A new species of cottontail rabbit (Lagomorpha: Leporidae: Sylvilagus) from Suriname, with comments on the taxonomy of allied taxa from northern South America

Abstract

Of the 19 currently recognized species of Sylvilagus Gray, 1867, 15 inhabit North America, and only 5 are recognized in South America: S. brasiliensis Linnaeus, 1758 (throughout most of the continent); S. varynaensis Durant and Guevara, 2001, restricted to the southern lowlands of Venezuela (states of Barinas, Portuguesa, and Guarico); S. andinus (Thomas, 1897) from the Andean páramos of Ecuador and potentially in a sporadic manner to the Colombian and Venezuelan páramos; and S. tapetillus Thomas, 1913, from the coastal plain in the region of Rio de Janeiro. In addition to these, putative subspecies of S. floridanus, primarily a North American taxon, nominally are recognized from the grassland plains areas of northwestern South America east of the Andes. While S. varynaensis and S. tapetillus are monotypic, S. brasiliensis contains at least 37 named taxa in synonymy, distributed in various habitats; S. andinus requires further study. As a result of the recent description of a neotype for S. brasiliensis, it is now possible to assess species limits and begin the process of illuminating formerly obscured biological diversity in South American cottontails. Here, I describe a new species of Sylvilagus from the lowlands of western Suriname, and excise S. sanctaemartaeHershkovitz, 1950 from synonymy with S. brasiliensis.

De las 19 especies de conejos actualmente reconocidas en el género Sylvilagus Gray, 1867, 15 habitan Norteamérica mientras que solo cinco se conocen de Suramérica. Estas son S. brasiliensis Linnaeus, 1758 (en la mayor parte de la región), S. varynaensis Durant y Guevara, 2001 (restringido a las llanuras del sur de Venezuela, en los estados de Barinas, Portuguesa y Guarico), S. andinus (Thomas, 1897) de los páramos andinos de Ecuador y esporádicamente hasta los páramos de Colombia y Venezuela, y S. tapetillus Thomas, 1913, de las planicies costeras en la región de Rio de Janeiro. Además de estas cuatro especies, se reconoce de forma nominal en las llanuras y pastizales del noroeste de Suramérica, al este de los Andes, a supuestas subespecies de S. floridanus, una especie mayormente norteamericana. Aunque S. varynaensis y S. tapetillus son monotípicas, S. brasiliensis en cambio comprende por lo menos 37 taxones en su sinonimia, distribuidos en numerosos y variados ambientes; se desconoce aún la taxonomía de S. andinus. Uno de los resultados de la reciente descripción de un neotipo para S. brasiliensis es que ahora es posible llevar a cabo una delimitación más certera de las especies de Sylvilagus en Suramérica. Con esto se puede así iniciar un proceso de descubrimiento de la diversidad biológica regional en el género, diversidad previamente entenebrecida. En el presente trabajo describo una nueva especie de Sylvilagus de las llanuras del oeste de Surinam, y extraigo a S. sanctaemartaeHershkovitz, 1950 de su sinonimia con S. brasiliensis.

biodiversity, center of endemism, Colombia, conservation, Guiana Refuge, Guyana Area of Endemism, Guyana Shield, Neotropics, Sierra Nevada de Santa Marta

A critical requirement of making informed conservation decisions is to have an accurate assessment of the taxonomy of the species under consideration. Insofar as the mammals of South America are concerned, biodiversity inventories and revisionary work are yielding an ever-increasing number of species in all groups of mammals: a cursory search yields at least 10 novel species described in 2015 from new material alone. Such enhanced vision of species limits can result in improved conservation recommendations: for example, the recent discrimination of Lonchophylla bokermanniSazima, Vizotto, and Taddei, 1978, into 2 species (L. bokermanni and L. peracchiiDias, Esbérard, and Moratelli, 2013) resulted in a range restriction (“extent of occurrence”) for the former L. bokermanni, from over 423,000 km² to a mere 1,506 km² (estimated using minimum convex polygon), solidly positioning the species in the IUCN’s “Endangered” category (Teixeira et al. 2015).

The issue of poorly circumscribed species is particularly acute among Neotropical cottontails, Sylvilagus. However, the recent designation of a neotype led from a biologically unrealistic singular species (Sylvilagus brasiliensis) occupying 1.09 × 10⁷ km², and containing upwards of 37 named subspecies, to a more manageable vision of a species ecologically restricted to 500–700 km² with a narrow distributional band along northern Atlantic coastal Brazil (Ruedas et al. 2017). In that work, the authors further suggested that the earliest name available for Sylvilagus species allied to S. brasiliensis (i.e., excluding the subspecies of “S. floridanus” extending into South America) and distributed north of the Amazon River would be Sylvilagus sanctaemartaeHershkovitz, 1950. Another possible name might be Sylvilagus orinociThomas, 1900, also treated as a valid species allied to S. brasiliensis by Tate (1933, 1939). However, Hershkovitz (1950) treated this taxon as a subspecies of S. floridanus. I have only cursorily examined the holotype of S. orinoci, and not in the context of whether it belonged in S. brasiliensis versus S. floridanus groups. Notwithstanding, it has recently been shown (Ruedas et al. 2017) that the S. brasiliensis versus S. floridanus groups as construed by Hershkovitz (1950) have mostly heuristic—rather than phylogenetic—value. Furthermore, S. orinoci is specifically distinct both from S. floridanus as well as from all the species treated herein; that matter will be dealt with in a subsequent work.

Sylvilagus sanctaemartae was described by Hershkovitz (1950) as a subspecies of S. brasiliensis, and applied to a form from the lower elevations of the Sierra Nevada de Santa Marta range of northeastern Colombia. Hershkovitz noted that there were ecological differences between the rabbits caught at 335 m in the Sierra, and those collected in the Río Guaimaral and Río César area, just somewhat lower, at 140 m. Regardless, all these primarily forest rabbits disappeared at the forest edge and were replaced in the savannas by taxa currently subsumed in S. floridanus. Notwithstanding this taxonomic replacement, Hershkovitz noted that the taxa allied with S. brasiliensis “…of the César undoubtedly have a wider, more continuous distribution with others in the lowlands of northern Colombia. Presumably they grade into messorius and gabbi of Panama and into nicefori of the forested slopes of the Colombian Andes” (Hershkovitz 1950:355).

A closer inspection of the geographical ecology of the Sierra Nevada de Santa Marta suggests an alternative hypothesis. The Sierra is an outcropping, some 12,000–14,000 km² in area (Tschanz et al. 1974; Alvear et al. 2015) isolated from the Andes chains of northern Colombia, and characterized as lying “far northwest of its proper position within the large-scale geotectonic framework of northwestern South America” (Tschanz et al. 1974:273). Geologically complex and composite, with formations of varying ages (Tschanz et al. 1974; Cardona and Ojeda 2010; Restrepo-Pace and Cediel 2010), it lies far from the Cordillera Central, but relatively close to the Serranía del Perijá, the northernmost Colombian branch of the Cordillera Oriental. To the north, the Sierra rises from the Caribbean seafloor, ca. −4,000 m, to an elevation of ca. 5,700 m above sea level (Pico Cristóbal Colón and Pico Simón Bolívar). As such, Pico Cristobal Colón is the fifth highest prominence on the planet and second in South America only to Aconcagua. To the west, the Sierra is isolated by the lowlands of the Magdalena province and the Magdalena River, as well as large coastal swampland, the Ciénaga Grande de Santa Marta, covering some 2,300 km², and the Santa Marta–Bucaramanga fault system. To the south lies the Magdalena River, into which flows the Río César, which in part forms the eastern boundary. Also to the east lies a valley formed by the southward flowing Río César and northward flowing Río Ranchería. Beyond the César–Ranchería river valley lies the Serranía del Perijá, with peaks higher than 3,600 m and reaching almost to the Gulf of Venezuela. East of the Serranía del Perijá is the Maracaibo Basin and beyond that, the Cordillera de Mérida, the northeasternmost branch of the Andes, reaching as high as 4,978 m (Pico Bolívar) and with extensive highland regions extending to the Caribbean at the Triste Gulf, then beyond Caracas as far east as the Unare River, at 65°11′W. The biological wealth of the area is almost unparalleled: Davis (1996:17) noted that “A naturalist need only spin the compass to discover plants and even animals unknown to science,” a sentiment echoed by Woodman (2002), who—in describing a new species of shrew from the adjacent Sarranía de Perijá—suggested that the Sierra Nevada de Santa Marta could be the next frontier in shrew species discovery. Alvear et al. (2015) provided an excellent summary of the current state of knowledge regarding geology and biology of the area. Mammals of the area were summarized by Solari et al. (2013).

To emphasize and summarize the data presented above, the Sierra Nevada de Santa Marta is ecologically distinct and isolated from nearby biomes and ecological regions. There exist a number of biogeographic obstacles in the nearly 2,000 km separating the Sierra Nevada de Santa Marta from the upland forests (100–200 m) in southwest Suriname, east of the Corantijn River. Any of these physical, ecological, or biogeographical obstacles could be insurmountable in isolation, let alone in combination, when confronted by a species with the biological characteristics of cottontails in the genus Sylvilagus. Thus, the probability of conspecificity for individuals of Sylvilagus distributed in those 2 locations was likely to be relatively low. I therefore used morphological attributes and characters in a comparative framework to test the hypothesis of conspecificity and, finding the hypothesis wanting, below name and describe a novel species of Sylvilagus from Suriname.

