More Discus: Three or Two (Or Even One)?Author: Wayne S. Leibel
In the last several months’ columns I have reviewed the recent work from the lab of Swedish ichthyologist Sven O. Kullander and associates on the systematic status of discus (Ready et al., 2006. Journal of Fish Biology 69 (Supplement B): 200–211). As I reported, Ready and associates now suggest that there are three species, including the Heckel discus Symphysodon discus; the brown/blue discus S. aequifasciatus Pellegrin, 1904; and the red-spotted green discus, which they re-describe in the paper as S. tarzoo Lyons, 1959.
The evidence for that conclusion includes not only the morphological distinctiveness of these latter, red-spotted green discus but also the genetic distinctness of this western population based on DNA sequencing data. This is in contrast to the apparent shared genetic heritage/interbreeding of central/eastern Amazonian brown and blue discus (S. aequifasciatus) and, surprisingly, Heckel discus (S. discus). While the latter two species are easily separable based on appearance (morphological traits), their shared DNA sequence coding for the respiratory protein cytochrome b found in mitochondria suggests that they either are one species or are becoming one species despite their morphological distinctness.
If this interpretation of the DNA sequencing/genetic data is true, that would reduce from three to two the number of good species of discus recognized by science. Some have argued, even more parsimoniously, that discus are a single species of interbreeding geographically/morphologically variable populations (see Mazeroll and Weiss, 1995, below, and also my December 2006 TFH article, “How Many Discus Species Are There?”). In this lumper’s interpretation, which is not without basis, there is really only one discus species, S. discus. Which interpretation is the correct one? Are there one, two, three, or even more species of discus?
Let’s look at the genetic data that were presented by Ready et al. (2006) and reviewed briefly in my March column. The DNA extracted from 23 discus specimens (including nominal S. discus and S. aequifasciatus from along their extensive distribution) was sequenced with respect to two genes/gene segments: rhodopsin and cytochrome b. Rhodopsin is one of the vertebrate visual pigments (proteins) important for color vision. In this study, a 514-base-pair segment of the gene (a small piece) was sequenced from each of the sampled discus. The aligned rhodopsin gene segments showed no variation between the 23 individuals. This high level of base sequence conservation—no differences at all—indicates that either rhodopsin is a very important protein and mutations (base changes) are not tolerated (are eliminated by natural selection), or the divergence of the three morphological species has occurred relatively recently—too recently for mutational base changes to have accumulated in this gene. Cytochrome b is an important protein in the energy-generating respiratory pathway of mitochondria, the powerhouses of cells. Here, 1134 nucleotide bases of the gene were sequenced (most of it), and the resulting sequences were aligned and compared with each other in pairwise fashion for all 23 individuals. These results were even more startling.
Ready et al. found cytochrome b sequence differences that sorted all the sampled discus into two (and not three) distinct genetic lineages. One of these genetic lineages comprised the western populations of what they have now chosen to call S. tarzoo, the red-spotted green discus (total of 9 individuals sampled). The second and larger genetic lineage lumped central and eastern S. aequifasciatus (blue/brown—total of 12 individuals sampled) with S. discus (Heckel—total of 2 individuals sampled) even though these are clearly different and separable morphologically.
So the DNA data along with the reliability of the red-spotting character suggest strongly that the western populations of discus are reproductively isolated and genetically divergent enough for them to be accepted as a distinct species, which Ready et al. did indeed choose to re-describe as S. tarzoo in their paper. The DNA sequence data additionally suggested that a genetically distinct far-eastern lineage of “S. aequifasciatus” might also exist, possibly a fourth (or third, depending how you are counting) species. However only two sample fish from this geographic distribution were sequenced, and that is far too few to reach this or any conclusion. Nevertheless, the observation remains as a hanging genetic curveball in need of more sampling.
So, while looking so different, how can S. discus and S. aequifasciatus share the same (near identical) cytochrome b sequence? It is the elephant in the room, impossible to ignore and difficult to reconcile. There are a couple of possibilities: 1.) the genetic separation/speciation between S. discus and S. aequifasciatus is of recent origin and mutations/base changes have not yet accumulated in their respective cytochrome b genes (making cytochrome b the wrong molecular marker to use here); or 2.) S. discus and S. aequifasciatus are sharing cytochrome b DNA—that is, they are hybridizing. The fact that the western S. tarzoo cytochrome b sequence is easily distinguished from that shared by central/eastern S. aequifasciatus and S. discus (whereas no such case exists for rhodopsin) indicates that cytochrome b is indeed a useful molecular marker here (as opposed to rhodopsin), though it may still be too well conserved to permit total resolution of the three species at the genetic/molecular level.
Even so, Ready et al. report that the genetic divergence of the cytochrome b sequences between S. tarzoo and S. discus/aequifasciatus is of the scale (produce branch lengths on evolutionary trees) found between some genera of African Rift Lake cichlids. In other words, it is sufficient to recognize genetic isolation at the species level. The shared cytochrome b sequences of S. discus and S. aequifasciatus therefore strongly suggest that these fish are interbreeding, particularly since mitochondria and their genes are inherited exclusively through the mother (matrilineally) and create an evolutionarily readable pedigree. (In much the same way that human origins can be tracked molecularly back to a “mitochondrial Eve.”)
