Different ontogenetic trajectories of body colour, pattern and crypsis in two sympatric intertidal crab species

Abstract

Animals frequently exhibit great variation in appearance, especially in heterogeneous habitats where individuals can be concealed differentially against backgrounds. Although background matching is a common anti-predator strategy, gaps exist in our understanding of within- and among-species variation. Specifically, the drivers of changes in appearance associated with habitat use and occurring through ontogeny are poorly understood. Using image analysis, we tested how individual appearance and camouflage in two intertidal crab species, the mud crab Panopeus americanus and the mottled crab Pachygrapsus transversus, relate to ontogeny and habitat use. We predicted that both species would change appearance with ontogeny, but that resident mud crabs would exhibit higher background similarity than generalist mottled crabs. Both species showed ontogenetic changes; the mud crabs became darker, whereas mottled crabs became more green. Small mud crabs were highly variable in colour and pattern, probably stemming from the use of camouflage in heterogeneous habitats during the most vulnerable life stage. Being habitat specialists, mud crabs were better concealed against all backgrounds than mottled crabs. Mottled crabs are motile and generalist, occupying macroalgae-covered rocks when adults, which explains why they are greener and why matches to specific habitats are less valuable. Differential habitat use in crabs can be associated with different coloration and camouflage strategies to avoid predation.

INTRODUCTION

The remarkable diversity of animal coloration has fascinated evolutionary biologists for centuries, being used as key examples of adaptation and natural selection (Darwin, 1859; Wallace, 1867; Kettlewell, 1955). The colour expressed by an animal can intercede in important processes throughout its life, including social signalling during mate choice, thermoregulation and defence against predators (Cuthill et al., 2017). One of the most common and widespread anti-predator strategies mediated by coloration in nature is camouflage, which works by reducing the probability of detection or recognition of prey by the visual system of predators (Stevens & Merilaita, 2009). Although many different camouflage strategies have already been described and tested, especially in artificial systems, including disruptive coloration (Cuthill et al., 2005), countershading (Rowland et al., 2007) and masquerade (Skelhorn et al., 2010), most of the work to date has focused on the most familiar and intuitive type of camouflage, known as background matching. In this strategy, there is a match between the general appearance of the individual, in terms of colour, brightness and/or pattern, and the background (Stevens & Merilaita, 2009). This leads to a reduction in the detection of well-concealed individuals by visual predators and, ultimately, increases the chances of survival of camouflaged prey (Duarte et al., 2018).

Colour is frequently used as a species-specific trait, but many animal populations exhibit considerable variation among individuals (i.e. intraspecific variation). In such cases, populations can be characterized by discrete or continuous colour variation, which can result in polymorphic (i.e. presenting genotypic and phenotypic variation) or polyphenic populations (i.e. exhibiting only phenotypic variation) (West-Eberhard, 1989). Intraspecific colour variation has been described for many species belonging to different taxa, from invertebrates (Reimchen, 1989; Krause-Nehring et al., 2010; Eacock et al., 2017) to vertebrates (Cheney et al., 2009; Calsbeek et al., 2010; Passarotto et al., 2018), and can be maintained in populations by different selective processes, such as assortative mating, differential niche use, environmental heterogeneity and frequency-dependent (i.e. apostatic) selection, usually guided by visual predation (Bond & Kamil, 2002, 2006; Gray & McKinnon, 2007). Differential coloration among individuals would be particularly important for species associated with heterogeneous backgrounds and could be driven by a wide range of evolutionary processes, such as differential concealment of colour types against contrasting microhabitats (Stevens et al., 2015; Nokelainen et al., 2017; Duarte et al., 2018). This could include the use of specific camouflage strategies on each type of background (Price et al., 2019) or as a means to defeat predator search images (Bond & Kamil, 2002). In the common shore crab Carcinus maenas, for example, juveniles are highly variable in terms of colour and brightness (Todd et al., 2006; Stevens et al., 2014). However, crabs inhabiting homogeneous mudflats are more uniform and match the background closely. In contrast, crabs from rock pools, where substrates are very colourful and heterogeneous, are more variable in brightness and do not match the background well, but instead exhibit much higher levels of disruptive coloration (Price et al., 2019).

