Sympatry and syntopy of cichlids ( Teleostei : Cichlidae ) in the Selva Central , upper Ucayali river basin , Peru

Sympatry and syntopy are rare among congeneric cichlids in the Neotropics. Among Peruvian Amazonian cichlids congeneric syntopy has so far been reported in only 1 case known from a single locality. Based on thorough exploration of a new locality I refute the syntopy of the 2 species involved: Bujurquina robusta (Kullander, 1986) is a mouth-bay species while B. labiosa (Kullander, 1986) is an upland species. I demonstrate the generality of this ecological allotopy with further examples from a second study area including a different set of species.

set of species of Bujurquina (Kullander 1986) and the sister genus Tahuantinsuyoa (Kullander 1986).Here, species are reported as allotopic except for one locality with possible syntopy (Kullander 1986).Among other cichlid genera in Peru sympatry is found in Crenicichla Heckel, 1840 where there are however species from 2 different species groups involved that, in ecological, morphological, and biogeographical sense, should best be treated as separate genera (Kullander et al. 2010, Piálek et al. 2012).In the studied area in Peru sympatry of species within the species groups of Crenicichla is absent, as is syntopy between the species groups (Kullander 1986, Říčan pers. obs.).Virtually all Peruvian cichlids are known chiefly from taxonomic contributions featuring their descriptions (Kullander 1986, Kullander 1991) and to my knowledge there are to this day no studies about the biology, habitat preferences, and, especially, large and small-scale biogeography of Bujurquina and Tahuantinsuyoa, the very common and dominant Peruvian cichlids.
Many presently known distributions reporting sympatry and syntopy are artifacts of sampling effort (e.g.Albert et al. 2011) and of not collecting systematically within a hypothesisdriven framework (for example, they do not seek out sufficient ecological and elevational gradients, such as is relevant to the present study).Using field observations from a new locality I demonstrate that the reported syntopy in the B. labiosa and B. robusta species pair is an observational and collection artifact since they occur syntopically in only a transition zone along the elevational gradient of the tributaries.Both species are, in fact, almost exclusively allotopic within the tributaries as observed along an elevational gradient within a stream in pristine condition in the vicinity of the Ashaninka village of Betania.This location also represents a new distributional record for both species along the Ucayali-Tambo river.Bujurquina robusta is a lowland lentic mouth-bay species while B. labiosa an upland lotic species.I further demonstrate the potential generality of this ecological allotopy in Peruvian Amazonian cichlids with further examples from a second study area around Puerto Bermúdez in the Rio Pichis-Pachitea river basin of the Ucayali drainage with a different set of species.
My comparisons between the two study sites show that ecologically mediated allotopy is likely a typical micro-distributional pattern of sympatric congeneric cichlids in the Selva Central of Peru.
At the 2 locations in the Rio Tambo-Rio Ucayali basin (Samaireni and Aerija) I observed the 2 Bujurquina species (B.labiosa and B. robusta) supposedly endemic to this stretch of the Ucayali river basin, plus another upper Ucayali endemic species, C. sedentaria, known to have a larger distribution (Kullander 1986).Other cichlids with wide distributions in Peru caught or observed at my localities are not treated in this study.The study site at Betania (Quebrada Samaireni) represents a new distribution record for all 3 species and widens their ranges into the Rio Tambo basin.The Quebrada Samaireni at Betania is in pristine condition (see Fig. 2) and I made observation of its cichlid fauna by walking the whole stream from its mouth into the Rio Tambo all the way to about 750 m above sea level (a.s.l.) (see Fig. 1).Throughout the stream I also made observations of the cichlid fauna by snorkling.Voucher (Fig. 2).The stream has only a short lentic mouth-bay area with semiturbid to turbid water and a gravelly-sandy-muddy bottom.I studied the stream for 6 days between 25-9-2015 to 30-9-2015, just before the start of the rainy season during the lowest water levels (again confirmed by my Ashaninka guide).
