Body size and allometric shape variation in the molly Poecilia vivipara along a gradient of salinity and predation

Classifying high-dimensional phenotypes with ensemble learning

Classification is a fundamental task in biology used to assign members to a class. While linear discriminant functions have long been effective, advances in phenotypic data collection are yielding increasingly high-dimensional datasets with more classes, unequal class covariances, and non-linear distributions. Numerous studies have deployed machine learning techniques to classify such distributions, but they are often restricted to a particular organism, a limited set of algorithms, and/or a specific classification task. In addition, the utility of ensemble learning or the strategic combination of models has not been fully explored.We performed a meta-analysis of 33 algorithms across 20 datasets containing over 20,000 high-dimensional shape phenotypes using an ensemble learning framework. Both binary (e.g., sex, environment) and multi-class (e.g., species, genotype, population) classification tasks were considered. The ensemble workflow contains functions for preprocessing, training individual learners and ensembles, and model evaluation. We evaluated algorithm performance within and among datasets. Furthermore, we quantified the extent to which various dataset and phenotypic properties impact performance.We found that discriminant analysis variants and neural networks were the most accurate base learners on average. However, their performance varied substantially between datasets. Ensemble models achieved the highest performance on average, both within and among datasets, increasing average accuracy by up to 3% over the top base learner. Higher class R2 values, mean class shape distances, and between- vs. within-class variances were positively associated with performance, whereas higher class covariance distances were negatively associated. Class balance and total sample size were not predictive.Learning-based classification is a complex task driven by many hyperparameters. We demonstrate that selecting and optimizing an algorithm based on the results of another study is a flawed strategy. Ensemble models instead offer a flexible approach that is data agnostic and exceptionally accurate. By assessing the impact of various dataset and phenotypic properties on classification performance, we also offer potential explanations for variation in performance. Researchers interested in maximizing performance stand to benefit from the simplicity and effectiveness of our approach made accessible via the R package pheble.

Population genetics and independently replicated evolution of predator-associated burst speed ecophenotypy in mosquitofish

Many species show replicated ecophenotypy due to recurring patterns of natural selection. Based on the presence or absence of pursuit predators, at least 17 species of fish repeatedly differentiated in body shape in a manner that increases burst swimming speed and the likelihood of predator escape. The predator-associated burst speed (PABS) ecophenotype is characterized by a small head and trunk and enlarged caudal region. Mechanisms promoting replicated phenotype-environment association include selection (without evolution), a single instance of adaptive evolution followed by biased habitat occupation, repeated instances of local adaptation, or adaptive phenotypic plasticity. Common garden rearing of mosquitofish, Gambusia affinis, demonstrated a likely heritable basis for PABS phenotypy, but it is unknown whether populations are otherwise genetically distinct or whether replicated ecophenotypy represents a single or replicated instances of adaptation. To genetically characterize the populations and test hypotheses of single or multiple adaptations, we characterized variation in 12 polymorphic DNA microsatellites in the previously studied G. affinis populations. Populations were genetically distinct by multilocus analysis, exhibited high allelic diversity, and were heterozygote deficient, which effects were attributed to G. affinis's shoaling nature and habitat patchiness. Genetic and phenotypic distances among populations were correlated for non-PABS but not PABS morphology. Multilocus analysis demonstrated ecophenotype polyphyly and scattered multivariate genetic structure which support only the replicated-adaptation model. As all of the diverse tests performed demonstrated lack of congruence between patterns of molecular genetic and PABS differentiation, it is likely that divergent natural selection drove multiple instances of adaptive evolution.

