Morphological Differentiation Between Introduced and Native Populations of Three Species of Cichlid (Perciformes)

Specimens examined.—

The specimens used in this study (Material Examined) are from the Nunnally Ichthyology Collection, Virginia Institute of Marine Science (VIMS, n = 129), the Field Museum of Natural History (FMNH, n = 128), the Museum of Comparative Zoology, Harvard University (MCZ, n = 53), the National Museum of Natural History, Smithsonian Institution (USNM, n = 45), the Louisiana State University Museum of Natural History, Louisiana State University (LSUMZ, n = 36), and the Cornell University Museum of Vertebrates, Cornell University (CUMV, n = 14).

Three species were studied: Pelmatolapia mariae (n = 111), Parachromis managuensis (n = 184), and Mayaheros urophthalmus (n = 110). They were distributed geographically as follows: Pe. mariae from Florida (n = 61), Cameroon (n = 49), and Ghana (n = 1); Pa. managuensis from Florida (n = 40), Guatemala (n = 128), Mexico (n = 10), Nicaragua (n = 4), and Honduras (n = 2); and M. urophthalmus from Florida (n = 28), Mexico (n = 36), Guatemala (n = 23), Belize (n = 10), Nicaragua (n = 8), and Honduras (n = 4). Specimens represented a broad range of sizes: Pe. mariae (native, 20.3–203.5 mm standard length [SL]; non-native, 22.2–155.6 mm SL), Pa. managuensis (native, 16.7–272.2 mm SL; non-native, 26.5–246.5 mm SL), and M. urophthalmus (native, 35.8–193.3 mm SL; non-native, 26.5–246.5 mm SL). Due to aspects of preservation, not all specimens were included in the final body-shape analysis (e.g., some specimens were preserved with open mouths or fins preserved out of position). Morphometric data were collected from all specimens for inclusion in linear morphometric analyses.

Qualitative species classifications.—

To consider the potential functional significance of observed morphological differences between native and introduced populations, we classified each species according to its broad trophic strategy. We paired this functional context with species-specific diet information from the literature and overall morphology to generate the following trophic classifications: Pe. mariae, herbivorous generalist; Pa. managuensis, predatory specialist; M. urophthalmus, predatory generalist.

Linear morphometric data collection.—

Specimens were photographed in left lateral view in a PhotoSimilie light box using a Nikon D3400 DSLR camera with an 18–55 mm lens. Measurements collected (Fig. 1A) are based on those from Bergmann and Motta (2005). Standard length (SL) was defined as the measurement from the anteriormost point of the lower jaw to the insertion of the central caudal-fin rays. Maximum body depth (MD) was measured by drawing multiple lines perpendicular to the anterior–posterior axis of the specimen, with the longest recorded as maximum depth. Head length (HL) was defined as the measurement from the anteriormost point of the lower jaw to the posteriormost margin of the fleshy opercular flap. In specimens in which the fleshy opercular flap was damaged, the margin was estimated. Snout length (NL) was measured from the anteriormost tip of the lower jaw to the anteriormost margin of the eye. Pectoral-fin length (FL) was measured from the base of the pectoral fin to the tip of the longest fin ray. Interorbital width (IO) was defined as the skull width at the midpoint of the eyes.

View Figure

• Download as Powerpoint
• Download Figure

Most linear morphometric data were collected from images using the software tpsDig (SL, MD, HL, NL; Fig. 2). Interorbital width and FL were collected using digital calipers to the nearest 0.01 mm. To measure IO, the calipers were positioned at the eyes on either side of the skull and closed until the tips contacted bone. Pectoral-fin length was measured from the left side unless the fin was obviously damaged, in which case the right side was measured.

View Figure

• Download as Powerpoint
• Download Figure

Analysis of linear morphometric data.—

To account for the potentially confounding effects of allometric growth on the population comparisons, we performed a growth trajectory analysis following the methods of Bergmann and Motta (2005). We generated linear models for each morphological variable-species-condition (condition: native/non-native) combination and compared each slope to a null value for isometric growth (m = 1), employing a Bonferroni correction (n = 30, α = 0.0017) to account for the high number of independent statistical tests.