Camouflage against spatially or temporally heterogeneous backgrounds can be improved either by oriented behavioural choices towards colour-matching substrates or by mechanisms of colour change (Kang et al., 2015; Duarte et al., 2017; Eacock et al., 2017; Green et al., 2019; Stevens & Ruxton, 2019), allowing the individuals to cope with environmental changes occurring over a range of spatial and temporal scales (Caro et al., 2016; Duarte et al., 2017). Besides flexibly altering coloration over multiple timescales for concealment (Duarte et al., 2017), many animal species also change appearance through ontogeny (Reimchen, 1989; Booth, 1990; Wilson et al., 2007; Todd et al., 2009; Hultgren & Stachowicz, 2010; Jensen & Egnotovich, 2015; Nokelainen et al., 2019). A variety of intertidal crab species, for example, exhibit remarkable colour and pattern variation as juveniles but not during the adult phase (Palma et al., 2003; Todd et al., 2009; Caro, 2018; Nokelainen et al., 2019). High colour variability can allow individuals to conceal themselves differentially against specific patches of the heterogeneous habitats on which they reside during this life stage (Palma & Steneck, 2001; Stevens et al., 2014) or to defeat search images in visual predators (Krause-Nehring et al., 2010; Karpestam et al., 2014). Ontogenetic colour change is commonly associated with a shift in habitat use, with small crabs occupying highly heterogeneous backgrounds but moving to less diverse substrates as adults (Palma & Steneck, 2001; Todd et al., 2006). Moreover, in many crab species, adults become more active than juveniles and frequently move across habitat types, potentially adopting a generalist camouflage strategy (i.e. compromise coloration; Hughes et al., 2019). Here, individual concealment is not directed towards a specific background but works as an efficient way to improve camouflage against a wide range of backgrounds (Nokelainen et al., 2019). Not mutually exclusively, ontogenetic colour shifts can also occur owing to a reduction in vulnerability to predation in the adult phase, because, after achieving a size refuge, larger crabs would have a reduced need for camouflage (Palma & Steneck, 2001; Krause-Nehring et al., 2010).

In crabs, phenotype–environment associations have been demonstrated in several species, across multiple spatial scales, especially as juveniles (Palma & Steneck, 2001; Palma et al., 2003; Todd et al., 2006; Krause-Nehring et al., 2010; Stevens et al., 2014; Nokelainen et al., 2017). However, whether such associations improve camouflage are still poorly explored (but see Jensen & Egnotovich, 2015; Russell & Dierssen, 2015; Nokelainen et al., 2017; Price et al., 2019). Additionally, most studies have focused on a restricted number of crab species, mainly from temperate areas, and used subjective estimates to measure individual coloration and camouflage (Palma & Steneck, 2001; Palma et al., 2003; Todd et al., 2006; Krause-Nehring et al., 2010). This results in a knowledge gap concerning intraspecific colour variation, crypsis and ontogenetic shifts in tropical and subtropical species (but see, for example, Hemmi et al., 2006; Detto et al., 2008; Stevens et al., 2013; Carvalho-Batista et al., 2015). Therefore, it is necessary to expand our knowledge of animal camouflage to those poorly known species and diverse ecosystems where predation is intense (Schemske et al., 2009; Roslin et al., 2017), as a way to understand whether the patterns of variation and camouflage, in addition to the underlying ecological processes described for temperate species, are commonplace in nature.

