Skip to main content
NCBI home page
Proceedings of the Royal Society B: Biological Sciences logoLink to Proceedings of the Royal Society B: Biological Sciences
. 2019 Oct 9;286(1912):20191879. doi: 10.1098/rspb.2019.1879

It's not all black and white: investigating colour polymorphism in manta rays across Indo-Pacific populations

Stephanie K Venables 1,2,✉, Andrea D Marshall 2, Elitza S Germanov 2,3, Robert J Y Perryman 2,4, Ricardo F Tapilatu 5, I Gede Hendrawan 6, Anna L Flam 2, Mike van Keulen 3, Joseph L Tomkins 1, W Jason Kennington 1
PMCID: PMC6790782  PMID: 31594509

Abstract

Intraspecific colour polymorphisms have been the focus of numerous studies, yet processes affecting melanism in the marine environment remain poorly understood. Arguably, the most prominent example of melanism in marine species occurs in manta rays (Mobula birostris and Mobula alfredi). Here, we use long-term photo identification catalogues to document the frequency variation of melanism across Indo-Pacific manta ray populations and test for evidence of selection by predation acting on colour morph variants. We use mark–recapture modelling to compare survivorship of typical and melanistic colour morphs in three M. alfredi populations and assess the relationship between frequency variation and geographical distance. While there were large differences in melanism frequencies among populations of both species (0–40.70%), apparent survival estimates revealed no difference in survivorship between colour morphs. We found a significant association between phenotypic and geographical distance in M. birostris, but not in M. alfredi. Our results suggest that melanism is not under selection by predation in the tested M. alfredi populations, and that frequency differences across populations of both species are a consequence of neutral genetic processes. As genetic colour polymorphisms are often subjected to complex selection mechanisms, our findings only begin to elucidate the underlying evolutionary processes responsible for the maintenance and frequency variation of melanism in manta ray populations.

Keywords: colour polymorphism, manta ray, melanism, Mobula alfredi, Mobula birostris

1. Introduction

Colour polymorphisms are found in a range of taxa and have long been a focus of research into the adaptive significance and evolutionary origin of polymorphic traits [1,2]. Melanism is a widespread and conspicuous colour polymorphism associated with the increased production of melanin pigments by the melanocortin system, resulting in darker coloured individuals [3]. A range of evolutionary processes contribute to the maintenance of colour polymorphisms within and between populations (reviewed in [4]). Selection by predation can lead to variation in morph frequencies in prey species, as particular coloration may offer more effective camouflage from predators in certain environments [5,6]. For example, the colour of pocket mice has been associated with substrate colour, with darker morphs found on dark lava flows and lighter morphs on lighter coloured rocky substrate [5]. Similarly, phenotypic variation can be best explained by habitat similarity in three lizard species, with morph colours closely matching the colour of different substrates [6]. In larger species at higher trophic levels such as felines, the adaptive benefits of melanism are more difficult to pinpoint [7]. Nevertheless, niche modelling suggests that frequencies of melanistic leopards (Panthera pardus) are higher in tropical and subtropical forests and lower in drier, more open habitats [8]. This indicates that melanism may be under selection in leopards, probably driven by camouflage and/or thermoregulation in certain landscapes [8].

In conjunction with predation, frequency-dependent selection (where rare morphs are favoured) can contribute to the maintenance of less common morphs [9]. Predators form a search image for common morphs, resulting in reduced predation of rare morphs as their appearance differs from the typical prey image [9,10]. Frequency-dependent selection can also be mediated by sexual selection to maintain polymorphisms [11], as observed in guppies (Poecilla reiculata) [12], and side-blotched lizards (Uta stansburiana), where male colour variation represents differing behavioural traits and mating strategies [13]. Melanism offers adaptive advantages through enhanced physiological processes including thermoregulation [14] and disease resistance [15]. In some species, however, the link between melanism and fitness benefits remains unclear [16,17]. Melanism may be a result of the pleiotropic effect of selection for certain physiological or behavioural traits, such as aggression and sexual activity, which are also regulated by the melanocortin system [18,19]. In some cases, frequency variation is thought to be a result of more random processes. Genetic drift, where allele frequencies vary owing to chance or random events, has been implicated in creating variations in morph frequencies among populations of northern leopard frogs (Rana pipiens) [20] and candy-stripe spiders (Enoplognatha ovata) [21]. However, it is generally expected that frequency variation occurs when genetic drift acts alongside other evolutionary forces, such as gene flow or different modes of selection [4]. Selection pressures can vary temporally in strength and may, at times, be weak and difficult to detect. It is during such periods that genetic drift is more likely to affect morph frequencies [21,22].

In contrast to terrestrial species where examples are common, melanism is rare in marine species, and as a result, the factors affecting melanism in the marine environment are less well understood [23]. Melanism has been reported in grey seals [24] and cetaceans (12 records in total) [25]. Arguably, the most prominent example of melanism in marine animals occurs in the two species of manta ray—the giant manta ray Mobula birostris (previously Manta birostris) and the reef manta ray Mobula alfredi (previously Manta alfredi) [26,27]. Typical colour morphs are black on the dorsal surface with white shoulder patches on the supra-branchial region. The ventral surface is predominantly white with a varying degree of darker pigmented spots and patches [27]. Melanistic morphs are completely black on the dorsal surface and predominantly black on the ventral surface except for a variably sized white blaze along the abdominal region and between the gill slits [27] (figure 1). A leucistic colour morph has also been documented for both species, which exhibits reduced pigmentation on the dorsal and ventral surfaces, resulting in a notably lighter appearance and lighter ventral markings [27,28]. These variable and distinctive ventral pigmentation patterns are unique to individuals and are used for identification [29,30]. For all morphs, these markings remain constant in size and shape in relation to the size of the individual, even over extended time periods (greater than 30 years) [29,31]. This has enabled the compilation of long-term regional photo identification (photo-ID) catalogues in numerous locations across the globe [32].

Figure 1.

Figure 1.

Open in a new tab

Examples of dorsal (top left photographs) and ventral pigmentation patterns exhibited by (a) melanistic morph M. alfredi, (b) melanistic morph M. birostris, (c) typical morph M. alfredi, and (d) typical morph M. birostris showing the variation in the percentage of white pigmentation observed within morphs. (Online version in colour.)

The typical manta ray colour pattern, where pigmentation is notably darker on the dorsal than the ventral surface, is an example of countershading [33]. Countershading is common among marine species and is considered to serve for camouflage from predators [34]. Although, as a result of their size and quick burst speed, manta rays have few known predators [35]. Lethal and non-lethal predatory interactions have been observed with grey reef sharks (Carcharhinus amblyrhynchos), Galapagos sharks (Carcharhinus galapagensis), bull sharks (Carcharhinus leucas), tiger sharks (Galeocerdo cuvier) and white sharks (Carcharodon carcharias) [35,36], as well as killer whales (Orcinus orca) [37]. Darker coloured ventral surfaces would be expected to enhance the conspicuousness of melanistic morphs from below, which may make these individuals more prone to predation. In cetaceans, melanism is suggested to be a disadvantage because darker coloration may affect an individual's ability to capture prey, impair visual/social communication and increase conspicuousness to predators [25]. The occurrence of melanistic morphs and the variation in the frequency of these morphs across global manta ray populations have been reported previously [27,38], yet it remains unknown why these variants have persisted and why they are more common in some regions than others.

Photo-ID data have been used in mark–recapture modelling to estimate demographic parameters for populations of marine megafauna including cetaceans [39,40], sharks [41,42] and manta rays [30,31,43]. Such studies typically focus on generating abundance estimates, yet this methodology is also valuable for estimating survivorship [44,45]. Here, we use photo-ID datasets to document the frequency of melanism across Indo-Pacific populations of M. alfredi and M. birostris. We determine whether colour variation partitions these populations into discrete morphs, and ask whether the maintenance of this trait in certain populations is owing to selection by predation or can be explained by selectively neutral processes. To do so we use mark–recapture modelling to compare survivorships of typical and melanistic colour morphs in three well-studied populations of M. alfredi. If melanism were under negative selection by predation, we would expect to observe lower survival rates for melanistic morphs. Alternatively, if this trait were selectively neutral, we expect survival rates to be equal across colour morphs.

2. Methods

(a). Study sites

Sighting records of M. birostris from five locations—Ecuador (1°16′40″ S, 81°4′7″ W), Mozambique (23°55′93″ S, 35°40′39″ E), Myanmar (11°24′6″ N, 97°50′44″ E), Raja Ampat, West Papua (0°34′47″ S, 130°32′32″ E) and Thailand (8°39′25″ N, 97°38′38″ E), and of M. alfredi from four locations—Mozambique, Komodo National Park (8°31′36″ S, 119°28′28″ E), Nusa Penida (8°43′57″ S, 115°26′51″ E) and Raja Ampat, West Papua, Indonesia (figure 2), were exported from MantaMatcher.org [46], a global, collaborative manta ray photo-ID database. Sighting records ranged between 2003 and 2018 and included researcher survey data as well as public submissions from dive operators, recreational divers and underwater photographers. We assessed a total of 3040 M. birostris individuals and 3799 M. alfredi individuals using MantaMatcher sightings data and included published data for M. alfredi populations in Japan (n = 303) [47] and Hawaii (n = 290) [48]. Unless otherwise stated, all analyses were conducted in R (v. 3.3.1) [49].

Figure 2.

Figure 2.

Open in a new tab

Map of study locations. Upper map shows populations in which melanism frequency was assessed, white circles represent M. birostris, orange circles represent M. alfredi, and split circles represent populations of both species. Lower maps show locations of populations included in mark–recapture modelling: (a) Inhambane Province, Mozambique, (b) Nusa Penida, (c) Komodo National Park, and (d) Raja Ampat, West Papua. Triangles represent main encounter sites.

(b). Morph distinction

To determine whether colour variation was structured into discrete morphs, we collated ventral identification photos from one population of each species (Raja Ampat for M. alfredi and Ecuador for M. birostris). We selected only photos that were clear, high resolution and included the entire manta ray with pectoral fins extended and ventral surface parallel to the camera. We then randomly selected 30 images for each species for analysis. Using the ‘magic wand’ selection and histogram tools in Adobe Photoshop (v. 16.0), we selected the outline of the manta ray and calculated the total number of pixels. As white pigmentation was common to all individuals, we then selected only the white pixels and used this to calculate the percentage of white pigmentation on the ventral surface of each individual. We fitted Gaussian mixture models to the percentages of white ventral pigmentation for each species to estimate the number of component distributions and identify the number of discrete morphs present. We used the package mixtools [50] to fit models via an expectation maximum algorithm. We fitted models with component distributions (k) of 1–4 and used the Bayesian information criterion (BIC) to assess the best-fit model for each population, with a lower BIC value indicating greater support for the model [51]. Finally, we compared individual morph assignments from this analysis to those recorded in the MantaMatcher database to ensure there were no discrepancies.

(c). Melanism frequencies

Ventral and, where available, dorsal pigmentation patterns were visually assessed by MantaMatcher database managers to assign each individual to a colour morph (typical or melanistic) based on the description adapted from Marshall et al. [27] (see §1 and figure 1). We did not consider leucistic morphs in our study owing to their rarity in our focal populations. We calculated melanism frequency for each population of each species, by dividing the number of melanistic individuals by the total number of identified individuals. To investigate the effect of location and sex on melanism frequency, we fitted generalized linear models (GLMs) to a separate dataset for each species in which each individual of known sex was only represented once, with colour morph as a binomial response variable (where 1 = melanistic and 0 = typical) and location and sex as categorical predictors. We fitted models separately for the two species with a binomial distribution and logit link function and tested all combinations of predictors and their interactions including the null model. The Akaike information criterion corrected for small sample sizes (AICc) was used to assess model fit, with a smaller AICc value indicating better model fit to the data [51].

If genetic drift was primarily responsible for the variation in melanism frequencies across populations, we would expect to see evidence of a relationship between melanism frequency and geographical distance under a stepping stone model of gene flow, where a higher proportion of migrants is assumed to come from neighbouring populations [52]. To test this we conducted Mantel tests using 9999 permutations in the ade4 package [53] and multiple matrix regression and randomization (MMRR) analysis [54] to assess the relationship between geographical and phenotypic distances between populations of each species. MMRR allows predictors to be tested individually and multiple predictors to be tested together to provide overall combined variance and significance values for each predictor. We calculated phenotypic distance using Simpson's index of diversity (D), the probability that two populations sampled at random will have different melanism frequencies, using the formula, D = 1 − Σ(pi)2, where pi is the proportion of melanistic morphs in a given population. Simpson's index of diversity was used to partition frequency variation among populations. The proportion of melanism frequency variability among populations was calculated as (DT − DS)/DT, where DT is the total frequency diversity, obtained by calculating D when all populations are pooled and DS is the mean frequency diversity within each population. We assessed geographical distance using three separate measures: Euclidean distance between sites derived using the package fossil [55] and over-water distance (OWD) derived based on the shortest across-water distance using the marmap package [56]. We assessed OWD without depth constraints and when distances between sites were calculated within a maximum depth constraint of 3000 m, based on satellite telemetry tracks from previous studies [57,58].

(d). Mark–recapture modelling

We used mark–recapture modelling to assess whether survivorship rates differed between colour morphs within populations, which would indicate whether melanism was under longevity selection in these populations. Sighting records of M. alfredi from aggregation sites in Praia do Tofo, Mozambique and three locations in Indonesia—Nusa Penida, Komodo National Park and Raja Ampat, West Papua (figure 2) were included in this analysis. For the purpose of this study, the term ‘mark’ refers to the earliest record of an individual (i.e. the first time it was photographed) and ‘recapture’ refers to a subsequent encounter of a previously identified individual. Owing to the migratory nature of manta rays [35] and study period durations, we assumed that populations had undergone additions from births/immigrations and reductions from deaths/emigration and populations could not be considered ‘closed’. Furthermore, violations of population closure assumptions were tested using the program Closetest [59], and we deemed it appropriate to use ‘open’ population models. We fitted Cormack–Jolly–Seber (CJS) models and the POPAN formulation of the Jolly–Seber (JS) models to encounter histories of individuals from each population in program Mark (v. 8.2) [60]. The estimation of demographic parameters using CJS and JS models are based on several assumptions, which are outlined in Marshall et al. [30]. These assumptions and the way we have addressed them are outlined in the electronic supplementary material, table S1.

(e). Parameter estimation and model selection

We grouped encounter histories by colour morph (typical or melanistic) for the Nusa Penida, Komodo and Raja Ampat populations. To date there has only been one reported sighting of a melanistic M. alfredi in Mozambique, we therefore pooled this individual with the typical morphs. For n encounter periods, CJS and POPAN models provide n − 1 estimates of apparent survival (Phi), the probability that an animal survives and does not permanently emigrate from the population and n estimates of recapture probability (p). POPAN models additionally provide n − 1 estimates of pent (the probability of entry into the population between encounter periods) and N (super-population size). We modelled Phi and p between encounter periods as constant over time (.), variable over time (t) and with group effects (morph) (morph and t). We modelled pent parameters as a variable over time (t) and with group effects (morph and t). All combinations of predictors were tested, and models were fitted using the logit link function for Phi and p, the identity link function for N, and the multinomial logit link function for pent to constrain the set of pent parameters to sum to one.

We used the quasi-likelihood Akaike's information criterion (QAICc) to assess model support and adjusted for small sample sizes and the variance inflation factor (ĉ). Lower QAICc values indicated greater support for the model and differences between QAICc values of <2 between models were interpreted to indicate approximately equal support of candidate models [51]. Parameter estimates and associated standard errors were obtained through model averaging across models adjusted using normalized Akaike weights in order to account for model variation in the accuracy of estimates [61]. We did not report values for super-population size (N), as estimating this parameter was not a central aim of our study.

(f). Data pooling and goodness-of-fit testing

Although sighting records for each population spanned extended time periods (greater than 9 years), only years with consistent survey effort (i.e. greater than 100 dedicated survey dives at Indonesian sites and greater than 50 in Mozambique) were included in analyses. This varied for each population, as did the seasonality of manta ray sightings. We therefore determined a six-month ‘peak season’ for each population based on monthly sighting totals and included only sightings that fell within the peak season. Owing to the variation in the number of survey years and seasonality of sightings in different locations, we adjusted the defined encounter periods within peak sighting seasons for each location to minimize overdispersion of data (see the electronic supplementary material, table S2). We implemented goodness-of-fit tests (TEST 2 and TEST 3) in U-Care [62] to test for overdispersion in the data. A detailed explanation of these tests can be found in Santostasi et al. [40]. Overall test statistics were obtained for the fully parametrized CJS model grouped by colour morph for each dataset. Wherever these tests were found significant, data were adjusted to account for lack of fit using ĉ which was calculated by dividing the overall χ2 statistic by the overall degrees of freedom [63] (electronic supplementary material, table S3). We conducted power analyses in the form of CJS model simulations in Mark to estimate the minimum difference in apparent survival that could be detected between typical and melanistic morphs given the current datasets (for detailed methods see the electronic supplementary material, table S11).

3. Results

(a). Colour variation between morphs

The percentage of white ventral pigmentation ranged from 0.06 to 99.3% for M. alfredi and 4.3 to 59.4% for M. birostris. Gaussian mixture models identified the best-fit number of components as k = 2 for both species, indicating two distinct colour morphs (electronic supplementary material, table S4; figure 3). For M. alfredi, the two identified peaks had means of 1.6% (±0.9% s.d.) and 79.8% (±14.8%). For M. birostris, the two identified peaks had means of 14.9% (±6.7%) and 53.1% (±5.1%; figure 3). Mobula alfredi exhibited a larger variation in the percentage of white ventral pigmentation for typical morphs (53.9–99.3%), and a smaller variation for melanistic morphs (0.06–3.7%), in comparison to M. birostris (typical 41.9–59.4%, melanistic 4.3–24.6%; figure 3). The dark pigmentation patches used for the identification of typical morphs are more variable in size and shape in M. alfredi than M. birostris. By contrast, the white blazes used for the identification of melanistic morphs are more variable for M. birostris (figure 1). Despite this variation within morphs, the bimodal distributions confirm that the two colour morphs can be separated statistically. All morph assignments by the quantification of ventral pigmentation matched those recorded in the MantaMatcher database, showing that visual categorization concurs with statistical discrimination.

Figure 3.

Figure 3.

Open in a new tab

Percentage of white pigmentation on the ventral surface of (a) M. alfredi and (b) M. birostris with overlaid components of best-fit Gaussian mixture models (k = 2). The first component (green) corresponds to melanistic morphs, and the second (orange) to typical morphs.

(b). Spatial variation in melanism frequencies

Melanism frequency in M. alfredi populations ranged from 0.0 to 40.7% (table 1). The highest frequency of melanism was recorded in the Raja Ampat population, mid-range frequencies were found in Nusa Penida and Komodo, and low frequencies in Hawaii, Japan and Mozambique (table 1). Melanism frequency in M. birostris populations ranged from 0.7 to 16.4%. The frequency was highest in the Ecuador population, whereas Indo-Pacific populations had lower frequencies of melanistic morphs (less than 2.5%; table 1). The best-supported GLMs for both species included location only and explained 26.6% and 6.2% of the total variance for M. alfredi and M. birostris, respectively. Sex had no apparent effect on melanism as an independent predictor (electronic supplementary material, table S5). Mantel tests revealed no significant association between phenotypic distance and any of the tested geographical distances for M. alfredi. Similarly, MMRR found none of the geographical distance measures to be significant predictors when included individually or in the combined model (electronic supplementary material, table S6). For M. birostris, the Mantel test showed a significant correlation between phenotypic distance and OWD with a maximum depth constraint of 3000 m, and the single-predictor MMRR found depth-constrained OWD to be a significant predictor, explaining 98.4% and 96.8% of the phenotypic variation, respectively (electronic supplementary material, table S6). The multiple-predictor MMRR model was not significant for M. birostris.

Table 1.

Frequency of typical and melanistic morphs across M. alfredi and M. birostris populations. (Data sourced from MantaMatcher.org unless otherwise indicated.)

location typical morph melanistic morph population total melanism frequency (%)
M. alfredi
Raja Ampat 422 290 712 40.7
Nusa Penida 619 66 685 9.6
Komodo National Park 1059 117 1176 9.9
Japan [47] 303 2 305 0.7
Mozambique 1225 1 1226 0.1
Hawaii [48] 290 0 290 0.0
M. birostris
Ecuador 1677 330 2007 16.4
Raja Ampat 261 6 267 2.3
Myanmar 128 3 131 2.3
Thailand 341 3 344 0.9
Mozambique 289 2 291 0.7
Open in a new tab

(c). Survival estimates of colour morphs within populations

Model-averaged estimates showed no difference in apparent survival between typical and melanistic morphs in the Nusa Penida population (table 2 and electronic supplementary material, table S7). Apparent survival did not vary over time or between morphs in the best-supported models, NP1 (CJS) and NP5 (JS; electronic supplementary material, table S7). Although model-averaged estimates for Komodo showed a 1% difference in apparent survival between melanistic and typical morphs (table 2 and electronic supplementary material, table S7), the 95% confidence intervals (CIs) for each morph overlapped, indicating this difference was not significant. The best-supported models (K1, CJS and K7, JS) included constant survival with no variation over time or between morphs (electronic supplementary material, table S8). Although K6 had a lower QAICc than K7, the difference in QAICc between these two models was less than 2, given the principal of parsimony the best-supported model was that with the lower number of parameters (K7) [64]. For Raja Ampat, model-averaged estimates of apparent survival were equal across colour morphs (table 2 and electronic supplementary material, table S9). The best-supported models, RA1 (CJS) and RA4 (JS), also included constant survival (electronic supplementary material, table S9). Model-averaged CJS and JS apparent survival estimates for Mozambique were notably lower than other populations (80.4% ± 2.9%), with the best-supported models (M1, CJS and M5, JS) indicating that apparent survival was constant over time (table 2 and electronic supplementary material, table S10).

Table 2.

Model-averaged apparent survival (Phi) estimations for CJS and JS models fitted to photo-ID datasets ± standard error (s.e.) and minimal detectable difference in Phi determined by CJS model simulations (see the electronic supplementary material, table S11).

population model Phi typical (±s.e.) Phi melanistic (±s.e.) difference in Phi (Phi melanistic − Phi typical) detectable difference in Phi
Nusa Penida CJS 0.959 (±0.006) 0.958 (±0.010) −0.001 0.06
JS 0.959 (±0.006) 0.958 (±0.010) −0.001
Komodo CJS 0.963 (±0.008) 0.974 (±0.017) 0.011 0.08
JS 0.963 (±0.007) 0.976 (±0.015) 0.013
Raja Ampat CJS 0.982 (±0.003) 0.984 (±0.003) 0.002 0.02
JS 0.983 (±0.003) 0.983 (±0.003) 0.000
Phi overall (±s.e.)
Mozambique CJS 0.804 (±0.029)
JS 0.804 (±0.029)
Open in a new tab

4. Discussion

Using long-term photo-ID datasets we show that, based on ventral pigmentation patterns, typical and melanistic morphs of M. alfredi and M. birostris are statistically discrete and demonstrate significant variation in the frequency of melanistic individuals among a subset of global populations of M. alfredi and M. birostris. To evaluate potential evolutionary processes that may be influencing melanism in manta rays, we tested for evidence of selection by predation and a relationship between phenotypic and geographical distance. There was no difference in apparent survival of typical and melanistic morphs in three M. alfredi populations, yet we found an association between phenotypic and geographical distances in M. birostris, but no significant association for M. alfredi.

We ask whether differences in the frequency of melanistic morphs among populations are because of selection by predation or if they can be explained by genetic drift on a selectively neutral trait. If melanism offers an adaptive advantage for manta rays in relation to predation, we would expect to find higher survivorship of melanistic morphs in populations with higher melanism frequencies. Mark–recapture modelling revealed no difference in apparent survival between colour morphs in populations in which melanism frequency ranged between 9.6 and 40.7%, suggesting that melanism is not under positive longevity selection in these populations. While we found no evidence of survivorship differences between morphs in populations where melanism is reasonably common (greater than 9%), we were unable to conduct similar analyses for populations where these morphs are rare or absent. Given their limited predators, we would expect that selection for camouflaging coloration in manta rays is weak. This would enable melanistic morphs to persist in certain populations, despite the potential selective disadvantage of a darker ventral surface, which would increase conspicuousness to predators hunting from below. Estimations of predator density within manta ray habitats would help to clarify levels of predatory pressure across different populations and may further explain the role of selection by predation on this trait.

An alternate scenario is that melanism is a selectively neutral trait, and differences between populations could have arisen from genetic drift and gene flow. Although previously implicated in the maintenance of morph frequencies in other species, the role of genetic drift is considered to be more significant when acting in synergy with selective forces, particularly during times when selection is weak [21,22]. If neutral genetic processes were primarily responsible for the variation in melanism frequencies across manta ray populations, we would expect to see evidence of a relationship between phenotypic and geographical distance under a stepping stone model of gene flow [52]. We found no correlation between phenotypic and geographical distance for M. alfredi, yet we found a strong association between phenotypic and over-water distance in M. birostris. Ongoing gene flow may therefore play a role in the maintenance of melanism within populations and the frequency differences between populations in M. birostris. If true, we would expect to find higher frequencies of melanistic individuals in other eastern Pacific populations and lower frequencies in Indian and Atlantic Ocean populations. By contrast, it would appear that variation in melanism frequencies across M. alfredi populations is primarily owing to genetic drift. It is possible, however, that a relationship between melanism frequency and geographical distance will become apparent for M. alfredi as data from more populations become available.

Genetic colour polymorphisms are often subjected to complex selection mechanisms [11], and other evolutionary processes that may be acting upon melanism cannot be ruled out here. Frequency-dependent selection and the ‘rare-male effect’ may be influencing melanistic morphs, where females shift their preference towards less common phenotypes, giving them a reproductive advantage [12]. Aside from coloration, the genes associated with the melanocortin system can affect numerous phenotypic traits including aggression, sexual activity, energy homeostasis, and stress and immunity responses [65]. It is therefore possible that melanism itself is not under selection in manta rays and may, in fact, be a pleiotropic result of selection for an associated behavioural or physiological trait [18]. Given their low reproductive rate and that wild birth has not yet been observed, genomic evidence would be required to properly unravel such scenarios.

Gene flow can play an important role in maintaining polymorphisms between adjacent populations regardless of selective processes acting within populations [66]. The population sizes reported here represent the total number of identified individuals to date and effective population sizes (Ne), the number of breeding individuals, are likely to be lower [67]. Species with smaller, fragmented populations and lower levels of gene flow are more susceptible to random changes in gene frequencies and exhibit more dramatic responses to selection [6,68]. Aside from population structure identified between Indo-Pacific M. birostris populations (Mexico and Sri Lanka) [58], little is known about the level of gene flow between manta ray populations and whether breeding occurs between locations. Both species are capable of long-distance movements [69,70], and irregular sightings far from known aggregation sites continue to expand our knowledge of species distribution [71]. However, reports of residency and site affinity to specific habitats are common, particularly for M. alfredi [32]. Nevertheless, gene flow should not be ruled out for populations within close geographical proximity, as those with documented interchange (Nusa Penida and Komodo) [72] exhibit similar melanism frequencies.

Manta rays are long-lived, with a lifespan thought to exceed 40 years [35]. Considering their longevity, the datasets for Indonesian populations represent a short period in the lifespan of a manta ray. Although power analyses indicate that a difference in apparent survival of as low as 2% could have been detected with our data, longer-term datasets will increase the capability of detecting differences in survivorship between morphs. Estimates of apparent survival have previously been generated using mark–recapture modelling for M. alfredi populations. Annual survival was estimated at approximately 1.0 in the eastern Australian population over a 4-year period, and this population consists of mostly mature individuals with strong site affinity to the study location [31]. Deakos et al. [48] reported survival rates of 0.68–1.0 in Maui, Hawaii. Annual survival rates previously estimated from 4 years of M. alfredi photo-ID data in Mozambique [30] were lower (0.6–0.7) than the estimates generated in our study (0.8). Mark–recapture modelling requires data collected over multiple capture events and robustness of parameter estimates increases with larger datasets. This could, therefore, be attributed to the additional 10 years of data included. Mozambique maintains the lowest apparent survival rate of studied populations. This aligns with an 88% decline in M. alfredi sightings in southern Mozambique reported by Rohner et al. [73]. Mortality in fisheries is likely to have contributed to this decline, with catches estimated at 20–50 individuals annually in the Inhambane Province [30]. Fishers in this region use non-selective methods including gill nets, which make it unlikely that the catch rate would differ between morphs. In areas with persistent anthropogenic mortality, it is likely that the adaptive importance of coloration and therefore selection pressures that may have been acting upon this population are reduced.

Here, we provide evidence of frequency variation of a discrete melanistic colour morph occurring in populations of two species of manta ray. Although we assessed two possible scenarios, selection by predation and selective neutrality, our findings only begin to unravel the evolutionary processes acting upon this trait. It is possible that multiple evolutionary processes are contributing to the maintenance of melanism in manta ray populations, and these processes may vary across geographical regions. Molecular studies to determine population structure and the genetic basis of melanism in manta rays are recommended as the next step in order to further our understanding of the adaptive significance and heritability of this distinctive trait.

Supplementary Material

ESM: It's not all black and white: investigating colour polymorphism in manta rays across Indo-Pacific populations. Tables S1 - S11
rspb20191879supp1.pdf (172.4KB, pdf)
Reviewer comments
rspb20191879_review_history.pdf (698.8KB, pdf)

Acknowledgements

We thank all supporters that provided invaluable in-kind support, particularly Peri-Peri Divers, Papua Explorers Dive Resort, Raja Ampat SEA Centre, Barefoot Conservation, Secret Garden, Lembongan Marine Association and the Komodo Dive Operator Community, Arenui liveaboard, Casa Barry Lodge, Exploramar, Diva Andaman and Off The Chart Expeditions. We are grateful to the citizen scientists who have submitted photos to MantaMatcher and the dedicated interns, volunteers and Ray of Hope Expedition participants who have contributed to data collection and management. Special thanks to H. Mitchell, P. Bassett and L. Ellevog for their tremendous contribution to this work, to T. Ko Gyi, A. Brival and L. Lawrance for their unwavering support and G. Winstanley and J. Holmberg for their instrumental support with MantaMatcher. We thank A. Rooney, J. Artendale and E. Cameron for the generous funding which supported the development and management of MantaMatcher and two anonymous reviewers for their valuable feedback on this manuscript.

Ethics

Photo-ID data collection were collected under permits issued by the Indonesian Ministry of Research, Technology & Higher Education (458/FRP/E5/Dit.KI/XI/2015, 2016, 2017 and 125/SIP/FRP/E5/Dit.KI/V/2016, 2017), and approval from the government of Mozambique. Animal ethics approvals were granted by the University of Western Australia (RA/3/100/1490) and Murdoch University (R2781/15). Public contributed photographic data were collected opportunistically by recreational divers and deposited into an online repository (MantaMatcher.org) developed explicitly to facilitate citizen science contributions.

Data accessibility

Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.8444pn0 [74]. Photo-ID catalogues are available at www.mantamatcher.org an online manta ray photo-ID repository [46].

Authors' contributions

S.K.V., W.J.K., J.L.T. and A.D.M. conceived the central idea of the manuscript. S.K.V., A.D.M., A.L.F., R.J.Y.P. and E.S.G. collated and managed datasets, A.D.M., R.F.T. and I.G.H. provided essential logistical support that enabled long-term monitoring of study populations. S.K.V. conducted statistical analyses and was the primary author of the manuscript. S.K.V., W.J.K., J.L.T. and A.D.M. contributed to the first draft of the manuscript and all authors contributed to editing and manuscript revisions.

Competing interests

We declare we have no competing interests.

Funding

This research was supported by the University of Western Australia, the Marine Megafauna Foundation, Ocean Park Conservation Foundation (4.516), Fortuna Foundation, PADI Foundation (14688) and Idea Wild. S.K.V. is supported by an Australian Government RTP scholarship (UWA) and the Bruce & Betty Green Foundation.

References

  • 1.True JR. 2003. Insect melanism: the molecules matter. Trends Ecol. Evol. 18, 640–647. ( 10.1016/j.tree.2003.09.2006) [DOI] [Google Scholar]
  • 2.Cook LM. 2003. The rise and fall of the carbonaria form of the peppered moth. Q. Rev. Biol. 78, 399–417. ( 10.1086/378925) [DOI] [PubMed] [Google Scholar]
  • 3.Kettlewell H. 1965. A 12-year survey of the frequencies of Biston betularia L. and its melanic forms in Great Britain. Entomol. Rec. J. Var. 77, 195–218. ( 10.1038/hdy.1958.4) [DOI] [Google Scholar]
  • 4.Gray SM, McKinnon JS. 2007. Linking color polymorphism maintenance and speciation. Trends Ecol. Evol. 22, 71–79. ( 10.1016/j.tree.206.10.005) [DOI] [PubMed] [Google Scholar]
  • 5.Nachman MW. 2005. The genetic basis of adaptation: lessons from concealing coloration in pocket mice. In Genetics of adaptation, Georgia genetics review III (ed. Mauricio R.), pp. 125–136. Dordrecht, The Netherlands: Springer. [DOI] [PubMed] [Google Scholar]
  • 6.Rosenblum EB. 2006. Convergent evolution and divergent selection: lizards at the White Sands ecotone. Am. Nat. 167, 1–15. ( 10.1086/498397) [DOI] [PubMed] [Google Scholar]
  • 7.Majerus ME, Mundy NI. 2003. Mammalian melanism: natural selection in black and white. Trends Genet. 19, 585–588. ( 10.1016/j.tig.2003.09.003) [DOI] [PubMed] [Google Scholar]
  • 8.da Silva LG, et al. 2017. Mapping black panthers: macroecological modeling of melanism in leopards (Panthera pardus). PLoS ONE 12, e0170378 ( 10.1371/journal.pone.0170378) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Olendorf R, Rodd FH, Punzalan D, Houde AE, Hurt C, Reznick DN, Hughes KA. 2006. Frequency-dependent survival in natural guppy populations. Nature 441, 633–636. ( 10.1038/nature04646) [DOI] [PubMed] [Google Scholar]
  • 10.Punzalan D, Rodd FH, Hughes KA. 2005. Perceptual processes and the maintenance of polymorphism through frequency-dependent predation. Evol. Ecol. 19, 303–320. ( 10.1007/s10682-005-2777-z) [DOI] [Google Scholar]
  • 11.Wellenreuther M, Svensson EI, Hansson B. 2014. Sexual selection and genetic colour polymorphisms in animals. Mol. Ecol. 23, 5398–5414. ( 10.1111/mec.12935) [DOI] [PubMed] [Google Scholar]
  • 12.Hughes KA, Houde AE, Price AC, Rodd FH. 2013. Mating advantage for rare males in wild guppy populations. Nature 503, 108 ( 10.1038/nature12717) [DOI] [PubMed] [Google Scholar]
  • 13.Sinervo B, Lively CM. 1996. The rock-paper-scissors game and the evolution of alternative male strategies. Nature 380, 240–243. ( 10.1038/380240a0) [DOI] [Google Scholar]
  • 14.Clusella TS, van Wyk JH, Spotila JR.. 2007. Thermal melanism in ectotherms. J. Therm. Biol. 32, 235–245. ( 10.1016/j.jtherbio.2007.01.013) [DOI] [Google Scholar]
  • 15.Wilson K, Cotter SC, Reeson AF, Pell JK. 2001. Melanism and disease resistance in insects. Ecol. Lett. 4, 637–649. ( 10.1046/j.1461-0248.2001.00279.x) [DOI] [Google Scholar]
  • 16.Eizirik E, Yuhki N, Johnson WE, Menotti-Raymond M, Hannah SS, O'Brien SJ. 2003. Molecular genetics and evolution of melanism in the cat family. Curr. Biol. 13, 448–453. ( 10.1016/S0960-9822(03)00128-3) [DOI] [PubMed] [Google Scholar]
  • 17.Theron E, Hawkins K, Bermingham E, Ricklefs RE, Mundy NI. 2001. The molecular basis of an avian plumage polymorphism in the wild: a melanocortin-1-receptor point mutation is perfectly associated with the melanic plumage morph of the bananaquit, Coereba flaveola. Curr. Biol. 11, 550–557. ( 10.1016/S0960-9822(01)00158-0) [DOI] [PubMed] [Google Scholar]
  • 18.Ducrest A, Keller L, Roulin A. 2008. Pleiotropy in the melanocortin system, coloration and behavioural syndromes. Trends Ecol. Evol. 23, 502–510. ( 10.1016/j.tree.2008.06.001) [DOI] [PubMed] [Google Scholar]
  • 19.Raia P, Guarino F, Turano M, Polese G, Rippa D, Carotenuto F, Monti DM, Cardi M, Fulgione D. 2010. The blue lizard spandrel and the island syndrome. BMC Evol. Biol. 10, 289 ( 10.1186/1471-2148-10-289) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Hoffman EA, Schueler FW, Jones AG, Blouin MS. 2006. An analysis of selection on a colour polymorphism in the northern leopard frog. Mol. Ecol. 15, 2627–2641. ( 10.1111/j.1365-294X.2006.02934.x) [DOI] [PubMed] [Google Scholar]
  • 21.Oxford G. 2005. Genetic drift within a protected polymorphism: enigmatic variation in color-morph frequencies in the candy-stripe spider, Enoplognatha ovata. Evolution 59, 2170–2184. ( 10.1111/0014-3820.2005.tb00926.x) [DOI] [PubMed] [Google Scholar]
  • 22.O'Hara RB. 2005. Comparing the effects of genetic drift and fluctuating selection on genotype frequency changes in the scarlet tiger moth. Proc. R. Soc. B 272, 211–217. ( 10.1098/rspb.2004.2929) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Stockin KA, Visser IN. 2005. Anomalously pigmented common dolphins (Delphinus sp.) off northern New Zealand. Aquat. Mamm. 31, 43–51. ( 10.1578/AM.31.1.2005.43) [DOI] [Google Scholar]
  • 24.Osinga N, Hart P, van Voorst Vader PC.. 2010. Albinistic common seals (Phoca vitulina) and melanistic grey seals (Halichoerus grypus) rehabilitated in the Netherlands. Anim. Biol. 60, 273–281. ( 10.1163/157075610X516493) [DOI] [Google Scholar]
  • 25.Visser IN, Fertl D, Pusser LT. 2004. Melanistic southern right-whale dolphins (Lissodelphis peronii) off Kaikoura, New Zealand, with records of other anomalously all-black cetaceans. N. Z. J. Mar. Freshwat. Res. 38, 833–836. ( 10.1080/00288330.2004.9517282) [DOI] [Google Scholar]
  • 26.White WT, Corrigan S, Yang L, Henderson AC, Bazinet AL, Swofford DL, Naylor GJ. 2018. Phylogeny of the manta and devilrays (Chondrichthyes: mobulidae), with an updated taxonomic arrangement for the family. Zool. J. Linn. Soc. 182, 50–75. ( 10.1093/zoolinnean/zlx018) [DOI] [Google Scholar]
  • 27.Marshall AD, Compagno LJV, Bennett MB. 2009. Redescription of the genus Manta with resurrection of Manta alfredi (Krefft, 1868) (Chondrichthyes; Myliobatoidei, Mobulidae). Zootaxa 2301, 1–28. ( 10.11646/zootaxa.2301.1.10) [DOI] [Google Scholar]
  • 28.Stevens GMW. 2016. Conservation and population ecology of manta rays in the Maldives. PhD thesis, University of York, York, UK.
  • 29.Pierce SJ, Holmberg JA, Kock AL, Marshall AD. 2018. Photographic identification of sharks. In Shark research: emerging technologies and applications for the field and laboratory (eds Carrier JC, Heithaus MR, Simpfendorfer CA), pp. 220–234. Boca Raton, FL: CRC Press. [Google Scholar]
  • 30.Marshall AD, Dudgeon CL, Bennett MB. 2011. Size and structure of a photographically identified population of manta rays Manta alfredi in southern Mozambique. Mar. Biol. 158, 1111–1124. ( 10.1007/s00227-011-1634-6) [DOI] [Google Scholar]
  • 31.Couturier LIE, Dudgeon CL, Pollock KH, Jaine FRA, Bennett MB, Townsend KA, Weeks SJ, Richardson AJ. 2014. Population dynamics of the reef manta ray Manta alfredi in eastern Australia. Coral Reefs 33, 329–342. ( 10.1007/s00338-014-1126-5) [DOI] [Google Scholar]
  • 32.Stewart JD, et al. 2018. Research priorities to support effective manta and devil ray conservation. Front. Mar. Sci. 5, 314 ( 10.3389/fmars.2018.00314) [DOI] [Google Scholar]
  • 33.Cott HB. 1940. Adaptive coloration in animals. London, UK: Methuen. [Google Scholar]
  • 34.Perrin WF. 2002. Coloration. In Encyclopedia of marine mammals (eds Perrin WF, Würsig B, Thewissen JCM), pp. 236–244. San Diego, CA: Academic Press. [Google Scholar]
  • 35.Couturier LIE, Marshall AD, Jaine FRA, Kashiwagi T, Pierce SJ, Townsend KA, Weeks SJ, Bennett MB, Richardson AJ. 2012. Biology, ecology and conservation of the Mobulidae. J. Fish Biol. 80, 1075–1119. ( 10.1111/j.1095-8649.2012.03264.x) [DOI] [PubMed] [Google Scholar]
  • 36.Marshall AD, Bennett MB. 2010. The frequency and effect of shark-inflicted bite injuries to the reef manta ray Manta alfredi. Afr. J. Mar. Sci. 32, 573–580. ( 10.2989/1814232x.2010.538152) [DOI] [Google Scholar]
  • 37.Alava J, Merlen G. 2009. Video-documentation of a killer whale (Orcinus orca) predatory attack on a giant manta (Manta birostris) in the Galápagos Islands. Latin Am. J. Aquat. Mammals 7, 81–84. ( 10.5597/lajam00139) [DOI] [Google Scholar]
  • 38.Barton O. 1948. Color notes on Pacific manta, including a new form. Copeia 1948, 146–147. [Google Scholar]
  • 39.Chabanne DB, Pollock KH, Finn H, Bejder L. 2017. Applying the multistate capture-recapture robust design to characterize metapopulation structure. Methods Ecol. Evol. 8, 1547–1557. ( 10.1111/2041-210X.12792) [DOI] [Google Scholar]
  • 40.Santostasi NL, Bonizzoni S, Bearzi G, Eddy L, Gimenez O. 2016. A robust design capture-recapture analysis of abundance, survival and temporary emigration of three Odontocete species in the Gulf of Corinth, Greece. PLoS ONE 11, e0166650 ( 10.1371/journal.pone.0166650) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Davies TK, Stevens G, Meekan MG, Struve J, Rowcliffe JM. 2012. Can citizen science monitor whale shark aggregations? Investigating bias in mark–recapture modelling using identification photographs sourced from the public. Wildl. Res. 39, 696–704. ( 10.1071/WR12092) [DOI] [Google Scholar]
  • 42.Dudgeon CL, Noad MJ, Lanyon JM. 2008. Abundance and demography of a seasonal aggregation of zebra sharks Stegostoma fasciatum. Mar. Ecol. Prog. Ser. 368, 269–281. ( 10.3354/meps07581) [DOI] [Google Scholar]
  • 43.Beale CS, Stewart JD, Setyawan E, Sianipar AB, Erdmann MV. 2019. Population dynamics of oceanic manta rays (Mobula birostris) in the Raja Ampat Archipelago, West Papua, Indonesia, and the impacts of the El Niño–Southern Oscillation on their movement ecology. Divers. Distrib. 25, 1472–1487. ( 10.1111/ddi.12962) [DOI] [Google Scholar]
  • 44.Bradshaw CJA, Mollet HF, Meekan MG. 2007. Inferring population trends for the world's largest fish from mark-recapture estimates of survival. J. Anim. Ecol. 76, 480–489. ( 10.1111/j.1365-2656.2006.01201.x) [DOI] [PubMed] [Google Scholar]
  • 45.Speed CW, Meekan MG, Rowat D, Pierce SJ, Marshall AD, Bradshaw CJA. 2008. Scarring patterns and relative mortality rates of Indian Ocean whale sharks. J. Fish Biol. 72, 1488–1503. ( 10.1111/j.1095-8649.2008.01810.x) [DOI] [Google Scholar]
  • 46.Marshall AD, Holmberg J.2019. MantaMatcher Photo-identification Library. See http://www.mantamatcher.org/ (accessed January 2018).
  • 47.Kashiwagi T. 2014. Conservation biology and genetics of the largest living rays: manta rays. PhD thesis, University of Queensland, Brisbane, Australia. [Google Scholar]
  • 48.Deakos MH, Baker JD, Bejder L. 2011. Characteristics of a manta ray Manta alfredi population off Maui, Hawaii, and implications for management. Mar. Ecol. Prog. Ser. 429, 245–260. ( 10.3354/meps09085) [DOI] [Google Scholar]
  • 49.R Core Team. 2016. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; See https://www.R-project.org. [Google Scholar]
  • 50.Benaglia T, Chauveau D, Hunter DR, Young D. 2009. mixtools: an R Package for analyzing finite mixture models. J. Stat. Softw. 32, 1–29. ( 10.18637/jss.v032.i06) [DOI] [Google Scholar]
  • 51.Burnham KP, Anderson DR. 2004. Multimodel inference: understanding AIC and BIC in model selection. Sociol. Method Res. 33, 261–304. ( 10.1177/0049124104268644) [DOI] [Google Scholar]
  • 52.Kimura M, Weiss GH. 1964. The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49, 561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Dray S, Dufour AB. 2007. The ade4 package: implementing the duality diagram for ecologists. J. Stat. Softw. 22, 1–20. [Google Scholar]
  • 54.Wang IJ. 2013. Examining the full effects of landscape heterogeneity on spatial genetic variation: a multiple matrix regression approach for quantifying geographic and ecological isolation. Evolution 67, 3403–3411. ( 10.1111/evo.12134) [DOI] [PubMed] [Google Scholar]
  • 55.Vavrek MJ. 2011. fossil: palaeoecological and palaeogeographical analysis tools. Palaeontol. Electron. 14, 1T. [Google Scholar]
  • 56.Pante E, Simon-Bouhet B. 2013. marmap: A package for importing, plotting and analyzing bathymetric and topographic data in R. PLoS ONE 8, e73051 ( 10.1371/journal.pone.0073051) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Jaine FRA, Rohner CA, Weeks SJ, Couturier LIE, Bennett MB, Townsend KA, Richardson AJ. 2014. Movements and habitat use of reef manta rays off eastern Australia: offshore excursions, deep diving and eddy affinity revealed by satellite telemetry. Mar. Ecol. Prog. Ser. 510, 73–86. ( 10.3354/meps10910) [DOI] [Google Scholar]
  • 58.Stewart JD, Beale CS, Fernando D, Sianipar AB, Burton RS, Semmens BX, Aburto-Oropeza O. 2016. Spatial ecology and conservation of Manta birostris in the Indo-Pacific. Biol. Conserv. 200, 178–183. ( 10.1016/j.biocon.2016.05.016) [DOI] [Google Scholar]
  • 59.Stanley TR, Burnham KP. 1999. A closure test for time-specific capture-recapture data. Environ. Ecol. Stat. 6, 197–209. ( 10.1023/A:1009674322348) [DOI] [Google Scholar]
  • 60.White GC, Burnham KP. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46, 120–139. ( 10.1080/00063659909477239) [DOI] [Google Scholar]
  • 61.Burnham KP, Anderson DR. 2003. Model selection and multimodel inference: a practical information-theoretic approach, 2nd edn, 487 p New York, NY: Springer Science & Business Media. [Google Scholar]
  • 62.Choquet R, Lebreton JD, Gimenez O, Reboulet AM, Pradel R. 2009. UCARE: Utilities for performing goodness of fit tests and manipulating CApture-REcapture data. Ecography 32, 1071–1074. ( 10.1111/j.1600-0587.2009.05968.x) [DOI] [Google Scholar]
  • 63.Cooch E, White GC. 2015. Program MARK: a gentle introduction, 14th edn, 1076 p Fort Collins, CO: Colorado State University. [Google Scholar]
  • 64.Vandekerckhove J, Matzke D, Wagenmakers EJ. 2015. Model comparison and the principle of parsimony. In The Oxford handbook of computational and mathematical psychology (eds Busemeyer JR, Wang Z, Townsend JT, Eidels A), pp. 300–319. Oxford, UK: Oxford University Press. [Google Scholar]
  • 65.Roulin A, Ducrest A-L. 2011. Association between melanism, physiology and behaviour: a role for the melanocortin system. Eur. J. Pharmacol. 660, 226–233. ( 10.1016/j.ejphar.2011.01.036) [DOI] [PubMed] [Google Scholar]
  • 66.Bittner TD, King RB. 2003. Gene flow and melanism in garter snakes revisited: a comparison of molecular markers and island vs. coalescent models. Biol. J. Linn. Soc. 79, 389–399. ( 10.1046/j.1095-8312.2003.0019.x) [DOI] [Google Scholar]
  • 67.Frankham R. 1995. Effective population size/adult population size ratios in wildlife: a review. Genet. Res. 66, 95–107. ( 10.1017/S0016672300034455) [DOI] [PubMed] [Google Scholar]
  • 68.Hare MP, Nunney L, Schwartz MK, Ruzzante DE, Burford M, Waples RS, Ruegg K, Palstra F. 2011. Understanding and estimating effective population size for practical application in marine species management. Conserv. Biol. 25, 438–449. ( 10.1111/j.1523-1739.2010.01637.x) [DOI] [PubMed] [Google Scholar]
  • 69.Armstrong AO, Armstrong AJ, Bennett MB, Richardson AJ, Townsend KA, Dudgeon CL. 2019. Photographic identification and citizen science combine to reveal long distance movements of individual reef manta rays Mobula alfredi along Australia's east coast. Mar. Biodivers. Rec. 12, 14–19. ( 10.1186/s41200-019-0173-6) [DOI] [Google Scholar]
  • 70.Hearn AR, Acuna D, Ketchurn JT, Penaherrera C, Green J, Marshall AD, Guerrero M, Schillinger G. 2014. Elasmobranches of the Galapagos marine reserve. In The Galapagos marine reserve (eds Denkinger J, Vinueza L), pp. 23–59. New York, NY: Springer; ( 10.1007/978-3-319-02769-2_2) [DOI] [Google Scholar]
  • 71.Arauz R, Chávez EJ, Hoyos-Padilla EM, Marshall AD. 2019. First record of the reef manta ray, Mobula alfredi, from the eastern Pacific. Mar. Biodivers. Rec. 12, 3 ( 10.1186/s41200-019-0162-9) [DOI] [Google Scholar]
  • 72.Germanov ES, Marshall AD. 2014. Running the gauntlet: regional movement patterns of Manta alfredi through a complex of parks and fisheries. PLoS ONE 9, e110071 ( 10.1371/journal.pone.0110071) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73.Rohner CA, Pierce SJ, Marshall AD, Weeks SJ, Bennett MB, Richardson AJ. 2013. Trends in sightings and environmental influences on a coastal aggregation of manta rays and whale sharks. Mar. Ecol. Prog. Ser. 482, 153–168. ( 10.3354/meps10290) [DOI] [Google Scholar]
  • 74.Venables SK, et al. 2019. Data from: It's not all black and white: investigating colour polymorphism in manta rays across Indo-Pacific populations Dryad Digital Repository. ( 10.5061/dryad.8444pn0) [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

  1. Marshall AD, Holmberg J.2019. MantaMatcher Photo-identification Library. See http://www.mantamatcher.org/ (accessed January 2018).
  2. Venables SK, et al. 2019. Data from: It's not all black and white: investigating colour polymorphism in manta rays across Indo-Pacific populations Dryad Digital Repository. ( 10.5061/dryad.8444pn0) [DOI] [PMC free article] [PubMed]

Supplementary Materials

ESM: It's not all black and white: investigating colour polymorphism in manta rays across Indo-Pacific populations. Tables S1 - S11
rspb20191879supp1.pdf (172.4KB, pdf)
Reviewer comments
rspb20191879_review_history.pdf (698.8KB, pdf)

Data Availability Statement

Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.8444pn0 [74]. Photo-ID catalogues are available at www.mantamatcher.org an online manta ray photo-ID repository [46].

Articles from Proceedings of the Royal Society B: Biological Sciences are provided here courtesy of The Royal Society

RESOURCES