Phylogenetic position and redescription of the endangered cichlid Nannacara hoehnei , and description of a new genus from Brazilian Cerrado ( Teleostei , Cichlidae , Cichlasomatini )

Rondonacara, gen. nov. is herein described as a member of the tribe Cichlasomatini, being considered as the sister taxon of Laetacara. The new genus differs from all the other genera of Cichlasomatini by having transversal streaks originated at the insertion of spines and softs rays of dorsal fin. Additionally, the new genus differs from the other genera of the Cichlasomatini by its unique combination of character states: longitudinal stripe of the flank horizontally oriented, ending directed to caudal-fin base; conspicuous suborbital bar; caudal-fin base spot on the middle of the fin and not ocellated; preopercle without scales; two supra-neural; and uniserial predorsal squamation. Rondona­ cara is monotypic, including only Rondonacara hoehnei, re-described herein. This species was originally described from a tributary of the Araguaia River, Mato Grosso state, with unprecise locality. Currently, Rondonacara hoehnei is known only from a single locality, a tributary of the das Mortes River, Araguaia River basin, in the Brazilian Cerrado. The restrict distribution area, and the widespread and accelerated destruction of natural habitat strongly indicate that Rondonacara hoehnei is critically endangered.

Nannacara hoehnei ribeiro, 1918 was firstly described on the basis of specimens collected during the "Expedição Científica Rondon-Roosevelt", an expedition led by Marechal Cândido Rondon andTheodore Roosevelt, between 1913 and1914.The type material was collected by the Brazilian botanist Frederico Carlos Hoehne, who was honored with the name of the species.Kullander (1983), in his revision of the Neotropical cichlid genus Cichlasoma swainson, 1839, provided a new arrangement for the taxa previously placed in Cichlasoma and Aequidens eigenmann & bray, 1894, and additionally N. hoehnei was transferred to the in certae sedis "Aequidens" guianensis group.Shortly after, Kullander & nijssen (1989) erected the genus Krobia Kullander & nijssen, 1989 to include species previously placed in the "Aequidens" guianensis group sensu Kullander, 1983 except "Aequidens" hoehnei and "Aequidens" potaroensis eigenmann, 1912, which remained incertae sedis.
The present paper aims to re-describe the poorly known species Nannacara hoehnei based both on types and recently collected material, as well as, to search for the phylogenetic position of N. hoehnei among Cichlasomatines using DNA sequences for the first time in order to evaluate its generic placement.

Morphological inspection
Morphological data were obtained from specimens fixed in formalin 10% for a period of 15 days, and then transferred to 70 % ethanol.
Measurements and counts were made according to ottoni et al. (2011) and ottoni et al. (2012).Counts of gill-rakers of the first branchial arch are expressed following the formula: gill-rakers on the epibranchial 1 + gill-rakers on the cartilage between the epibranchial 1 and the ceratobranchial 1 + gill-rakers on the ceratobranchial 1.In the description and tables, the number of specimens exhibiting a character state is presented in parentheses.Description of the colouration in life was based on five photographed specimens and on observation of all the collected specimens during the field work.
Measurements are presented as percentages of standard length (SL), except for those related to head morphology, which are expressed as percentages of both standard length (SL) and head length (HL).Measurements were taken on the left side of each specimen with digital callipers under a binocular microscope.Osteological studies were made on five cleared and counterstained (c&s) specimens prepared according to taylor & van dyKe (1985); the osteological nomenclature follows costa (2006).Nomenclature related to colour patterns follows Kullander (1983).The lateral band sensu Kullander (1983) is named "longitudinal stripe" here.