The Araçá Bay, located in the northern coast of São Paulo, Brazil, is a large tidal flat of high biological diversity (Nucci et al., 2001; Amaral et al., 2010; Dias et al., 2018), composed of different types of substrates, ranging from large rocks, frequently covered by ephemeral green algae (Ulva spp.), to coarse sand and fine silt/clay sediments (Amaral et al., 2010). Two of the most common crab species found at this site are the narrowback mud crab, Panopeus americanus Saussure, 1857 (Decapoda, Panopeidae) (hereafter, ‘mud crab’; Fig. 1A), and the mottled shore crab, Pachygrapsus transversus Gibbes, 1850 (Decapoda, Grapsidae) (henceforth, ‘mottled crab’; Fig. 1B), which exhibit remarkable variation in colour and pattern on their carapace. Both species occupy the intertidal zone of rocky substrates and muddy beaches in estuarine and mangrove areas, exhibiting ontogenetic changes in habitat use, with juveniles occupying different substrate types from adults (Abele et al., 1986; Flores & Negreiros-Fransozo, 1999; Vergamini & Mantelatto, 2008a, b; García et al., 2016). There are marked differences in habitat use and life-history traits between mud and mottled crabs (Table 1), but ontogenetic habitat shifts might broadly affect concealment to background colour and texture in individuals of both species. Here, we used digital image analysis to test whether the appearance and crypsis of mud and mottled crabs inhabiting different microhabitats (Fig. 1C) vary according to the size (i.e. age) of the individuals. We predict substantial changes in both species, but high background similarity in mud crabs, because they are less active, not moving far from their shelter (Micheli, 1997; Vergamini & Mantelatto, 2008a; Carvalho-Batista et al., 2015), and low overall concealment in mottled crabs, which move faster and over larger foraging areas, making them habitat generalists (Abele et al., 1986; Christofoletti et al., 2010).

MATERIAL AND METHODS

Sampling and photography

Sampling was conducted at Araçá Beach and Ilha Pernambuco at the Araçá Bay, located in Southeast Brazil (23°48′78.1″S, 45°24′46.9″W). We concentrated our sampling on a single day in August 2016 during the low tide, when we searched for mud crabs (Panopeus americanus; Fig. 1A) and mottled crabs (Pachygrapsus transversus; Fig. 1B) in an area of ~200 m². Crabs of all sizes were collected manually while removing boulders, pebbles and searching over gravel beds. Collected individuals (61 mud crabs and 53 mottled crabs) were housed in plastic containers filled with seawater and small rock pieces and taken to the laboratory, where they were photographed (see details below). In order to quantify the colour range of the different microhabitats used by the two crab species, we obtained in the field 24 photographs of three clearly distinct background types that are found in the study area (Fig. 1C): (1) large rock covered by green algae (hereafter ‘alga’, N = 8); (2) muddy areas containing small rocks (hereafter ‘mud rock’, N = 8); and (3) sandy areas covered by small pebbles (hereafter ‘pebbles’, N = 8).

We used a Nikon D80 digital camera coupled with a Nikkor 60 mm lens and a UV-blocker filter (62 mm; Tiffen, USA) to photograph backgrounds in the field from a fixed distance of 1.5 m and crabs in an external area of the Centre for Marine Biology (CEBIMar-USP, São Sebastião, Brazil) using a copy stand for photography under natural illumination conditions. Images were taken in RAW format, with manual white balancing and fixed aperture settings to avoid over-exposure (Stevens et al., 2007), and included a ruler, in addition to black (7.5%) and white (91%) Spectralon reflectance targets (Labsphere, Congleton, UK), following current standard procedure (Troscianko & Stevens, 2015). Carapace width (CW) was measured using the ruler included in the photographs.

All images were first linearized based on curves modelled from eight Spectralon standards ranging from 2 to 99% of reflectance to correct for camera non-linear responses to light intensity (Troscianko & Stevens, 2015). Next, photographs were equalized for changes in light conditions using the black and white standards and saved as 32-bit multispectral images. Image channels were then scaled to reflectance values, where an image value of 255 on an eight-bit scale equals 100% reflectance (Stevens et al., 2007). After all these procedures, images corresponded to the physical reflectance properties of crabs and backgrounds in three parts of the spectrum [long wavelength (LW), medium wavelength (MW) and short wavelength (SW)] and could be used for analysis of coloration and pattern. All routines were performed using functions from the ‘Multispectral Image Calibration and Analysis Toolbox’ implemented in ImageJ software (Rasband, 1997; Troscianko & Stevens, 2015).

Measurement of crab colour and pattern

For each multispectral image, we selected regions of interest (crab carapace and entire background, avoiding areas of specular reflectance), from which we extracted reflectance values that were used to calculate several metrics of appearance based on brightness, colour and pattern. Similar to other studies on crabs (Detto et al., 2008; Stevens et al., 2013, 2014), we preferred to work with normalized reflectance data instead of using visual modelling because there is a wide range of potential predators from different taxonomic groups (e.g. other crabs, fish and birds) consuming crabs in the Araçá Bay, and existing information about their identity is scarce and mostly anecdotal (Carvalho-Batista et al., 2015).