Impacts of the biocide chlorothalonil on biomarkers of oxidative stress, genotoxicity, and sperm quality in guppy Poecilia vivipara

Chlorothalonil is a fungicide present in antifouling paints and other formulations used in agriculture, although studies have shown this chemical to be toxic to fish species. To clarify the deleterious effects of chlorothalonil for these non-target organisms, the present study evaluated the toxic effects of this biocide for the estuarine guppy Poecilia vivipara in terms of an acute mortality test (96 h) and the analysis of biomarkers of oxidative stress, genotoxicity, and sperm quality. The LC₅₀ calculated for P. vivipara was 40.8 μg/L of chlorothalonil. For the analysis of biomarkers, fish were exposed (96 h) to 1 and 10 μg/L of chlorothalonil. It was observed that chlorothalonil alters the levels of pro- and antioxidants towards oxidative stress. In the gills, a negative effect on total antioxidant capacity (ACAP) was detected, while there was a reduction in the activity of glutathione S-transferase (GST) in the liver. However, levels of glutathione (GSH) and the activity and glutamate-cysteine-ligase (GCL) increased in both tissues, as a possible detoxification response. Following chlorothalonil exposure, oxidative damage measured by lipoperoxidation (LPO) significantly increased at the cellular level only (red blood cells (RBCs) and sperm cells). An increase in fluidity of membranes, reactive oxygen species concentration and micronuclei (MNs) incidence were also seen in RBCs. In sperm cells, LPO increased, while membrane and mitochondrial functionality as well as sperm motility decreased. Based on these results, chlorothalonil can be considered as a toxic compound for fish, causing genotoxicity and affecting the RBCs physiology and the fertility of males of P. vivipara.

Geographic variation of Moenkhausia bonita (Characiformes: Characidae) in the rio de la Plata basin, with distributional comments on M. intermedia

ABSTRACT Moenkhausia bonita occurs in numerous additional localities from the Bermejo, Paraná, Paraguay, and Uruguay river basins. Given that this finding greatly expands the distributional range of M. bonita, we carried out an intraspecific comparison, using multivariate methods for 18 morphometric and eight meristic characters taken from a comprehensive sample of 536 specimens. All localities were distributed in four major geographic groups as follows: Bermejo, Paraná, Paraguay, and Uruguay. Results of the morphometric comparisons showed significant differences among the studied groups except between the Paraguay and Uruguay groups. Statistical differences in meristic values were found for most between-group comparisons, especially in those resulting from discriminant canonical analyses (DCA). Specimens from the Bermejo basin were the most distinct group in most morphological comparisons. However, the overall subtle differences found in body morphology likely reflect intraspecific variation within M. bonita and seem to be mainly influenced by spatial and environmental features of drainages. As M. bonita was previously identified as M. intermedia in the río de La Plata basin, distributional comments on the latter species in that basin are provided.

Predation risk and abiotic habitat parameters affect personality traits in extremophile populations of a neotropical fish (Poecilia vivipara)

Understanding whether and how ambient ecological conditions affect the distribution of personality types within and among populations lies at the heart of research on animal personality. Several studies have focussed on only one agent of divergent selection (or driver of plastic changes in behavior), considering either predation risk or a single abiotic ecological factor. Here, we investigated how an array of abiotic and biotic environmental factors simultaneously shape population differences in boldness, activity in an open‐field test, and sociability/shoaling in the livebearing fish Poecilia vivipara from six ecologically different lagoons in southeastern Brazil. We evaluated the relative contributions of variation in predation risk, water transparency/visibility, salinity (ranging from oligo‐ to hypersaline), and dissolved oxygen. We also investigated the role played by environmental factors for the emergence, strength, and direction of behavioral correlations. Water transparency explained most of the behavioral variation, whereby fish from lagoons with low water transparency were significantly shyer, less active, and shoaled less than fish living under clear water conditions. When we tested additional wild‐caught fish from the same lagoons after acclimating them to homogeneous laboratory conditions, population differences were largely absent, pointing toward behavioral plasticity as a mechanism underlying the observed behavioral differences. Furthermore, we found correlations between personality traits (behavioral syndromes) to vary substantially in strength and direction among populations, with no obvious associations with ecological factors (including predation risk). Altogether, our results suggest that various habitat parameters simultaneously shape the distribution of personality types, with abiotic factors playing a vital (as yet underestimated) role. Furthermore, while predation is often thought to lead to the emergence of behavioral syndromes, our data do not support this assumption.

