Understanding morphological variability in a taxonomic context in Chilean diplomystids (Teleostei: Siluriformes), including the description of a new species

Following study of the external morphology and its unmatched variability throughout ontogeny and a re-examination of selected morphological characters based on many specimens of diplomystids from Central and South Chile, we revised and emended previous specific diagnoses and consider Diplomystes chilensis, D. nahuelbutaensis, D. camposensis, and Olivaichthys viedmensis (Baker River) to be valid species. Another group, previously identified as Diplomystes sp., D. spec., D. aff. chilensis, and D. cf. chilensis inhabiting rivers between Rapel and Itata Basins is given a new specific name (Diplomystes incognitus) and is diagnosed. An identification key to the Chilean species, including the new species, is presented. All specific diagnoses are based on external morphological characters, such as aspects of the skin, neuromast lines, and main lateral line, and position of the anus and urogenital pore, as well as certain osteological characters to facilitate the identification of these species that previously was based on many internal characters. Diplomystids below 150 mm standard length (SL) share a similar external morphology and body proportions that make identification difficult; however, specimens over 150 mm SL can be diagnosed by the position of the urogenital pore and anus, and a combination of external and internal morphological characters. According to current knowledge, diplomystid species have an allopatric distribution with each species apparently endemic to particular basins in continental Chile and one species (O. viedmensis) known only from one river in the Chilean Patagonia, but distributed extensively in southern Argentina.

Different interpretations concerning the taxonomic composition of the family exist. According to the phylogenetic analysis of  Fig. 1A herein), Diplomystidae includes two genera, Diplomystes Duméril, 1856 (with three valid species, D. chilensis (Molina, 1782), D. nahuelbutaensis D. camposensis Arratia, 1987), and Olivaichthys (with O. viedmensis Mac Donagh, 1931, which are found in freshwaters of Chile and Argentina, respectively (see also Ferraris, 2003;Ferraris, 2007;López et al., 2008;Cussac et al., 2016). In contrast, Azpelicueta (1994) based on certain morphological characters, interpreted the genus Olivaichthys as a synonym of Diplomystes and recognized the three subspecies previously described for Argentina as species (D. cuyanus, D. mesembrinus, and D. viedmensis;see also Ferraris, 2003;López et al., 2008). The most recent molecular study, which included Argentinean diplomystids, listed only D. viedmensis as a valid species for the country (Muñoz-Ramírez et al., 2014;Fig. 1B herein), confirming the morphological phylogenetic hypothesis of Figs. 1A-1B). Considering that in both morphological and molecular hypotheses Diplomystes and Olivaichthys are monophyletic, we interpret Olivaichthys as a valid taxon, an approach that we follow here.
According to current morphological information, four nominal species inhabit the freshwaters of Central and South Chile: D. chilensis, D. nahuelbutaensis, D. camposensis Arratia & Huaquín, 1995;Dyer, 2000;herein), and Olivaichthys viedmensis, which is in the Baker River, the southern-most known distribution of diplomystids (Centro de Ecología Aplicada, 2008;Muñoz-Ramírez et al., 2014;present paper). Tentatively, we name this taxon Olivaichthys viedmensis until our morphological study in progress that includes specimens from Argentina and Chile is published. Additionally, a group identified as Diplomystes spec. from central Chile (Rapel and Maule Basins) was briefly described by Arratia (1987: 63) and Arratia & Huaquín (1995: 44-46, Figs. 20-21) and left without taxonomic assignment until more material would become available. Arratia & Huaquín (1995: 46-50, Figs. 21-24) described the skin of another potential new diplomystid from Copequén River (a northern tributary of Rapel Basin) that was identified as Diplomystes aff. chilensis, because of its geographic proximity to Maipo Basin, the type locality of D. chilensis (Fig. 2). These diplomystids are described here as a new species (see below).
A recent phylogeographic analysis (Muñoz-Ramírez et al., 2014) of Diplomystidae resulted in a topology of the tree identical to the phylogenetic hypothesis based on morphological characters (compare Figs. 1A with 1B) proposed by Arratia (1987: Fig. 38). Few differences between morphological and molecular data and interpretations exist. For instance, Muñoz-Ramírez et al. (2014) placed all species in Diplomystes instead of characteristically diagnostic of D. nahuelbutaensis and D. camposensis (Campos, Arratia & Cuevas, 1997). All diagnoses of the species of Diplomystidae are heavily based on osteological characters that require study of cleared and stained specimens under microscopes. Thus, any taxonomic assignment of specimens could be incorrect without this morphological analysis. Since the species appear to have an allopatric distribution, information regarding the basin specimens collection is critical to assign a possible identification. However, any taxonomic assignment needs to be confirmed with morphological characters, as done here. Additionally, the skin of diplomystids is highly specialized and diagnostic Arratia & Huaquín, 1995;herein), but again, the skin requires observation under microscopes (i.e., SEM) that makes its description difficult.
Previous morphological studies (e.g., Arratia, 1992;Arratia & Huaquín, 1995) of Chilean diplomystids were based on relatively a few specimens due to their endangered status. In those studies on diplomystids, Arratia observed marked external body variation that appears to be connected with individual size (age) complicating the identification of particular specimens. She could not study such variability due to the limited sample size availability, which could not be evaluated statistically. During the last 25 years, larger numbers of specimens were collected by non-governmental environmental organizations. The main goals for such collections were to (i) find new localities to establish the overall geographic distribution of Diplomystidae, (ii) clarify the endangered status of the different species, and (iii) collect molecular data from the different populations. Those agencies donated their specimens to us and therefore, our specimens are not the same as those used in the molecular studies by Muñoz-Ramírez et al. (2014) and Muñoz-Ramírez, Victoriano & Habit (2015). These new specimens plus older ones deposited in museums worldwide, are the foundation of this contribution whose main goals are to (1) study and analyze for the first time the little-known ontogenetic variation of body morphology; (2) search for external diagnostic morphological characters to facilitate specific identification; (3) communicate new morphological information that we have compiled; (4) describe a new diplomystid species; and (5) re-evaluate a few diagnostic features characterizing the now five nominal species of the Chilean diplomystids, as well as create an identification key.

