Paretroplus petiti Pellegrin, 1929

Sparks, J. S., 2008, Phylogeny Of The Cichlid Subfamily Etroplinae And Taxonomic Revision Of The Malagasy Cichlid Genus Paretroplus (Teleostei: Cichlidae), Bulletin of the American Museum of Natural History 2008 (314), pp. 1-151 : 72-77

Paretroplus petiti Pellegrin, 1929 Figures 34–35; plate 1C; table 4

HOLOTYPE: MNHN 1928.282 , 82.8 mm SL; northwestern Madagascar: Province of Mahajanga (Majunga): Riviere de Maintimaso ; G. Petit.

DIAGNOSIS: Based on the holotype, a juvenile, and only known specimen. Paretroplus petiti is a deep-bodied Paretroplus and a member of Clade I. It is distinguished from all other members of this clade, except P. menarambo , by dark brown coloration in combination with the lack of a strong lateral barring pattern. It is distinguished from P. menarambo by the absence of a pinstriped pigmentation pattern (in preservation), which results in the latter species due to conspicuous spotting on scales. Paretroplus petiti can be distinguished from all congeners not included in Clade I by a deep, disk-shaped body and the presence of lateral ridges of scales (‘‘scale sheathing’’) that are free from the fin membranes over the entire length of the soft dorsal and anal fins. It should be noted that this diagnosis is tentative in that it is based on a single juvenile specimen and that live coloration is not known (i.e., lack of distinguishing marks could be an artifact of long-term preservation).

DESCRIPTION: (Note: The following description is based on features that can be visualized (including radiographically) in the intact holotype, the only known specimen, without causing damage to the specimen.) Morphometric and meristic data presented in table 4. Morphological characteristics and general pigmentation pattern in preservative can be observed in figure 34. A deep-bodied, laterally compressed Paretroplus belonging to Clade I, which comprises all deep-bodied and essentially disk-shaped members of Paretroplus (also including P. dambabe , P. maculatus , P. maromandia , P. menarambo , and P. polyactis ) (fig. 1). Head relatively blunt and steeply sloping in lateral view. Snout mildly convex. Predorsal profile mostly straight and only slightly rounded. Caudal peduncle short, deep, and laterally compressed.

Total vertebral count 33, with a formula of 15 + 18 precaudal and caudal vertebrae, respectively.

Jaws isognathous. Single row of spatulate, unicuspid teeth in both upper and lower jaws. Teeth laterally expanded, flattened at crown, and procumbently implanted. In upper jaw, tooth on either side of premaxillary symphysis greatly enlarged, remaining teeth graded in size laterally. Lower-jaw damaged and counts not possible. Teeth in upper jaw number six on one side and seven on the other, and total 13. Upper- and lower-jaw teeth widely set.

Upper and lower pharyngeal tooth plates well developed and dentition robust. Nine elongate and triangular gill rakers arrayed along lower limb of first gill arch. Rakers appear edentate in holotype (but difficult to determine on intact specimen). All other lower-limb rakers (i.e., those on gill arches 2–4) short and strongly denticulate dorsally. Epibranchial rakers on first gill arch elongate.

Body covered with large, regularly imbricate, cycloid scales. Well-developed ridges of scales (5 scale sheathing) present along dorsal- and anal-fin bases. Scale ridges free from both spiny and soft portions of dorsal and anal fins. Pelvic axillary scale present and well developed. Interpelvic scale elongate, robust, and rounded distally. Lateral-line scales number 39. Chest scales only slightly smaller than lateral body scales and weakly embedded. Belly scales small compared to other body scales and strongly embedded, those along ventral midline smallest. Five (left) to six (right) rows of scales on cheek. Opercle, subopercle, and interopercle scaled. Preopercle scaled (including shaft) except ventrally. Snout, lacrimal, and anterior portion of interorbital region asquamate. Scales on caudal fin reduced in size and extending posteriorly 2/3 to 3/4 length of fin on dorsal and ventral lobes, and 1/4 to 1/3 length of fin medially.

Dorsal with XVII spines, 18 soft rays. Anal with IX spines, 14 soft rays. Origin of dorsal fin at about level of vertical through

TABLE 4 Morphometric and meristic data for holotype of Paretroplus petiti .

pectoral-fin insertion. Caudal fin emarginate, but damaged terminally. Pectoral fin broad and rounded at distal margin. Pelvic fins extending beyond anal-fin origin, to about level of third anal-fin spine, when adducted.