Geometric morphometrics, scute patterns and biometrics of loggerhead turtles (Caretta caretta) in the central Mediterranean

We investigate for the first time allometric vs. non-allometric shape variation in sea turtles through a geometric morphometrics approach. Five body parts (carapace, plastron, top and lateral sides of the head, dorsal side of front flippers) were considered in a sample of 58 loggerhead turtles (Caretta caretta) collected in the waters around Lampedusa island, Italy, the central Mediterranean. The allometric component was moderate but significant, except for the plastron, and may represent an ontogenetic optimization in the case of the head and flippers. The predominant non-allometric component encourages further investigation with sex and origin as potential explanatory variables. We also reported the variation of marginal and prefrontal scutes of 1497 turtles, showing that: variation of marginals is mostly limited to the two anteriormost scutes, symmetry is favored, asymmetry is biased to one pattern, and the variation of marginal and prefrontal scutes are linked. Comparisons with other datasets from the Mediterranean show a high variability, more likely caused by epigenetic factors. Finally, conversion equations between the most commonly used biometrics (curved and straight carapace length, carapace width, and weight) are often needed in sea turtle research but are lacking for the Mediterranean and are here estimated from a sample of 2624 turtles.

Morphological variation among populations of Hemigrammus coeruleus (Characiformes: Characidae) in a Negro River tributary, Brazilian Amazon

ABSTRACT We explored patterns of phenotypic variation in Hemigrammus coeruleus from the Unini River basin, a blackwater river in the Brazilian Amazon. Geometric morphometrics was used to evaluate variation in body shape among populations from four tributaries (UN2-UN5). We found no evidence for sexual dimorphism in body size and shape. However, morphological differences among populations were detected as the analyses recovered significant groups corresponding to each sub-basin, with some overlap among them. The populations from UN2, UN3 and UN5 had more elongate bodies than fish from UN4. The most morphologically divergent population belonged to UN4, the tributary with the most divergent environmental conditions and the only one with seasonally-muddy waters. The morphological variation found among these populations is likely due to phenotypic plasticity or local adaptation, arising as a product of divergent ecological selection pressures among sub-basins. This work constitutes one of the first to employ a population-level geometric morphometric approach to assess phenotypic variation in Amazonian fishes. This method was able to distinguish subtle differences in body morphology, and its use with additional species can bring novel perspectives on the evaluation of general patterns of phenotypic differentiation in the Amazon.

Multivariate heritability of shape in June sucker (Chasmistes liorus) and Utah sucker (Catostomus ardens): shape as a functional trait for discriminating closely related species

Molecular genetic data suggest that June sucker (Chasmistes liorus) is only shallowly diverged from the co-occurring but phenotypically distinct Utah sucker (Catostomus ardens) in Utah Lake. Phenotypes representing both extreme morphologies (June sucker and Utah sucker) are observed in the small wild population, but relatively large numbers of intermediate phenotypes are also present. This relatively continuous variation between the two putative species could result from extensive hybridization (including reticulate evolutionary patterns) of genetically based phenotypes or incomplete divergence among lineages and extensive phenotypic plasticity with little genetic basis. To help inform the evolutionary history of June sucker and to provide critical information for management and restoration of June sucker populations, we evaluated the distribution of shape phenotypes among purebreds of each species and their hybrids and determined the heritability of shape and ecological performance between June sucker (C. liorus) and its sister species Utah sucker (C. ardens). Mouth shape of purebred June sucker and Utah sucker are located at the extremes, and hybrids are located midway between the purebreds. Multivariate heritability was relatively high for mouth shape at 0.27. Heritability for growth rate was high at 0.32-0.42, but variation was not associated with cross type. Genetically based variation in mouth shape has evolved fairly rapidly or has been maintained in the face of ongoing hybridization between the two species. Currently, there seems to be little evidence for differential selection between species that would maintain shape variation.

