Selection towards different adaptive optima drove the early diversification of locomotor phenotypes in the radiation of Neotropical geophagine cichlids

Abstract

Background: Simpson envisaged a conceptual model of adaptive radiation in which lineages diversify into "adaptive zones" within a macroevolutionary adaptive landscape. However, only a handful of studies have empirically investigated this adaptive landscape and its consequences for our interpretation of the underlying mechanisms of phenotypic evolution. In fish radiations the evolution of locomotor phenotypes may represent an important dimension of ecomorphological diversification given the implications of locomotion for feeding and habitat use. Neotropical geophagine cichlids represent a newly identified adaptive radiation and provide a useful system for studying patterns of locomotor diversification and the implications of selective constraints on phenotypic divergence in general.

Results: We use multivariate ordination, models of phenotypic evolution and posterior predictive approaches to investigate the macroevolutionary adaptive landscape and test for evidence of early divergence of locomotor phenotypes in Geophagini. The evolution of locomotor phenotypes was characterized by selection towards at least two distinct adaptive peaks and the early divergence of modern morphological disparity. One adaptive peak included the benthic and epibenthic invertivores and was characterized by fishes with deep, laterally compressed bodies that optimize precise, slow-swimming manoeuvres. The second adaptive peak resulted from a shift in adaptive optima in the species-rich ram-feeding/rheophilic Crenicichla-Teleocichla clade and was characterized by species with streamlined bodies that optimize fast starts and rapid manoeuvres. Evolutionary models and posterior predictive approaches favoured an early shift to a new adaptive peak over decreasing rates of evolution as the underlying process driving the early divergence of locomotor phenotypes.

Conclusions: The influence of multiple adaptive peaks on the divergence of locomotor phenotypes in Geophagini is compatible with the expectations of an ecologically driven adaptive radiation. This study confirms that the diversification of locomotor phenotypes represents an important dimension of phenotypic evolution in the geophagine adaptive radiation. It also suggests that the commonly observed early burst of phenotypic evolution during adaptive radiations may be better explained by the concentration of shifts to new adaptive peaks deep in the phylogeny rather than overall decreasing rates of evolution.

Figure 1

Geophagini Maximum Clade Credibility (MCC) tree. The MCC tree has been pruned to include species used in comparative analyses and scaled to a length of 1. Crenicichla sveni and Teleocichla gephyrogramma were used to approximate age and phylogenetic position of closely related C. saxatilis and T. sp. “preta” respectively which were not included in the original phylogeny. Photos are included to illustrate some of the phenotypic diversity in Geophagini. From the top photos are: Mikrogeophagus altispinosus, Crenicara punctulatum, Geophagus aff. dicrozoster, Biotoecus dicentrarchus, Crenicichla sp. “Orinoco-wallacii, Teleocichla sp. ‘preta’, Crenicichla lugubris, Mazarunia charadrica, Taeniacra candidi, and Satanoperca daemon. The coloured branches correspond to the geophagine adaptive peaks identified by the SURFACE model. Green branches: Crenicichla-Teleocichla adaptive peak. Blue branches: benthivorous/epibenthivorous adaptive peak. The MCC tree is modified from [23]. Photos were taken by H. López-Fernández, J.H. Arbour, K.M. Alofs, N.K. Lujan and C.G. Montaña.

Figure 2

Phylogenetic Principal Components Analysis (PCA) for locomotor morphology based on the Maximum Clade Credibility (MCC) tree. Refer to Figure 1 for complete species names. The green and blue convex hulls indicate the areas of morphospace occupied by lineages belonging to the two geophagine adaptive peaks as identified by SURFACE. Green: Crenicichla-Teleocichla adaptive peak. Blue: benthivorous/epibenthivorous adaptive peak. Colours correspond to those on Figure 1. Text on the top and right margin of the plot indicate the trait complexes and functional implications of these trait complexes based on published literature (refer to text and Additional file 2 for details) at the extremes of PC1 and PC2. Numbers in brackets indicate the percent variance explained by each of the critical PC axes. Photos are included in the plot to show the variation in functional morphology along the axes and correspond to species present near the extreme of each axis. Starting with the photos at the top left corner and going clockwise, the species represented are: Crenicichla sp. “Orinoco-wallacii”, Geophagus aff. dicrozoster, Mikrogeophagus altispinosus, Guianacara dacrya, Teleocichla sp. “preta”, and Crenicichla lugubriscp The photo in the centre of the plot is Mazarunia charadrica. Photos were taken by H. López-Fernández, J.H. Arbour, K.M. Alofs, N.K. Lujan.

Figure 3

Disparity through time (DTT) plots for PC1 and PC2 axes. Grey lines show a random subset of 10 000 Brownian motion (BM) simulations that fall within the 95% confidence interval of the 1000 simulations performed for each of the 1000 posterior distribution trees. The dotted line is the mean change in disparity across all simulations and the solid black line shows the mean of the actual change in disparity across the 1000 trees. The observed DTT curve for PC1 was compared to simulations under the A) SURFACE model, B) early burst model, and C) BM model. D) The observed DTT curve for PC2 was compared to simulations under a single-peak Ornstein Uhlenbeck model.