Materials and Methods

Morphological characters

Terminology of cranial characters and features generally follows Wible (2007) and Ruedas (1998); measurements were defined by White (1987) and Ruedas (1998), and were extensively detailed by Ruedas et al. (2017). Mensural characters included: GLS, greatest length of skull; POSTORB, width of postorbital constriction; BROSTR and DEPROSTR, breadth and depth (height) of rostrum; BBRAIN, breadth of braincase; ZYGO1, greatest width across the masseteric spine; ZYGO2, zygomatic breadth; LZYGO, length of zygomatic arch; NASALL, greatest length of nasal bone; NASALW, greatest width across left and right nasal bones; I2P2, least alveolar length of I2–P2 diastema; P2M3, greatest alveolar length of P2–M3 toothrow; HBRAIN, height of braincase; HBULLA, height of bulla; CONDL, condylopremaxillary length of cranium; LPALFOR, WPALFOR, length and width of incisive foramina; PALONG, palatal length; PALBRDGE, greatest anteroposterior dimension of palatal bridge; BASIOC, anteroposterior length of basioccipital; WIDBULL, width of auditory bulla; ANTBULL, anteroposterior length of auditory bulla, from the most anterior projection of the ectotympanic to the most posterior point between the occipital and the paracondylar processes of the exoccipital; INTBD, least breadth across the basioccipital between the ectotympanic bones; OCCOND, width across the occipital condyles; INTBOC, length between the posteriormost edge of the palatal bridge and the suture between the basioccipital and basisphenoid bones; CHOANA, breadth of choanae; MASTOID, greatest breadth across the mastoid exposure of the petrosal; DEPZYGO, least anteroposterior length across the maxillary bone at the base of the masseteric spine on the maxillary portion of the zygomatic arch; ip3, least alveolar length of i–p3 diastema; MANDEP, depth of mandibular body; p3m3, greatest alveolar length of p3–m3; HMAND, height of the mandible; HPTT, distance from ventral aspect of angular process (labial to pterygoid shelf) to most dorsal aspect of pterygoid tuberosity; BCON, breadth of condyloid process; LMAND, length of mandibular body. Characters of lower premolar 3 follow White (1987), Ruedas (1998), and Ruedas et al. (2017). Dental characters recorded from the third lower premolar included: wpostlof, greatest width of the posterior loph; wpm3, greatest width; aplpostl, greatest anteroposterior length of posterior loph; wantloph, greatest width of anterior loph; lantloph, greatest anteroposterior length of anterior loph; lpm3, greatest anteroposterior length; antlofar, area circumscribed by enamel in the anterior loph; pstlofr, area circumscribed by enamel in the posterior loph. Statistical analyses were performed using the Statistical Analysis System (SAS) software, version 9.4 (2002–2012—SAS Institute Inc. 1988a, 1988b), generally following Ruedas (1995, 1998). Significance in all analyses was set at α = 0.05; because of the restricted sample size, 0.10 ≥ α > 0.05 also are reported (Moyé 2000). Similarly, no attempt could be made to determine presence or extent of sexual dimorphism in the taxa examined, although sexual dimorphism has been reported in measurements of Sylvilagus (Orr 1940) and could affect results of multivariate analyses (Reyment et al. 1984) given the small sample sizes of the present study (S. brasiliensis, 2; S. sanctaemartae, 13; Sylvilagus species nova, 8: see Appendix I). Univariate statistics (mean, SD) were calculated using the UNIVARIATE procedure of SAS. Analysis of variance was carried out using the GLM procedure, enabling the MEANS routine with option REGWQ, which controls for type I error (SAS Institute Inc. 1988b; Day and Quinn, 1989). A principal component analysis (procedure PRINCOMP) was carried out on the covariance matrix of log-transformed normalized measurement values data. Such a posteriori grouping methods are preferred over a priori grouping methods (multiple range tests, canonical discriminant analysis) because there is no prior hypothesis as to the putative identity of specimens examined. These data are useful to examine ontogenetic growth patterns, which in the sample covariance matrix can be construed as the dispersion of points along the major axis of each sample, with the first eigenvector representing Huxley’s allometric equation (Voss et al. 1990).

Dental characters

Drawings of p3 were made by tracing from photographs taken using a Canon EOS 30D digital camera mated to a Canon MP-E 65 mm f/2.8 1-5X Macro Photo lens, or a Canon EOS 6D mated to the same lens or an AmScope CA-CAN-SLR-III camera adapter for microscopes, shooting either through a camera tube on a binocular dissecting microscope or an ocular tube with the ocular removed, also on a binocular dissecting microscope. Among leporids, p3 generally constitute the most informative dental elements for taxonomic and systematic purposes (Hibbard 1963; Dalquest 1979; Palacios and López Martínez 1980; White 1987; Dalquest et al. 1989; White 1991; White and Morgan 1995; Ruedas 1998; Winkler and Tomida 2011; Ruedas et al. 2017). Discrete characters were deemed the most important in this particular research; accordingly, resulting figures were oriented and scaled to the same size in linear dimensions to carry out size-independent comparisons of interspecific characters. Characters considered follow the standard terminology of Palacios and López Martínez (1980), were described in Appendix I of Ruedas (1998), and illustrated in Ruedas et al. (2017). Additional characters useful in distinguishing among lagomorph species were extracted from Palacios (1996) and Palacios et al. (2008).

Results

Morphological comparisons of specimens examined in each of the populations of Sylvilagus suggest that the Suriname population represents a new species (Table 1). Results of a principal component analysis (Table 2; Fig. 1), undertaken using only specimens with all the cranial features complete, fail to falsify this hypothesis. This taxon is described here and compared with S. sanctaemartae, the name currently applied to lowland populations of Sylvilagus allied with S. brasiliensis inhabiting areas north of the Amazon (Ruedas et al. 2017), and with the neotype of S. brasiliensis, recently restricted to coastal areas in the state of Pernambuco, Brazil, which currently is represented only by the neotype and 1 referred specimen; specimens other than these 2 from other parts of South America cannot be referred to S. brasiliensis without a more in-depth assessment, within a taxonomic and evolutionary framework, of morphological or molecular variation, undertaken across the range of Sylvilagus in South America (Ruedas et al. 2017).

Table 1.

Univariate statistics for species under consideration herein. The data include mean ± SD, and range (minimum–maximum). Superscripted letters indicate statistically significant groups, established using the Ryan–Einot–Gabriel–Welsch multiple range test in SAS 9.2 (procedure GLM MEANS statement, option REGWQ). Groups with the same letter are not statistically distinct, and are ranked by size class (a > b). Readers are cautioned that sample sizes are limited (Appendix I), and include juveniles (particularly in Sylvilagus sanctaemartae); these circumstances could compromise the analyses. Stars next to the statistics indicate statistically significant sexual dimorphism; caution again is advised however, as sample size is at best 7 females and 6 males (S. sanctaemartae), or 4 of either sex (S. parentum). For S. brasiliensis sensu stricto, only the neotype’s measurements are presented: I consider only 2 known specimens to constitute this taxon at the present time (Ruedas et al. 2017). For S. parentum, measurements are those of the holotype followed by data as described above.

Character	S. sanctaemartae (n = 13)	S. brasiliensis	S. parentum (n = 8)
Greatest length of skull	67.20a ± 5.33, 59.7–77.0	71.4a	76.7, 75.30a ± 2.46, 72.1–78.2
Length of nasals	*25.31b ± 1.80, 22.7–29.2	31.4a	33.8, 31.69a ± 2.39, 28.6–34.7
Width of nasals	*11.93b ± 1.05, 9.9–13.2	15.4a	17.0, 15.05a ± 1.79, 12.2–17.4
Zygomatic breath at spine	30.23b ± 1.41, 28.2–33.0	33.8a	35.9, 35.15a ± 1.10, 33.7–36.8
Zygomatic breadth	31.65b ± 1.35, 30.2–34.3	34.6a	36.5, 35.73a ± 1.27, 34.0–37.2
Zygomatic length	27.47b ± 2.12, 24.6–32.7	31.4a	33.3, 33.02a ± 1.06, 31.9–34.5
Breadth of braincase	*23.55b ± 0.82, 22.0–24.6	26.3a	25.6, 25.08a,b ± 0.93, 24.0–26.2
Breadth of exoccipital	22.68b ± 1.04, 21.5–24.2	24.0a,b	24.2, 25.00a ± 0.72, 24.0–25.8
Breadth of mastoid process	*21.29b ± 0.48, 20.6–21.9	21.5b	25.8, 24.36a ± 0.56, 23.5–24.9
Breadth of bulla	5.55a ± 0.35, 5.0–6.2	6.3a	6.2, 5.98a ± 0.38, 5.4–6.3
Interorbital breadth	12.16a ± 0.72, 11.4–14.0	12.9a	12.2, 11.76a ± 0.77, 10.6–12.8
Height of rostrum, caudal	14.22a ± 1.09, 12.7–16.6	14.6a	16.8, 16.24a ± 1.23, 15.2–18.5
Height of rostrum, rostral	12.99a ± 0.81, 11.9–14.9	12.9a	14.2, 14.29a ± 0.99, 13.0–16.3
Width of rostrum	16.90a ± 1.21, 15.1–18.6	19.0a	19.8, 18.46a ± 1.30, 16.0–19.8
Length of incisive foramina	15.84a ± 1.68, 13.5–20.0	17.5a	18.8, 17.97a ± 1.34, 15.7–19.5
Width of incisive foramina	5.24a ± 0.62, 4.6–6.5	6.5a	5.8, *5.56a ± 0.40, 5.0–6.0
Palatal bridge length	6.81a ± 0.68, 5.4–8.0	6.7a	7.4, 7.28a ± 0.49, 6.6–7.9
Breadth of choana	6.16a ± 0.73, 5.0–7.2	6.9a	6.5, 6.73a ± 0.41, 6.3–7.2
Width of palatine constriction	8.8a ± 0.80, 7.6–10.2	9.8a	10.2, 9.76a ± 0.27, 9.4–10.2
Width of basisphenoid constriction	5.67a ± 0.46, 5.0–6.3	5.7a	6.1, 5.96a ± 0.17, 5.6–6.1
Alveolar length of maxillary toothrow	*12.63b ± 1.05, 11.3–15.0	14.6a	14.4, 14.27a ± 0.60, 13.5–15.2
Diastema length	18.64a ± 1.62, 16.2–21.9	19.9a	23.3, 21.76a ± 1.37, 19.8–23.3
Mandibular toothrow length	13.00a ± 0.80, 11.4–14.0	14.0a	14.8, 14.8a ± 0.72, 14.1–15.9
Mandibular diastema length	14.64a ± 1.54, 12.7–18.1	15.0a	18.4, 17.20a ± 0.68, 16.4–18.4
Height of mandibular ramus	*31.00b ± 1.96, 27.8–34.6	35.1a	38.5, *37.18a ± 1.57, 35.1–39.2
Length of articular condyle	*7.59b ± 0.57, 6.8–8.3	8.9a	10.6, *9.77a ± 0.68, 8.6–10.6
Least length of articular notch	6.43b ± 0.42, 5.8–7.1	7.7a	9.1, 8.25a ± 0.64, 7.0–9.1
Ventral length of mandibular ramus	*20.44b ± 1.74, 17.5–22.9	24.0a	26.6, 25.58a ± 0.82, 24.6–26.6