Euthanasia
The specimens were euthanized according to the guidelines of the Journal of the American Veterinary Medical As so ciation (AVMA Guidelines), and European Commission DGXI consensus for fish euthanasia.Specimens were submerged in a buffered solution of ethyl-3-aminoben zoat-methansulfonat (MS-222) at a concentration of 250mg/l, for a period of 10 min.or more, till the complete cessation of opercular movement.This substance acts on the nervous system, interrupting sensory input or muscle contractions till deactivation of the central nervous system.
Polymerase chain reaction (PCR) was performed in 30µl reaction mixtures containing 5 × Green GoTaq Reaction Buffer (Promega), 3.2 mM MgCl 2 , 1 µM of each primer, 75 ng of total genomic DNA, 0.2 mM of each dNTP and 1 U of Taq polymerase.Thermocycling profile was: (1) 1 cycle of 4 minutes at 94°C; (2) 35 cycles of 1 minute at 92°C, 1 minute at 58 -60°C and 1 minute at 72°C; and (3) 1 cycle of 4 minutes at 72° C. Negative controls were used to check DNA contamination in all PCR reactions.PCR results were visualized by means of agarose gel electrophoresis (1.8/1%) in TBE buffer.Amplified PCR products were purified using Wizard SV Gel and PCR Clean-Up System (Promega).Sequencing reactions were made using BigDye Terminator Cycle Sequencing Mix (Applied Biosystems).Cycle sequencing reactions were performed in 10 µl reaction volumes containing 1 µl BigDye 2.5, 1.55 µl 5 × sequencing buffer (Applied Biosystems), 2 µl of amplified products (10 -40 ng), and 2µM of primer.Thermocycling profile was 35 cycles of 10 seconds at 96°C, 5 seconds at 54°C and 4 minutes at 60°C.The sequencing reactions were conducted directly from purified and denatured PCR reaction products.Those samples were run on an ABI 3130 Genetic Analyzer.List of species and respective Gen-Bank accession numbers are shown in Tables 1  and 2.

Phylogenetic analysis
Four character partitions were defined for this study: 16S, Cytb, Rag1 and S7.Each partition block was singly edited using MEGA 5 (tamura et al., 2011) and aligned using Muscle algorithm (edgar, 2004) with following opening parameters G = -500 and extension parameter W = 0 for all genes fragments.The alignment obtained was checked visually.The complete molecular partition includes all sequence data merged into a single data matrix with a total of 3.901pb.
Phylogenetic relationships were estimated by Max imum Parsimony (MP), Maximum Likelihood (ML) and Bayesian analysis methods.The same data matrix was submitted to all the analysis.It was composed by 2.799 constant, 350 variable but parsimony-uninformative, and 711 parsimony-informative characters.MP was performed with TNT 1.1 (goloboFF et al., 2008), using the 'traditional' search and setting random taxon-addition replicates to 10, tree bisection-reconnection branch swap-ping, multi-trees in effect, collapsing branches of zerolength, characters equally weighted, and a maximum of 10,000 trees saved in each replicate.MP tree branch support was given by bootstrap analysis (Felsenstein, 1985), using a heuristic search with 1,000 replicates and the same settings used in the MP search, and saving a maximum of 1,000 trees in each random taxon-addition replicate; and them strict consensus tree was calculated.Acaronia nassa was defined as out-group.
The appropriated evolutionary model for each partition was estimated using JModelTest version 2. 1.3 (posada, 2008; darriba et al., 2012), through the Akaike criteria, AIC (aKaiKe, 1974).The best scored models found to each gene was: GTR + I + G for 16S and Cytb, GTR + I for RAG1, and HKY + G for S7.The ML analysis was performed in Garli 2.0 (zwicKl, 2006), which allow to apply different evolutionary models to each data partition.ML tree branch support was calculated with 1000 nonparametric bootstrap replicates using the same settings.
Bayesian analysis was performed in MrBayes v. 3.1.2 (ronquist et al., 2012).The same evolutionary models for each partition utilized in ML analysis were applied.Two independent runs of 3 million generations using four Markov chains and saving one tree every five hundred generations were performed.The program Tracer v 1.5 (rambaut, 2009; rambaut et al., 2013) was utilized to verify and evaluate the effective sample size (ESS) for all parameters for the analysis.Posterior probabilities were calculated and evaluated as a majority-rule consensus of the saved trees.

Phylogenetic analysis
Among the analysed characters, 2.799 were constant, 350 variable but parsimony-uninformative, and 711 parsimony-informative.The MP analysis generated one most parsimonious tree (total length 3014; consistency index 0.551; retention index 0.560; Fig. 1), which topology agreed with the results found for ML and Bayesian  (Figs. 1 and 2).All phylogenetic analyses recovered Nan na cara hoehnei as the sister group of Laetacara, with bootstrap support value of 99 in MP, and maximum support values in ML and Bayesian analysis (Figs. 1 and 2).However, the remaining species of the genus Nannacara, including the type species N. anomala, clustered together, and it was recovered as the sister-group of Ivanacara, with 99 of bootstrap support value in MP, and maximum support values in ML and Bayesian analysis (Figs. 1 and  2).
Monophyly of Laetacara was corroborated by MP, ML and Bayesian analysis.The clade L. fulvipinnis + L. dor sigera + L. curviceps + L. flamannellus was recovered as the sister group of the remaining species of Lae ta cara with elevated support values (Figs. 1 and 2).Relationships between this group and L. thayeri and L. araguaiae showed no resolution, resulting in a polytomy.In addition, other two clades were recovered: An dinoacara + Bujurquina + Tahuantinsuyoa, with maximum or 99 bootstrap support value (Figs. 1 and 2); and Cich la soma + Aequidens + Krobia, with lower bootstrap support value (55 in MP, 88 in ML) and posterior probability of 99 in the Bayesian analysis (Figs. 1 and 2).
Thus, the fact that Nannacara hoehnei failed to cluster with any historically allocated genera nor any available genus provided molecular evidence that it has been incorrectly allocated in Nannacara and that it belongs to a new separated genus.Therefore, we here propose the new genus Rondonacara (see below) to allocate the species Nannacara hoehnei.

