In this study, we show what might be considered a paradox: a high level of morphological variability among and within Percichthys populations, sufficiently high that variants from different drainages have been deemed to be different species, yet we find no evidence of deep divergence in the entire haplotype tree for Percichthys east of the Andes, and only a very shallow phylogeographic structure for the region. We argue here that this pattern has been produced by two very different processes, operating on different time scales.
The morphological trait identified as best distinguishing groups of populations, and the only trait that differed more among than within populations, was the length of the dorsal fin spine. Long dorsal spines have been associated with predator avoidance in fish (Januszkiewicz & Robinson, 2007), and we found that variation among populations in spine length correlates with the density of conspecifics and introduced salmonids. Small Percichthys are vulnerable to predation by larger conspecifics, and perhaps also by some salmonids. Longer spines could be a defensive trait that has evolved in populations subject to high predation pressure, or could be a developmental response to predators, as has been shown for sunfish (Januszkiewicz & Robinson, 2007). Superimposed on the phenotypic differences among populations, there was strong evidence of marked intrapopulation phenotypic diversity involving mainly gill raker length, a trait often linked to resource acquisition in fish. Thus, it appears that the relative importance of the various processes responsible for the generation of within-population diversity may differ from those involved in allopatric differentiation among populations (see also Calsbeek & Cox, 2010). Regardless of which processes have produced phenotypic diversification within and among Percichthys populations east of the Andes, our molecular analysis indicates that the phenotypic differences do not correlate with genetic differentiation (as assessed from variation at the mitochondrial control region) (Fig. 1). Below we discuss the potential reasons for, and implications of, our results.
Absence of phylogeographic structure within thePercichthys truchacomplex
As we found previously with a smaller data set (Ruzzante et al., 2006), analysis of 21 populations of Percichthys, encompassing their full distributional range east of the Andes as well as some western populations, demonstrates little to no phylogeographic structure. The very shallow structure east of the Andes suggests that there must have been relatively recent mixing of populations throughout the region. We suspect that this mixing took place via two non-exclusive mechanisms. Firstly, there may have been exchange among now disjunct lakes that were part of larger proglacial lakes formed in front of the ice during the retreat of the glaciers. Secondly, and probably more importantly, individuals were probably able to move between current drainages via the many braided and deltaic connections that formed on the exposed continental shelf during glacial periods (Martínez & Kutschker, 2011; Ponce et al., 2011).
Species respond in different ways to repeated glacial advances and retreats, depending, in part, on characteristics such as cold tolerance and dispersal ability. Recent studies suggest that the various terrestrial Patagonian taxa did show distinctive responses, with some species surviving glacial periods in one to a few southern refugia, whereas others survived in and subsequently recolonized from northern areas (Sersic et al., 2011, Pardiñas et al., 2011). Likewise, there is evidence that some aquatic species appear to have survived in southern refugia (Zemlak et al., 2008, 2010), but it is likely that others were driven extinct locally. Percichthys is a relatively warm water-adapted Patagonian fish that reaches very high densities in warmer steppe lakes and reservoirs. If it survived in southern drainages through the glacial cycles, the refugia must have been on the expanded Patagonian surface that included exposed areas of the currently submerged continental shelf (Cavallotto et al., 2011; Ponce et al., 2011).
Several lines of evidence suggest that the exposed continental shelf may have provided refugial habitat as well as opportunities for the movement of Percichthys and other freshwater fauna among river drainages, from the Colorado River in the north to the Santa Cruz and Gallegos Rivers in the south. Bathymetric images provide evidence for the presence of endorheic basins with a circular morphology and a deeper centre than periphery in the present-day San Jorge, San Matías, and San José gulfs (Ponce et al., 2011). These areas would have been exposed during glacial periods, and were probably filled with shallow, relatively warm water. Shallow, productive basins are an ideal habitat for Percichthys. The exposed shelf is relatively homogeneous and flat, with a west–east gradient of generally < 1%, and is thus conducive to channel shifts and overflow during wetter periods. In addition, there is sedimentary evidence for the formation of braided river systems (which typically have unstable, shifting channels) and deltaic fronts (e.g. the Colorado and Negro systems; Martínez & Kutschker, 2011). Thus, there was considerable potential for the large, currently disjoint, Patagonian river systems on the continental shelf to have merged continually or intermittently during the extended full glacial periods that lasted tens of thousands of years, and probably also during the relatively short glacial termination periods, when river flows were high, up to ten times greater than today (Cavallotto et al., 2011; Martínez & Kutschker, 2011). The shallow phylogeographic structure that we see in Percichthys is thus likely to have been maintained through periodic mixing, produced by large-scale landscape changes that occurred as a function of changing climate through the Quaternary (Rabassa, Coronato & Salemme, 2005; Rabassa, 2008). The extent of mixing probably varied among freshwater species, and we speculate that species that inhabited the Patagonian steppe and/or the shallow lake and river environments on the now submerged continental shelf were most susceptible.
