http://www.worlddiscus.com/swillis/papers/Willis_thesis_Ch2_Casiquiare_for_Cichla.pdf
Interesting read.
Discussion
Historical Demographics
Cichla intermedia
Cichla intermedia has the smallest distribution of the Cichla species distributed in
both the Orinoco and Negro drainages, (Figure 2.4A). Sampling did not extend into the
Negro river beyond the Casiquiare drainage, but C. intermedia have not been reported
previously from beyond the lower Casiquiare. It appears that the majority of C. intermedia
populations, from the upper Orinoco, Casiquiare, and the middle Orinoco, are characterized
by isolation by distance (Table 2.3), meaning that although dispersal occurs, it is reduced as
the distance between the populations increases. Judging by lower genetic diversity and lack
of exclusive haplotypes in the middle Orinoco (Cinaruco), this area may have experienced a
colonization wave that provided only a limited sample of the genetic diversity present
elsewhere in the upper Orinoco (Figure 2.4C). Therefore, it appears that C. intermedia
genetic diversity is strictly of Orinoco origin, and the Casiquiare has been a corridor for
dispersal of Orinoco diversity into the upper Negro (Amazonas) in recent history.
The population in the Caura river, however, appears to have been allopatrically
isolated for enough time to have developed an exclusive haplotype lineage that is
reciprocally-monophyletic with respect to the remainder of the localities. Additionally,
observations of fishes from the Caura have suggested that they may show divergent
morphology (Willis & Montaña, unpublished data). However, the sequence divergence of
this lineage relative to the rest of the population is relatively limited, and only a single
haplotype for this portion of the d-loop was observed (Figure 2.4C). Additionally, these
Caura specimens showed the same cytochrome b haplotype for which the rest of the C.
intermedia populations were found to be monomorphic (not shown here). The low genetic
diversity in this particular locality may indicate that the population has recently undergone a
genetic bottleneck, though additional analyses will be needed to confirm this hypothesis
(e.g. analysis of variation in nuclear markers).
Cichla intermedia and C. orinocensis sensu stricto are sister species (Figure 2.2), and
therefore by definition of equal age. Thus, it is curious to observe a disparity in their
respective genetic diversities (25% genetic diversity for C. orinocensis sensu stricto, but
18.8% genetic diversity for C. intermedia, Table 2.2; for a graphical representation of this
disparity, compare Figures 2.4B,C and 2.6B,C). Cichla intermedia apparently has stricter
habitat requirements than C. orinocensis, the former requiring rocky habitats in quickly
flowing water with high visibility, whereas the latter prefers more littoral, slow flow habitats,
often with lower visibility (Jepsen et al. 1997, Winemiller 2001). Rocky, clearwater
conditions are found mostly in the main channel of the Orinoco tributaries that radiate off of
the Guyana shield before they encounter the large, meandering, sediment-laiden main
channel of the Orinoco. This pattern has caused the populations of C. intermedia to occur in
disjunct pockets where appropriate habitat is available (Winemiller 2001). This disjunct
pattern (Figure 2.4A), and the subsequent isolation-by-distance that has apparently
resulted from it, may increase the susceptibility of these populations to loss of haplotypes
from genetic drift as an effect of lower effective population size (Wright 1931). Littoral
habitats with low visibility, however, are very common in the form of floodplain lagoons
along the main channel of the middle and lower Orinoco and tributaries in the llanos (plains)
and Orinoco delta, and consequently C. orinocensis are widespread throughout the Orinoco
and Negro river drainages (Figure 2.6A). This vast availability of habitat may have helped
maintain a larger effective population of C. orinocensis, thereby promoting the maintenance
of higher levels of genetic diversity.
41
Cichla monoculus
Relatively few clades of haplotypes from C. monoculus showed distributions that
were statistically different from panmixia, that is, than if they had been scattered at random
over the landscape (only 10 of 37 nested clades showed spatially correlated distributions
according to GeoDis, Appendix B). This stands in contrast to Figure 2.5C which shows
geographical localization of clade distributions.. However, the low number of significantly
localized distributions identified by GeoDis is likely an artifact of the distribution of sampling
localities over such a large region. The relatively large number of mutational steps between
haplotypes required in the nested clade analysis of C. monoculus may be a result of the
number of missing connecting haplotypes that might be found in geographically
intermediate areas. Because of the significant genetic distance between many haplotypes,
a great number of lower level nesting clades did not have genetic and geographic variation,
prerequisites for geographical analysis, and this resulted in an inflation of the number of
insignificant results,(i.e. distributions not different from random scatter). However,
compared with C. temensis (Figure 2.7C) and particularly C. orinocensis (Figure 2.6C), C.
monoculus actually showed fewer localities with haplotypes from multiple nesting clades,
except for the lower Amazonas river where genetic diversity was highest (Figure 2.5C). This
suggests that populations of C. monoculus are actually more structured that either of the
former two species.
