and alternative defensive strategies. Am Nat . 173: 566– 578. Google Scholar CrossRef Search ADS PubMed Hultgren KM, Stachowicz JJ. 2011. Camouflage in decorator crabs. Integrating ecological, behavioural and evolutionary approaches. In: Stevens M, Merilaita S, editors. Animal camouflage mechanisms and function . Cambridge, UK: Cambridge University Press. p. 212– 236. Hultgren KM, Thanh PD, Sato M. 2006. Geographic variation in decoration selectivity of Micippa platipes and Tiarinia cornigera in Japan. Mar Ecol Prog Ser . 326: 235– 244. Google Scholar CrossRef Search ADS Jackson RR, Pollard SD. 2007. Bugs with backpacks deter vision-guided predation by jumping spiders. J Zool . 35: 358– 363. Google Scholar CrossRef Search ADS Limm MP, Power ME. 2011. The caddisfly Dicosmoecus gilvipes: making a case for a functional role. J N Am Benthol Soc . 30: 485– 492. Google Scholar CrossRef Search ADS Mariappan P, Balasundaram C, Schmitz B. 2000. Decapod crustacean chelipeds: an overview. J Biosci . 25: 301– 313. Google Scholar CrossRef Search ADS PubMed Martin TG, Wintle BA, Rhodes JR, Kuhnert PM, Field SA, Low-Choy SJ, Tyre AJ, Possingham HP. 2005. Zero tolerance ecology: improving ecological inference by modelling the source of zero observations. Ecol Lett . 8: 1235– 1246. Google Scholar CrossRef Search ADS PubMed Martinelli M, Calcinai B, Bavestrello G. 2006. Use of sponges in the decoration of Inachus phalangium (Decapoda, Majidae) from the Adriatic Sea. Ital J Zool . 73: 347– 353. Google Scholar CrossRef Search ADS McClintock JB, Janssen J. 1990. Pteropod abduction as a chemical defence in a pelagic Antarctic amphipod. Nature . 346: 462– 464. Google Scholar CrossRef Search ADS Mondy N, Cathalan E, Hemmer C, Voituron Y. 2011. The energetic costs of case construction in the caddisfly Limnephilus rhombicus: direct impacts on larvae and delayed impacts on adults. J Insect Physiol . 57: 197– 202. Google Scholar CrossRef Search ADS PubMed Mondy N, Rey B, Voituron Y. 2012. The proximal costs of case construction in caddisflies: antioxidant and life history responses. J Exp Biol . 215: 3453– 3458. Google Scholar CrossRef Search ADS PubMed Otto C. 2000. Cost and benefit from shield cases in caddis larvae. Hydrobiologia . 436: 35– 40. Google Scholar CrossRef Search ADS Ruxton GD. 2009. Non-visual crypsis: a review of the empirical evidence for camouflage to senses other than vision. Phil Trans R Soc B . 364: 549– 557. Google Scholar CrossRef Search ADS PubMed Ruxton GD, Stevens M. 2015. The evolutionary ecology of decorating behaviour. Biol Lett . 11: 20150325. Google Scholar CrossRef Search ADS PubMed Pinheiro J, Bates D, DebRoy S, Sarkar D; R Core Team. 2017. nlme: linear and nonlinear mixed effects models. R package version 3.1–131. Available from: https://CRAN.Rproject.org/package=nlme Stachowicz JJ, Hay ME. 1999. Reducing predation through chemically mediated camouflage: indirect effect of plant defences on herbivores. Ecology . 80: 2085– 2101. Google Scholar CrossRef Search ADS Skelhorn J, Rowland HM, Speed MP, Ruxton GD. 2010. Masquerade: camouflage without crypsis. Science . 327: 51. Google Scholar CrossRef Search ADS PubMed Szebeni T, Hartnoll RG. 2005. Structure and distribution of carapace setae in British spider crabs. J. Nat. Hist . 39: 3795– 3809. Google Scholar CrossRef Search ADS Thanh PD, Wada K, Sato M, Shirayama Y. 2003. Decorating behaviour by the majid crab Tiarinia cornigera as protection against predators. J Mar Biol Assoc UK . 83: 1235– 1237. Google Scholar CrossRef Search ADS Thanh PD, Wada K, Sato M, Shirayama Y. 2005. Effects of resource availability, predators, conspecifics and heterospecifics on decorating behaviour by the majid crab Tiarinia cornigera. Mar Biol . 147: 1191– 1199. Google Scholar CrossRef Search ADS Venables WN, Ripley BD. 2002. Modern applied statistics with S . New York: Springer. Google Scholar CrossRef Search ADS Wicksten MK. 1983. Camouflage in marine invertebrates. Oceanogr Mar Biol . 21: 177– 193. Woods CMC, Mclay CL. 1994. Use of camouflage materials as a food store by the spider crab Notomithrax ursus (Brachyura: Majidae). NZ J Mar Fresh Res . 28: 97– 104. Google Scholar CrossRef Search ADS Zeileis A, Kleiber C, Jackman S. 2008. Regression models for count data in R. J Stat Softw . 27: 8. © The Author(s) 2017. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: email@example.com
Behavioral Ecology – Oxford University Press
Published: Jan 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera