3.2 Demographic parameters
In an attempt to correlate heterozygote deficiency to demographic data, we have retrieved the population size and level of habitat fragmentation of colour variable populations (Table 1). The values of 19 variable populations were statistically compared to similar data of non-variable populations that were less constrained by population or habitat size (an overview of all tests and results are available in Table S3). Low population size generally yielded higher FIS values, but this relation was non-significant (Kruskal–Wallis; X2= 7.25; P = 0.06). The correlation was tested significant when comparing threatened populations (<250 individuals) to non-threatened populations (>250 individuals), in which threatened populations had higher inbreeding coefficients (Mann-Whitney; W = 194; P = 0.01; Figure 3c). Although we found a weak positive correlation, a higher level of habitat fragmentation did not yield significantly higher inbreeding coefficients (Kruskal–Wallis; X2 = 4.96; P = 0.08). We did find that body weight negatively correlates with allelic richness (Spearman; R = 0.36; P = 0.04; Figure 3d) and expected heterozygosity (Spearman; R = 0.40; P = 0.02). To account for phylogenetic nonindependence of our species data due to shared ancestry, we used a PGLMM to test the effect of body weight on the presence of rare colour variants. The results suggests that variants do not vary in their occurrence in relation to body weight (PGLMM coefficient: 0.069 ± 0.088 SE, Z = 0.579, P = 0.42). The intercepts for species were tested significant, meaning that close relatives had similar values (PGLMM coefficient: 0.93 ± 0.45 SE, Z = 2.04, P = 0.04). Thus, the results imply an overall phylogenetic effect, and no (or little) body weight effect. The R2 value suggests that body weight alone explains approximately 15.9% of colour variation in populations, whereas this is reduced to 11.5% if phylogeny is incorporated.
Discussion
The aim of our study was to test whether rare colour variants in the wild may relate to a deficiency of heterozygotes, possibly resulting from habitat fragmentation and reduced effective population size. We found that variable populations displayed a significant reduction in heterozygosity and allelic richness compared to non-variable populations across species. We also found a significant correlation between population size and inbreeding coefficients, in which small populations suffer from larger heterozygote deficiencies. This supports the hypothesis that heterozygote deficiency in small populations might have an influence on the prevalence of rare colour variants due to the expression of rare recessive alleles. The high frequency of the rare colour variants in some populations also implies that genetic drift is occurring in small populations.
In 19 variable populations for which we found genetic data, there was some indication of a heterozygosity deficiency. This was particularly evident in charismatic species such as black tigers (Kolipakam et al., 2019), red leopards (Tensen et al., 2022), and white cougars Puma concolor (Saranholi et al., 2017). Non-variable populations of many species also suffered from heterozygote deficiencies due to demographic losses, but the difference with non-variable populations was nonetheless profound in some cases and significant across species. The direct comparison between variable and non-variable populations remains challenging, because stochastic events or loss of rare variants will interact with selective pressures (Caro & Mallarino, 2020). Furthermore, as our data is based on literature, we were not always able to include the populations of choice or retrieve historic frequency and presence / absence data. Studies with different sample sizes were also compared, which introduces the risk of bias, and the PGLMM showed an influence of phylogeny on colour variant presence. Despite these limitations, the loss of genetic diversity in populations where rare colour variants prevail is remarkable, and adds to the body of knowledge on the phenotypic effects of inbreeding in wild populations. The level of habitat fragmentation did not appear to influence the presence of rare colour variants, possibly because our dataset did not allow for enough fine-scale detail. The relationship between habitat fragmentation and genetic diversity remains indisputable (Templeton et al., 1990; Keyghobadi, 2007; Lino et al., 2019), and has been linked to rare colour variants before, such as for Glacier and Spirit bears Ursus americanus in Alaska (Lewis et al., 2020), where fjords act as barriers to gene flow. Likewise, spatial variation appears to be the primary driver of colour morph variation in dragon lizard Ctenophorus decressi in Australia (McLean et al., 2015).
Another effect of geographic dispersal barriers is that local effective population sizes are smaller, increasing the effect of genetic drift in relation to migration and selection as an evolutionary force (Runemark et al., 2010). A striking example of genetic drift in small populations is the tiger population in the isolated Similipal tiger reserve, India, where pseudomelanism reached a frequency of 78%, and in parallel an exceptionally low heterozygosity was observed (HO = 0.28) (Sagar et al., 2021). It has been revealed that nearly 14% of homologous alleles, both neutral and damaging, were fixed in small and isolated tiger populations (Khan et al., 2021). India has a particularly high degree of habitat fragmentation, and 56 mammalian colour variants have been recorded between 1886 and 2017 (Mahabal et al., 2019). A similar situation appears to occur in the highly fragmented Atlantic rainforest, where melanism in jaguars Panthera onca is increasing, reaching frequencies of up to 38% (Haag et al., 2010a;b). A leucistic cougar has been spotted in the same area, which was speculated to be a sign of inbreeding (Cronemberger et al., 2018). Likewise, in Eurasian lynx Lynx lynx a wide range of natural coat patterns exists, associated with geographic location (Darul et al., 2022), which frequencies have recently altered due to population bottlenecks resulting from habitat fragmentation (Kubala et al., 2020). In South Africa, red leopards are spreading in northern provinces, where effective population sizes are low due to high local population offtake (Tensen et al., 2022). Even though this population is not isolated, high population turnover has led to natal philopatry, increasing the overall relatedness in the population and chances of recessive alleles being expressed.
We expect this trend to hold true for many other wild carnivores, in which genetic drift is locally overruling natural selection in small effective populations. For instance, phenotypic differences in Arctic foxes Vulpes lagopusreflect distinct selective advantages across their range, however, a recent increase of blue homozygotes in Arctic foxes Vulpes lagopus Scandinavia is considered the result of genetic drift (Tietgen et al., 2021). Furthermore, it appears that melanistic and leucistic genets Genetta genetta and mongoose Herpestes ichneumonare much more common in Europe, where they have been introduced, compared to their natural range in Africa (Delibes et al., 2013; Descalzo et al., 2013). Although it has been suggested that the phenotypic variation in Spain may relate to a relaxation of selective pressures (Descalzo et al., 2013), the founder effect has been overlooked. Likewise, melanistic wolves are common in Yellowstone National Park, United States, where they were introduced from in and around Jasper National Park, Canada, where melanism is also common (Hedrick et al., 2014). Melanism has reached frequencies of approximately 50% in these areas, which is unlikely to relate to a concealment advantage (Dekker, 2009) or reproductive fitness (Stahler et al., 2013), and may instead be the result of genetic drift. The binary appearance of melanistic wolves, as well as coyote, across their range (Hinton et al., 2022) supports the notion that neutral, rather than selective, processes are at play. In European badgers Meles meles , a population has recently established in central Norway where leucism now seems abundant, likely as a result of a founder effect (Hofmeester et al., 2021). The same could be the case for a leucistic coyote Canis latrans in Costa Rica (Arroyo-Arce et al., 2019), and melanistic bobcats at the periphery of their range (McAlpine, 2021), although this is speculative.
A general trend of an increase in melanistic morphs towards the tropics and leucistic morphs towards the poles, as predicted by the Gloger’s rule (Caro & Mallarino, 2020), could not be detected in this study. For instance, melanistic bobcats Lynx rufus are readily found in the northern tip of the United States (McAlpine, 2021) and Florida (Regan & Maehr, 1990), and a melanistic Canada lynx was found in the northern tip of Alaska (Jung, 2023). Although the Canada lynx appears to be a single sighting and is likely to suffer a selective disadvantage due to increased visibility, melanistic bobcats have become fairly common (McAlpine, 2021). When looking at the map of colour morphs, we can also see the presence of many leucistic animals in the tropical forests of Mid-America, such as leucistic coyote Canis latrans , coatiNasua narica , and neotropical otters (Lontra longicaudis ). Furthermore, leucistic tayra Eira barbara have become abundant in the Guyana shield in Brazil (Mendes Pontes et al., 2020). This may further strengthen the theory that rare colour variants are a random, sometimes maladaptive, occurrence, instead of a selective driver in response to spatially or temporally variable environments (Svensson, 2017). When not influenced by genetic drift, rare colour variants can only be maintained through balancing or frequency-dependent selection (Gray & McKinnon, 2007), which seems unlikely for the before-mentioned examples. Furthermore, colour variants will only be fixed throughout the range of a species when they have a selective benefit, at which stage they are normally referred to as colour polymorphism (Svensson, 2017). Ecogeographical rules, as such, are more likely to imply to the latter category, and not to rare colour variants that seems more selectively neutral or disadvantageous.
It has previously been suggested that leucism mainly prevails in the Mustelidae, based on the number of published studies (Olson & Allen, 2019), for which we found no evidence during this study. In general, it appeared that colour variants are most abundant in the Felidae family (37.5% of cat species), representing 20 out of 59 colour variants recorded in this study. To compare, the proportion of species that contain rare colour variants is 23% in Canidae, 15% in Mustelidae, 12.5% in Ursidae, and 3% in Herpestidae and Viverridae (Table S1). Therefore, the high frequency of variable cats can be considered somewhat unusual. Furthermore, the PGLMM has illustrated a strong effect of phylogeny on the presence of colour variants, implying shared inherited traits among closely related taxa. Because phenotypic variation is a driver of genetic variation, it is possible that Felidae species contain more protein polymorphisms associated with pigment-type switching genes (Andersson & Purugganan, 2022). A similar range of coat patterns is observed in domestic cats, suggesting it is a conserved mechanism that can be altered by selection (Kaelin et al., 2012). Wild cat species exhibit the greatest diversity in colours and spots of all terrestrial carnivores, likely because they are more reliant on camouflage compared to for instance canids, which rely more on the chase (Schneider et al., 2012; Darul et al., 2022). However, it is worth noting that even though polymorphisms in wild canids are rare, domestic dogs show more phenotypic variation than any other domesticated animal (Kaelin & Barsh, 2013). This is due to genome-wide slippage mutations and pure repetitive sequences in domestic dogs, which occur at a much greater rate than in other carnivores, likely resulting from strong population bottlenecks associated with the origin of dog-breeds (Laidlaw et al., 2007).
Although we found that body size is generally a poor predictor of colour variation in populations, we did find a correlation between allelic richness and body weight. Furthermore, a high proportion (33%) of the colour variants occurred in large carnivore species (>10 kg), even though they have a low species diversity (representing 18% of all carnivore species). For instance, six of the listed colour variants were found in large cats (Panthera spp .), of which only five species exist. The potential bias towards large carnivores could also partially result from the fact that they are more extensively studied and more easily observed (Tensen, 2018). They also suffer more from demographic and genetic deterioration compared to medium-sized and small predators, based on relatively low effective population sizes (Lino et al., 2019). Even though we found no significant trend that apex predators contain more colour variants, this was previously found in birds. While polymorphisms occur in approximately 3.5% of all bird species, an estimated 30% of raptorial species are known to be polymorphic (Hugall & Stuart-Fox, 2012). In birds, colour polymorphism was associated with accelerated speciation in avian apex predators, due to a larger niche breath, consistent with theoretical models where speciation is driven by the fixation of one or more morphs (Hugall & Stuart-Fox, 2012). Polymorphic bird species also tended to be younger than monomorphic species, with shorter phylogenetic branches, which can further increase speciation rates (Gray & McKinnon, 2007; Forsman et al., 2008). For future research, it would be interesting to test whether this holds true for polymorphic carnivores as well.
In conclusion, we have collected some compelling evidence that heterozygote deficiency might contribute to the prevalence of rare colour variants in the wild, due to the expression of recessive alleles following reduction in effective population size. As such, it can be added to the list of phenotypic consequences of population bottlenecks in the wild, alongside fluctuating asymmetry and bone deformities. Further investigation of phenotypically variable populations is needed to determine the genetic basis and the adaptive and evolutionary significance of rare colour variants to help preserve genetic diversity. Ideally, a more comprehensive future study should directly sample, genotype, and compare variable and non-variable populations of species. With an increase in genomic studies, we may also gain more insights into random processes versus directional selection in small and fragmented populations, and their relationship to colour variation in the wild.