During my boat journey to Betania and Atalaya I further explored the whole stretch of the Rio Tambo down from Puerto Prado.I found all tributaries of this stretch of the river down to Poyeni (where the river exits the Tambo canyon) dry during the dry season, which my Ashaninka guide confirmed as well.Cichlid distribution in these tributaries is thus ruled out; B. labiosa, due to its habitat requirements (see below), is likely not found upriver from Poyeni.Bujurquina robusta is more specimens were collected at 3 localities along the stream (localities P15-01 to P15-03; Fig. 2 and Table 1).The uppermost locality within the stream (P15-01) marks the end of perennial water in the stream.Above this point, the river, during the dry season, holds water only in several pools isolated by hundreds of meters of dry stream bed.The pools are fed by water seeping from the rocks.Between the pools the water flows beneath the river bed.The uppermost locality is thus, based both on my observations and the character of the stream, the upper limit of cichlid distribution within the stream (confirmed by my Ashaninka guide).For the majority of its length the stream is fast flowing and crystalline with isolated pools and the stream bed composed of boulders, rubble, coarse gravel, and sand  where I studied localities along the elevational gradient to the spur running from the Bosque de Protección San Matías-San Carlos located on the opposite side of the Cordillera El Sira (Fig. 1).See Table 1 for geographic (GPS) coordinates, elevation and other data on the localities.Within the Rio Pachitea basin of the Ucayali drainage several narrow endemic species of Bujurquina and the closely related Tahuantinsuyoa have been reported together with C. sedentaria and additionally with A. patricki.Most of the species are reported as allotopic within the the Pachitea basin (Kullander 1986).Detailed habitat preferences for the species have again not been reported, but at at least 1 site several species have been reported in possible syntopy (Kullander 1986).In the study area around Puerto Bermúdez I studied several tributaries and localities (P15-05 to P15-12) during 5 days from 2-10-2015 to 6-10-2015 at the end of the dry season during lowest water levels.The streams and localities were again explored by both fishing and observation (including snorkling where possible).The elevational gradient was studied by comparing the different localities and also along the largest tributary with the largest gradient (between the lowland locality P15-5 and the upland localities P15-9 to P15-11; Figs. 1, 3; Table 1).The elevational gradient was again studied by walking the length of the largest stream, in this case not in its entirety but only from the mouth into the transitional stage between the lowland and upland habitats, and then in the upland localities (P15-9 to P15-11).The lowland-upland boundary is in this stream much less pronounced compared to the Rio Tambo-Rio Ucayali left-hand tributaries (Fig. 1).The upland localities P15-9 and P15-11 are again characterized by strongly flowing clear water with a bouldery bottom and rare pools (Fig. 3) and are also at the uppermost limit of cichlid distribution within this tributary since the streams above these points carry very little water and are extremely shallow during the dry season.Localities P15-07, P15-10, and P15-12 are on shaded slow-flowing rainforest streams with sandy-muddy bottom and lots of submerged wood (I had no success in collecting cichlids at P15-07 despite the river looking healthy and suitable for them), while P15-05 and P15-08 are lowland stretches with exposed sky, close to mouths into the Rio Pichis, with gravellysandy-muddy bottoms and semiturbid to turbid water.Locality P15-06 is a human-made pool in a deforested cattle patch on a very small (originally rainforest) stream.I observed and collected in the study area around Puerto Bermúdez 1 Bujurquina species (B.megalospilus) and 1 Tahuantinsuyoa species (T.chipi) together with C. sedentaria and A. patricki.My B. megalospilus, T. chipi, and C. sedentaria records from around Puerto Bermúdez represent the first records of these species in the Rio Pichis drainage.Previously they were reported this far upstream only from the Rio Palcazú (Kullander 1986), which together with the Rio Pichis forms the Rio Pachitea.
Fishes were collected under a permanent permit to Hernán Ortega, Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (MUSM), Lima, Peru.Specimens were obtained by fishing using cast nets and gill nets in all available microhabitats at all localities.Localities were chosen specifically so that they capture the full regional ecological habitat variation of the two main studied areas and so that they especially capture elevational gradients within single tributaries at both studied areas.At localities with sufficient visibility (all upland and rainforest streams and some lowland locations) direct observation of the fish fauna was carried out using mask and snorkel.
Collected specimens (Table 1) were individually and permanently tagged and a DNA tissue sample taken from their right pelvic fin.DNA tissue samples were preserved in 1.5 ml Eppendorf Safe lock tubes in absolute ethanol and deposited at the fish tissue collection of the Department of Zoology, Faculty of Science, University of South Bohemia in České Budějovice, Czech Republic.Specimens were preserved in 10% formalin and later transferred to 70% ethanol and deposited at the fish collection of the MUSM, Lima, Peru.