Table 2.

Eigenvector loadings from the principal component analyses. Column 1 lists the variables. Listed are the first through fourth eigenvectors of the principal component analysis (PCA) carried out on the covariance matrix of the log-transformed normalized measurement values; these eigenvectors jointly account for 80.6% of the variation (PC1–PC7 = 92.2%). The number of characters is somewhat reduced from those presented in Table 1 to increase the sample size for the analysis. Individuals removed due to missing characters include: Sylvilagus sanctaemartae: USNM 279976, 279979, 279990, 279992, and 279994; S. parentum: RMNH 30798, 30799, and 31154.

Character	PC1	PC2	PC3	PC4
Length of nasals	0.304	0.019	0.061	0.031
Width of nasals	0.389	0.156	0.032	−0.581
Zygomatic breath at spine	0.089	0.183	0.004	0.076
Zygomatic breadth	−0.005	0.229	−0.021	−0.026
Zygomatic length	0.140	−0.056	0.081	−0.024
Breadth of braincase	−0.073	0.388	0.003	0.042
Breadth of bulla	−0.053	0.392	−0.111	0.159
Interorbital breadth	−0.303	0.492	0.307	0.080
Height of rostrum, caudal	0.041	−0.018	0.223	0.118
Height of rostrum, rostral	−0.028	0.009	0.206	0.348
Width of rostrum	0.078	0.307	−0.104	−0.266
Length of incisive foramina	0.132	−0.211	0.177	0.132
Palatal bridge length	−0.268	0.145	−0.537	0.061
Breadth of choanae	−0.076	0.061	0.597	−0.117
Alveolar length maxillary toothrow	0.052	0.131	0.184	0.018
Diastema length	0.159	0.088	0.176	−0.246
Mandibular toothrow length	0.006	−0.077	−0.007	0.085
Mandibular diastema length	0.070	0.152	−0.008	−0.166
Height of mandibular ramus	0.224	−0.183	0.029	0.011
Length of articular condyle	0.429	0.123	0.022	0.138
Least length of articular notch	0.398	0.182	0.011	0.319
Ventral length of mandibular ramus	0.277	0.169	−0.079	0.376
Eigenvalue	0.008	0.003	0.001	0.001
Proportion of variance explained	0.495	0.182	0.076	0.053
Cumulative proportion	0.495	0.677	0.753	0.806

Fig. 1.

Graphical representation of the results of a principal component analyses carried out on the mensural data, derived from the covariance matrix of the log-transformed normalized measurement values for skulls and mandibles of specimens with the complete suite of measurements available for the analysis, as summarized in Table 2: Sylvilagus brasiliensis, diamond; S. sanctaemartae, squares; S. parentum, circles. Open symbols represent holotypes (neotype in the case of S. brasiliensis). Horizontal axes, principal component 1; vertical axes, principal component 2.

Sylvilagus parentum, species nova

Suriname Lowland Forest Cottontail, bakrakondre konkoni, langa jesi konkoni

Figs. 1–6

Fig. 2.

Overview of external appearance of the species under consideration herein. Dorsal and ventral views of (left to right) Sylvilagus parentum, RMNH 31149, ♀, holotype; S. sanctaemartae, USNM 279993, ♀, holotype; S. brasiliensis, UFPE 1740, ♂, neotype. Scale bar (left) equals 30 cm.

Fig. 3.

Skulls of type material of Sylvilagus under consideration herein, with crania scaled to the same size to highlight shape differences. Top row, left to right, dorsal, ventral, and lateral views of Sylvilagus parentum, RMNH 31149, ♀, holotype; greatest length of skull (GLS), 76.7 mm. Middle row, S. sanctaemartae, USNM 279993, ♀, holotype; GLS, 77.0 mm. Bottom row, S. brasiliensis, UFPE 1740, ♂, neotype; GLS, 71.4 mm.

Fig. 4.

Lateral perspective of the left mandibles, scaled to the same anteroposterior dimensions, of (top to bottom): Sylvilagus parentum, RMNH 31149, ♀, holotype; height of mandible (HD), 38.5 mm; S. sanctaemartae, USNM 279993, ♀, holotype; HD, 31.1 mm; S. brasiliensis, UFPE 1740, ♂, neotype; HD, 35.1 mm.

Fig. 5.

Diagrams of the occlusal surface enamel patterns of third lower molars of selected specimens examined herein. Top 2 rows, holotype and paratypes of Sylvilagus parentum; bottom 3 rows: paratypes and holotype of S. sanctaemartae. Top row, left to right: RMNH 31149, holotype, ♀; RMNH 30798, ♂; RMNH 30799, ♂; RMNH 30800, ♀. Second row, left to right: RMNH 30801, ♀; RMNH 31153, ♀; RMNH 31154, ♂; RMNH 31909, ♂. Third row, left to right: USNM 279975, ♂; USNM 279976, ♀; USNM 279979, ♂; USNM 279980, ♂. Fourth row, left to right: USNM 279981, ♀; USNM 279982, ♀; USNM 279983, ♀; USNM 279989, ♂. Bottom row, left to right: USNM 279990, ♀; USNM 279991, ♂; USNM 279993, holotype, ♀; USNM 279994, ♀.

Fig. 6.

Dorsal (right 2 columns), ventral (middle 2 columns), and lateral (left 2 columns) aspects of the crania of selected specimens of the species discussed herein. All crania scaled to the same size to highlight shape differences. In each column doublet, specimens are the following: right, from top to bottom, paratypes of Sylvilagus parentum: RMNH 30798, 30801, 31153, 30800, 31909. Bottom of second column from right, neotype of S. brasiliensis, Universidade Federal de Pernambuco UFPE 1740; remaining 4 crania, from top to bottom, S. sanctaemartae: USNM 279989, 279981, 279982, 279975. Of these, USNM 279989 is inferred to be a paratype; the remainders are presumed referred specimens (Hershkovitz [1950] did not explicitly list specimens, rather he indicated the number of paratypes, presumably also topotypes, while referred specimens came from a different locality).

• Lepus brasiliensis Linnaeus, 1758:58. Part. Type locality: “America meridionali” [South America]

• Sylvilagus brasiliensisHoogmoed, 1983:35. Part, not Linnaeus, 1758. Hoogmoed (1983:41) noted that “…ik het niet gewaagd om de Surinaamse konijnen tot een bepaalde subspecies te rekenen.” [I have not dared to assign the Surinamese rabbits to a particular subspecies [of S. brasiliensis].]

Holotype

The holotype is a specimen of a fully adult female, prepared as a skin and skull, Rijksmuseum van Natuurlijke Historie—Naturalis Biodiversity Center no. 31149, collected by D. G. Reeder, of the RMNH, field number DGR 81-37, on 15 May 1981, with 2 embryos preserved in alcohol. External measurements (recorded from skin tag) are: head and body, 390 mm; tail, 25 mm; hind foot, 85 mm; ear: 60 mm; mass, 1,460 g.

Type locality

Collected in Suriname, Nickerie District (N.B.: currently in Kabalebo Resort [administrative division] of Sipaliwini Dist.), on the Avanavero–Amotopo road, 110 km from Avanavero airstrip. Based on figure 2 of Hoogmoed, 1983, I estimate the geographic coordinates as ca. 4.32°N, 57.69°W (WGS84; no datum provided by Hoogmoed, [1983]), elevation ca. 110–130 m. Ecologically, all of the specimens examined inhabit the Guianan Moist Forest ecoregions (Olson and Dinerstein, 1998; Olson et al. 2001), a region that remains relatively unknown biotically but quite rich insofar as those studies that have been undertaken (Tate, 1939; Husson, 1978; Voss and Emmons, 1996; Simmons and Voss, 1998; Lim et al. 1999; Simmons et al. 2000; Voss et al. 2001; Lim and Engstrom, 2001, 2005; Lim and Tavares, 2012; Brûlé and Touroult, 2014; Lim and Catzeflis, 2014; Lim 2016).

Paratypes

I also examined at RMNH—Naturalis specimen numbers 30798 ♂ (207 km from Avanavero airstrip), 30799 ♂ (skull crushed; 210 km), 30800 ♀ (215.5 km), 30801 ♀ (205 km), 31149 ♀ (110 km), 31150 ♂ (skull missing; between 114.5 and 123 km), 31153 ♀ (skull plus head skin only; 218 km), 31154 ♂ (skull broken; 219 km), and 31909 ♂ (219 km; this specimen was not listed in Hoogmoed, 1983).

Additional specimens

Hoogmoed (1983) listed additional specimens, preserved complete in fluid unless otherwise specified, that I did not examine: 31151 ♂, 31152 ♂, 31773 (sex unknown, skull missing), 1 of 2 embryos of 30801, 2 embryos of 31149 (holotype), and 2 embryos of 31153. Ancillary materials also include RMNH 19677 (fecal pellets; source animal lagomorph but not identified; likely the new species described herein). An additional specimen, field number DGR 81-47, collected 214 km from Avanavero, was listed by Hoogmoed (1983) as being deposited in the University of Suriname; I have not been able to confirm the existence of this latter specimen.

Specimen RMNH 18229, a juvenile-age female weighing 75 g and preserved in fluid, is from the Kayser airstrip, near the Kayser Mountains (variously Kayser Gebergte, or Käysergebergte), highlands of southern Suriname (3°05′35″N, 56°28′25″W, 260 m), and constituted the first record of Sylvilagus from Suriname (Husson, 1978:369). I am explicitly not designating this specimen as a paratype of S. parentum. Given the tight correspondence previously documented between taxonomy and ecology in Neotropical Sylvilagus (Ruedas et al. 2017), and the strong ecological differences between the locations along the Kabalebo River drainage where holotype and paratypes of S. parentum were collected, including such factors as precipitation, soils, elevation, etc., I hypothesize that RMNH 18229 may represent a species of Sylvilagus distinct from those currently described. Collection of adults from this area should be a priority in order to undertake a more informed assessment of the hypothesis of conspecificity between Sylvilagus in these 2 areas.

Etymology

Genitive plural of the Latin noun parens, for my parents, Patricio Ruedas Younger (11 January 1931–22 February 2014) and Paloma Martín Daza (b. 25 January 1937), who supported me in so many ways during my life. Ordinarily, I prefer indigenous names; however, in the present instance, there are none such (E. B. Carlin, Section of Languages and Cultures of Native America, Leiden University Centre for Linguistics, in litt.). The Maroon name is konkoni, a name clearly derived from the Dutch “konijn,” and a term also applied to the agouti, Dasyprocta sp. (D. cristata, of uncertain taxonomic status, and D. leporina both occur in the region), hence lacking in information content.

Nomenclatural statement

A life science identifier (LSID) number was obtained for the new species Sylvilagus parentum: urn:lsid:zoobank.org:pub:F0D021E1-99B0-45E9- AF50-6C67F3982228.

Diagnosis

Sylvilagus parentum is distinguishable from other species of Sylvilagus by the combination of having extremely long and slender maxillary processes of the frontal bones; posterodorsal process of premaxilla extending only barely caudad to nasal bones; nasal bones ending caudolaterally in a point; relatively long, thin process on frontal bone extending between posterodorsal process of premaxillae and nasal bones; large, prominent postorbital processes, ending caudally in a blunt, almost squared off manner; mediocaudal aspect of postorbital processes in contact with, but not fused, to frontal bone; single, generally minute, craniopharyngeal foramen in basisphenoid; and lack of premolar foramen.

Description

Body relatively large for a South American Sylvilagus (Fig. 2), body measurements (from fresh specimen): head and body, 390 mm; tail length, 25 mm; hind foot, 85 mm; ear length, 60 mm; mass, 1,460 g. Pelage (Fig. 5) somewhat lighter in overall appearance than habitual for tropical forest Sylvilagus species. Colors assessed by direct comparison with Smithe (1975). Nuchal patch Orange Rufous to Mahogany Red in general appearance; upon closer (microscopic) examination, predominant tones between Flesh Ocher and Yellow Ocher, approximately 8–10 mm in length, distal 2 mm or so tipped with Jet Black, and sparser, bicolored emergent hairs, ca. 15 mm long, also between Flesh Ocher and Yellow Ocher, with distal half Jet Black. Middorsal fur approximately 22 mm in length, with thick, fine, Pale Neutral Gray underfur approximately 6–8 mm in depth, emergent hair coarse and agouti with basal band ca. 13–14 mm closest to Vandyke Brown (sample 121, not 221, of Smithe, 1975), followed by a central band, closest to Clay (sample 123B, not 26, of Smithe, 1975), ca. 5–6 mm in length, and distal tip Jet Black, 4–5 mm in length. Midventral hair ca. 17 mm in length, with thick, primarily white to Pearl Gray underfur constituting basal ~6 mm. Remaining emergent fur in midventral area consisting of coarse predominantly white hair, with a slight tinge of Chamois. Fur from border between ventral and lateral aspects with thick, fine, Grayish Horn underfur, 5–7 mm thick, with emergent hair closest to Cream; total length of emergent hairs in ventrolateral transition area mostly ca. 15 mm, with some sparser hairs up to 20 mm. Fur in gular area about 25 mm in length, basal 15 mm consisting of thick, fine, Light Neutral Gray underfur; emergent hair with a ca. 7 mm band close to Clay, with remaining distal band almost Jet Black, with a tinge of brown. Hair from tail about 8–14 mm in length, with an almost inconspicuous basal layer of thick, fine, underfur Drab Gray to Glaucus (latter color 80, not 79, of Smithe, 1975), emergent hair mostly Robin Rufous, with scattered sparse agouti hairs varying in color distribution. Hind feet Warm Buff dorsally, getting progressively lighter distally; plantar surface thickly furred, between Prout’s Brown and Brussels Brown.

Skull (Fig. 3) relatively large for a Sylvilagus (see “Comparisons,” below): greatest length of skull, 76.7 mm; basilar length (measured after the fact from photographs), about 61 to 62 mm; zygomatic breadth, 36.5 mm. Other measurements listed in Table 1. Nasal bones relatively long (33.8 mm, r, and 33.1 mm, l), posterolateral margin of nasal bones terminating fairly acutely, with a long, thin process on frontal bone extending between posterodorsal process of premaxillae and posterolateral margin of nasal bones. Nasal bones begin to diverge about 19 mm from their most rostral edge, about 11.5 mm down the midline of the cranium from an imaginary line drawn from the caudal edge of either nasal bone. Nasal bones diverging from each other caudally at a very acute angle; angle, from point of divergence to caudal extreme of right nasal bone, about 143°. Left nasal bone with a caudad extension down the midline starting where the 2 nasal bones diverge; extension likely corresponding to unpaired internasofrontale, or nasofrontale, imperceptibly fused in holotype with left nasal (present in isolation in paratype RMNH 30801). Lacrimal tubercles quite prominent, extending 3.7 mm (l) to 2.9 mm (r) from lacrimal and maxillary bones. Supraorbital shelf essentially lacking antorbital processes; postorbital process strong, only barely in contact with the frontal at the tuberculum frontoparietale (I note in passing that this structure is named such for its position in other leporids: in the species described herein, the tuberculum frontoparietale is in fact on the suture line between the frontal and squamosal bones, somewhat rostral to the closest point where the parietal meets the frontal and squamosal bones).

Frontal bones mostly smooth anteriorly, to about midway between the posterior supraorbital incisures, with some pitting rostrally along the sides of the frontal bones to about the rostral extent of the supraorbital shelf. Pinhole-sized fenestra along the suture of the frontal bones, located approximately even anteroposteriorly with the least interorbital breadth. Frontoparietal suture heavily crenellated, with strong interdigitation between frontal and parietal bones, as well as between parietal and squamosal bones. Parietal bones largely smooth except along either side of their midline suture. Two parietal ridges present: prominent one beginning as a caudad projection from zygomatic process of squamosal bone then transitioning from squamosal to squamoso–parietal suture, to terminus of suture at supraoccipital. Less prominent ridge beginning at the tuberculum frontoparietale and dorsally paralleling the previously described ridge, ending diffusely in the parietal bones just rostral of the interparietal bone. Latter ridge marked dorsally with some pitting on the parietal.

Dorsolateral wing of rostral aspect of maxillary bone progressively more fenestrated in a dorsal direction, with some initial lightening of the bone just dorsal to the level of the dorsal rim of the infraorbital foramen, first true fenestration beginning about 3 mm dorsal to infraorbital foramen (ca. 7 mm dorsal to the palate), and progressing to an intensely latticed portion for the last ~4 mm until the dorsal margin of the maxilla (ca. 9–10 dorsal to the palate); latticed area about 53 mm² on the right, left not measured. Zygomatic arch quite robust, greatest dorsoventral depth 5.3 mm, zygomatic fossa relatively inconspicuous. Entopterygoid crest of pterygoid prominent, projecting ventrally, with large, caudally recurved hamulus.

Incisive foramina long and relatively narrow, 18.8 and 5.8 mm, respectively. Incisive foramina ending caudally about where an imaginary line between the rostral margin of the alveolus of the first upper premolar (technically upper premolar 2, PM2). Palatal bridge, formed by the maxillary and palatine bones, extends from the aforementioned line to—rostrally on either side of the midline—another extending from between PM4 and M1. Premolar foramen indistinct if present, area rostral and medial to PM2 displaying some pitting, major palatine foramina relatively small but conspicuous. Single, small but conspicuous carniopharyngeal canal. Foramen ovale relatively prominent, length along longest anteroposterior axis approximately 2.4 mm (left) and 2.8 mm (right).

Dentaries (Fig. 4) largely unremarkable. Diastema (alveolar length of i–p3) very long, 18.4 mm; dorsoventrally relatively thin. Coronoid process relatively prominent; condyloid process high (greatest vertical length of mandibular ramus 38.5), and long anteroposteriorly at the articular head of the condyle: 10.4 mm across articular head from front to back, 9.3 mm from front to back across “neck” of condyle, between the anterior and posterior mandibular incisures. Angle relatively very large, both long and high, with very strong masseteric line on the inferior margin projecting ventrolaterally, extending weakly posterodorsally; thin, weak pterygoid tuberosity, with a more prominent masseteric tuberosity extending ventrally to merge on the ventral margin of the angle with the masseteric line. Small fenestra present near the ventrocaudal bend of the angle, just rostral to the ventral projection of the masseteric tuberosity. Some pinhole fenestrae present between ventral projection of pterygoid tuberosity and ventral projection of masseteric tuberosity.

Dental terminology follows Palacios and López Martínez (1980; see also figure 9 in Ruedas et al. 2017). Third lower premolar (Fig. 5) about 2.8 mm in greatest rostrocaudal dimension, 2.5 mm from most lingual extreme of tooth (on lingual margin by lingualmost invagination of hypoflexid) to labial margin of protoconid, and 3.1 mm from most lingual extreme of tooth to labial margin of hypoconid. Single, marked anteroflexid and second, less prominent invagination on anterior aspect of labial anteroconid. Enamel of anteroflexid and rostral margin of tooth thick throughout. Depth of the anteroflexid from the external surface of the enamel to an imaginary line drawn between the labial and lingual anteroconids approximately 0.2 mm. Paraflexid present but minute. Protoflexid constituted by single entrant; marked, relatively deep (approximately 3 mm from an imaginary line drawn between most labial margin of protoconid and most labial margin of anteroconid). Enamel thickness tapering gradually from the labial anteroconid and the protoconid to the deepest point of the protoflexid, where it is at its thinnest, then increasing very rapidly in thickness onto labial margin of protoconid. Enamel on protoconid and anterior margin of hypoflexid very thick; central angle prominent but relatively open, angle formed is approximately 135°. Enamel on posterior margin of hypoflexid (rostral margin of posterior loph) relatively thin and smooth, with only a slight waving. Labial margin of hypoconid with thick enamel, tapering gently on the caudal margin of the posterior loph. Lingual and linguocaudal margins of entoconid relatively featureless, with thin enamel.

Variation

Although sexual dimorphism has been documented in other species of leporids (Orr, 1940), its extent in S. parentum is unclear on account of the small sample size. In terms of external measurements, data were available for 3 males and 3 females; only mass differed significantly (P = 0.043): mean of females = 1,420 g (1,250, 1,550, 1,460 [holotype]), mean of males = 1,077 g (1,230, 1,000, 1,000). However, 2 of the 3 females (30801, coll. 22 May 1981; 31149, coll. 15 May 1981) had 2 embryos each, doubtlessly contributing to their mass. An additional female (31153, coll. 23 May 1981) was preserved as a head skin plus skull only, with no external measurements recorded, but also was carrying 2 embryos.

Three of the cranial measurements displayed significant sexual dimorphism (Table 1): width of incisive foramina (females: 5.77 ± 0.26, 5.4–6.0; males: 5.13 ± 0.24, 5.0–5.3; P = 0.0469), height of mandibular ramus (females: 38.24 ± 0.08, 37.3–39.2; males: 35.77 ± 1.06, 35.1–37.0; P = 0.0165), and length of articular condyle (females: 10.20 ± 0.42, 9.8–10.6; males: 9.19 ± 0.49, 8.6–9.5; P = 0.0319). Length of mandibular diastema displayed a P-value of 0.0921 (females: 17.57 ± 0.65, 16.8–18.4; males: 16.70 ± 0.34, 16.4–17.0). In all instances, females were larger than males. There were no evident differences in discrete characters between males and females.

Dentition varied among individuals: although most individuals shared characters in pm3 with the holotype as described above, the crenellation of the anteroflexid in particular was variable. RMNH 30798 in particular displays an additional rostral invagination in the lingual anteroconid, almost of the magnitude of a second anteroflexid, and a second shallow invagination in the labial anteroconid. This individual also differs from remaining individuals of S. parentum in displaying a very complex, crenellated caudal margin of the hypoflexid. This individual also displayed more pitting on the cranium (caudal portion of frontals and entire parietal bones) than remaining individuals, and a somewhat sinistral lean to the distal portion of the rostrum. In every other respect, it was otherwise consistent with the general gestalt of the species. One individual (30800) does not have an anteroflexid, only a shallow open folding to separate lingual from labial anteroconids. Another (30799) has very shallow invaginations on the rostral surface of pm3, rather than a definite anteroflexid. In contrast, another individual (30801) has 2 distinct, albeit thin, anteroflexids. Most individuals lack a paraflexid, which is only present in the holotype and in RMNH 30799. While the hypoconid generally has thick enamel on its labial aspect, 1 individual (30801) instead has relatively thin enamel.

Comparisons

In comparison with species of Sylvilagus for which such data are readily available, S. parentum is relatively large; most of the comparator species have basilar length reported, rather than greatest length of skull. I calculated this measurement after the fact as ca. 61 or 62 mm on the holotype, making it smaller than the mean for (in magnitude rank, largest to smallest) S. aquaticus, S. palustris, and S. graysoni, but larger than S. insonus, S. floridanus, S. transitionalis, S. audubonii, S. bachmani, S. nuttallii, or S. mansuetus (data from Orr 1940; Chapman 1974a, 1974b, 1975; Lowery 1974; Chapman and Willner 1978, 1981; Chapman et al. 1980; Chapman and Feldhammer 1981; Thomas and Best 1994; Cervantes 1997; Cervantes and Lorenzo 1997; data on S. transitionalis obtained from Chapman [1975] presumably includes—or indeed, may exclusive refer to—individuals of S. obscurus, between which species he was quite unable to distinguish). Mean zygomatic breadth in S. parentum is smaller than S. aquaticus, S. graysoni, S. transitionalis, and S. floridanus, and larger than S. audubonii, S. bachmani, S. nuttallii (not recorded from S. palustris, S. insonus, or S. mansuetus). Other comparisons similarly place S. parentum among the larger of Sylvilagus species.

More meaningful and detailed comparisons may be undertaken with South American Sylvilagus species (Fig. 6). Recently, S. andinus and S. tapetillus were excised from synonymy with S. brasiliensis (Ruedas et al. 2017). In addition to those species, here I undertake comparisons between S. parentum and S. sanctaemartae. Because the latter was described as a subspecies of S. brasiliensis (Hershkovitz 1950), a brief note on the assessment of S. sanctaemartae as a valid species in its own right is perhaps worth a passing mention.

Hershkovitz’ description of S. sanctaemartae (Hershkovitz 1950), and its distinction from S. brasiliensis, apparently was made exclusively on the basis of coloration and distributional and behavioral characters. Perhaps because he felt it was distinct merely at the subspecific level, no mention was made by Hershkovitz of craniodental characters and how their disposition may have affected his taxonomic decision. In examining material ascribed to this taxon, I found a number of distinct character differences between the S. sanctaemartae material at the United States National Museum—the type series—and the neotype of S. brasiliensis. In particular, the posterodorsal process of premaxillae extends well beyond the caudal margin of nasals, a feature heretofore unique among Neotropical Sylvilagus. There is evidence of an antorbital process at least on certain of the specimens, and the relatively massive postorbital process is almost invariably tightly fused to the frontal bone over a considerable extent of its length, such that the posterior supraorbital incisures are become foramina rather than incisures. As in all Sylvilagus, the caudal aspect of the palatal bridge is constituted by the palatine; however, the palatine in S. sanctaemartae is extremely broad, essentially running from left to right alveoli beginning rostrally midway on premolar 4. Such is the breadth of the palatine that in most individuals, it stretches laterally to some extent between molars 1 and 2. Most notably, although not evident on the holotype because of the presence of tissue obstructing the view, all specimens of S. sanctaemartae appear to have 2 foramina on the basisphenoid. The large size and smooth, ventrocaudally oriented caudal aspect of the caudal foramen suggests that it is homologous with the craniopharyngeal canal, although only a dissection of soft tissue will indubitably ascertain this assertion. A single foramen, or collection of multiple localized pinhole foramina, on the rostral aspect of the basisphenoid appears to be a widespread phenomenon among certain Neotropical Sylvilagus. For example, it is present in S. g. gabbi (USNM 37794/11371), S. g. messorius (USNM 179569), and to some extent in S. g. truei (USNM 34878/6357); also present as a pinhole in S. g. consobrinus (AMNH 36793) and diffusely present as a linear series of pinhole foramina in S. g. tumacus (AMNH 28409), wherein the craniopharyngeal canal is broken in 2 by a lateral strut, and in S. kelloggi (AMNH 60515), it is present as a series of generally localized pinhole foramina. However, this secondary basisphenoid foramen is absent from all specimens of S. parentum, as well as S. brasiliensis and S. andinus.