To determine if introduced populations diverged morphologically from specimens collected in their native ranges, we organized the data into a set of ratios: SL/MD, SL/HL, SL/NL, SL/IO, and SL/FL. We analyzed the ratio data instead of the raw measurement data to remove any effect of differences in body size of the specimens. We assessed whether the variables were normally distributed both visually, by generating a series of histograms, and statistically, using Shapiro-Wilk tests for normality. Few variables were normally distributed, so we used non-parametric tests to analyze these data. We performed a series of Mann-Whitney U tests to compare native and non-native populations for each ratio. A Bonferroni correction was used due to the large number of repeated tests (n = 15, α = 0.0033).

Geometric morphometric analyses.—

To further explore divergence between non-native and native populations, we employed geometric morphometric analyses to quantify and compare external body-shape morphology. We placed a series of landmarks (n = 15, Fig. 1B) in repeatable, homologous points based on the landmarks used by Fruciano et al. (2016), with the goal of capturing major axes of variation, particularly those with high potential for functional relevance (e.g., oral jaw position, fin position, eye position, head size, etc.). We generated thin-plate spline (TPS) files composed of specimen images using tpsUtil (Rohlf and Slice, 1990), with one species per file. We randomized the orders of the specimens within the TPS files using tpsUtil to minimize bias in landmark placement. Landmarks were placed and digitized using tpsDig (Rohlf and Slice, 1990).

Some specimens were bent, and in these cases, we estimated landmark positions and noted which specimens involved landmark estimates. After landmarking all specimens, we reviewed landmark placement and removed all specimens for which landmarks could not be confidently placed (Pe. mariae, n = 11; Pa. managuensis, n = 15; M. urophthalmus, n = 5). Outliers were identified using the R package “geomorph” and defined as specimens for which the Procrustes distance from the mean shape was beyond the upper quartile for the species sample. We removed those outlier specimens that were distorted (Pe. mariae, n = 0; Pa. managuensis, n = 2; M. urophthalmus, n = 1). Outlier specimens that were identified to be accurately landmarked and physically acceptable were not removed from the analyses (Pe. mariae, n = 3; Pa. managuensis, n = 1; M. urophthalmus, n = 1). We performed these analyses twice: once treating native populations as a single group, and once splitting native populations by their geographic origin (locality analyses). However, the locality analyses revealed no significant, unique information. We therefore only report the grouped analyses herein.

Analyses of morphological variation.—

All geometric morphometric analyses were performed on data after outliers were removed or otherwise addressed. Prior to landmark data analysis, shape data were aligned for each species with a generalized Procrustes analysis. An allometric regression was also performed on each species to test whether there was an effect of size on shape variation and to remove this effect, if found. Principal component analyses were performed for each species. To test for differences in morphospace occupation between native and introduced populations, MANOVA tests were performed testing the null hypothesis that there is no difference in shape between native and introduced populations.

To test for differences in morphological variation between native and introduced samples for each species, Procrustes variances were calculated for each source condition for each species. For each set of Procrustes variances within a species, the ratio between variances of the native and introduced populations were calculated as follows: $$\mathit{Variance}\mathit{Ratio}=\frac{\mathit{Procrustes}\mathit{Variance},\mathit{Native}}{\mathit{Procrustes}\mathit{Variance},\mathit{Introduced}}$$

We then performed pairwise statistical tests on the null hypothesis that there is no difference in variance between native and introduced populations for each species.

Analyses of linear morphometric data.—

The growth trajectory for all species examined was found to be consistent with isometric growth (Table 1). There was no significant difference between observed and null slopes against the Bonferroni-corrected α.

Table 1.Growth trajectory analysis results for linear variables taken from each species; Bonferroni-corrected t-tests indicate growth trajectories for all variables not significantly different from isometry.