Diganosis.
A typical cichlasomatine cichlid possessing only four pores on the dentary (vs.five in others tribe of the Cichlidae, except Heroini) and just three analfin spines (vs.more in Heroini).Rondonacara differs from all other genera of Cichlasomatini by presence of transversal streaks originated at the insertion of spines and softs rays of dorsal fin in live specimens, preserved ones retains just the proximal portion of the transversal streaks (Fig. 3) (vs.absence).Additionally, it differs from Acaronia (my ers, 1940), Bujurquina, Krobia, Andinoacara and Ta huan tinsuyoa by having the longitudinal stripe of the flank horizontally oriented, ending directed to caudal-fin base (Fig. 3) (vs.dorsally displaced, ending directed to the end of dorsal-fin base); from "Aequidens" paloemeuen sis Kullander & nijssen, 1989, "A".potaroensis, Cichlasoma and Aequidens by having the caudal-fin base spot on the middle of the fin (Fig. 3) (vs.dorsally displaced with major part of spot occurring above lower lateral line); from Cichlasoma and Aequidens by presence of a conspicuous suborbital bar (Fig. 3) (vs.suborbital bar transformed into a spot located below orbits [cheek spot sensu Kullander (1983)], and caudal-fin base spot not ocellated (Fig. 3) (vs.ocellated); from Ivanacara römer & hahn 2007, Laetacara Kullander, 1986, Nannacara regan, 1905and Cleithracara Kullander & nijssen 1989 by having unscaled preopercle (vs.with scaled); from Cleithracara, Ivanacara, and Nannacara by presence of two supraneuralia (vs.1), and from Laetacara by having uniserial predorsal squamation (vs.triserial).
Etymology.Rondonacara, a contraction of Rondon, in honour to Marechal Cândido Rondon, leader of the "Expedição Científica Rondon-Roosevelt", expedition during which the type series of this species was collected; and Acara, a vernacular name for cichlid fishes in Brazil.The name is masculine gender.
Description.Morphometric data are summarized in Table 3, meristic data in Table 4. Dorsal profile slightly convex from snout to caudal-peduncle origin, more conspicuous between tip of snout and dorsal-fin origin.Ventral profile slightly convex from snout to caudalpeduncle origin.Caudal peduncle approximately straight ventrally and dorsally.Body profile oval, laterally compressed.Mouth isognath.Three rows of teeth on both upper and lower jaws.Jaw teeth caniniform, hyaline, reddish at tip.Opercle not serrated.
Anterior portion of dorsal fin rounded, posterior region pointed.Tip of dorsal fin reaching vertical through ⅓ to half of caudal fin.Anal fin rounded anteriorly, pointed posteriorly.Tip of anal fin reaching vertical through ⅓ to half of caudal fin.Caudal fin rounded, with anterior ⅓ covered by cycloid scales of half size of flank scales.Pectoral fin with posterior margin rounded.Pectoral-fin  Trunk covered by ctenoid scales; scales above upper lateral line with few ctenii.Predorsal scales cycloids or with few ctenii; squamation pattern uniserial, composed by eight scales and occasionally more one smaller.Prepelvic scales cycloids.Region between dorsal and anal-fins spines and rays unscaled.Pectoral and pelvic fins without scales.Two scales between lateral lines, upper and lower lateral line scales not overlapping vertically, separated by one or two vertical scale rows.Three large scales between upper lateral line and dorsal-fin origin; one or two scales, upper with half of size of lower one, between end of upper lateral line and dorsal-fin base.Opercular, interopercular and subopercular scales cycloids.Three vertical rows of scales on opercle plate, and one row on subopercle and interopercle.Preopercle without scales.Cheek scales cycloid or with few ctenii, rarely typical ctenoid.Cheek with three scale rows on anterior portion (sometimes two), and two scale rows on posterior portion.
Ceratobranchial 4 with one or two tooth plates.Ceratobranchial 5 partly medially sutured and relatively robust, with 7 teeth along midline and 19 to 21 teeth along posterior margin.Posterior teeth usually more compressed.Posterior and medial teeth larger than lateral and anterior teeth.All teeth bicuspid, second cuspid more evident in larger teeth (Fig. 4).Presence of two supraneuralia, rectangular lacrimal (Fig. 5), infraorbitals with conspicuous laminar expansions (Fig. 5), and anterior ceratohyal without any depression on its margin or with depression very inconspicuous (Fig. 6).
Colour in alcohol (Fig. 3b -e).Side of body light brown; seven dark brown bars with two or three flank scales wide, between posterior limit of caudal peduncle and posterior margin of opercle.Bars from both sides of body not merging ventrally.First four bars located from end of caudal peduncle to beginning of anal fin.Trunk bar 7 only on region above mid-portion of trunk.Dorsal region of body darker than ventral region.Belly and chest pale brown, without bars.Two black spots; first spot not ocellated on middle of caudal fin, relatively elliptical in adults and sometimes rounded on juveniles, about one flank scale wide, crossed by lower lateral line on its half; second spot rounded, located on junction between longitudinal stripe and vertical trunk bar 5, about three flank scales wide.Interrupted longitudinal stripe dark brown between half of trunk bar 1 and posterior margin of opercle, lighter and inconspicuous between bars, as well as from bar 1 to bar 5.
Side of head of same ground colouration as trunk, darker on opercle and dorsal region.Dark brown blotch on nape.Suborbital bar from lower margin of orbit to posterior margin of preopercle, slightly inclined to posterior portion.No supraorbital bar.One black bar on head anterior of eyes.Three conspicuous black interorbital stripes; first positioned on half of eyes length, second on ⅔ of eyes length and third on anterior margin of orbits, colouring mid-dorsal region of upper jaw.
Dorsal fin without spot located on its base above trunk bar 5. Dorsal and anal fins grey-hyaline, with small light brown interrupted bars or dots on posterior portion of fins.Dorsal fin spines and sometimes rays with dark brown or black pigmentation on its insertion.Caudal fin grey-hyaline, without any dots or small interrupted bars.Pectoral fin grey-hyaline.Pelvic fin grey-hyaline, darker on rays.
Colouration in life (Fig. 3a).Body ground colouration light brown, darker on dorsal region.Ventral portion of trunk pale brown.Trunk bars dark brown; bars 2 -4 reaching anal-fin base, bars 5 -6 not reaching chest, bar 7 only above longitudinal stripe.Interrupted longitudinal stripe black, inconspicuous from trunk bar 1 to trunk Ground colour of dorsal-fin greyish hyaline, with yellow stripe on margin; posterior region of fin with yellow pigmentation.Few white dots present on yellow posterior portion of fin.Presence of transversal streaks origi-nated at the insertion of spines and softs rays of dorsal fin.Anal-fin yellow, with few white dots on base and posterior portion.Caudal fin yellow without dots, with brown coloration near base.Inconspicuous blue stripe present on posterior margin of fin.Pectoral fin yellowish hyaline.Pelvic fin yellow, with pale yellow colouration near base.