Species were determined using the original taxonomic descriptions (Kullander 1986, Kullander 1991) that include the only review work (Kullander 1986) on the cichlid fauna of Peru.Morphological data are also compared with Kullander (1986).Detailed comparative ecomorphological analyses and molecular phylogenetic analyses are deferred to special publications.
Bujurquina labiosa and B. robusta were originally known from only 2 and 8 specimens, respectively, collected at a single locality (with the exception of 1 B. robusta specimen), a backwater on the R. Chinipo near Chicosa in the upper R. Ucayali (Kullander 1986).Bujurquina labiosa is actually solely known from this locality while specimens referred to as B. robusta have been reported from a wider area reaching into the Urubamba tributary of the upper Ucayali (Carvalho et al. 2011) and is so far known from 3 separate areas (localities; Fig. 1).
Live coloration of B. labiosa was previously unknown.The live coloration of my specimens (Fig. 4) differs from the description of preserved specimens by Kullander (1986) chiefly in having prominent horizontal opalescent blue lines, with 2 lines of these per 1 longitudinal series of scales (1 opalescent line through the dorsal part of the scale and 1 through the ventral part of the scale with the central part of the scale darkly pigmented).This is the same pattern as in B. robusta, where it is, however, more distinctly and symmetrically developed.In both species the opalescent lines are chiefly limited to below the upper lateral line.The blue coloration in B. labiosa is always darker than in B. robusta, in which it grades either into silvery or light golden.The opalescent blue head markings are also darker than in B. robusta, as are the leading edges of the ventral fins.In coloration B. labiosa is most different from B. robusta in almost completely lacking any traces of the suborbital stripe, which at best is visible only as a very diffuse blotch just below the eye.This condition corresponds to the only well preserved specimen of Kullander (1986).Another coloration character that well separates B. labiosa from B. robusta are the vertical bars, which in B. labiosa are much narrower and more distinct than in B. robusta (Fig. 4), again corresponding to the specimen shown in Kullander (1986).
The diagnosis of B. robusta given by Kullander (1986) also corresponds well to my specimens: 24 scales in a horizontal series above that including the lower lateral line (E1 scales); jaws equal anteriorly; lips thin; pectoral fin short, 30-33% of SL; lateral band usually continuous, running toward dorsal-fin; bars 5 and 6 separate; vertical suborbital stripe faint but not completely faded with increasing size; no preopercular spot; nape band strong; bar 7 blotch-like; posterior dorsal-fin lappets black with white edge; and otherwise spinous dorsal fin immaculate.
Live coloration of B. robusta was previously unknown from specimens confidently identified as B. robusta.A Bujurquina species photographed underwater by Vierke (1983a, b) was referred to as B. robusta by Kullander (1986).The description of the live coloration by Kullander (1986) mostly corresponds to my specimens.Bujurquina robusta live coloration (Fig. 4) differs from the preserved coloration chiefly in having prominent horizontal opalescent blue lines with 2 lines of these per 1 longitudinal series of scales (as in B. labiosa; see above).The prominent buccal lines (1 and 2) and dots are also bright, light blue, as in most other Bujurquina.The anterior edge of the pelvic fin is contrastingly white, again as in most Bujurquina.Contrary to Kullander (1986), observing from the color photographs of Vierke (1983a, b), both the spinous dorsal fin as well as the soft dorsal fin in my specimens are edged orange (Fig. 4) versus "spinous dorsal fin appears white-edged, the soft dorsal fin red-edged."All fins are light orange and the cheek has more orange tones than the body.The diagnostically very important suborbital stripe (Kullander 1986) in B. robusta is highly diffuse, not well developed, and long.In adults it is almost absent but more visible in juveniles.This is in strong contrast with the blotch-like remains of the suborbital stripe in B. labiosa (see above).
Bujurquina megalospilus, Tahuantinsuyoa, and A. patricki have previously been reported only from several localities within the Rio Pachitea basin and the adjacent Rio Aguaytia, both tributaries of the upper Ucayali.Tahuantinsuyoa has been split into two species, each supposedly endemic to each of the two tributary basins (Pachitea and Aguaytia; Kullander 1991), while B. megalospilus and A. patricki are treated as single species in both these non-communicating tributaries (Kullander 1986).