Brightness was calculated as (LW + MW + SW)/3 and is a simple achromatic measure of how dark or bright crabs are across the entire spectrum (Stevens et al., 2014). The colour metrics saturation and hue were also calculated. Saturation is generally assumed to be the amount of a given colour in relationship to white light (i.e. the colour richness) and was calculated by transforming the standardized LW, MW and SW reflectance values (i.e. proportional value to the summed reflectance across the entire spectrum) to x–y coordinates of a trichromatic colour space (i.e. Maxwell triangle). Saturation was considered as the shortest distance from the given colour point to the achromatic centre of the space, with larger values representing greater saturation (Kelber et al., 2003). For hue, we first conducted a principal components analysis to define the main axis of colour variation for both crab species and used this to determine a logical colour channel (as in the studies by Green et al., 2019; Nokelainen et al., 2019). Principal components analysis was applied to the standardized reflectance values from the three reflectance channels, and hue was defined further as the ratio MW/(SW + LW), which is a simple measurement of medium wavelength vs. the two extremes of the light spectrum, broadly analogous to an opponent-style colour channel (Spottiswoode & Stevens, 2011).

Besides colour, mud and mottled crabs also differ greatly in pattern, exhibiting colour spots of different shapes and sizes on their carapace (Fig. 1A, B). We adopted the widely used and well-established ‘granularity’ analysis approach to measure pattern in the two crab species. This method is based on the decomposition of an image into a series of different spatial frequencies (i.e. granularity bands) using Fourier analysis and band-pass filtering (Barbosa et al., 2008; Stoddard & Stevens, 2010; Stevens et al., 2014), followed by the determination of the relative contribution of different marking sizes to the overall pattern. In this analysis, the amount of information (‘energy’) is calculated from markings of different sizes, starting with small markings (i.e. formed by few pixels) and increasing in size to larger markings. We used a log-scale set-up, with a starting size of two pixels and a log multiplier of 1.414 increment up to a maximum of 4096 pixels, where no pattern energy was observed. Next, for each granularity band, we calculated the overall pattern ‘energy’ as the sum of the squared pixel values divided by the total number of pixels (Barbosa et al., 2008). Finally, after processing all images, we calculated three different metrics of pattern from each granularity spectrum (i.e. each decomposed image), namely: (1) maximum frequency (i.e. the spatial frequency with the highest energy, which corresponds to the dominant marking size); (2) summed energy (i.e. the energy summed across all scales, which is a measure of pattern contrast); and (3) proportional energy (i.e. the energy at the maximum frequency divided by the summed energy, which is a measure of pattern diversity) (Chiao et al., 2009; Stoddard & Stevens, 2010).

Measurement of crab background matching

In order to measure the degree of concealment of both mud and mottled crabs against their main habitats in the Araçá Bay, we used the colour values extracted from the carapace and from backgrounds, categorized into the three broad types: ‘alga’, ‘mud rock’ and ‘pebble’ (Fig. 1C). For that, we initially converted the camera colour channel values of crabs and backgrounds to x–y coordinates of a two-dimensional colour space, where each colour is expressed as a single point (Kelber et al., 2003). We then calculated the Euclidean distance between each crab coordinate and the eight replicate values of each of the three backgrounds, which were averaged to a single value. This provides a receiver-independent estimate of the degree of background match of the two crab species across the different habitat types, for which lower distances indicate more similar coloration between crabs and backgrounds.

Classification of size groups

In order to understand how the appearance and crypsis of crabs of both species change with ontogeny, we classified individuals as small or large based on the size at which individuals spend more time on alternative foraging habitats and, in the case of mottled crabs, become more active. This behavioural change takes place, approximately, when individuals reach 14 mm CW in the case of mud crabs (Carvalho-Batista et al., 2015) and 13 mm in the case of mottled crabs (Abele et al., 1986). We then compared how effectively crabs fell into these categories based on the appearance metrics we measured (e.g. brightness, saturation, hue, maximum frequency, summed energy and proportional energy) using discriminant function analysis (DFA).