Taxonomic remarks
Herein Rondonacara, gen.nov. is described as a member of the tribe Cichlasomatini, considered to represent the sister group of Laetacara (Figs. 1 and 2).The new genus is a typical member of the Cichlasomatini, sharing the two most important diagnostic character states with the tribe: four pores on the dentary, instead of five in other Cichlidae tribes (except Heroini), just three anal-fin spines, instead of more than three anal-fin spines in species of the Heroini (Kullander, 1998).
Morphologically, the clade Laetacara + Ron do na cara shares with the clades Cleithracara + Ivanacara + Nan na cara [also partially recovered by musilová et al. (2009)], and Cichlasoma + Aequidens [also recovered by mu silová et al. (2009)] a unique colour pattern character state, in which there is a longitudinal stripe horizontally oriented, ending directed to caudal-fin base (e.g.: Fig. 3; Kul lan der, 1986, plates XXXV, XXXVI, XXXVII and XXXVIII; Kullander & nijssen, 1989, Figs. 73, 122 and 123;ottoni & costa, 2009, Fig. 1;ottoni et al., 2012, Figs. 1 and 2), whereas in the other Cichlasomatini, that stripe is dorsally displaced, ending directed to end of the dorsal-fin base (e.g.: Fig. 8; Kullander, 1986, plates VI, XXIX, XXXII and XXXIII;Kullander & nijssen, 1989, Figs 80 and 81;steele et al., 2013, Fig. 2).However, all phylogenetic analyses performed in this work recovered Krobia as the sister group to Cichlasoma and Aequidens (Figs. 1 and 2) [also recovered by musilová et al. (2009)], despite Krobia showing the flank longitudinal stripe displaced in dorsal direction.Thus, the hypothesis that these three clades together form a monophyletic group is not supported here.In addition, a clade comprising these three clades, and Krobia guianensis as the sister group to Cichlasoma dimerus and Aequidens tetramerus was supported by lower bootstrap support value (67) in the ML tree, but had a high posterior probability value (98) in the Bayesian analysis.However, it is beyond the scope of the present work establish the relationship of the tribe Cichlasomatini.
The clade including Laetacara and Rondonacara is clearly morphologically distinguished from the clade including Cichlasoma and Aequidens by the caudal-fin base spot on the middle of the fin (Fig. 3) (transformed in a complete bar in some species of Laetacara), whereas the caudal fin spot is dorsally displaced, with major part of spot occurring above lower lateral line in Cichlasoma and Aequidens (Fig. 7).In addition, Cichlasoma and Aequidens have the suborbital bar transformed into a spot below orbit [cheek spot sensu Kullander (1983)] (Fig. 7), whereas Rondonacara and Laetacara thay eri show a conspicuous suborbital bar (Fig. 3), while the others Laetacara have it lost.Moreover, the clade Laetacara + Rondonacara is also clearly distinguished from the clade Cleithracara + Ivanacara + Nannacara by species of the prior having two supraneuralia instead of one.Rondonacara possess a unique character state among species of Cichlasomatini: transversal streaks originated at the insertion of spines and softs rays of dorsal fin in live specimens, preserved ones retains just the proximal portion of the transversal streaks (Fig. 3).It also shares with Laetacara fulvipinnis a rare colour pattern characteristic among species of Cichlasomatini characterized by the absence of a pattern of dots on caudal fin (Fig. 3;staecK & schindler, 2007, figs. 1, 3, 4 and 5).In addition, Rondonacara easily differs from its closely related genus, Laetacara, by two additional character states: first by uniserial predorsal squamation, whereas Laetacara shows trisserial squamation; and second by unscaled preopercle, whereas Laetacara have scaly preopercle.