The arrangement of genetic diversity for C. monoculus in the Amazon (Figure
2.5B,C) suggests a history of range expansion according to Templeton et al.s (1998)
criteria. Range expansion has been observed in other taxa when previously unavailable
habitat becomes available (e.g. after species introductions), and this may explain the
expansion of the C. monoculus population if this species had historically been restricted
from some parts of its current Amazonian range. The presence of only a single species of
Cichla in the upper Amazon is also an intriguing pattern, as it is the only part of the
distribution of the genus where only a single species inhabits a region without obvious
biogeographic barriers limiting dispersal of other Cichla species (see Willis et al. [Chapter 1]
for the overall generic pattern). Additionally, this pattern appears to be repeated in other
groups of fishes. Geophagus proximus is the only species of its genus known from the
Peruvian (upper) Amazonas, but it is also widespread in the lower Amazonas and Negro
rivers where it occurs with congeners (Reis et al. 2003). The upper Amazonas shows
similar water chemistry to the lower Amazonas and it is unlikely that an ecological gradient
would explain the inability of other Cichla species to colonize the upper Amazon.
Alternatively, the Purus arch, a subsurface topographic feature which borders the Solimões
sedimentary basin and lies approximately perpendicular to the course of the river just
upstream of the confluence of the Negro and Amazonas rivers (Lundberg et al. 1998b), may
have posed a historical barrier which does not currently restrict gene flow. At times of preerosion
and/or sea-level lows (e.g. Pleistocene), the arch may have been a more significant
barrier to dispersal into the upper Amazon. Breakdown of the Purus arch as a barrier to
dispersal into the upper Amazonas may have allowed recent range expansion into this
region. Finally, the monomorphism of haplotypes in the Orinoco and close relationships
with populations in the middle and lower Negro suggest this may represent a relatively
recent range expansion event for C. monoculus from the middle/upper Negro into the
Orinoco via the Casiquiare.
Cichla orinocensis sensu stricto
C. orinocensis sensu stricto, the Orinoco/Casiquaire population, showed the greatest
relative panmixia of species in this study (Table 2.3), with a high frequency of cases where
one nested clade was present in localities also inhabited by or predominated by other
nested clades (Figure 2.6C). Following the inferences from the GeoDis results (Table 2.3),
after the divergence of clade 3-3 in the northern llanos (localities Arichuna, Manipicito,
Cunavichito, Delta) and haplotypes in clades 3-1 and 3-2, a range expansion for what is
42
now the most common clade (clade 3-1) seems to have occurred. This range expansion
probably originated in the upper Orinoco and reached more distant habitats (the Pasiba,
middle and lower Orinoco localities), as evidenced by the relatively old genetic diversity that
occurs there (clade 3-2). Subsequent fragmentation and lineage sorting may also have
occurred for these populations. The Atures rapids near Ayacucho (see Ayacucho locality on
Figure 2.3) separate the upper from the middle and lower Orinoco, and have previously
been hypothesized as a biogeographic barrier to explain the restriction of some Orinoco
diversity to the upper Orinoco (Chernoff et al. 1991, Lovejoy & de Araújo 2000). The
majority of localities in which haplotypes in the northern llanos clade (3-3) were found lie in
or near the floodplain of the Apure river, a lower Orinoco tributary (Figure 2.1). This clade
may show an association with the high density of lagoon habitats in this area. However,
better sampling in the northern llanos and eastern north bank Orinoco tributaries are
needed to confirm that the Apure floodplain is the area of highest density and diversity for
this clade. Finally, demographic hypotheses for C. orinocensis suggest that recurrent
dispersal between different geographic areas may be occurring or have occurred recently
(e.g. long distance dispersal/colonization of haplotypes from Cinaruco/Capanaparo into
Arichuna/Manipicito and vice versa), causing the breakdown in historical genetic structure
among geographic areas.
The range expansion event from the upper Orinoco, in which the lower and middle
Orinoco and Casiquiare region were colonized, may also have been responsible for the
original presence of C. orinocensis morphotypes in the middle and lower Negro drainages.
Willis et al. (Chapter 1) proposed a hybridization event to explain the fact that individual
fishes from the Negro river that were morphologically identified as C. orinocensis had
haplotypes with close relationships to C. monoculus haplotypes. Hybridization is not
surprising between species diverged in allopatry where no selection pressure for
reproductive isolation per se would have been in action (Endler 1982, Hewitt 2001,
Templeton 2001). However, fragmentation of the Negro and Orinoco populations of C.
orinocensis has probably occurred since that initial dispersal event as no genetic admixture
was observed, and the age of the event (as evidenced by genetic diversification of the
Negro clade) would suggest isolation of significant duration.
Cichla temensis
The closest relatives of C. temensis occur in the Amazonas drainage, suggesting an
Amazonian affinity for this species (Willis et al. [Chapter 1]). Demographic inferences from
the current study and biogeographic context for this species from Willis et al. (Chapter 1)
indicate a pattern of range expansion from the Amazonas into the Orinoco (Table 2.3).