As previously mentioned (Leibel, TFH December 2006), there have been reports with persuasive photo-documentation of apparent natural hybridizations occurring between S. discus and S, aequifasciatus in the wild. Mazeroll and Weiss (1995; “The State of Confusion in Discus Taxonomy,” in: The Cichlids Yearbook Volume 5, Cichlid Press) argue that there is only one single discus species and that all discus are simply S. discus. Their primary evidence is/was some newly imported discus from the Rio Madeira that showed color patterns that were clearly intermediate between S. discus and S. aequifasciatus, having diagnostic elements of both to varying degrees. (Color photos of these fish appeared with the article.) These fish had the basic colorational appearance of brown/blue discus but had unequal vertical barring and expressed a very prominent fifth “Heckel bar.” They also had highly variable iridescent body vermiculations.
Marc Weiss was kind enough to send me video of some of these Madeira discus back in 1995, and I concur with his description and interpretation of their appearance (having just re-viewed the tape today in 2007). Indeed, a recent photo of these phenotypically hybrid fish by Oliver Lucanus accompanied my December TFH article (p. 107, bottom photo). In contrast, Ready et al. report that S. aequifasciatus and S. tarzoo overlap in distribution slightly in the Madeira River: one individual in their sample from this location had the phenotype (overall appearance) and cytochrome b sequence (haplotype) typical of western populations (i.e., S. tarzoo). Is it possible that S. tarzoo and S. aequifasciatus are interbreeding occasionally at this midpoint of their distribution, much as may be the case for S. discus and S. aequifasciatus in central/eastern Amazonia where the two come into contact? None of the fish they sampled speak to that, but maybe there really is only one species, as Mazeroll and Weiss strongly advocate.
For what it is worth, such opportunistic and promiscuous discus crossings have been observed in aquaria where discus breeders have set up mixed pairs in no-choice situations. Bleher (2006) in his book observes (based largely on observations made by legendary German discus breeder Eduard Schmidt-Focke) that all color varieties of blue and brown discus (S. aequifasciatus sensu Ready et al.) cross easily and their offspring have always proved fertile. (In fact, many of the cultivar strains are derived from such crosses coupled with line-breeding.) He also acknowledges that Heckel discus will, on occasion, cross with brown/blue discus to produce fertile intermediate offspring, but these break down (become sterile) after the F3 or F4 generation. The aberrant intermediate Heckel/blue forms he indeed reports from the wild at several locations, he writes, have all proved to be sterile natural hybrids. Finally, greens (S. tarzoo sensu Ready et al.) sometimes overlap with blue/brown discus in the wild, but never with Heckels, and they don’t cross successfully with either.
So, while instances of phenotypically intermediate individuals occurring in the wild are tantalizing, as are the cytochrome b DNA sequencing data, the genes we really would like to identify and sequence to answer the question of “how many species?” are those actually associated with the morphological changes classical ichthyologists use to discover and describe species. Alas, no one has yet achieved this. Once the control/regulatory genes can be identified that cause changes in the expression of color pattern and changes in the anatomical relationships of the highly changeable cichlid mouthparts that permit feeding specialization and niche diversification, we will have the real tools we need to chart the genetic divergences associated with morphological speciation and not simply unrelated genes that establish the timing of the speciation event (molecular clock), though this is useful too. Indeed, Ready et al.’s selection of rhodopsin as a second marker gene here makes great sense, since the morphological changes involved in discus speciation involve color pattern changes, and rhodopsin is an important protein of the vertebrate visual pigment system. Regrettably, however, there are no sequence differences noted between the three discus types/species for the small gene segment they analyzed.
Given our current inability to survey these morphologically relevant control genes, we eagerly await the further genetic analyses that have been promised by Heiko Bleher (Stolting, K., Salzburger, W., Bleher, H., and A. Meyer, 2006. Preliminary Revision of the Genus Symphysodon Heckel. Aqua, Journal of Ichthyology and Aquatic Biology: Special Publication, in press), which are supposedly near completion and publication to shed more light on this very curious discus situation. There is strength in data congruence: if multiple genes/data sets yield the same deduced evolutionary trees/relationships, then we have a better handle on and confidence in the findings. We also hope that his series of sampled discus will be greater in sheer numbers of individuals and represent a wider distributional set as well.
Failing that, Ready et al., or others, need to journey back to the Amazon and fill in some of the blanks with more collecting (easier said from the comfort of my computer, I know). I believe the Bleher/Meyer team is our best current hope for resolving the question of “how many species,” simply because of Heiko’s obsessional and exceedingly valuable sampling of obscure discus sites. Here’s hoping I am right!How many discus species are there? Maybe three; maybe two; maybe four (maybe more); or maybe just one. It all depends on what data you choose to accept as most powerful and conclusive and how you elect to interpret it. The story is far from over yet, and neither is my continuing treatment of it…sorry. Next time: why so few species and how did they arise?