Conservation status
Rondonacara includes only Rondonacara hoeh nei, described from a tributary of the Araguaia River, Mato Grosso state, with unprecise locality.Currently, Rondonacara hoehnei is known only from a single locality at a tributary of the das Mortes River, Araguaia River basin (Fig. 9).Despite recent field work efforts, this species was collected in this single locality only (costa pers. comm.).
Natural habitats of the Brazilian Cerrado, one of the global hotspots according to myers et al. (2000), under-     go accelerated and severe destruction, as modern agriculture rapidly develops.The landscape is increasingly taken not only by extensive soya bean cultivation, but also by fields of maize and rice, and cattle pastures.Other activities such as charcoal production also add to the devastation suffered by the Cerrado.Evaluation of satellite photographs available from Google Earth (www.earth.google.com)allows the observation that large areas on both sides of the riverbed of the das Mortes River are engaged in agricultural plots (Fig. 10).Thus, the Brazilian Cerrado is currently reduced to about 50% of its original area (de luca et al. 2009).The restricted distribution and the widespread and accelerated destruction of natural habitats strongly indicate, that Rondonacara hoehnei has to be stated to be a "critically endangered" species.

Fig. 2 .
Fig. 2. Topology founded in Maximum Likelihood (ML) and Bayesian analysis through a combined data set.Numbers branches are bootstrap values above 50% and below branches posterior probability <98.

Fig. 3 .
Fig. 3. Rondonacara hoehnei:UFRJ 9408; tributary of the das Mortes River, Araguaia River basin, Central Brazil.A -life specimen with 53.4 mm SL (taken on April 6 th 2013); B -preserved specimen with 53.4 mm SL; C -dorsal fin of preserved specimen with 53.4 mm SL; D -preserved juvenile specimen with 29.9 mm SL; E -dorsal fin of preserved juvenile specimen with 29.9 mm SL.
through about first and second dorsal-fin spine.Tip of pectoral-fin reaching vertical through middle of flank spot.Pelvic fin pointed.Pelvic-fin base on vertical through third or fourth spine of dorsal fin.Tip of pelvic fin approximately reaching vertical through first anal-fin spine or near it.

Fig. 10 .
Fig. 10.Satellite image of the area where the recent material was collected, taken from Google Earth.Satellite image taken on August 17 th 2004.Red circle represents the precise locality the recent collected material.

Table 2 .
Accession codes from GeneBank sequences of the other species used in this work.