Geographical distribution
According to current information, the Chilean species of Diplomystidae have allopatric distributions (Fig. 2), and seem to be endemic to specific basins, with the possible exception of the widespread Olivaichthys viedmensis in the Baker Basin of southern Chile and Argentina.
Diplomystes chilensis (Molina, 1782), with Maipo Basin as the type locality, was recorded in the literature from a few localities in this basin (e.g., Colina and Paine;Leybold, 1859;Philippi, 1866), and most frequently from the ''rivers of Santiago,'' which refers to area rivers of the Maipo Basin. References to specimens from the rivers of Valparaiso (Aconcagua Basin) have been repeated in the literature following LaCepede (1803: 114); however, no single specimen from this basin in any museum or an illustration from a specimen supposedly collected in Aconcagua Basin has been found. It is doubtful if D. chilensis ever lived in the Aconcagua Basin. Consequently, Fig. 2

Material studied
Considering the endangered status of Diplomystes, museum specimens that were collected long ago are included as well as a significant number of specimens collected from 1989 to 2015 by private environmental organizations. Both and new specimens were collected in the central valley up to the Andean region in Chile, with the exception of one specimen from the Nonguén River (part of the coastal Andalién Basin). The survey of diplomystids in coastal rivers performed by CQ-R during the last nine years-from Aconcagua to Puerto Montt-has given negative results. Most specimens are fixed in ethanol. Some have been cleared and stained (c&s), and one is a dry skeleton (ske). Radiographs were obtained for most specimens. Institutional abbreviations are listed in Sabaj Perez (2014) except for PC, which refers to specimens that are under the care of the first author and will be deposited in the Collections of Fishes of the National Museum of Natural History, Santiago, Chile after completion of the study. All specimens have been kept separated by locality and data collection. Each specimen was measured following standard procedures (outlined below) and provisionally identified following the diagnoses in  and characterization of cephalic sensory canals, pores, and neuromast lines in Arratia & Huaquín (1995).