Statistical analyses

All statistical analyses were undertaken using the software R v.4.0.0 (R Core Team, 2020). We first used the ‘lda’ function from the ‘MASS’ package in R (Venables & Ripley, 2002) to run DFA in order to validate the size categories we had chosen (i.e. small or large) for both crab species based on individual appearance. Brightness and all colour (saturation and hue) and pattern metrics (maximum frequency, summed energy and proportional energy) were compared individually between crab species (mud or mottled crabs) and size classes (small or large) using a two-way analysis of variance (ANOVA) to test for possible species-specific ontogenetic trajectories of colour and pattern. A linear mixed-effects model was applied to the estimates of background similarity (i.e. colour distances between crabs and backgrounds), with background types (alga, mud rock or pebble), crab species (mud or mottled crabs) and size classes (small or large) as fixed between-subjects factors, and crab identity as a random factor to control for repeated measurements on the same individual, because each crab was compared with all background types. The ANOVA model was fitted using the ‘aov’ function in the ‘base’ package, whereas the mixed-effects model was fitted using the ‘lmer’ function in the ‘lme4’ package (Bates et al., 2015) and the associated significance tests through the ‘anova’ function in the ‘lmerTest’ package (Kuznetsova et al., 2017). Model residuals were checked visually for normal error distribution using histograms and q-q plots, and the homogeneity of variances was tested using the Bartlett test in R, for which brightness and all pattern metrics required natural log transformation to meet model assumptions. For the estimates of hue and background similarity, variances remained heterogeneous even after natural log transformation. However, given that our sample size is large (minimum sample size = 20), making the models robust to variance heterogeneity (Underwood, 1997), we decided to run both models anyway using the raw data. Finally, in the case of significant effects, Tukey’s post hoc test was applied to compare mean differences between factor levels using the ‘emmeans’ function from the ‘emmeans’ package (Lenth, 2019).

RESULTS

Distinctiveness in the appearance of small and large crabs

The size of mud crabs ranged from 5.8 to 32.2 mm CW (mean ± SD: 14.7 ± 6.2 mm) and the size of mottled crabs from 3.6 to 23.5 mm CW (11.5 ± 4.4 mm). The discriminant function analyses based on appearance metrics validated the classification of crabs of both species into the two size classes previously determined. In the case of mud crabs, the proportion of correct classification was 0.820, with small individuals (N = 32) correctly assigned at a proportion of 0.750 and large crabs (N = 29) at 0.897. For mottled crabs, the proportion of correct classification was 0.755, with small individuals (N = 33) being correctly assigned at a proportion of 0.909 and large crabs (N = 20) at 0.500.

Ontogenetic variation in crab colour and pattern

The brightness and colour metrics (saturation and hue) of the carapace of each crab were significantly different between size classes and/or species (Table 2). For mud crabs, small individuals were on average brighter than large crabs (mean ± SEM: small = 11.22 ± 0.79%, large = 7.07 ± 0.32%; t₁₁₀ = 5.19, P < 0.001; Fig. 2A), but for mottled crabs there was no difference in brightness between size classes, with both small and large crabs being equally dark (small = 4.62 ± 0.24%, large = 4.52 ± 0.27%; t₁₁₀ = 0.16, P = 0.871; Fig. 2A), resulting in an interaction between species and size class. In addition, small mud crabs exhibited a marked variation in brightness compared with larger individuals (coefficient of variation: 40.09% for small and 24.61% for large crabs), indicating that small individuals of this species are more diverse in brightness compared with larger ones.

Regardless of size, mottled crabs exhibited carapaces with more saturated coloration than mud crabs (mud crab = 0.103 ± 0.004, mottled crab = 0.123 ± 0.005; t₁₁₀ = 3.41, P < 0.001; Fig. 2B; Table 2). In contrast, different from brightness, hue was significantly larger for large mottled crabs compared with small individuals (small = 0.519 ± 0.005, large = 0.550 ± 0.009; t₁₁₀ = 4.13, P < 0.001; Fig. 2C), indicating that crabs tend to become greener as they grow. For mud crabs, however, there was no difference in hue between size categories, with individuals maintaining similar values over ontogeny (small = 0.511 ± 0.003, large = 0.519 ± 0.002; t₁₁₀ = 1.18, P = 0.239; Fig. 2C), explaining the significant interaction between main factors. Therefore, mud crabs become darker as they grow but remain the same colour, whereas mottled crabs remain similarly dark but become greener.