However, the relationships among populations suggest a complex biogeographic pattern
(Figure 2.7). Populations in the Negro (e.g. Unini) and upper Orinoco (e.g. Atabapo)
drainages are closely related, while the middle Orinoco (e.g. Cinaruco) and mainstem
Amazonas (Igapo-Açu) populations, near the opposite ends of C. temensis distribution,
group together. The sampling locality from the Casiquiare (Pasiba), in the middle of C.
temensis distribution, contained genetic representatives from four 1-step clades in the
nested haplotype network that were largely geographically localized in other areas. Put
another way, the Pasiba locality contained haplotypes that were also distributed in or closely
related to other haplotypes from the upper Orinoco, middle Orinoco, and mainstem
Amazonas (Figure 2.7C). This suggests that genetic representatives from throughout the
range of C. temensis may migrate through the Casiquiare corridor.
We expect that the overall pattern of geographic localization of haplotype clades
(multiple representation in the Casiquiare aside) suggests that C. temensis was the first of
the three Cichla species to disperse between the Amazonas and Orinoco basins. If
colonization of the Orinoco basin was particularly ancient with respect to other Cichla
species, a pattern of greater lineage sorting and local adaptation would be expected among
43
some populations of C. temensis subsequent to the establishment of stable rates of gene
flow, mutation, and genetic drift (Wright 1931).
Recurrent patterns
Dispersal between drainages
For each of the species under study a hypothesis of general panmixia, or unrestricted
interbreeding among all populations, was rejected (Table 2.3). However, deep genetic
divergences corresponding to the division between the Amazonas and Orinoco basins were
not observed (Figures 2.4-2.7). Had they been found, such deep divergences would have
suggested that the Casiquiare did not function as a corridor for movement of fishes between
the drainages. In fact, three of these species were inferred to have attained their current
patterns of distribution in the two basins as a result of dispersal or range expansion from
one basin to the other via the Casiquiare connection. The inferred events, in chronological
order, are
(1) the range expansion of C. temensis from the Negro river (Amazonas) to the
Orinoco basin; (2) the dispersal of C. orinocensis from the Orinoco to the Negro river; and
(3) the recent range expansion of C. monoculus into the Casiquiare and Orinoco basin from
the Negro river. Thus, this pattern conflicts with earlier studies that found limited
biogeographic connections via the Casiquiare (Lovejoy & de Araújo 2000, Turner et al.
2004). However, biological characteristics of Cichla may explain this difference. These
fishes are large-bodied and strong swimmers, potentially capable of dispersing over long
distances. Thus, they may have been some of the taxa most capable of crossing the
drainage divide. Fishes which are less labile for ecological or mechanical reasons may not
show the pattern of movement through the Casiquiare. Rather, their biogeography may
reflect the more ancient biogeographic connection between the Amazon and Orinoco that
existed prior to the late Miocene (Hoorn et al. 1994, Lundberg et al. 1998b).
Upper Orinoco and Caura/Caroní
It is interesting that the 3 species widely present in the Orinoco basin (C. intermedia,
C. temensis, and C. orinocensis) show a concordant biogeographic pattern. Cichla
intermedia show 2 major clades: the first is found mostly in tributaries of the upper Orinoco,
and the other in the middle Caura river (Figure 2.4C). Cichla orinocensis sensu stricto
(Figure 2.6C) and C. temensis (Figure 2.7C) show a similar pattern, with related haplotypes
in the upper Orinoco and Caura rivers (as well as Caroní for C. temensis). These concordant
patterns suggest that similar external biogeographic forces have shaped genetic diversity in
these species, probably with effects tempered by the biological characteristics (e.g. habitat
preferences vs. availability) of each species.
The pattern of close relationships between haplotypes in the upper Orinoco and
Caura/Caroní rivers to the exclusion of geographically intermediate Orinoco tributaries is
also interesting because it has been observed before. Fish species from several unrelated
groups (e.g. Leporinus brunneus, Chernoff et al. 1991; Geophagus taeniopareius, Kullander
et al. 1992) and freshwater invertebrates (e.g. Fredius estevisi, Rodriguez & Campos 1998)
show distributions restricted to northern and western Guyana shield tributaries (i.e. Siapa,
Ocamo, Ventuari, Caura, Caroní) to the exclusion of many geographically-intermediate
Orinoco tributaries. This suggests that the genetic diversity of Cichla species may have
been heavily affected by historical Guyana shield hydrography which may have differed
from the current arrangement. Although we cannot directly comment on a historical
connection between the upper Orinoco and Caura river (presumably through the Ventuari),
given the current dataset, we believe it is a possibility.
I think the purest pbass without hybridization in the wild belongs to azul but they do have some cheeks markign. 
he looks very different from my other two RNs