View Fullscreen

Linear morphometric data (relative to SL) yielded significant differences between native and introduced populations of each species (Table 2). In Pe. mariae (Fig. 3A–C), non-native specimens were found to have larger HL, smaller NL, and larger IO than native specimens, relative to SL (P < 0.05). In Pa. managuensis (Fig. 3D–F), non-native specimens were found to have larger MD, NL, and IO than native specimens, relative to SL (P < 0.05). In M. urophthalmus (Fig. 3F–H), non-native specimens were found to have larger MD and smaller IO than native specimens, relative to SL (P < 0.05).

View Figure

• Download as Powerpoint
• Download Figure

Table 2.Meristic ratio analysis results; all variables are represented as ratios between the variable and standard length (e.g., “Max depth” = MD/SL). Bonferroni-corrected Mann-Whitney U-Tests revealed eight significantly diverged variables across the three species.

View Fullscreen

Allometric growth analyses for geometric morphometric data.—

A significant effect of size variation on shape was detected in all three species (Pe. mariae, P < 0.0005; Pa. managuensis, P < 0.0005; M. urophthalmus, P < 0.0005; Table 3).

Table 3.Growth trajectory analysis results for geometric morphometric data; a significant effect of size variation on shape was detected in all three species.

View Fullscreen

Principal component analyses.—

In Pe. mariae, PC1 and PC2 accounted for 26.0% and 14.6% of the total morphological variation, respectively. Along the PC1 axis, specimens from introduced populations occupy morphospace beyond that of native specimens (Fig. 2A, Table 4). PC1 captures shape variation concentrated in the anterior portion of the specimens (e.g., in the jaw, snout, pectoral-fin position, and the posterior opercular margin; Landmarks 1, 3, 14, 15), as well as in the insertion point of the pelvic fin (Landmark 11). Both native and introduced specimens co-occupy morphospace along the PC2 axis, which captured variation primarily related to the anterior–posterior position of the pelvic-fin insertion. Variation along PC1 is plotted in Figure 4A. Moving positively along the axis describes a general shortening of the body and upturning of the mouth; specimens from their native range extended into the negative region, whereas specimens from introduced populations extended into the positive region.

View Figure

• Download as Powerpoint
• Download Figure

Table 4.Morphospace occupation results; there is support for significant differences between the morphology of native and introduced populations for each of the three focal species.

View Fullscreen

In Pa. managuensis, PC1 and PC2 accounted for 32.3% and 12.8% of the total morphological variation, respectively. The shape variation captured by PC1 (Fig. 2B) is concentrated around the jaw and midbody (Landmarks 1, 4, 10–15). PC1 also captures variation in general curvature along the anterior–posterior axis when comparing the representative specimens for the minimum and maximum values of the PC. PC2 primarily captures variation in the landmark at the convergence of the snout and forehead (Landmark 3) and the anterior insertion of the dorsal fin (Landmark 4; Fig. 4B). These two landmarks are closer to each other among specimens from their native range on the positive end of the PC2 axis and are more spread apart among specimens from Florida on the negative end of the axis. Introduced specimens of Pa. managuensis occupy regions of morphospace along PC2, which are unique relative to the morphospace occupation of native specimens (Table 4).

In M. urophthalmus, PC1 and PC2 accounted for 23.9% and 15.1% of the total morphological variation, respectively. The shape variation captured by PC1 is spread throughout the body but is most extreme at the anterior margins of the anal and dorsal fins and in the snout and jaw (Landmarks 1, 4, 10, 13; Fig. 2C). The axis of variation for PC1 relates to streamlining and elongation of the body and the anterior extension of the oral jaws on the negative end versus a generally deeper body and shorter head toward the positive end of the axis. Specimens from both native and introduced regions are represented among the positive values along this PC, but only native specimens are represented toward the negative end of this axis. The variation captured by PC2 is concentrated on the ventral edge of the specimens (Fig. 4C): at the posteroventral margin of the head, the pelvic-fin insertion, and the anterior insertion of the anal fin (Landmarks 10, 11, 12). The variation appears to be related to being narrower along the dorsal–ventral axis (i.e., upward movement of the ventral landmarks) versus a deepening of the body along those points.

Testing for differences in morphological variation.—

Variation in body shape was significantly different between native and introduced samples for each of the three species examined (Pe. mariae, P < 0.0005; Pa. managuensis, P < 0.0005; M. urophthalmus, P < 0.01; Table 5). For Pe. mariae, Pa. managuensis, and M. urophthalmus, the variance ratios (VR) are 1.47, 1.71, and 1.43, respectively. In all three species, morphological variation was greater among native samples than those from introduced populations.

Table 5.Morphological variation results; for each species, the variation in external morphology across specimens is greater for native populations than their introduced counterparts.

View Fullscreen

Morphological variation of introduced populations.—

Species introductions offer unique opportunities to explore the relationship between the morphology of an organism and its ecological context. Much like natural colonization events, in which populations evolved in one context may be introduced to a novel set of niches to which they must respond and adapt to survive, introduced populations encounter novel opportunities for diversification. Although the morphological response to introduction among plants has historically received much attention (Caplan and Yeakley, 2013; Martín-Forés et al., 2017), the morphological response to introductions among vertebrates is less well studied. Lima et al. (2012) observed phenotypic divergence among introduced populations of House Sparrows in Brazil associated with relatively weak genetic differentiation between populations. Another study observed morphological variation among an invasive bird (Acridotheres tristis) in South Africa and found that patterns in morphological variation were driven by spatial sorting, with dispersal-related traits found near the expanding edge of the introduced populations (Berthouly-Salazar et al., 2012). Surprisingly little direct study of the morphological differentiation between native and non-native populations has been made for fishes.

In this study, we complement the study by Gilbert et al. (2020), who compared the morphology of cichlid species before and after the damming of a clearwater river in Brazil. Their study found morphological divergence in six species from five genera spanning a range of ecological and morphological conditions. The observed morphological changes were attributed to changes in ecology associated with the impoundment of the river, and they were detectable as few as 34 years post-construction. Furthermore, their findings contrasted a similar study that reported minimal morphological response to damming in characids (Geladi et al., 2019). Our study investigates the morphological response to introduction among three cichlid species across three genera and a range of ecomorphs, with evidence of morphological divergence after as few as 25 years after introduction, thereby advancing the potential rate at which cichlids can respond morphologically to changes in environmental conditions.

Cichlids have undergone multiple notable adaptive radiations (Salzburger et al., 2005; Henning and Meyer, 2014) and are well known for their capacity to exhibit rapid phenotypic responses to novel conditions, especially through changes in head and trophic morphology in response to novel feeding opportunities (Meyer, 1987; Crispo and Chapman, 2010; Parsons et al., 2016). Introduced cichlids are established throughout the fresh waters of Florida (Bergmann and Motta, 2005; Harrison et al., 2013; Lukas et al., 2017; Robins et al., 2018), having achieved near ubiquity in the southern regions of Florida. Therefore, the Floridian populations of cichlids offer an ideal system to study the morphological effects of introduction on a species. The morphological differences we report herein between specimens from introduced populations of three species of cichlids from native populations and specimens from introduced populations from Florida suggest they have each undergone rapid morphological change. Further, these species represent three different feeding guilds, suggesting that morphological differentiation is not restricted to occupants of a single ecological niche, as these changes have occurred across a range of trophic strategies (i.e., both benthic feeders and piscivores experience morphological changes in non-native populations).