MISCELLANEOUS OSTEOLOGY AND ANAT- OMY: Large, well-developed exoccipital excavations present. Prominent excavation (5 supraoccipital notch of Stiassny et al., 2001) along posterior margin of supraoccipital. Supraoccipital extending anteriorly over median frontal pores of neurocranium (nlf 0 of Barel et al., 1977). An additional, fully ossified, anal- and dorsal-fin pterygiophore, not associated with any fin rays, present terminally in both fins.

COLORATION IN LIFE: Unknown. Only known specimen is the formalin-fixed holotype. As in other deep-bodied members of Paretroplus restricted to western drainages (Clade J), it is clear that, in life, P. petiti is characterized by the presence of conspicuous alternating, light and dark horizontal stripes on the flank.

COLORATION IN PRESERVATIVE: Base body coloration dark golden brown. Alternating dark (dark brown) and light (golden brown) horizontal striping pattern evident on flanks (fig. 34). Very faint horizontal barring pattern also evident on flanks, but hardly noticeable. No additional distinguishing markings present on flanks. Snout and interorbital region dark grayish brown. Otherwise, head region more or less a uniform dark brown. Unpaired fins golden brown to grayish brown. Spiny dorsal with black pigment along distal margin. Pectoral fins light golden brown. Pelvic fins grayish brown and golden brown.

VISCERA AND DIET: Based on examination of a radiograph of the holotype, the gut is packed with crushed shells, indicating that P. petiti is primarily a molluscivore.

DISTRIBUTION AND HABITATS: There has been some uncertainty surrounding the type locality of P. petiti Pellegrin, 1929 , which Pellegrin reports as ‘‘Rivière de Maintimaso (province de Majunga): G. Petit’’. From the limited amount of information that is available, it appears that the type specimen of P. petiti (MNHN 1928-282) was collected from the Maintimaso River, a tributary of the westward flowing Betsiboka River, near the town of Maintimaso (16 ° 10 9 60 0 S, 46 ° 43 9 00 0 E), which is situated roughly 69 km (37 nm) southeast of Majunga (5 Mahajanga) in northwestern Madagascar (fig. 35). The MNHN database, however, reports the collection locality of this specimen as ‘‘Ambila’’. No town with this spelling could be located in the vicinity of Majunga, although a town with the name of Ambilo is located near the Betsiboka River, about 35 km (19 nm) south of Majunga and to the northwest of Maintimaso. To add to the confusion, an old tag with the species name and also including the word ‘‘Ambanja’’ was found in the jar with the holotype at the MNHN (Paris). Ambanja is a large town located several hundred km to the north of Mahajanga and Maintimaso. No additional information with regard to either the Ambila or Ambanja localities could be located, and the most prudent course of action is to accept the locality information presented by Pellegrin (1929).

In response to this interpretation de Rham and Nourissat (2004: 102) note that they believe that ‘‘Ambanja’’ refers not to the large town located far to the north, but instead to Lake Ambanja, which is located near the town of Ambato-Boeni, which itself is located about 96 km (52 nautical miles) to the southeast of Mahajanga and also within the Betsiboka drainage basin. However, it seems that de Rham and Nourissat and their Malagasy colleagues never collected in Lake Ambanja, but instead in nearby Lake Marajory. From Lake Marajory they obtained specimens that they refer to as P. cf. maculatus (see discussions under P. maculatus and P. menarambo below), which are clearly referable to P. maculatus given their possession of a large, black humeral blotch. They also mention that their Malagasy colleague apparently collected another species of Paretroplus from the lake in 2003, but these specimens were lost due to attacks by bandits and export problems.

Despite this uncertainty, nevertheless, it is clear from the information that is available that the holotype of P. petiti was not collected from the Mahavavy du Sud drainage basin, to which P. dambabe is endemic, but instead was obtained from a tributary of the Betsiboka River, one of Madagascar’s largest drainage basins (fig. 35). Although P. dambabe was only recently described ( Sparks, 2002a), since at least the early 1960s individuals of this species collected from Lac Kinkony, a large floodplain lake located within the Mahavavy du Sud drainage basin, have erroneously been referred to P. petiti (e.g., Kiener, 1963; Kiener and Thérézien, 1963; Kiener and Mauge´, 1966). Based on results presented by Sparks (2002a) and herein, all specimens previously assigned to P. petiti that were collected in Lac Kinkony are instead referable to P. dambabe (see Relationships and Discussion for P. dambabe below). Although the conservation status of P. petiti is unknown, the possibility remains that P. petiti is extant in the Betsiboka River drainage basin.

LOCAL NAME: Unknown.

ETYMOLOGY: Named in honor of G. Petit who collected the holotype that was described by Pellegrin (1929).