Methods
Specimens cleared and stained for both bone and cartilage were prepared following the technique of Dingerkus & Uhler (1977) with modifications outlined in Arratia & Schultze (1992).
Morphometric character terminology used herein, is illustrated in Fig. 5, with the following exceptions: interorbital width (distance between dorsal border of orbits), head width (taken at the level of opercle); body width (taken below dorsal fin origin, at mid flank), and peduncle width (taken at the level where peduncle depth was measured). Body measurements were obtained from all studied specimens on the left side of each specimen with a digital caliper reading to 0.1 mm. Different body ratios were calculated and expressed in percent of SL and HL.
D. nahuelbutaensis and D. camposensis to achieve proper statistical validity. Due to the limited museum sample size D. chilensis and Diplomystes incognitus sp. nov. were excluded. We used an analysis of covariance (ANCOVA) with SL as the covariate, to evaluate whether the population means of each morphometric character were statistically different between species, while controlling for the effects of SL. In other words, we calculated linear regressions for each morphometric character per species separately, and then with the ANCOVA, to test for differences in slopes and intercepts between regression lines of both species. In this analysis, if difference in slopes are significant, this implies that the rate of growth for a trait is different between species. If the intercepts differ (but not the slopes), it indicates that the mean of the trait differs consistently through ontogeny but there are no differences in the growth rate. To compare individuals of the same size range, smaller (SL < 90 mm) and larger (SL > 180 mm) individuals were eliminated from this analysis. Normality was checked with q-q plots.
Secondly we evaluated whether morphometric characters can discriminate among the three species of Diplomystes and Diplomystes incognitus sp. nov., excluding younger specimens (SL < 90 mm). For this, a linear discriminant analysis (LDA) was used. This standard statistical method provides a linear combination of morphological variables that allows the most efficient discrimination among groups, e.g., species (Legendre & Legendre, 2012). To maximize the number of individuals with complete measurements, the following characters were selected for the LDA: (1) head length, head width, head depth, predorsal-fin length, prepelvic-fin length, preanal-fin length, pectoral-fin length, pelvic-fin length, and caudal-fin length in percentage of SL; and (2) barbel length, preorbital length, mouth width, interorbital width and eye diameter in percentage of HL. The LDA was performed in R v 3.1.2 (R Core Team, 2016) with the MASS package (Venables & Ripley, 2002), reporting the jackknifed posterior classification of species (CV = TRUE) (Borcard, Gillet & Legendre, 2011). Graphical operations were performed with the ade4 package (Dray & Dufour, 2007) with default options.
The electronic version of this article in Portable Document Format (PDF) will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: urn:lsid:zoobank.org:pub:D22D3881-ED19-470C-8BAF-0CA306EA073B. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS.

Morphometric analyses
Considering individuals 90 mm SL and greater: summary statistics of morphometric measurements conducted on 14 specimens of Diplomystes chilensis (132-191 mm SL), 10 specimens of Diplomystes incognitus sp. nov. (98.3-179.5 mm SL), 50 specimens of D. nahuelbutaensis (96.9-222.2 mm SL), and 59 specimens of D. camposensis (94.9-202.5 mm SL) are shown in Table 1. Although some mean values differ between species, there is an overlap in most ratios and as such, they are not diagnostic characters. To avoid misinterpretations of means and ranges with reduced number of specimens, changes along the ontogeny are described only between D. nahuelbutaensis and D. camposensis.
Comparisons between D. nahuelbutaensis and D. camposensis suggest that for a given length D. nahuelbutaensis has shorter head and pectoral spine length, a narrower caudal peduncle, and smaller eye diameter (Figs. 6A-6D). In these cases, the slope of the regression lines is equal (ANCOVA F 1,99 = 1.29, p > 0.25 for head length; ANCOVA F 1,85 = 0.009, p > 0.92 for pectoral spine length; ANCOVA F 1,77 = 0.13, p > 0.71 for caudal peduncle width; and ANCOVA F 1,99 = 0.64, p > 0.42 for eye diameter), but the intercept is different (ANCOVA F 1,100 = 40.22, p < 0.001 for head length; ANCOVA F 1,86 = 18.45, p < 0.001 for pectoral spine length; ANCOVA F 1,78 = 59.54, p < 0.001 for caudal peduncle width; and ANCOVA F 1,100 = 41.52, p < 0.001 for eye diameter). These results show that although the rate of growth of these characters is similar across the species examined here, there are consistent differences throughout ontogeny for these morphological features between species.
The first linear discriminant (LD) explained 67.9% of the variance, whereas the second LD explained 26.2% of the variance, showing that the LDA displays good discriminatory power among species (Fig. 7). The coefficients of the linear discriminants are shown in Table 2: variables with higher linear discriminant coefficients are head length, pelvic-fin  length and eye diameter on the LD1; whereas pelvic-fin length, head width and preanal-fin length score highest on LD2. Overall, posterior species classification is highly accurate -88 of 109 individuals were correctly classified (81%; Table 3). In summary, although the ranges in most characters show overlap between species, a finer analysis reveals significant differences in morphological traits among Diplomystes species when analyzed through ontogeny. This implies that most external measures are not diagnostic but change for most species in the same way, maintaining differences through ontogeny. When combined in a standard multivariate analysis, our results show that the   LDA permits the classification of individuals of three nominal species of Diplomystes and the newly described D. incognitus sp. nov. using a set of external morphometric characters.

Body shape, fins, and ontogenetic changes
Some body proportions of the three species of Diplomystes become markedly different throughout ontogeny, especially with regard to the size and position of the fins (see above) and the position of the anus and urogenital pore, which are positioned closely in diplomystids . All fins are large and separated by a short distance; it appears as if the fins are continuous in small specimens (under 40 mm SL) of D. nahuelbutaensis, Diplomystes incognitus sp. nov., and D. camposensis. Note, for instance, the large size and almost rounded shape of the paired and unpaired (dorsal, adipose, and anal) fins of Diplomystes nahuelbutaensis illustrated in Fig. 8A (see also Lundberg, Berra & Friel, 2004). The pectoral fins are so large that they almost overlap with the pelvic fins. The pelvic fins reach the origin of the anal fins and may extend further caudad. The posterior margin of the adipose and anal fins almost reach the expanded anterior margin of the caudal fin, which, in the smallest specimens, has an almost straight posterior margin (Fig. 8A). As a consequence of the large size of the pelvic fins, the anus and urogenital pore lie between the two pelvic fins in the smallest specimens. As similar features are observed in small specimens of D. nahuelbutaensis, D. camposensis, Diplomystes incognitus sp. nov. (Figs. 8B-8C), and Olivaichthys viedmensis, we hypothesize that D. chilensis has a similar early development of the fins. There is one available specimen under 133 mm SL (CAS 27839, 132.0 mm SL).
Throughout growth, changes in the distance between fins and the length of the fins in specimens of Diplomystes nahuelbutaensis (Figs. 9A-9D) and D. camposensis (Figs. 10A-10D) are observed. It is almost impossible to distinguish both species when the fishes are small in size (under 140 mm SL), but after sexual maturity in D. nahuelbutaensis, the separation between pelvic and anal fins increases progressively and the anus and urogenital pore are posterior to the distal margins of the pelvic fins (Figs. 9C-9D; 11B). The posterior margins of the pelvic fins in D. camposensis stay near the anal fin so that the position of anus and urogenital pore is between both pelvic fins (Figs. 10C-10D, 11C). A similar pattern is observed in Diplomystes incognitus sp. nov. The position of the anus and urogenital pore is just posterior to the posterior margins of the pelvic fins in D. chilensis (Fig. 11A). The anus and urogenital pore are between the posterior tips of the pelvic fins in Olivaichthys viedmensis from Baker River. becomes larger in D. camposensis through growth, and remains small in D. nahuelbutaensis. According to our observations, external differences in pelvic fin shape, as well as its position relative to the anus and urogenital pore, are reached in large specimens over 150 mm SL. Sexually mature ovaries and testes have been observed in individuals of 110 mm SL and above in D. nahuelbutaensis (Vila, Contreras & Fuentes, 1996). Such size (110 mm TL) is relatively small in comparison with the largest sizes recorded for D. nahuelbutaensis (300 mm TL; Vila, Contreras & Fuentes, 1996) and for D. camposensis (242 mm TL; material studied by Curoto (2015) and measured by us).