The dentition of S. sanctaemartae also is distinct from that of S. brasiliensis (Fig. 5). There is sufficient intraspecific variation that the differences are not diagnostic, but trends are evident, and certain individuals are diagnosably distinct. Most of the individuals examined, including the holotype, have a relatively complex rostral margin to the anterior loph, including numerous invaginations, or crenels, that render impossible the identification of the anteroflexid, or even a distinction between lingual and labial anteroconids. In addition, there is a tendency for the protoflexid to be shallow.

Mensurally, the specimens examined of S. sanctaemartae differ from the neotype of S. brasiliensis in a number of measurements (Table 1). Of the 28 measurements recorded, 12 were different, almost 9 times the number that would have been expected by chance alone at α = 0.05. This also is highlighted by the principal component analyses (Fig. 1). In the principal component analysis derived from the covariance matrix of the log-transformed measurement values (Fig. 1), when the size component is removed, clear distinctions become evident not just between S. sanctaemartae and S. brasiliensis. The present analysis also suggests some degree of distinction between the holotype and paratypes on the one hand (the 3 specimens plotted in the lower left of Fig. 1), and the remaining specimens listed in the type series on the other, which are from a different locality on the southeast lower slopes of the Santa Marta massif, in the César–Ranchería river valley (possibly ca. 9°56′14.25″N, 73°36′44.81″W [WGS84], elev. ca. 53 m), in contrast to the type locality that lies on the western slope of the Santa Marta massif, only a few km first from the alluvial plain of the “Zona Bananera” (elevation ca. 40 m), followed by the alluvial plain enveloping a large marshland known as the Ciénaga Grande de Santa Marta (ca. 0 m). The range of this taxon was estimated by Ruedas et al. (2017) to be located ca. 10°22′14.16″N, 73°55′26.35″W (datum: WGS84).

Examination of a broader geographic sample within the putative range of S. sanctaemartae (Fig. 7) suggests that this taxon may be more widespread than previously thought. Ruedas et al. (2017) suggested that S. sanctaemartae was “most likely restricted to the Sierra Nevada de Santa Marta” although their ecological niche modeling analysis supported a more widespread vision, particularly under climate change scenarios (Ruedas et al 2017: Appendix A27–A28). The data presented herein likewise support a more widespread distributional concept for S. sanctaemartae. The 3 specimens illustrated in Fig. 7 are drawn from 3 distinct populations separate from the type locality and the locality of the paratypes (see Appendix I: Specimens examined). However, all have features distinctive of S. sanctaemartae as defined herein, including a relatively robust—in the transverse plane—posterodorsal process of the premaxilla, antorbital present, massive postorbital process fused to frontal bones, posterior of palatine in contact with alveoli of molars, and in particular, 2 foramina in the basisphenoid (in the case of FMNH 68938, this is represented by 1 foramen and a series of pinhole foramina).

Fig. 7.

Dorsal (top row), ventral (middle row), and lateral (bottom row) aspects of the crania of selected specimens of Sylvilagus sanctaemartae from localities other than the type or paratype localities. All crania scaled to the same size to highlight shape differences. Left column: FMNH 68937, ♀, from Colombia: Bolívar; San Juan Nepomuceno (est.: environs of 9°57′4″N, 74°4′46″W, ca. 172 m); GLS = 61 mm. Middle column: FMNH 68938, ♀, from Colombia: Cordoba; Socorré, upper Rio Sinú (est.: environs of 7°51′3″N, 76°16′56″W, ca. 155 m); GLS = 77 mm. Right column: FMNH 68950, ♀, from Colombia: Córdoba; Catival, upper Rio San Jorge (possibly ca. 8°17′2″N, 75°41′0″W, ca. 60 m); GLS = 73 mm.

Collectively, these lines of evidence lead me to conclude that S. sanctaemartae constitutes a species clearly and diagnosably distinct from S. brasiliensis. Because it was recently suggested (Ruedas et al. 2017) that S. sanctaemartae was the oldest available name for Neotropical Sylvilagus species allied to S. brasiliensis southeast of the Darién Gap, East of the Andes, and north of the Amazon, I herein compare S. parentum to both S. sanctaemartae and S. brasiliensis, as these comparisons appear to be the most meaningful.

Externally (Fig. 2), S. parentum differs most obviously from both S. sanctaemartae and S. brasiliensis in the absence of any white pelage on the dorsal surface of the hind feet. Similarly, S. brasiliensis has white extending along the forelegs to the wrist; this is absent in S. sanctaemartae and S. parentum. The latter 2 differ in the forelegs in that S. parentum transitions in the foreleg from grayish tinged with buff nearer the pectoral area to creamy buff distally, whereas S. sanctaemartae is uniformly orange brown in the forelegs. In addition, the gular patch appears to be more extensive in S. parentum (although that may be an artifact of preparation), and its color, described above in detail, differs markedly both from that of S. sanctaemartae as well as that of S. brasiliensis (described in detail in Ruedas et al. 2017). The nuchal patch also appears to be larger in extent than either S. sanctaemartae or S. brasiliensis although that too may be an artifact of preparation. The coloration of the nuchal patch is distinct from that of S. sanctaemartae and S. brasiliensis. Overall, the appearance of S. parentum is somewhat less dark than S. sanctaemartae and S. brasiliensis, which latter display the typically dark appearance of Neotropical forest Sylvilagus.

The skull of S. parentum is most distinctive in having a process extending rostrally on the frontal bone between the posterodorsal processes of the premaxillae and the nasal bones. This process is absent from either S. sanctaemartae or S. brasiliensis. In addition, the nasal bones end caudally in a rather acute manner, whereas they are more rounded in S. sanctaemartae and S. brasiliensis. In the individual (RMNH 31909) with the most rounded caudal margin to the nasal bones, hence most similar to S. brasiliensis, this rostral process of the frontal remains, thereby maintaining its diagnostic identity. Most individuals have a caudally oriented process originating at the midline suture of the nasal bones extending from where they diverge caudally into the midline between the frontal bones. In the holotype and RMNH 30798, it is on the left nasal bone; in RMNH 31153, it is on the right nasal bone; in RMNH 31909, the medial margin of the left nasal is somewhat caudally reflected but without a distinct caudally directed process; and it is absent on RMNH 30800. Inspection of RMNH 30801 suggests that this is an unpaired internasofrontale, or nasofrontale, that is fused to the frontal in other individuals of S. parentum. There is no evidence of this feature in S. sanctaemartae; in the neotype of S. brasiliensis, the left frontal is reflected caudally to a greater extent than in RMNH 31909.

Every individual of S. parentum that I examined had a pinhole fenestra along the midline suture between the frontal bones, about even with the anterior terminus of the supraorbital shelf. This fenestra, if present in S. sanctaemartae, is so in a more indistinct manner (absent on holotype and USNM 279981; unclear in USNM 279989; likely in USNM 279982; and near the midline suture on the right frontal in USNM 279975). In the neotype of S. brasiliensis, there is thinning of the frontal in the middorsal region between the putative antorbital processes, possibly with fenestrae in both frontal bones immediately adjacent to the midline suture. Such fenestrae have been documented in rodents, most particularly ctenomyids, phyllotines, octodontids, and caviids (Gardner and Anderson 2001), but the extent of their presence across lagomorph species remains unclear at this time.

Antorbital processes are essentially absent in S. parentum, although some individuals have a slight uni- or bilateral invagination or notch marking the putative location of this feature, possibly homologous with the supraorbital notch documented in Sciuridae by Wahlert (1974). In this respect, they are similar to S. brasiliensis, but differ from S. sanctaemartae in that the latter, while lacking a well-developed antorbital process, nevertheless displays a distinct and marked emargination in that location of the cranium. While there may be some interdigitation between the mediocaudal aspect of the postorbital process and the frontal bone at the tuberculum frontoparietale (e.g., RMNH 30798), the postorbital process is never fused to the frontal, as is the case in S. sanctaemartae and S. brasiliensis.

The basisphenoid bone is similar in conformation across the 3 species, with the exception that S. sanctaemartae has 2 foramina in the basisphenoid, as noted above, rather than the single craniopharyngeal foramen present in most Sylvilagus species and exemplified in S. parentum and S. brasiliensis. However, all specimens of S. parentum display a small foramen in the presphenoid. This foramen is absent in S. sanctaemartae and S. brasiliensis. The basioccipital of S. parentum has a sharp medial narrowing, or constriction, between the auditory bullae, approximately at the level of the external carotid foramen, approximately at the rostral terminus of the occipital condyles. The angle formed between the most lateral point rostral to the constriction, the constriction, and the most posterolateral point of the occipital condyle in S. parentum is 124° ± 5.1 (mean ± SD); in S. sanctaemartae, it is 149° ± 4.9 (USNM 279993: 147° left, 149° right); in the neotype of S. brasiliensis, left and right, respectively, are 142° and 145°. An analysis of variance indicates that the mean of this angle in S. parentum is significantly different from those of S. sanctaemartae and S. brasiliensis (P < 0.0001).