Bujurquina megalospilus has been diagnosed (Kullander 1986) by 23 E1 scales; jaws equal anteriorly; lips thin; pectoral fin long, 37.2-40.6% of SL; lateral band usually discontinuous; lateral blotch dominant flank marking; bars 5 and 6 separate; suborbital stripe vertical, indistinct in adult; no preopercular spot; nape band indistinct; bar 7 blotch-like; posterior dorsal-fin membranes dark with hyaline base, spinous dorsal fin otherwise immaculate.This diagnosis is well reflected by my specimens except that in none of my (live) specimens the lateral blotch (that gave the species its name) is well developed and so cannot be named as a dominant flank marking (see below that comparisons of live coloration versus preserved coloration do clarify this conflict).Since my specimens also represent a new distribution record, I intend to collect at the type locality and other close localities in coming years to further ascertain the species status of the specimens reported here.Kullander (1986) described the live coloration of his specimens but did not provide color photographs and thus comparisons can only be made with the description of the coloration.The holotype, when freshly collected, was white on chest, abdominal sides, and lower sides of head.My specimens are also whitish in these areas but this is uninformative since most Bujurquina (indeed, cichlids and fish in general) are whitish ventrally.My specimens (Fig. 4) have a distinctive orangish to bright orange background coloration with golden margins (or dots) of the midlateral blotch and the dark stripe anterior from the blotch.Kullander does not report any of this interesting and diagnostic coloration but reports that "near lateral spot, which not well discernible in live specimens, three scales with large silvery spots."This pattern refers to what I consider a diagnostic feature of my specimens and hence of B. megalospilus.The species name megalospilus is thus rather inappropriate since the large midlateral blotch is only a feature of preserved specimens when all other markings have faded.Most of my preserved specimens indeed have a prominent midlateral blotch.Kullander's specimens have head, back, and sides overall light café-au-lait (Kullander 1986).This coloration is not that different from my specimens (which are orange) and could be attributed to a slightly different part of the dry season of Kullander's specimens, water conditions (e.g., turbidity), and/or especially geographical distance, since my specimens are those most geographically removed.Buccal stripes 1-2 are bluish in my and in Kullander's specimens, as is the black iris, with light, reddish spot dorsally (in my specimens anterodorsally), and pupil ring golden (as in most Bujurquina).Suborbital stripe barely indicated in my specimens and similar to the other two featured Bujurquina species, but more similar to B. labiosa in being limited to just below the eye.Lateral band is dull black and vertical bars are faintly grey, showing on back only in my specimens.The usual opalescent horizontal lines of most Bujurquina are also light blue in my specimens, but arranged differently than in B. labiosa and B. robusta described above.In my specimens they run throught the center of the scales, and thus there is only 1 opalescent line per 1 horizontal scale series.This corresponds to what Kullander reported as "middle lateral line scales with silvery dot near tube."Also unlike the previous two Bujurquina the opalescent lines are only present ventrally from the mid-lateral blotch (versus ventrally from the upper lateral line).The overall expression of the opalescent markings is much less than in B. labiosa and in B. robusta (where they dominate the live coloration) both on the body and head.Fins are light oranage, as in both of the previous Bujurquina (and as in Kullander's specimens).The ventral fin also has a white leading edge.
Tahuantinsuyoa is, based on present knowledge (Musilová et al. 2008, Musilová et al. 2009), the sister genus of Bujurquina.Tahuantinsuyoa chipi is very similar to its sister species T. macantzatza (Kullander, 1986), differing in having a dark spot on the head (adjacent to the posterosuperior orbital rim), instead of a caudal-inclined dark nape band (connected with the transversal band posteriorly), and in having a long dark suborbital stripe below the eye of about uniform width, instead of one which is distinctly wider ventrally.This diagnosis of T. chipi well corresponds to my specimens (Fig. 4) even though the separate species status of the population might be seen questionable since the differences are very slight.My work, including phylogenomic analyses, will soon shed more light on this question.The long and well-pigmented suborbital stripe is the chief coloration pattern distinguishing Tahuantinsuyoa from Bujurquina in the area studied here (Fig. 4).
Live coloration of T. chipi has not been reported in ichthyological literature so far, but the species is very similar to its sister species based on my specimens and on the original description.The coloration pattern of the 2 species, besides 2 distinguishing features, is virtually identical.These distinguishing features become best evident in breeding coloration in which Tahuantinsuyoa become almost completely white (unlike any Bujurquina species where breeding coloration is known) with the suborbital stripe and nape band contrasting black.The posterior inclination of the nape band in T. Macantzatza also becomes much more visible (versus anterior inclination in T. chipi) (see also http://www.cichlidae.com).