Skin
The skin is one of the most distinctive characters of Diplomystidae Arratia & Huaquín, 1995). Histological and SEM studies of the skin of Diplomystes and Olivaichthys have shown that the skin is characterized by a high number of taste buds all the over the body, including fins and inside of the mouth, and different pit-lines represented by a few, large neuromasts. However, the middle pit-line trunk is represented by a line of small neuromasts along the mid-flank.
The skin of diplomystids is covered by a variable number of papillae (depending on the age/size) carrying different types of receptors, especially taste buds. Young and juvenile diplomystids have a smooth surface that develops a variable number of papillae during ontogeny. Thus, the descriptions (and illustrations) of the skin presented here correspond to large specimens. Diplomystes chilensis is unique among catfishes in the presence of long papillae that give the impression of a ''hairy'' skin (Figs. 12A, 13). The only known description of a fresh specimen of D. chilensis presents it as having its entire body covered by long, gray ''hairs'' that become slightly brownish toward the caudal region and whitish in the ventral region (Leybold, 1859). We have observed long papillae along the whole body, and they are longer in the barbels (Fig. 13) and flanks, close to the lateral line pathway. Similar long papillae have not been observed in any of the studied diplomystids (compare Figs. 12A, collected south of Maipo Basin (Fig. 2), the type locality of D. chilensis.
The skin of the body and mouth of D. camposensis is densely covered by papillae that are especially conspicuous on the lips, barbels, and gular region (Fig. 12C). Fewer papillae are present in D. nahuelbutaensis (Fig. 12B) and O. viedmensis, especially on the dorsal region of the head and flanks. When the skin of these catfishes was observed under SEM (Arratia & Huaquín, 1995), the skin of D. camposensis has the unique presence of a reticulated pattern that has not been observed in other species. The skin of Diplomystes incognitus sp. nov. (Fig. 14) is characterized by the presence of large and rounded papillae all over the body, that give the skin a ''blackberry'' or verrucose aspect. The amount of papillae (and size) is so great that the pores of the sensory canals and pit-lines are concealed.

Neuromast lines
Although the presence of superficial neuromast lines may be obscured by the development of the papillae, we have gathered new information that may aid in species' identification. For instance, the rostral line is placed anterior to the anterior nostril, close to the anterior margin of the snout in Diplomystes chilensis and D. camposensis, whereas in D. nahuelbutaensis the anterior nostril is placed almost at the anterior margin of the snout so that the rostral line is placed anteriorly, on the anterodorsal curvature of the snout. The anterior nostril is relatively distant to the anterior margin of the snout in Diplomystes incognitus sp. nov., but the rostral line cannot be identified among the rounded papillae covering the region.

