Figure 1.
Percent volume for the three dominant diet categories of the 23 species studied.
Categories are: benthic prey, epibenthic prey and fish. Species are arranged in decreasing order of percent benthic diet to illustrate the apparent pattern of exclusion between benthic and fish diets. Green stars show the distribution of the EBL, and orange species names indicate mouthbrooding lineages. See text and Table 1 for more details.
Table 1.
Percent volume of diet categories in 23 species of Neotropical cichlids and their phylogenetic signal.
Figure 2.
Graphic representation of the first three principal components of morphology.
PCA analysis included Ln(SL) and 9 size-corrected morphological variables for 55 species in 21 genera of South American cichlids, including 15 of 17 geophagine genera (missing genera are Teleocichla and Crenicara). Genera are labeled by color as per the inset legend. Polygons represent the total morphospace for each genus, with small points depicting each individual score and large points their centroid. Scores for morphological variables increase in the direction of the arrows. (See Table S3 for species list, sample sizes, morphological eigenvector values and score values for each individual fish).
Figure 3.
Pruned tree with 23 taxa used for diet and morphology analyses.
Photographs depict representatives of all genera analyzed in this study (except Dicrossus) and illustrate morphological variation among genera of Neotropical cichlids (see also Fig. 2). Species shown do not necessarily represent the same species studied. Inset is a diagrammatic representation of the first gill arch with the geophagine epibranchial lobe (EBL) highlighted within the circle. Green branches show the distribution of the EBL, and orange species names indicate mouthbrooding lineages. Photographs by H López-Fernández, K. M. Alofs and A. Lamboj.
Figure 4.
Procrustean superimposition of three first principal components of morphology and diet.
PROTEST permutation [62] revealed a significant association between diet and morphology without phylogenetic correction (m12 = 0.8015, p<0.03). The analysis was not significant after phylogenetic correction (m12 = 0.9572, p = 0.94, not shown). a and b, Procrustes superimposition plots of morphological variables (blue dots) and diet categories (end point of solid lines) for PC1 v. PC2 and PC1 v. PC3 scores, respectively. Solid lines represent residuals after the Procrustes superimposition procedure. Species codes correspond to the two-letter codes in Tables 1 and 2. c and d, directions of variation of morphological variables (blue) and diet categories (red) as indicated by their eigenvectors in the PCA analysis. The original diet eigenvectors have been transformed using the rotational matrix from the Procrustes procedure to maximize the association with the morphological eigenvectors. Photographs depict genera representing the associations between diet and morphology in their region of combined multivariate space. Photographs by H. López-Fernández, K. M. Alofs and A. Lamboj.
Table 2.
Principal components analysis eigenvectors for 7 diet categories and 10 morphometric variables associated with feeding.
Table 3.
Rotational matrix and rotated diet PCA eigenvectors from Procrustes superimposition of diet and morphology.
Table 4.
Mean values and phylogenetic signal of the 10 untransformed morphometric variables used in comparative analyses.