Aequidens patricki is hard to confuse with any other Peruvian cichlid and easily identified using the original description (Kullander 1984) which also features pictures of live specimens.My specimens (Fig. 4) reflect well the diagnosis and the photographs of those live specimens.
Crenicichla sedentaria is also a distinctive species among Peruvian Crenicichla that is hard to confuse (Kullander 1986); it is readily separated by large flank scales (less than 70 in C. sedentaria versus 100 in C. johanna, Heckel 1840, and C. cincta, Regan 1905) and dental shape and colour features (inner teeth depressible, head compressed, absence of humeral blotch and suborbital stripe present in C. saxatilis (L.1758) species group).My specimens (Fig. 5) fit well to the diagnosis and the photographs of the live specimens, which, though not reported by Kullander in the original description, are available in several publications including online sources (e.g., http:// www.cichlidae.com) and are hard to confuse since this is such a distinctive species.Kullander (1986) gave the name sedentaria because tentativelly identified aquarium specimens (without locality information and general area of origin, but most likely from Colombia) were seen resting on the bottom.The species name sedentaria based on my results is rather misleading since I exclusively observed normal swimming behaviour in the species in the wild as in most other Crenicichla (unlike the benthic specialists from the Tocantins and Xingu rivers) and since the species is found to be ecotopically generalistic in all habitats studied.
Bujurquina labiosa has been reported as syntopic with B. robusta (at the previously only known locality in the Rio Chinipo, from Staeck & Linke 1985), from which it differs strikingly in mouth structures.This led Kullander (1986) to suggest that the co-existence is associated with different trophic habits.Kullander himself did not visit the site and all information was taken from Staeck & Linke (1985) who provided the specimens as well as locality and habitat observations.My data support the distinctiveness of the 2 species (see above and Fig. 4) but I did not find or observe them in syntopy in my elevational study along the pristine Quebrada Samaireni by Betania in the Rio Tambo drainage (Figs. 1, 2; localities P15-01-03; Table 1).I have only found them in syntopy in the human-polluted Quebrada Aerija close to the town of Atalaya (locality P15-04; see Table 1).Within the Quebrada Samaireni at Betania, B. robusta is limited to the lowermost portion of the stream with reduced to no flow close to the mouth of the Rio Tambo.This stretch of the stream is characterized by turbid slow flowing water with a gravelly-sandy-muddy bottom (Fig. 2).On the other hand I observed B. labiosa throughout the rest of the stream up to locality P15-01 (vouchers and DNA samples were also collected at Figs. 1,2).This whole stretch of the stream is characterized by long sections of fast flowing clear water with bouldery bottoms interspersed with rather rare pools, in which the concentration of B. labiosa was the highest (Fig. 2).Crenicichla sedentaria was observed throughout the stream from its mouth to the uppermost locality and does not seem to have as strict habitat requirements as either of the 2 Bujurquina species.
The Quebrada Samaireni near Betania is typical for the whole stretch of left-hand tributaries of the Rio Tambo-Upper Rio Ucayali.All these left-hand tributaries (including Quebrada Samaireni, the Rio Chinipo and Quebrada Aerija) show the same marked elevational gradient (Fig. 2) with markedly different habitat characteristics of the lowermost and the remaining stretches.The streams have short lowland stretches close to the mouth into the Rio Tambo-upper Rio Ucayali while most of their lengths have upland (highland) characteristics.Both B. labiosa and B. robusta appear to be endemic to this portion of the upper Ucayali river basin (including the lower Urubamba in B. robusta), with B. labiosa being an upland species while B. robusta is a lowland species.Following my observations the reported syntopy of the 2 species does not exist or at least is very rare in pristine streams with sharp lowland-upland transition zones (such as Quebrada Samaireni near Betania).However, in human-modified streams (such as Quebrada Aerija), or in longer streams with more protracted lowland-upland transitions, the 2 species can be found in syntopy in the transition zone.The previously known locality of both species on the Rio Chinipo is said to be on a farm and the stream was not surveyed in the same systematic manner as in my study of Quebrada Samaireni near Betania.The syntopy of the 2 Bujurquina species thus appears rare and the different trophic habits (as postulated from the different morphology) are not the result of co-existence but of different ecological habitats that each of the species inhabits.These habitats occur in allotopy along the elevational gradients of the tributaries of the Rio Tambo-upper Rio Ucayali.