Regarding the pattern metrics, the dominant size of markings on carapace of crabs (i.e. the maximum frequency) differed significantly between species and size classes (Table 3), being on average larger for mud crabs (mud crab = 1.986 ± 0.170 mm, mottled crab = 0.641 ± 0.058 mm; t₁₁₀ = 8.52, P < 0.001) and larger individuals (small = 0.971 ± 0.100 mm, large = 1.878 ± 0.208 mm; t₁₁₀ = 3.97, P < 0.001; Fig. 3A). The overall pattern contrast (i.e. the summed energy) varied between size categories but depended on the species (Table 3). Small mud crabs exhibited more contrasting pattern markings than large individuals (small = 10.503 ± 0.582, large = 8.093 ± 0.366; t₁₁₀ = 3.30, P = 0.001), indicating that carapace markings become less contrasting (Fig. 3B). Conversely, mottled crabs showed small and similar pattern contrast between size classes (small = 4.858 ± 0.224, large = 5.513 ± 0.308; t₁₁₀ = 1.60, P = 0.112). Finally, although not different between species (Table 3), the diversity of markings (i.e. the proportional energy) was higher for small individuals (small = 0.090 ± 0.001, large = 0.085 ± 0.001; t₁₁₀ = 2.88, P = 0.005), indicating that larger crabs of both species tend to exhibit less diverse pattern markings on their carapace (Fig. 3C).

Ontogenetic variation in background matching of crabs against different backgrounds

The degree of background matching, measured as the colour distance between crabs and substrates, differed between species, size categories and background habitats (Table 4). In mud crabs, regardless of size, crabs were better concealed against mud rock and pebbles compared with algal substrates (Fig. 4A). In mottled crabs, however, small individuals were better concealed against mud rock and pebbles, but large crabs showed equally low concealment against all substrates (Fig. 4B). Therefore, small and large mud crabs are consistently more cryptic against mud rock and pebbles, whereas large mottled crabs become generally conspicuous, and thus poorly concealed against background habitats.

Discussion

Here, we show that ontogenetic changes in animal appearance and crypsis can occur through modifications of different colour and pattern metrics and, apparently, linked to other species-specific traits. In the less motile mud crab (Panopeus americanus), the variability in brightness and the degree of pattern contrast on the carapace of small crabs reduces with ontogeny, resulting in darker and smoother large individuals. These changes probably stem from the adjustment of crabs to the gradual reduction in the chromatic heterogeneity of the backgrounds they are exposed to when older. In the more active mottled crab (Pachygrapsus transversus), however, ontogenetic changes occur through a modification in carapace colour, with individuals changing from dark to green tones as they grow, which is probably a result of changes in habitat use and behaviour with age. The modification of appearance of mud crabs with ontogeny is not followed by a reduction in background matching, as would be expected in species that can escape predation by achieving a size refuge, when effective camouflage would be less important. On the contrary, both small and large mud crabs show higher background similarity than mottled crabs, in which large individuals present poor camouflage. Therefore, our findings suggest that other life-history traits of the studied species might interfere in the ontogenetic colour and pattern changes. For more resident species, individuals would rely continuously on camouflage to escape predation, whereas for more active species, background similarity is probably less relevant because individuals can move fast and escape from most predators.

High intraspecific colour variability of many animal species is frequently associated with concurrent variation in the chromatic aspects of the backgrounds with which they are associated (Caro et al., 2016), resulting in phenotype–environment matching and an increase in camouflage effectiveness (Todd et al., 2006; Stevens et al., 2015; Nokelainen et al., 2017). Therefore, the colour diversity of the different backgrounds occurring in our sampling site (e.g. rocks covered by algae, boulders and pebbles on both muddy and sandy substrates), together with the differential use of those by small and large crabs, should select for specific patterns of coloration and camouflage at the different size categories. Indeed, we observed species-specific ontogenetic trajectories in both crab colour and crypsis: in mud crabs, small individuals are very variable in brightness and in the contrast of carapace markings compared with larger crabs, which are darker and less patterned. In mottled crabs, both size classes exhibit low variability in all colour and pattern metrics, but large crabs become distinctly coloured. The ontogenetic change observed in mud crabs is common to many other crab species, in which juveniles are associated with colour heterogeneous backgrounds but inhabit homogeneous substrates when adults (Palma & Steneck, 2001; Palma et al., 2003; Todd et al., 2009; Krause-Nehring et al., 2010; Stevens et al., 2014; Jensen & Egnotovich, 2015). In contrast, ontogenetic variation in colour but not in the diversity of pattern markings, as observed for mottled crabs, is common in species in which individuals occupy habitats of different coloration as they grow (Booth, 1990; Wilson et al., 2007; Hultgren & Stachowicz, 2010). Therefore, different ecological processes related to changes in habitat use probably underlie the ontogenetic colour and pattern variation in the two crab species we studied here.