RELATIONSHIPS AND DISCUSSION: Paretroplus petiti is recovered as a member of Clade I, comprising all of the deep-bodied and more or less disk-shaped members of Paretroplus (also including P. dambabe , P. maculatus , P. maromandia , P. menarambo , and P. polyactis ), which is supported by three unambiguously optimized morphological features, two of which are unique and unreversed (features discussed and presented above) (fig. 1). Apart from differences in pigmentation pattern and coloration, all members of Clade I are morphologically very similar.

In the combined analysis of morphological features and nucleotide characters, P. petiti is recovered in an unresolved polytomy with all other deep-bodied species of Paretroplus that have distributions restricted to western basins (fig. 35: Clade J). Monophyly of Clade J, comprising P. dambabe , P. maculatus , P. maromandia , P. menarambo , and P. petiti , is supported by two unambiguously optimized morphological features, the second of which is unique and unreversed: the presence of a blunt snout with a steeply sloping profile in lateral view and a distinctive lateral pigmentation pattern comprising alternating light and dark horizontal stripes. A lack of resolution within Clade J is due to the fact the P. petiti (known only from the formalinfixed holotype) is included on the basis of morphological features only, which are insufficient for resolving relationships within this anatomically conservative clade.

Since at least the early 1960s, individuals that are now placed in P. dambabe Sparks, 2002 were erroneously referred to P. petiti (e.g., Kiener, 1963; Kiener and Thérézien, 1963; Kiener and Mauge´, 1966). Based on the results presented by Sparks (2002a) and discussed above, all specimens previously assigned to P. petiti that were collected in Lac Kinkony, Mahavavy du Sud drainage basin, are determined instead to be members of P. dambabe , and P. petiti is, therefore, currently known only from the holotype described by Pellegrin (1929), presumably collected from the Betsiboka drainage basin. Paretroplus petiti is readily distinguished from P. dambabe in preservative by overall coloration (dark brown vs. light yellow-olive in P. dambabe ), the absence of dark brownish spotting on the flanks (present in P. dambabe ), and a deeper body (57.9% SL vs. mean 52.8% [ranges from 48.3%–57.1%] in P. dambabe ). Sparks (2002a) noted that the holotype of P. petiti , a juvenile, is not only considerably deeper bodied than P. dambabe of similar standard length, but deeper bodied than any individual of P. dambabe examined, whereas it is quite similar in body depth to specimens of P. menarambo of similar standard length.

As discussed in detail by Sparks (2002a), the possibility exists that P. petiti and P. menarambo Allgayer, 1996 are conspecific. Given that only a single juvenile of P. petiti (MNHN 1928-282, holotype) is known, Sparks (2002a) deferred judgment on the status of P. menarambo until more material from the region of the putative type locality of P. petiti became available for comparison. The holotype of P. petiti is well preserved in terms of pigmentation pattern, and is similar in body shape and pigmentation pattern, in preservation, to P. menarambo (fig. 21D), a species restricted to the Bemarivo River floodplain lakes within the Sofia River basin in northwestern Madagascar, and which until very recently was presumed extinct. Paretroplus petiti and P. menarambo are both dark brown with a conspicuous pattern of thin, alternating light and dark horizontal stripes on the flanks (figs. 21D and 34). In the original description of P. petiti, Pellegrin (1929) noted this horizontal striping pattern, unique to members of Clade J, yet did not mention any additional distinctive markings. Apart from P. petiti and P. menarambo , no other species within the genus are dark brown overall in preservative and lack additional distinctive lateral flank markings, with the exception of a faint vertical barring pattern. These vertical bars are very faint in the holotype of P.petiti , and although not mentioned by Pellegrin (1929) in the original description may have faded considerably over time in preservative. Whereas coloration in life for P. petiti is unknown, in life P. menarambo is characterized by a pinstriped lateral pigmentation pattern, a pattern which is still quite obvious (and distinctive) in preservative in many specimens (see de Rham and Nourissat, 2004: 98–100).

One final point merits mentioning. Deepbodied members of Paretroplus that occur in northwestern Madagascar (Clade J) are narrowly endemic in distribution, and as far as is known, all are restricted to a single drainage basin ( Sparks, 2002a, 2004a) and no two species occur sympatrically. Paretroplus menarambo was endemic to the Bemarivo River drainage basin, whereas the holotype of P. petiti was collected from the Betsiboka River drainage basin, assuming the collection locality reported by Pellegrin (1929) is correct. The Betsiboka and Bemarivo are not adjacent basins in northwestern Madagascar, which might cast doubt on the possibility that these two nominal taxa are conspecific.