Cephalic sensory canals and main lateral line
The cephalic sensory canal system, including sensory tubules and pores opening to the skin surface, is similar among the species of Diplomystes and it has been illustrated for the three species . However, newly observed differences are reported here and  (9) is identical for the species of Diplomystes as illustrated in Fig. 15; however, there are differences in the position of the nine branches of the infraorbital canal among the three species. Whereas in D. chilensis pores 4-9 are regularly distributed in the ventral and ventroposterior regions of the canal, the distribution of pores 7-9 is different in the three species and in Diplomystes incognitus sp. nov. (see Figs. 15A-15D). All pores are concentrated predominantly in the ventral region of the infraorbital canal with pore 9 emerging at the posteroventral corner (and occasionally between the last infraorbital bone and cranial bones) in D. nahuelbutaensis, and pore 8 emerging at the posteroventral corner and pore 9 emerging at the posterodorsal region of the canal in D. camposensis. Additionally, there are differences in the number and position of the infraorbital bones in the three species, as shown in Figs. 15A-15D (the antorbital Arratia & Huaquín, 1995) is included in the count of the infraorbital bones). Among these, D. chilensis has seven or eight infraorbital bones, whereas D. nahuelbutaensis has nine or ten, but the dorsalmost infraorbital bone is formed by fusion or partial fusion of two or three bones (Fig. 15C). Diplomystes camposensis has eight or nine infraorbitals, but the most posterodorsal one is not a compound bone. Diplomystes incognitus sp. nov. has 10 infraorbitals, as in D. nahuelbutaensis, but a dorsalmost compound bone is absent (Fig. 15B). The posterior part of the circumorbital series is formed mainly by a narrow, elongated infraorbital bone in Diplomystes incognitus sp. nov., D. nahuelbutaensis, and D. camposensis, yet it is comparatively shorter in D. nahuelbutaensis.
The main lateral line and the middle trunk line of neuromasts are continuous along the flank in species of Diplomystes. However, Olivaichthys viedmensis from Baker River has an interrupted lateral line (see Fig. 4E) between the dorsal and anal fins, and the middle trunk line is complete and composed of small neuromasts.

Vomerine tooth plates
The presence of two tooth plates is a condition of young diplomystid specimens. From this early ontogenetic stage, the tooth plates may stay separated during growth (e.g., D. camposensis and D. incognitus sp. nov.) or become fused (e.g., D. chilensis and D. nahuelbutaensis). Since this process involves a fusion of elements, some large specimens still show both tooth plates partially fused. In these specimens, this condition has been counted as one element. Comparisons among 19 individuals of over 100 mm SL of D. chilensis reveal that one vomerine tooth plate is found in 15 specimens (79%) and two tooth plates are present in four specimens (21%). In contrast, two vomerine tooth plates are found in 19 specimens (83%) of a total of 23 large individuals of D. camposensis and only one tooth plate is present in four specimens (17%). Only one tooth plate is present in 23 individuals (90%) among 25 large specimens of D. nahuelbutaensis and two tooth plates are present in two specimens (10%). Among the largest available specimens (four) of Diplomystes incognitus sp. nov., three are present with two tooth plates. Two tooth plates are also present in the two larger specimens of Olivaichthys viedmensis from the Baker River.
to discuss the branchial arches of diplomystids but to report an interesting feature that we have observed during this study.
An elongated median cartilage extending anteriorly to the hypohyal region was illustrated in young specimens of Olivaichthys viedmensis  Fig. 16 herein). The position of this cartilage in front of the first ossified basibranchial and its dorsal extension to the hypohyal region could be interpreted as a glossohyal cartilage. Because the absence of a bony glossohyal is considered a synapomorphy of Siluriformes, this cartilage was not labeled by . We agree with the common interpretation that the absence of the glossohyal (Fig. 17A) is a feature characterizing the order, thus, we were surprised to find a small median bone in front of the basibranchials and above the dorsal hypohyals Abb.: a.ce, anterior ceratohyal; ANT, anterior; bb1-2, basibranchials 1-2; cb1-2, ceratobranchials 1-2; gh, glossohyal; hb1-2, hypobranchials 1-2; dh, dorsal hypohyal; vh, ventral hypohyal. in D. nahuelbutaensis (Fig. 17B). We interpret this bone as a glossohyal that may be an atavism that is occasionally present in diplomystids.
More primitive ostariophysans, such as gonorynchiforms and cypriniforms, have a well-developed glossohyal.