Bujurquina megalospilus, B. hophrys (Kullander, 1986), B. apoparuana (Kullander, 1986), T. chipi, A. patricki, A. tetramerus (Heckel, 1840) and C. sedentaria have previously been reported as sympatric in the Rio Pachitea river basin (Kullander 1986).However, at most localities only 1 of the species was collected suggesting allotopy (or collection bias).At 1 location, Panguana in the Rio Llullapichis, a tributary of the middle Pachitea, Kullander (1986Kullander ( , 1991) ) reported 3 species (B.megalospilus, B. hophrys and T. chipi).The species were collected by Lüling (1981) and not by Kullander himself, who did not visit this site or the area, and, most important here, Lüling (1981) provides information that the specimens were collected in a wider area than just at the research station at Panguana.The syntopy of the species at Panguana is thus only conjectural.At my study site in the Rio Pichis tributary basin around Puerto Bermúdez I only found a subset of these species (B. megalospilus, T. chipi, A. patricki, and C. sedentaria).The missing species (B.hophrys, B. apoparuana, A. tetramerus) thus very likely do not reach this high up into the Rio Pachitea basin.I collected most of the species in allotopy (in agreement with the results of previous studies).Based on my study syntopy of the species is thus indeed rare and not a collection bias.The only habitat in which all the 4 species could be found in syntopy were small, shaded rainforest streams with moderate current velocities and lots of habitat heterogenity in the form of rubble, sand, and mud on the bottom, fallen logs, wood, and both small riffles and pools.By far the dominant species here is A. patricki; I have not been able to collect or observe this species in my study area in any other stream type.
On the other hand, I only found B. megalospilus in lowland open-sky semi-muddy mouth areas of tributaries.It will probably be also present in pools in the main river (though my collections there were not successful).
Tahuantinsuyoa chipi is found virtually only in upland stretches of the tributaries in pools and in the strongly flowing clear water over bouldery bottoms.I collected a few specimens in the rainforest streams, but not in the lowland mouth areas.Tahuantinsuyoa chipi and B. megalospilus are thus completely allotopic to the same extent and with the same habitat preferences and morphologies as B. labiosa and B. robusta (Fig. 4).
As in the Rio Tambo area I observed and collected C. sedentaria throughout the various habitats and localities without clear habitat preferences.
My comparison between the 2 study sites (Betania and Puerto Bermúdez) shows that allotopy is a typical micro-distributional pattern of cichlids within individual streams in the Selva Central of Peru.In my study area, upland and lowland species are separated at elevational and ecological transition zones within single tributaries which are well reflected in the ecomorphology of the lowland (lentic) and upland (lotic) species, each of which are represented by a distinct morphology (Figs.2-4).Upland species living in fast-flowing, clear-water streams with bouldery bottoms have thick lips, long heads, large mouths, longer jaws, preorbital distance, streamlined body, long caudal peduncle, and short streamers on pelvic fins (lotic ecomorphology), while lowland species living in proximity to mouths into large (here, always turbid) rivers in slow flowing to stagnant semi-transparent to muddy streams with gravelly-sandy-muddy bottoms have thin lips, short heads, smaller mouths, shorter jaws, preorbital distance, deep body, short caudal peduncle, and long streamers on pelvic fins (lentic ecomorphology; Fig. 4).The long head and thick lips in Tahuantinsuyoa are associated with stout pharyngeal teeth and the species are thought to be at least partially molluscivorous.The condition of the pharyngeal teeth in B. labiosa is unknown and presently under investigation to determine whether the supposed parallelism is also functional (but see notes below on phylogenetic relationships).
Small short streams with steep gradients (such as Quebrada Samaireni by Betania) have very sharp and short transition zones between lentic and lotic habitats (and hence cichlid ecomorphs) while in longer tributaries the transition zone is less pronounced and more diffuse, enabling overlap between the lentic and lotic ecomorphs.
Rainforest streams at intermediate elevations (in between the upland portions and the mouths) and ecologies (moderate flow, mixture of lowland and highland bottom types, etc.) have rich faunas that have additional cichlids (e.g., Aequidens species in the studied case) as well as both the upland and lowland species at low densities (Figs. 3,4).