Trophic variation.—

Once a species is reproductively established within a region, trophic strategy, predator avoidance, and reproduction become among the most important factors in determining species success in the new environment (Ludsin et al., 2014). Pelmatolapia mariae is primarily an herbivorous trophic generalist from Africa, feeding mostly on a diet of algae and plant matter, but also opportunistically consumes all life-history stages of invertebrates (Oboh et al., 2019). In contrast, both M. urophthalmus and Pa. managuensis are primarily carnivorous, although M. urophthalmus feeds on a broad range of small fishes and invertebrates and may also feed on plant matter (Bergmann and Motta, 2005). Parachromis managuensis is a predatory specialist, using a highly protrusible upper jaw and caniniform teeth to capture small fishes (Del Moral-Flores et al., 2020). All three species are monogamous, pair-forming substrate spawners with similar levels of parental care, investment, and fecundity (FWC, 2021). The three species also attain similar adult sizes, which suggests they do not likely encounter substantially different predation pressures. Because of these broad similarities among the species, as well as that of their shared environmental conditions (e.g., water flow regimes), we focus on their disparate trophic ecology as their primary axis of diversity and consider the observed differences in morphology through this lens.

Geometric morphometric analyses revealed differences between native and introduced populations in Pe. mariae associated with the mouth and body length. Although there was considerable overlap between the morphospace of the two groups along this axis, native specimens extended beyond introduced specimens toward the negative end of the axis, i.e., associated with a longer body and more downturned mouth, whereas specimens from introduced populations have shorter bodies and more upturned mouths. A benthic ecology, where fishes primarily feed near or on the benthos, has been associated with a downturned mouth in multiple cichlid adaptive radiations (Cooper et al., 2010) and is consistent with the morphological patterns displayed by other fishes with benthic ecology. In contrast, more generalist trophic ecology is associated with forward-facing oral jaws in cichlids. The differences found between specimens from native and non-native populations that were detected in Pe. mariae could suggest a dietary shift from a specialized diet of algae and plant matter to a more generalist diet that includes more mobile prey. Such niche shifts in response to introduction to novel environments have been reported in other cichlids, including M. urophthalmus in Florida, which shifted from predominantly detritivory in its native environment toward greater consumption of animal prey in Florida (Bergmann and Motta, 2005). If this shift reflects greater animal prey availability in Florida compared to other regions, or easier prey capture of mobile prey due to a lack of appropriate predator avoidance adaptations in response to novel predator–prey interactions with cichlids, then the morphological changes observed in Pe. mariae might be explained by an increased reliance on mobile prey due to a changing cost-benefit relationship in the novel environment compared to their native range. Alternatively, changes in relative body length may be related to differences in flow regime between native and introduced environments. A similar disparity in body dimension was reported in another cichlid, Oreochromis mossambicus, and was correlated with local water flow conditions, with deeper-bodied fishes associated more strongly with low-flow environments and more streamlined fishes associated with higher flow environments (Firmat et al., 2012). Introduced specimens of Pa. managuensis occupy a more restricted morphospace compared to native specimens as a single group along the first principal component (i.e., body length) but occupy novel morphospace along the second principal component, which corresponds to snout length, (i.e., significantly larger snout length in introduced specimens compared to native-range specimens). Changes in snout length could reflect changes in the length of the ascending process of the premaxilla, which is related to the degree of jaw protrusion. Jaw protrusion as a mechanism for prey capture in suction-feeding fishes has been demonstrated to increase prey-capture efficiency in teleosts (Holzman et al., 2008), and an increase in total protrusion distance may confer further advantages in prey capture. This could support the idea that prey capture forms a more significant component of the trophic arsenals of Floridian cichlids in response to some change in the dynamics of predator–prey interactions between their native and non-native environments. We also observed an increased maximum body depth in introduced populations, which could be related to differences in flow regime between the two geographic regions. Alternatively, if a greater maximum depth corresponds to a deeper caudal peduncle, this could reflect a greater capacity for quick bursts of speed in fishes (Bejarano et al., 2017), which could allow for increased prey capture capacity and efficiency. This may reflect a greater reliance on elusive prey base in Florida, with selection among introduced populations favoring individuals with greater capacity for rapid acceleration.