Pelvic radial
A free cartilaginous pelvic radial, independent of the posterior articular surface of the basipterygium and of its medial posterior cartilaginous process, was described and illustrated for Olivaichthys viedmensis by   Fig. 37B). It was interpreted as a synapomorphy of the genus Olivaichthys. However, a cartilage that is part of the articular posterior surface of the basipterygium was interpreted as a pelvic radial in some specimens of Diplomystes nahuelbutaensis and D. camposensis by Azpelicueta (1994). According to our observations, pelvic radials are independent cartilaginous elements in a few teleosts and extant neopterygians where these elements occur. No independent pelvic radials were observed in any specimen of Diplomystes studied herein. We suggest further research on the variability reported for the Argentinean diplomystids (Azpelicueta, 1994).

Pores of axillary gland
Members of the family Diplomystidae are characterized by the presence of an axillary gland that is placed just dorsomedial to the insertion of the pectoral spine. It may open to the skin surface by a variable number of pores. Diplomystes chilensis presents one to three pores, commonly two in both sides of the body , a count that is distributed as follows among 12 specimens: Two pores in eight specimens (67%), one pore in three specimens (25%), and three pores in one specimen (8%). In contrast, D. nahuelbutaensis was reported to have one to three pores, commonly three on each side of the body . The larger sample of specimens studied has changed such information and has provided new findings: two pores in both sides of the body are present in 22 individuals (44%) among a total of 50 examined specimens, the absence of pores is found in 19 specimens (38%), one pore is found in four specimens (8%), and three pores in one or both sides of the body are found in five specimens (10%).
Diplomystes camposensis was previously reported as having no pores and/or a single on one side of the body . The new results, based on 57 examined specimens, show slightly more variability with 36 individuals (63%) demonstrating no axillary pores opening to the skin surface, nine specimens (16%) possessing either one pore on one or both sides of the body, 11 specimens (19%) with two pores in either one or both sides of the body, and one specimen (2%) with three pores. Diplomystes incognitus sp. nov. shows no axillary pores opening to the surface in the seven studied specimens, but four pores are found on one side of the body in one individual studied. The two large specimens of Olivaichthys viedmensis from the Baker River show no axillary pores, similarly to the condition observed in five larvae. Pores of the axillary gland are present at all growth stages. Therefore, ontogenetic variability is refuted for all species.