This complexity of habitats (even within individual streams) and species restricted to them enables the existence of the large sympatric diversity of Amazonian cichlids.This biogeographical complexity needs to be taken into account when conducting and interpreting phylogenetic, phylogeographic, and biogeographic studies on cichlids in the Amazon.
From a systematic and taxonomic point of view the ecomorphological dichotomy between lotic and lentic ecomorphs in the fishes studied is reflected at 2 levels: between the lotic genus Tahuantinsuyoa and the predominantly (and likely ancestrally) lentic genus Bujurquina, and within the genus Bujurquina, most strongly between B. labiosa and its sister group or sister species, which are currently unknown but pending a molecular phylogenetic analysis currently being conducted (Říčan et al. in prep.).Bujurquina labiosa has not been studied for those anatomical characters that supposedly separate Bujurquina and Tahuantinsuyoa (Kullander 1986, Kullander 1991), nor was it included in the previously conducted molecular phylogenetic studies that found the 2 genera as sister groups (Musilová et al. 2008, Musilová et al. 2009).Apart from very few anatomical characters, virtually all external characters separating Tahuantinsuyoa from Bujurquina are ecomorphological characters not reflecting phylogeny but only habitat preferences (Říčan et al. 2016, this study).All of these are identical between Tahuantinsuyoa and B. labiosa (Kullander 1996), as already demonstrated by morphological phylogenies (Musilová et al. 2009).Morphological phylogeny with ordered multistate characters, which emphasizes ecomorphological characters (since most are meristic or body shape characters, hence, multistate), place B. labiosa as the sister species of Tahuantinsuyoa, which is an inner group of Bujurquina (Musilová et al. 2009).Only analyses with unordered multistate characters find the 2 genera monophyletic with the similarity between Tahuantinsuyoa and B. labiosa interpreted as convergence (Musilová et al. 2009).The few critical anatomical characters separating Tahuantinsuyoa and B. labiosa have not been checked in B. labiosa and thus it remains to be demonstrated that they are not ecomorphological characters as well and whether there is merit in keeping Tahuantinsuyoa as a separate genus from Bujurquina.The breeding coloration of B. labiosa is presently unknown but is very important to investigate and compare it with the supposedly diagnostic white breeding coloration of Tahuantinsuyoa.In Middle American cichlids it has been found that the breeding colorations with the greatest amount of pure white coloration are also found in species from highland clear-water streams while lowland species from more turbid waters have more obscured breeding colorations (Říčan et al. 2016).
The lentic-lotic ecomorphological dichotomy within Bujurquina (and between Bujurquina and Tahuantinsuyoa) in a clear sympatric setting is so far only known from the studied area in the upper Ucayali river basin (Kullander 1986, this study).Whether the absence of this phenomenon elsewhere within the large distribution of Bujurquina is real or just a sampling artifact remains to be investigated within this hypothesisdriven framework.If it should turn out as real, its presence within the upper Ucayali then remains to be explained.

Figure 1 .
Figure 1.Maps of studied areas and collection records.A. Overview map showing location of the studied areas.B. Detail of the location showing the 2 collection areas around the Cordillera El Sira.Colored dots show previous collection records of the studied species.Dark green -T.chipi; light green -B.megalospilus; orange -B.labiosa and B. robusta; yellow -B.robusta.C. Topographic map and the studied locations in the Rio Ucayali-Rio Tambo.D. Satellite map of the studied area at Betania with collecting localities.Red bar marks boundary of lentic and lotic habitats; white bar marks the end of the explored part of the stream.E. Topographic map of studied locations in the Rio Pichis area around Puerto Bermúdez.F. Satellite map of studied area around Puerto Bermúdez with collecting sites.

Figure 2 .
Figure 2. Studied localities and encountered Bujurquina species in Quebrada Samaireni at Betania.A, B. High elevation locality P15-01 (370 m a.s.l.; photos taken 26-9-2015) showing the semi-isolated pools within the steep river bed.The only Bujurquina species present was B. labiosa.C, D. Middle elevation locality P15-03 (260 m a.s.l.; photos taken 28-9-2015) showing the fast running clear-water stream and a pool.The only Bujurquina species present was B. labiosa.E, F. Low elevation locality P15-02 (242 m a.s.l.; photos taken 28-9-2015) close to the mouth into the Rio Tambo showing the slow flowing, turbid stream.The only Bujurquina species present was B. robusta.