The remaining question, then, is what processes were likely to have produced such divergent morphology among and within populations. The lakes we surveyed span the latitudinal and elevational range of Percichthys in Patagonia east of the Andes, and differ greatly in the abiotic conditions that are associated with productivity. Deep Andean lakes of glacial origin are ultraoligotrophic–oligotrophic, with nitrogen levels sufficiently low so as to limit productivity (Soto et al., 1994; Diaz et al., 2007). Steppe lakes, on the other hand, tend to be warmer, to have higher levels of dissolved nutrients, and to be much more productive (Quirós & Drago, 1999; Diaz et al., 2007). Differences in productivity are likely to lead to differences in both resource/competitive and predation regimes.
By far the most important morphological character differentiating populations was the length of the dorsal fin spine, suggesting that predation (predator defence or predator avoidance) may underpin much of the among-population morphological diversity (see also Reimchen 1983 and Reimchen & Nosil 2002, and references therein). Percids typically erect their dorsal spine in response to piscivorous fish (Ylönen et al., 2007): longer spines presumably reduce predation risk, perhaps in part through increases in apparent size to gape-limited predators. Predators have been shown to induce morphological changes in fish through water-born chemicals, including the induction of longer dorsal spines in sunfish (Januszkiewicz & Robinson, 2007), and greater body depth in sunfish and other species (Brönmark & Miner, 1992; Langerhans et al., 2004; Januszkiewicz & Robinson, 2007). Body depth can also be related to predation risk through its effects on swimming performance and escape success (Domenici et al., 2008). The principal predation threats to young Percichthys in the study lakes are conspecifics (large Percichthys are partially piscivorous) and, perhaps, introduced salmonids (Macchi et al., 1999). The much longer spines in high-density populations of Percichthys, such as the steppe Lake Musters, probably result from an induced or evolutionary response to high predation intensity. The lack of genetic differentiation of this population, together with a plausible ecological explanation for its morphological divergence argues that the Percichthys species described for this lake, P. colhuapensis, is no more than a morphotype of P. trucha that develops under particular environmental conditions, i.e. high densities of conspecific and salmonid predators and/or competitors.
The other characters that differed somewhat among populations (head and upper jaw lengths and, to a lesser extent, mean gill raker length) are usually related to feeding, and differences probably resulted from variation in resources among the lakes. Resource availability and type can induce variation in trophic morphology in fish: individuals that feed primarily on zooplankton tend to develop more streamlined bodies, and a head morphology that can efficiently consume small pelagic prey (many long, closely-spaced gill rakers), whereas those that feed on benthic prey tend to develop deeper bodies, and sometimes longer, more robust jaws (Adams, Woltering & Alexander, 2003; Andersson et al., 2005; Yonekura, Kohmatsu & Yuma, 2007; Berner et al., 2008). In oligotrophic Andean lakes, Percichthys feed primarily on benthic macroinvertebrates, although small crustaceans (e.g. cyclopoid copepods and cladocerans) are consumed by juveniles (Ruzzante et al., 2003). Lake productivity affects the age at which young percids begin feeding on larger benthic invertebrates (Persson, 1987; Huss, Byström & Persson, 2008), and variation in the timing of diet shifts can lead to differences in trophic morphology. The most productive lake in our study (Lake Musters) was the only lake in which small Crustaceans (Cladocera) were a significant part of the diet of adult Percichthys, perhaps because of a greater availability of plankton or perhaps because of more intense competition for benthic resources, and the very distinctive morphology of individuals in this lake might therefore be resource related as well as predation related.