Lateral to the anterolateral aspect of the basioccipital, the neotype of S. brasiliensis has on its left side, clearly defined, foramen lacerum and foramen ovale, and on the right, a single foramen, a confluent foramen lacerum of Craigie (1948) absent the bony splint demarcating foramen lacerum and foramen ovale. Craigie (1948) suggested that the more ventral aperture, immediately in front (rostral) of the bulla was the foramen lacerum, divided into caudal (foramen lacerum) and rostral (foramen ovale sensu stricto) portions by a slender bony splint. Wible (2007) called this aperture in Romerolagus the foramen ovale, noting that the bar of alisphenoid separating the foramen ovale from the piriform fenestra (caudal aperture) was absent, leading to a single aperture constituted by confluent foramen ovale from the piriform fenestra in Oryctolagus and Romerolagus (see also Gabbert 2004:185, on the “sphenotympanic fissure”). In contrast to the condition of S. brasiliensis, both S. parentum and S. sanctaemartae have a large, singular foramen ovale. More lateral to this foramen, all specimens of S. parentum display some fenestration of the ascending portion of the alisphenoid (best viewed in the lateral perspectives of the crania; Figs. 3 and 6). This fenestration is present in certain of the specimens of S. sanctaemartae, absent in others, and indistinct in some due to tissue remaining over the bone. The fenestration of the alisphenoid is absent in the neotype of S. brasiliensis.

Specimens of S. parentum have a separation between the alisphenoid and the ectotympanic bone (auditory bulla), the piriform fenestra; in some specimens, the piriform fenestra extends between the squamosal and ectotympanic. This fenestra is completely absent from S. brasiliensis. In some specimens of S. sanctaemartae, there are some vacuities between the caudal aspect of the squamosal and the ectotympanic, but not a full fenestra.

Another species recently excised by Ruedas et al. (2017) from synonymy with S. brasiliensis is S. andinus (Thomas, 1897). This species, besides being a highland species likely restricted to elevations above 3,500 m in the Andes, and possibly in Ecuador only, differs from all others treated here. Sylvilagus andinus has a smooth dorsal surface of the cranium, versus pitted; the postorbital processes end in an acute angle, rather than being rounded or squared, and are distinctly not fused to the frontal, being separated from the latter by a marked posterior supraorbital incisures, rather than fused and displaying a posterior supraorbital foramen. The holotype of S. andinus (MNH 1897.11.7.54) has distinct antorbital processes; other specimens of the species do not have antorbital processes differentiated to such a degree, but in all specimens there is a definite emargination to the anterior region of the supraorbital shelf, with marked anterior supraorbital incisures. There are distinct premolar foramina adjacent lingually to upper premolar 4 that are absent in S. brasiliensis, S. sanctaemartae, or S. brasiliensis. These features make S. andinus distinctly diagnosable from the latter 3 species.

Morphometric analyses

I carried out 2 principal component analyses on the mensural data including the 3 species discussed in detail herein. In the principal component analysis undertaken on the correlation matrix of the measurements, I interpret the first component as a size component, based on the relative homogeneity of the magnitude of the eigenvectors constituting it, and the dispersion of points in multivariate space (results not shown). The first 4 principal components accounted for 90.6% of the variance in the data set. Sylvilagus parentum and S. sanctaemartae are generally well separated in principal component 1, accounting for 75% of the variance in the data set, with the exception of the holotype of S. sanctaemartae, which is situated almost within the cloud of S. parentum. It should be noted that the holotype of S. sanctaemartae is an unusually large representative of that species: its greatest length of skull is 77.0 mm; the next largest individual is 70.4, with the rest falling between 61.6 and 69.2 mm. The discrete characters are consistent in discriminating between S. sanctaemartae and S. parentum, but the fact remains that the holotype of S. sanctaemartae is unusually large. Principal component 2, accounting for 6% of the variance, discriminates between S. brasiliensis, and S. sanctaemartae and S. parentum. This is a more shape-based component based on the heterogeneity of the magnitudes of the eigenvectors constituting it. The largest eigenvalue corresponds to interorbital breadth measured at the narrowest point of the posterior orbital incisures. Interestingly, although contributing to shape in principal component 2, this measurement was not significantly different across species in the analysis of variance (Table 1), and had a very small eigenvalue in principal component 1. Width of incisive foramina and breadth of choanae are the next 2 major contributors to principal component 2; these also do not discriminate among the species in the analysis of variance.

The second principal component analysis was carried out on the covariance matrix of the log-transformed normalized measurements (Fig. 1; Table 2), resulting in a shape and growth based assessment of the data; in this analysis, it took 7 principal components to account for > 90% of the variance in the data set (92.2%). Principal component 1, accounting for 50% of the variance, distinguished S. sanctaemartae from S. brasiliensis and S. parentum, while principal component 2, accounting for 18% of the variance, distinguished S. brasiliensis from S. parentum. Length of articular condyle, least length of articular condyle notch, and width of nasals were the principal contributors to principal component 1, all of which were significantly different in the analysis of variance, distinguishing between S. sanctaemartae on one hand, and S. brasiliensis and S. parentum on the other. Interorbital breadth, width of bulla, and breadth of braincase were the primary contributors to principal component 2.

Perhaps of greater interest in this analysis was that the orientation of the principal axis in multivariate space differed between S. sanctaemartae and S. parentum. This suggests a distinct pattern of general growth in either species, given that their principal axes are not parallel (Voss et al. 1990). In addition, as mentioned above, the holotypes and paratypes of S. sanctaemartae differed from the referred specimens of the same species, while maintaining a parallel major axis. This is a suggestive distinction that may bear future scrutiny with enhanced sample sizes.

Distribution

All known specimens of S. parentum were collected from a fairly circumscribed geographic area, adjacent to or near the road between the Avanavero and Amotopo airstrips, between km 108 and km 224 as measured from the Avanavero airstrip. Satellite data gleaned from Google Earth suggest that all these localities are approximately between 100 and 120 m in elevation. The specimens are divided into 2 clusters: the first from between km 108 and km 123, the second from between km 205 and km 219 (Amotopo airstrip is at km 255). The second cluster is in the upper reaches of the Linker Kabalebo River, which also drains into the Kabalebo. Hoogmoed reported that Sylvilagus had been seen by the Amotopo airstrip, but no specimens were collected from that location. The first cluster is in the upper reaches of the Dalbana River drainage, a north-flowing stream that drains into the Kabalebo River just SW of Avanavero. No collections appear to have been made in the environs of the Corantijn River, into which drains the Kabalebo River. Hoogmoed (1983) suggested—and was puzzled by—a distribution restricted to western Suriname. The current distribution of this species therefore appears to be upper reaches of the Kabalebo River drainage system. Whether the species truly is restricted to so circumscribed a geographic provenance remains unclear.

Ecology and habitat

Hoogmoed (1983) reported that S. parentum were more common during the rainy season than the dry. Only 1 specimen was obtained during the beginning of the 1980 dry season, on the night of 2 October; another was observed on the night of 1 October. In contrast, 9 specimens were collected during the rainy season collecting of 1981, between 27 April and 9 June. No specimens were obtained during the dry season collecting period of 1981 (18 October–1 December), although Hoogmoed (1983) noted that the latter was primarily focused on botanical fieldwork. Hoogmoed (1983) indicated that, although the road cut through primary forest, all specimens were collected within 50 m of the road, in what he described as secondary forest; no cottontails were observed in primary forest by Hoogmoed and coworkers.

Although the road between Avanavero and Amotopo airstrips traverses apparently homogeneous lowland forest habitat, the geological substrate is variable. The specimens at RMNH from the 2 collection clusters all are from areas of Lower Proterozoic granitoid rock (National Planning Office of Suriname 1988). Geological formations intervening between the 2 collection clusters include principally Permo–Triassic Apatou dolerites and Precambrian Avanavero dolerites, as well as some Lower Proterozoic Dalbana rhyolites. More detailed recent work by Kroonenberg et al. (2016) classifies the granitoid rock at the sites of collections as Wonotobo Granite and Sipaliwini Leucogranite (ca. 1,980 ± 6 Ma and 1,980 ± 4 Ma, respectively) and the intrusions as somewhat younger Avanavero and Käyser Dolerites (ca. 1,794 ± 1.6 Ma and 1,501 ± 5 Ma, respectively). The Dalbana Formation elements, a complex of felsic metavolcanic rocks, date from 1,987 ± 4 Ma. It is unclear at this point whether the clumped distribution of the RMNH specimens is an artifact of collecting, or whether the differing Proterozoic basements have developed somewhat different floras, thereby potentially resulting in the exclusion of Sylvilagus from ecologically unsuitable areas.

Remarks

The specimen representing a juvenile (RMNH 18229; 13 September 1965) from the Kayser airstrip (Hoogmoed 1983) is unlikely to represent S. parentum. There are deep and extensive ecological and geographical distinctions between the 2 localities that most likely preclude their constituting a single panmictic population. In addition, the underlying geological formation of the Kayser airstrip is Dalbana Formation, which suggestively constitutes a break between collection clusters of S. parentum. Be that as it may, the morphological features of juvenile Sylvilagus are not conducive to irrefragable identification. Additional specimens from the area of the Kayser airstrip, of adults, would therefore likely prove most stimulating.