Phylogeny and biogeography of Poecilia (Cyprinodontiformes: Poeciliinae) across Central and South America based on mitochondrial and nuclear DNA markers

Poeciliids are a diverse group of small Neotropical fishes, and despite considerable research attention as models in ecology and evolutionary biology, our understanding of their biogeographic and phylogenetic relationships is still limited. We investigated the phylogenetic relationships of South and Central American Poecilia, by examining 2395 base pairs of mitochondrial DNA (ATPase 8/6, COI) and nuclear DNA (S7) for 18 species across six subgenera. Fifty-eight novel sequences were acquired from newly collected specimens and 20 sequences were obtained from previously published material. Analyses of concatenated and partitioned mitochondrial DNA and nuclear DNA sets resulted in a well-supported phylogeny that resolved several monophyletic groups corresponding to previously hypothesized subgenera and species complexes. A divergence-dating analysis supported the hypothesis of the genus Poecilia dispersing into Central America in the early Pliocene (ancestors of Psychropoecilia+Allopoecilia+Mollienesia: 7.3-2.0Mya) from predominantly South America. Subsequently, one lineage (subgenus Allopoecilia: 5.1-1.3Mya) expanded deeper into South America from Lower-Central America, and one lineage expanded from Nuclear-Central America into South America (subgenus Mollienesia: 0.71-0.14Mya). The subgenus Mollienesia diverged into three monophyletic groups that can be identified by nuptial male dorsal fin morphology and inner jaw dentition. A subclade of the unicuspid short-fins (subgenus Mollienesia) was the lineage that expanded into South America during the middle Pleistocene. Species in this subclade are now distributed across northern South America, where they are partially sympatric with Allopoecilia. However the P. (A.) caucana complex was not monophyletic, with P. (A.) wandae clustering in the Mollienesia subclade that expanded into South America. It is apparent that characters (body size, scale count, pigmentation, and gonopodium morphology) used to define the P. (A.) caucana complex are homoplastic and suggestive of rapid convergence in northern South America. Our improved taxon sampling and divergence-time calibration allowed for insights into the timing and direction of dispersals, and provides an improved understanding of the biogeographic history of an enigmatic group of fishes. Furthermore, we provided strong evidence for the monophyly of the subgenus Mollienesia and further substantiated its species complexes; therefore, we advise a taxonomic re-evaluation for the P. (A.) caucana complex to maintain monophyly of both Mollienesia and Allopoecilia.

Body size and allometric shape variation in the molly Poecilia vivipara along a gradient of salinity and predation

Background

Phenotypic diversity among populations may result from divergent natural selection acting directly on traits or via correlated responses to changes in other traits. One of the most frequent patterns of correlated response is the proportional change in the dimensions of anatomical traits associated with changes in growth or absolute size, known as allometry. Livebearing fishes subject to predation gradients have been shown to repeatedly evolve larger caudal peduncles and smaller cranial regions under high predation regimes. Poecilia vivipara is a livebearing fish commonly found in coastal lagoons in the north of the state of Rio de Janeiro, Brazil. Similar to what is observed in other predation gradients, lagoons inhabited by P. vivipara vary in the presence of piscivorous fishes; contrary to other poeciliid systems, populations of P. vivipara vary greatly in body size, which opens the possibility of strong allometric effects on shape variation. Here we investigated body shape diversification among six populations of P. vivipara along a predation gradient and its relationship with allometric trajectories within and among populations.

Results

We found substantial body size variation and correlated shape changes among populations. Multivariate regression analysis showed that size variation among populations accounted for 66% of shape variation in females and 38% in males, suggesting that size is the most important dimension underlying shape variation among populations of P. vivipara in this system. Changes in the relative sizes of the caudal peduncle and cranial regions were only partly in line with predictions from divergent natural selection associated with predation regime.

Conclusions

Our results suggest the possibility that adaptive shape variation among populations has been partly constrained by allometry in P. vivipara. Processes governing body size changes are therefore important in the diversification of this species. We conclude that in species characterized by substantial among-population differences in body size, ignoring allometric effects when investigating divergent natural selection’s role in phenotypic diversification might not be warranted.

Electronic supplementary material

The online version of this article (doi:10.1186/s12862-014-0251-7) contains supplementary material, which is available to authorized users.