Other marked differences among lakes in the type of benthic prey consumed by Percichthys probably also reflect variation among lakes in resource availability/abundance. (As all fish were collected in the summer, in January and February, differences in diet across lakes are not likely to be greatly confounded by seasonal differences in composition of prey community). Some of the variation in diet could be associated with variation in morphology: for instance, populations that relied heavily on Odonata tended to have relatively short gill rakers and jaws compared with those that did not feed on Odonata. We do not know the nature of any links between diet and trophic morphology for Percichthys: adult morphology is almost certainly influenced by the diet of early developmental stages, and diet can also be affected by predation regime. Thus, some combination of differences in predation and resource regimes, both related to lake productivity, are likely to be responsible for the morphological diversity within the species. What is very clear is that very different morphologies have emerged in different aquatic environments without concomitant genetic divergence.
Percichthys is also known to have variable morphology within lakes (Ruzzante et al., 1998, 2003; López-Albarello, 2004). Our analysis showed that two characters (gill raker length and caudal peduncle depth) were much more variable within than among populations. Both traits are commonly linked to resource use. Competition for resources may promote diversification of resource use and associated divergence in phenotype (Lack, 1947; Schluter & McPhail, 1992; Schluter, 1994), although other processes such as predation can also play a significant role (Jablonski & Sepkoski, 1996; Rundle, Vamosi & Schluter, 2003; Langerhans et al., 2004; Nosil & Crespi, 2006; Meyer & Kassen, 2007). The population with the greatest interindividual variation in trophic morphology, the steppe Lake Musters (D. E. Ruzzante, unpubl. data), was also the most productive lake with the highest density of potential predators (Percichthys and salmonids). Thus, ecological factors (resource competition and/or predation) are likely to be responsible for within-population morphological variation as well as the differences among populations.
Speciation within PatagonianPercichthyidae?
Several of the morphological characters examined in this study (upper jaw length, body depth, and dorsal spine length) have been used to define different species within the genus Percichthys (Ringuelet et al., 1967; López-Albarello, 2004). We found similarly high levels of within- and among-population variability in these traits, and identified three broad morphological groups: one encompassing all Percichthys populations from the Limay and Futalaufquen river basins; one encompassing Perichthys from lakes Argentino, Pueyrredón, and Puelo; and one type from Lake Musters (Fig. 3A). In a recent and thorough attempt to sort out species designations within Argentine Percichthys, López-Albarello (2004) collapsed most of the previous species into a single species, P. trucha, but designated the Percichthys from Lake Argentino as P. laevis, whereas those from Lake Musters were presumed to be P. colhuapensis. We also found these populations to be morphologically distinct. However, the molecular differences (mtDNA and nuclear sequences) among individuals and populations that we describe here and in previous studies (Ruzzante et al., 2006, 2008) are much smaller than would be expected for species, or even subspecies, designation. All Percichthys populations in Patagonia east of the Andes thus appear to belong to the same species: P. trucha. We argue that different processes produced the spatial patterns of genetic versus morphological variation. Shifting aquatic landscapes during the Quaternary mixed the populations, producing a very shallow phylogeographic structure east of the Andes, whereas ecological factors (perhaps differences in predation and competition regimes) most likely account for the current morphological differences. We expect that populations of P. trucha do diverge over time in response to different ecological pressures in different environments, and that current P. trucha populations in Patagonia may be in some intermediate stage of an adaptive radiation. Divergence at the molecular level appears to be considerably short of speciation, however, perhaps because of the relatively young age of individual populations, or perhaps because of inconsistency through time in the direction or strength of selection pressures (e.g. Svanbäck & Persson, 2009). We suspect that these processes may have been repeated multiple times in the past: populations in different environments underwent partial but incomplete divergence caused, at least in part, by natural selection, but the divergence was then lost as climate change altered the landscape, allowing haplotypes from different populations to mix.