Revised diagnoses of species of Diplomystes
Diplomystes chilensis (Molina, 1782) (Figs. 3A,4A,11A,12A,13,15A) Diagnosis (emended from . Diplomystid that is distinguished from all congeners by the possession of ''hairy-like'' skin, covered with long, simple or lobulated papillae (vs. granulose skin with short papillae); acuminated pectoral fins (vs. distally rounded fins); fused anterior processes of autopalatine (vs. separated processes); lateral line enclosed by ossified, tube-like ossicles along the whole flank (vs. few ossicles at the anterior third of flank in other Diplomystes). With few maxillary teeth (8-13, commonly 9) vs. 11-13 (commonly 11) in D. nahuelbutaensis, 12-19 in D. camposensis (commonly 15), and 7-9 in Diplomystes incognitus sp. nov. Diplomystes chilensis can further be differentiated Diplomystes aff. chilensis Arratia & Huaquín, 1995: 46-50 Diplomystes cf. chilensis: Muñoz-Ramírez et al., 2014: tb. 4. Diplomystes cf. chilensis: Muñoz-Ramírez, Victoriano & Habit, 2015: genetic structure Diagnosis. Diplomystid that is distinguished from all congeners by the possession of the skin of head, body, and fins densely covered by round, short papillae giving the skin a blackberry-like or verrucose aspect in large individuals; with a short head, slightly squarish and as long as broad (versus slightly longer more triangular-shaped head); high dorsal fin, ca. 20% of SL (range 17-25%) and triangularly-shaped (versus slightly rhomboidal); maxilla with 7-9 teeth (vs. 8-13 in D. chilensis, 11-13 in D. nahuelbutaensis, and 12-19 in D. camposensis); with 10 infraorbital bones, as in D. nahuelbutaensis, but the dorsalmost compound bone is absent; urogenital pore and anus placed between posterior tips of pelvic fins as in D. chilensis (vs. urogenital pore and anus placed between pelvic fins or in between the distal tips of pelvics and anal fin); and absence of pores of axillary gland with occasionally four on one side of body (vs. two or three pores). Etymology. The specific name incognitus is in reference that recognition of the species was obscured by the assumption that Diplomystes chilensis also extended south of Maipo Basin. Geographical distribution. In Rapel, Mataquito, Maule, and Itata Basins (Fig. 2).  Coloration. The skin of live, young and juvenile individuals (Fig. 18) is greenish in the dorsal part of the head and body and dorsal and adipose fins, with minuscule black and golden spots. The greenish color is also observed irregularly in the tail. The ventral body is of brown-reddish color, with a yellow or creamy colored belly. The lateral aspect of the head is of a gray and reddish color mixture. Large individuals have similar coloration when alive, which turns almost dark red-brown or dark gray-brown at death. Fishes fixed in ethanol become a uniformly gray or brown color. Comments. For morphometric data see Table 1 and for number of branchiostegals, vertebral elements, and fins rays see Table 4.  (3) 2b -Lateral line interrupted, divided in two sections; maxilla with more than two rows of teeth anteromedial, two or more rows posterolaterally; with more than 20 maxillary teeth;

Key to diplomystids
.

FINAL COMMENTS
This study illustrates the difficulties in the identification of species of Diplomystes, which possess similar external aspects, especially in young and juvenile specimens. Because of this fact, we concentrated our efforts on characters that could identify and separate these species. During this research, it became obvious that sexually immature specimens and specimens below 150 mm SL belonging to different species are very similar in their external morphology. This includes similarities in body proportions, fin shape and size, and a smooth skin with few or no papillae. These external morphological characteristics become significantly different in large Diplomystes chilensis, D. nahuelbutaensis, D. camposensis, and Diplomystes incognitus sp. nov., however, not in the morphometric information including body proportions, the ranges of which overlap despite differences in means (Table 1). These findings provide evidence that members of D. nahuelbutaensis and D. camposensis (Figs. 9 and 10) undergo some significant changes in certain body features throughout growth, especially in the relationships between pelvic and anal fins and the position of the anus and urogenital pore. These changes are unknown in other siluriforms at this present time.
All known specimens of D. chilensis are large. No young specimens are available for study. We predict that they are also indistinguishable from the other two species when the populations found in the Bío-Bío Basin would represent a new species, but not D. nahuelbutaensis. Our findings do not support such a claim. We have expanded our analysis to include large specimens collected from the upper part of the Bío-Bío Basin, a region not sampled by those authors. Muñoz-Ramírez et al. (2014) proposed a multispecies hypothesis concerning diplomystids (e.g., individuals from Toltén Basin would not correspond to D. camposensis but to a new sister species of D. nahuelbutaensis). We consider this hypothesis premature; it should be investigated further, incorporating diplomystid populations living in between Bío-Bío and Valdivia Basins, a region including the Toltén Basin, which is incompletely sampled.