Selection, hybridization, and the evolution of morphology in the Lake Malaŵi endemic cichlids of the genus Labeotropheus

Abstract

The cichlid fishes of Lake Malaŵi are the paramount example of adaptive radiation in vertebrates. Evidence of their astounding diversity is perhaps most visible in their adaptations for obtaining food; the genus Labeotropheus, due to their prominent snouts, are an interesting example of an extreme adaptation for feeding. Two different body types are found in this genus: a deep-bodied form (e.g., L. fuelleborni) found most often in turbulent shallow water; and a slender bodied form (e.g., L. trewavasae) found in structurally-complex deep water habitats. Here we test the hypothesis that L. trewavasae should suffer a loss in fitness, measured as growth rate, if raised in turbulence; additionally, we examined growth and morphology of L. fuelleborni and L. fuelleborni x L. trewavasae hybrids under these conditions. We did find the predicted loss of fitness in turbulent-raised L. trewavasae, but found no loss of fitness for L. fuelleborni in either condition; hybrids, due to an unusual morphology, performed better in turbulent as opposed to control conditions. Fitness in turbulent conditions was dependent upon morphology, with deeper bodies and upturned neurocrania allowing a greater growth rate under these conditions. Directional selection on morphology was crucial in the evolution of morphology in the Labeotropheus.

Figure 1

The overall dimensionality of morphology among all three species. RW 1 explains 21.89% of the observed variation in body shape, and primarily accounts for the difference between a slender body with a downwardly-pointing head and a deep body with an upwardly-tilted head. RW 2 explains 14.41% of the variation in body shape, and accounts for the difference between a slender body with a dorsoventrally-compressed head and a deep body with an expanded craniofacial region. The deformation grids show the morphologies associated with the extremes of each axis, and are positioned accordingly.

Figure 2

Differences in body shape among treatment groups at Day 45. This is the same RW plot as shown in Fig. 1, but the treatment groups have been identified by symbol (circle = control treatment; triangle = turbulent treatment) and color (green = L. fuelleborni control; red = L. fuelleborni turbulent; pink = hybrid control; aqua = hybrid turbulent; blue = L. trewavasae control; orange = L. trewavasae turbulent); the colors of the deformation grids also correspond to these treatment groups. In the turbulent-raised L. fuelleborni andL. trewavasae, note the expanded and upturned heads, especially in comparison to their respective controls. In the hybrid fishes, note the deep body but exaggerated breadth of the craniofacial area in the control cohort, especially in comparison to the L. fuelleborni and L. trewavasae controls; conversely, the turbulent-raised hybrids are much more similar to the turbulent-raised cohorts of the purebred parental species.

Figure 3

Comparisons of neurocranium and jaw angles among treatment groups, and examinations of possible feeding angles resulting from these morphological angles. Colors identify treatment groups, and are the same as in Fig. 2. From left to right: Panel (a) compares the jaw and neurocranium angles of L. fuelleborni control (green) and L. trewavasae control (blue), and further illustrates the landmarks that comprise each angle (jaw angle: landmarks 1, 14, 13; neurocranium angle: landmarks 1, 18, 3; see Fig. 5 in the Methods for a description of all landmarks). Panel (b) superimposes all landmarks of the control groups of L. fuelleborni and L. trewavasae as a heuristic to compare an approximate feeding position these fishes would have assumed in this experiment. Panel (c) comparison of jaw and neurocranium angles of L. fuelleborni control (green) and L. fuelleborni turbulent (red), including the average angle and standard error (SE) for both groups; Panel (d) superimposition of all landmarks of L. fuelleborni control and L. fuelleborni turbulent specimens. Panel (e) comparison of angles of hybrid control (pink) and hybrid turbulent (aqua); Panel (f) superimposition of all landmarks of hybrid control and hybrid turbulent. Panel (g) comparison of angles of L. trewavasae control (blue) and L. trewavasae turbulent (orange); Panel (h) superimposition of all landmarks of L. trewavasae control and L. trewavasae turbulent. Note the increased feeding angles in turbulent-raised L. fuelleborni and L. trewavasae, especially in comparison to their control-raised counterparts. See Table 2 for analyses of differences of angles among treatment groups.

Figure 4

Fitness curves of control- and turbulent-raised specimens. Panel (a) depicts the fitness curve for body shape in the control condition; Panel (b) is the deformation grid that depicts the morphology with the highest fitness in this control treatment, which is a straight, somewhat deep body with no curvature to the neurocranium, similar to that found in L. fuelleborni. Panel (c) displays the fitness curve for the turbulent condition, and Panel (d) is a deformation grid depicting the morphology of maximum fitness depicted in the turbulent condition, which consists of a deep, somewhat curved body that results in an upturned neurocranium, which was found in L. fuelleborni and to a lesser degree in hybrids. The probability curves on the top and right sides of panels (a) and (c) show the distributions of RW1 scores and fitness, respectively. Panel (e) is a superimposition of the fitness curves, clearly displaying the directional selection against slender, straight bodies in the turbulent condition.

Figure 5

Homologous landmarks used for geometric morphometric analysis. For clarity, the landmarks are displayed on an adult L. fuelleborni from Chidunga Rocks. The landmark positions are described in greater detail in the Methods.