The juvenile phase is known to be the most vulnerable life stage for predation risk in many crab species (Palma & Steneck, 2001; Krause-Nehring et al., 2010), and this phase is when stronger selection for camouflage is expected (Caro, 2018). As crabs grow, they can eventually reach a size refuge from predation (Gosselin & Qian, 1997), at which individuals are capable to defend themselves or to escape more rapidly from predators, and camouflage becomes less crucial (Todd et al., 2006). The results presented here for mud crabs do not support this hypothesis, because colour distances of small and large individuals against backgrounds are very similar, indicating better camouflage against mud rock and pebble substrates regardless of size. In many cases, prey are exposed to different predators along their ontogeny, with later-life predators sometimes being large enough to consume adults at a similar rate to that at which juveniles are preyed on, which can result in prey never achieving an escape size (Eggleston et al., 1990; Pessarrodona et al., 2019). The same scenario is expected when predation on juveniles is so intense that it could work as a bottleneck for future ontogenetic stages (Beck, 1995). Therefore, given that mud crabs are characterized by lower mobility throughout their entire lifetime (Micheli, 1997), a continuous need for efficient camouflage over ontogeny is expected (Hughes et al., 2019). Small crabs (juveniles) are found near to the lower limit of the intertidal zone (Vergamini & Mantelatto, 2008a), where desiccation stress is lower but predation potentially higher, because exposure to fish increases. Large bird species, such as herons and spoonbills, are frequently seen feeding in the middle and upper part of the intertidal region in the Araçá Bay during low tide (Amaral et al., 2010; Mancini et al., 2018) and could easily prey on even the largest mud crabs. Similar predation pressure is expected throughout the geographical distribution of mud crabs, and therefore strong selection for crypsis is expected through ontogeny.

An alternative explanation for the differences we observed in mud crab appearance between size categories is that the high variability found in small individuals could result from juveniles using more diverse microhabitats than adults, which would require differential coloration for camouflage, either by means of matching the general appearance of specific background patches or by showing highly disruptive markings (Todd et al., 2006, 2009; Wilson et al., 2007; Nokelainen et al., 2017; Price et al., 2019). Given that small mud crabs are associated with highly heterogeneous backgrounds, such as crushed-shell substrates or coarse sandy areas (Carvalho-Batista et al., 2015), it might be challenging for the same crab individual to conceal itself on many backgrounds at once. Therefore, in addition to using background matching for effective concealment against rocky substrates, as we show, it is also possible that small mud crabs could optimize their camouflage efficiency through disruptive coloration (Cuthill et al., 2005). The high contrasting markings located on the carapace of juvenile crabs could contribute to create false edges and boundaries around the body, preventing the recognition of individuals by predators (Webster et al., 2013). Given that large adult mud crabs are smoother, exhibiting few and less-contrasting markings on their carapace, they probably optimize their concealment only by matching the general appearance of backgrounds. Finally, it is also possible that the highly variable coloration of small mud crabs could contribute to the defeat of predator search images, resulting in a mechanism of apostatic selection, in which predators attack more common colour types disproportionately often, with selection favouring rarer types and maintaining colour variation at the population level (Bond & Kamil, 2002, 2006). Future studies are necessary to understand what camouflage strategies are used by the different life stages of mud crabs, possibly indicating a change from disruptive coloration to background matching along their ontogeny. In addition, studies testing whether the high individual variability in brightness and pattern of juvenile mud crabs could effectively defeat predator search images are also needed.