Discussion

The area of distribution of S. parentum was identified by Haffer (1969) as lying within the approximate bounds of what he called the Guiana Refuge. Haffer (1969) suggested that the diversity of the current Amazonian fauna is due to contraction of forest fragments during dry spells in the Pleistocene and Holocene, followed by expansion during intervening humid periods with an accompanying fauna that would by then have been putatively isolated by genetic, behavioral, or other, means. Cracraft (1985) in contrast suggested that differentiation of the South American fauna was better understood using the concept of centers of endemism, pointing out that “scarcely any evidence exists […] to indicate where refuges might be located” and that “Hypotheses about refuges and their historical significance are only as good as our understanding of areas of endemism and their history, because the concept of refuges is a derivative of our knowledge about endemism” (Cracraft 1985:50). In the Cracraft (1985) model, the area of distribution of S. parentum was included in a more expansive Guyanan Center of Endemism characterized by some 33 endemic species of birds. As Cracraft (1985) observed, however, there is no necessary incompatibility between the 2 concepts: the ages of the regional biotas need not be restricted to Pleistocene events: increasingly, data show cyclical oscillations in climate at fairly regular intervals (theoretical model—Paillard 1998; empirical data—Petit et al. 1999; Jouzel et al. 2007; Cramer et al. 2011). Considered together, the biogeographic hypotheses and climate data suggest that episodes of speciation should occur anywhere in deep time, rather than being singularly associated with a distinct event (Blois et al. 2013; Leyte and Rogers 2013). Indeed, current evidence does not support the hypothesis of a singular refuge in the Guiana region (Boisselier-Dubayle et al. 2010).

With respect to the Guiana region in particular, the confluence of climate and biogeography leading to numerically exceptional speciation can be appreciated by recent molecular work on a number of vertebrate taxa. This is borne out in the amphibian genus Atelopus (Noonan and Gaucher 2005, 2006), and multiple species in multiple genera of frogs (Fouquet et al. 2012, 2016; Kaefer et al. 2013), in the lizard genera Plica (De Oliveira et al. 2016) and—adjacent to the Guiana region—in lizards of the genera Enyalius (Rodrigues et al. 2014) and Gymnodactylus (Domingos et al. 2014), and birds (Ribas et al. 2012). Perhaps most spectacularly, however, recent molecular work on primates has demonstrated a remarkable endemism in the Guiana region, with a striking number of cryptic species present, and their relevance to biogeography and evolution of the group. In particular, Ateles has a species endemic to the region (A. paniscus—Morales-Jimenez et al. 2015), as does Saguinus (S. midas, S. bicolor, and S. martinsi—Buckner et al. 2015), and similarly, Saimiri sciureus (Lynch Alfaro et al. 2015), which further demonstrates the role of rivers in speciation in Saimiri (Mercês et al. 2015). Some of these taxa are temporally congruent with the hypothesized timing of cladogenesis in Sylvilagus and its diversification in South America: basal ancestors of South American clades at ca. 4.8–5.1 Ma (Ruedas et al. 2017). For example, the most recent common ancestor of all Ateles exclusive of A. marginatus, 4.5 Ma, 95% highest posterior density: 2.7–6.6 Ma (Morales-Jimenez et al. 2015), or the basal radiation of the Saguinus midas and S. oedipus groups (Buckner et al. 2015; summary of New World monkeys [Platyrrhini] data in Aristide et al. 2015). The confluence in timing of cladogenesis across diverse taxonomic groups of mammals and across geography in the data sets described above underscores the uniformity in processes operating on speciation in, and the importance of, the Guianas region of northern South America.

Patterson (2001), in examining trends in biodiversity of Neotropical mammals, noted 2 phenomena with respect to biological species discovery: 3 times as many species were being discovered unappreciated in museum collections than from the wild, and 3 times as many names were coming out of synonymy (since 1982) as were becoming synonyms. The recent history of taxonomy in Sylvilagus appears to bear out those predictions. Until 2016, only 3 species were known for the entire South American continent (S. brasiliensis, S. floridanus, and S. varynaensis), as compared to 15 in North America. However, the number for South America has since grown to 7 by excision of S. andinus and S. tapetillus (Ruedas et al. 2017), and S. sanctaemartae (present manuscript) from S. brasiliensis, and the de novo description of S. parentum. The uncertain application of old names, frequent absence of types for older material, and use of names without prior examination of type materials has been a persistent obstacle to clarity in the taxonomy of Neotropical mammas (Voss and Angermann 1997; Leite et al. 2011): the recent designation of a neotype for S. brasiliensis is predicted to lead to a stable taxonomy and more rational designation of species and assessment of species limits in Neotropical Sylvilagus. Patterson (2001) further noted that about 25 new species of mammal were being annually described at the time; Pimm et al. (2010) refined that estimate for Brazil, suggesting that no more than 38 species of mammals remained to be discovered from Brazil. The ongoing work on Sylvilagus suggests that a considerable proportion of the species predicted on an annual basis by Patterson (2001) will come from this genus, and that the estimate of Pimm et al. (2010) is a vast undervaluation of the true biological reality of the region.

Acknowledgments

Portions of this work were carried out under the auspices of NSF grant DEB-0616305. I am grateful to S. van der Mije and P. Kamminga, Rijksmuseum van Natuurlijke Historie—Naturalis Biodiversity Center, for allowing me access to the specimens under their care, and their gracious help during my visit to their institution in 2015. I thank A. L. Gardner and D. P. Lunde, US National Museum of Natural History—Division of Mammals, for permission to examine holotypes under their care, their invaluable assistance during my visit to the USNM in 2016, and most particularly accommodating my unorthodox work schedule. As usual, discussions with A. L. Gardner proved insightful and fruitful. I also thank E. Hoeger of the American Museum of Natural History for her help during my visit to that institution in 2016, and access to the holotypes, and particularly R. S. Voss, for fascinating discussions on all aspects of the biology and nature of Neotropical mammals, and for letting me use his fine microscope; discussions and microscope both proved inestimable. I very much enjoyed discussing cranial foramina with J. H. Wahlert; some of what I may purportedly have written above on the matter likely was lifted from his engrossing e-mails on the subject. J. M. Mora commented on a previous version, thereby improving it. Comments from A. Feijó on a late version of this manuscript (and South American mammals in general) were gratefully received and immensely improved its quality. In addition, A. Feijó examined and photographed the materials from the Field Museum of Natural History detailed herein, which proved invaluable to the study. His integrity and friendship are very much appreciated.

Appendix I

Specimens examined.—Acronyms for institutions housing specimens used in the analyses presented above are as follows: FMNH, Field Museum of Natural History, Chicago; RMNH, Naturalis Biodiversity Centre, Rijksmuseum van Natuurlijke Historie, Leiden, The Netherlands; UFPE, Laboratório de Mastozoologia, Departamento de Zoologia, Universidade Federal de Pernambuco, Recife; USNM, United States National Museum.

Sylvilagus sanctaemartae.—Colombia: Magdalena; Southern Slope Sierra de Santa Marta, Colonía Agrícola de Caracolicito, Río Ariguaní, 335 m (estimated geographic coordinates, Datum WGS84: 10°46′2.25″N, 74°2′58.83″W): USNM 279993 ♀ (holotype). Magdalena; Southern Slope Sierra de Santa Marta, Colonía Agrícola de Caracolicito, 400 m, USNM 279975 ♂, 279989 ♂, 279990 ♀, 279991 ♂, 279992 ♂, 279994 ♀ (all paratypes). Magdalena; Distrito Valledupar, Río Guaimaral, 140 m, USNM 279976 ♀, 279979 ♂, 279980 ♂, 279981 ♀, 279982 ♀, 279983 ♀. Localities mapped in Hershkovitz (1950: figure 43, as localities 1 and 2). Bolívar; San Juan Nepomuceno (estimated geographic coordinates: environs of 9°57′4″N, 74°4′46″W, ca. 172 m): FMNH 68937 ♀. Córdoba; Catival, upper Rio San Jorge (possibly ca. 8°17′2″N, 75°41′0″W, ca. 60 m, based on Paynter 1997, who listed elevation of this site as 120 m): FMNH 68947 ♀, 68948 ♀, 68949 ♂, 68950 ♀, 68951 ♂, 68952 ♂, 68953 ♀. Cordoba; Socorré, upper Rio Sinú (estimated geographic coordinates: environs of 7°51′3″N, 76°16′56″W, ca. 155 m; also fidePaynter 1997, who, however, listed the elevation as 110 m): FMNH 68938 ♀, 68939 ♀, 68940 ♂, 68941 ♂, 68942 ♂, 68943 ♂, 68944 ♀, 68945 ♀, 68946 ♀, 68954 ♀. FMNH specimens were examined, but not measured, for this study.

Sylvilagus parentum.—Suriname: Nickerie Dist.; Road to Amotopo, Km. 207: RMNH 30798 ♂. Nickerie Dist.; Road to Amotopo, Km. 210: RMNH 30799 ♂. Nickerie Dist.; Road to Amotopo, Km. 215.5: RMNH 30800 ♀. Nickerie Dist.; Road to Amotopo, Km. 205: RMNH 30801 ♀. Nickerie Dist.; Road to Amotopo, Km. 110: RMNH 31149 ♀. Nickerie Dist.; Road to Amotopo, Km. 218: RMNH 31153 ♀. Nickerie Dist.; Road to Amotopo, Km. 219: RMNH 31154 ♂, 31909 ♂. Localities mapped in Hoogmoed (1983: figure 2).

Sylvilagus brasiliensis.—Brazil: Pernambuco; Municipality of Paudalho, Mata da Privativa, forest fragment CIMNC (Centro de Instrução Marechal Newton Cavalcanti), 7°50′38.4″S, 35°6′7.3″W (datum: WGS84), elevation: ca. 137 m: UFPE 1740 ♂ (neotype). Locality described in detail and mapped in Ruedas et al. (2017).

Author notes

Associate Editor was Patrick Zollner.

Version of Record, first published online May 17, 2017, with fixed content and layout in compliance with Art. 8.1.3.2 ICZN.