Specimens of M. urophthalmus from introduced populations occupy a morphospace entirely within that of specimens from native populations. Only native-range specimens are represented at the negative end of PC1, which is associated with a relatively streamlined fish with a more upturned snout and oral jaws. In contrast, the positive end of the PC1 axis describes a deeper-bodied fish with a more forward-facing head and mouth. This is reflected in the MD measurement in the linear data analysis, which shows a greater MD among introduced specimens. As in Pe. mariae and Pa. managuensis, the difference in body dimension reflected in both geometric morphometric and linear data (MD) may reflect a difference in flow regime. However, it may also reflect a change in predatory behavior, as suggested for Pa. managuensis.

Possible basis for differentiation.—

The adaptive radiations of East African cichlids are examples of evolution through exploitation of novel niches, in which riverine species successfully colonized lacustrine environments through rapid ecological and morphological diversification (Salzburger et al., 2005). Species introductions represent an analogous situation: for an introduced species, a novel environment may present a range of novel niche opportunities, especially in the form of changes to the costs and benefits of trophic opportunities. These opportunities may also include release from competition, predators, or parasites that might limit foraging efficiency in the native range (Hulme, 2008). We demonstrate that these three species of cichlids are capable of differentiating in as few as 25 years since initial introduction to Florida. Gilbert et al. (2020) reported morphological disparity in cichlids that was putatively induced by the damming of a clearwater river in Brazil within as few as 34 years. Our study suggests that the timeframe in which observable morphological differences can be observed in a cichlid population is even fewer years following a major environmental change (i.e., introduction in a novel environment). Although we cannot conclude that such rapidly evolving characteristics are reflective of adaptive radiation, species introductions offer unique opportunities to better understand the mechanisms underlying speciation and adaptive radiations by providing selective pressures to which introduced species must respond to survive. Several mechanisms may serve as the basis for this response.

One possible explanation for differences observed between specimens from native and non-native ranges is phenotypic plasticity, which is the capacity of a single genotype to produce multiple phenotypes via environmental induction. Phenotypic plasticity has been well documented in the Cichlidae (Meyer, 1987; Magalhaes et al., 2009), especially as it is induced in trophic morphology by differences in diet. Phenotypic plasticity occurs within the lifetime of an individual, so it would be reflected in the samples collected from Florida. Both our study and that of Gilbert et al. (2020) are consistent with the rate at which phenotypic plasticity is known to act. Furthermore, we observed differences in functional traits, most notably the position of the insertions of the fin, broad body dimensions, and head morphology. Phenotypic plasticity may allow members of introduced populations to express novel, environmentally induced phenotypes, relative to their native kin. If the observed changes in morphology are due to phenotypic plasticity yielding morphological features beyond the range expressed in the native environment, this would suggest that the waterways of Florida possess different—perhaps more functionally or otherwise biologically demanding—characteristics, biotic or abiotic, from those found in the native ranges of Central America and Africa. Phenotypic plasticity has also been long suggested to support successful plant introductions into novel environments by expanding the range of conditions under which species can maintain positive population growth (Hulme, 2008). If phenotypic plasticity underlies the observed differences observed in this study, it would suggest that plasticity might play a significant role in successful vertebrate species introductions. Further study is needed to investigate the role that phenotypic plasticity has in how successful a vertebrate species may be when introduced into a novel environment.

Although phenotypic plasticity is a recurring theme in cichlid biology and potentially underlies the morphological variation observed here, it is also possible that the introduced populations in Florida were established by a morphologically distinct subset of the larger native meta-population. This is particularly plausible for M. urophthalmus, as it has been suggested that some populations of this species should be recognized as separate species based on their morphological differences (Barrientos-Medina, 1999). Previous work has also suggested that M. urophthalmus was likely introduced from multiple sources (Harrison et al., 2014). Moreover, Pa. managuensis was likely introduced multiple times through the aquarium trade, suggesting that it, too, may contain genetic material from multiple source populations. The role of the source population in observed variation will be important for future studies of the phenotypic diversity of invasive species.