In comparison to mud crabs, mottled crabs are generally darker and less patterned, exhibiting low variation in all colour and pattern metrics within each size class. Besides possessing more colour-saturated carapaces, crabs exhibited ontogenetic variation in colour but not in brightness, with large crabs being greener than small individuals. This differential coloration between size categories is probably a result of specific patterns of habitat use in juvenile and adult mottled crabs (Booth, 1990). During low tide, small crabs are active but remain close to refuges, such as rocky crevices and holes, including the interior of dead mussels and barnacles, where they execute short foraging excursions to feed on detritus (Abele et al., 1986). It is also common to find small crabs hidden under rocks. This would explain their dark coloration and better concealment against mud rock and pebble substrates. As they grow larger, crabs increase mobility and spend significantly more time feeding at flat rocks covered with macroalgae during low tide (Abele et al., 1986; Christofoletti et al., 2010), on which a greenish coloration would increase background matching. However, we found comparable background similarity of adult mottled crabs against all substrates we considered, suggesting a generalist camouflage strategy (Hughes et al., 2019). This strategy is based on the concept of compromise coloration (Merilaita et al., 1999), in which the individual matches several background types somewhat, but matches none perfectly. Considering the perspective of the crab as prey, a low dispersal rate of individuals would favour local adaptation and specialization, similar to what we observed for mud crabs, whereas higher mobility would promote a generalist strategy, as in adult mottled crabs (Hughes et al., 2019). Future work could aim to determine the scales of phenotype–environment matching in juvenile and adult mottled crabs and whether the high mobility of individuals allied to a compromise coloration might work together to reduce predation pressure.

It is likely that the differences in colour and pattern observed between size categories of the two crab species result from changes in colour after moulting events through developmental plasticity (Duarte et al., 2017), which are common to many crab species (Detto et al., 2008; Hultgren & Stachowicz, 2010; Jensen & Egnotovich, 2015; Nokelainen et al., 2019; Carter et al., 2020). Alternatively, if crab appearance is fixed, it is also possible that such ontogenetic differences result from differential predation on small individuals that are not well concealed against the background to which they are exposed during ontogenetic changes in habitat (Stevens et al., 2014). Considering that the variance of colour distances against backgrounds is similar between small and large individuals of both species, the second hypothesis is less likely to explain ontogenetic changes. However, future studies are necessary to describe the relative importance of these processes. Regardless, unlike other studies with similar species (Freire et al., 2011; Carvalho-Batista et al., 2015; Jensen & Egnotovich, 2015), our study used an objective method to quantify coloration (i.e. standardized reflectance) and considered that mud and mottled crabs did not exist as discrete morphs, but instead exhibited continuous colour variation.

Considering the limitations of our study, future work on wider size ranges and based on larger sample sizes and more prolonged sampling periods is necessary to elucidate further how individuals of both species change their appearance along ontogeny and how this would affect concealment against different backgrounds. Taking into account the scarcity of studies testing different ecological theories in tropical areas (Martin et al., 2012), especially related to animal coloration and crypsis, which are almost exclusive to temperate regions (but see Hemmi et al., 2006; Detto et al., 2008; Stevens et al., 2013), our work highlights an important avenue of research on evolutionary and behavioural ecology. Intertidal crabs are a highly representative group in tropical regions, where predation pressure is known to be higher than in temperate areas (Schemske et al., 2009; Roslin et al., 2017). As a result, we expect higher selection for the evolution of anti-predatory strategies in the tropics, where the increased diversity of life histories would also contribute to the existence of yet unexplored patterns of animal coloration and crypsis in nature.

SHARED DATA

The data are available from: https://github.com/rafaduarte87/crab_colour_pattern_crypsis.

ACKNOWLEDGEMENTS

This study was funded by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), which granted a post-doctoral fellowship to R.C.D. (CAPES, Finance Code 001; FAPESP #2019/01934-3) and a visiting professor grant to M.S. (FAPESP #2015/22258-5). G.M.D. is supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (Grant/Award Number: 308268/2019-9) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP #2019/15628-1). We are grateful to Tim Caro and two anonymous reviewers for their valuable comments on an earlier draft of this manuscript. We especially thank the technical staff at the Center for Marine Biology for helping in the field survey. The authors declare no competing interests.

REFERENCES