(Figs. 1 B, 2 B–C, E–F, 4 E–J, 5 B–C, 6 C–D, 7 C–D, 8; Table 1)
Scolopendra alternans — Leach, 1815: 408 [Type locality: unknown, see discussion below].
Type locality. Leach (1813) stated that the habitat was unknown for S. iaequalis and in 1815 and 1817 Leach had no habitat listed for S. alternans. The original type specimen and the locality for S. alternans remains unknown (see Shelley, 2002). Unable to find a type specimen for S. alternans, Shelley (2002) designated a neotype for the species from the British Virgin Islands because Leach was British, and he felt the type may have come from a British territory in the Caribbean. NEOTYPE: British Virgin Islands, Tortola, Fat Hog’s Bay, 12 March 1984 collected by A. Penn. The neotype specimen has not been examined for this study and Shelley did not provide any illustrations or morphological details. The suitability of this neotype specimen should be reevaluated in the future in light of the hypothesis that S. alternans records represent a species group rather than a single species.
Material examined. Dominican Republic: East of La Romana, north of large sugar plantation, under rocks along road in humid forest, N 18.29. 443 W 68.55065, ele: 208 ft., coll: J. Huff, July 2004 (1, AMNH); Dominican Republic: Azna, Sto. Domingo, 3-13 - 13, P.G. Russel (1, USNM); Santo Domingo, Colegio De La Salle, Bro. Basilio Augusto (1, USNM); Boca del Inferno, Samana & B., Feb. 29, 1928 (1, USNM); Haiti, Lake Assuei, Mar. 11, 1918 (1, USNM); Etang Saumâtre, April 6, 1920, Dr. W.L. Abbott (1, USNM); Planisance, Nov.– Mar. 1925, caught eating large snail, Coll. E. C. Leonard (1, USNM); Trou Caiman, Feb. 18, 1943, A. Curtiss (5, USNM); Haiti (4, RJM); Puerto Rico: Mayaqüez, June 21, 1948, J.A. Rivero (2, FMNH); San Juan, Nov. 3, 1899, on battlement under old bricks, Coll. Cook, Collins, Gall? (2, USNM); San Juan, 1899, August Busck (1, USNM); Lares, January 25, 1899, August Busck (2, USNM); Cayey, June 1902, D.M.C.? (1, USNM); Puerto Rico (2, RJM); N. Antilles: Saba Island, Windward side, Elev. 400 m, Jan. 13 –15, 1968, B. Malkin (20, FMNH).
Remarks. In 1815 and 1817, Leach listed S. alternans as a species with the general characters of “Corporis segmenta transversa alternantia, quinto et sexton subaequalibus,” which translates to body segments transversely alternating, 5 th and 6 th subequal. The maximum length of S. alternans is on the order of 150–190 mm (Shelley 2002). The maximum adult length and weight observed in this study for live S. alternans from Haiti ~ 150 mm long and weighed ~ 10.7 g.
The coxosternal tooth-plates exhibit some fairly consistent variation from region to region (Figure 4 E–J). There are differing degrees of medial teeth fusing, where the invagination takes place between teeth (sometimes differing in the same specimen on either side of the tooth-plate), and the extent which the medial teeth protrude beyond that of the outer teeth. The leading edges of the teeth vary from blunt, crushing, molar-like as that in the Haitian, Trou Caiman specimen (Figure 4 F), while those with piercing and sharp knife-like edges are seen in the Puerto Rican, Mayaqüez specimen (Figure 4 I). Tooth-plate structure is perhaps indicative of the preference for a particular prey item, where the need for crushing, piercing or slicing is more appropriate for puncturing or opening their prey to gain access to the more easily imbibed liquid contents. Overall, the anterior borders of tooth-plates show some subtle but noticeable variability throughout the Caribbean region. For example, the anterior tooth-plate border of the S. alternans specimen from Mayaqüez, Puerto Rico is more curved with the fused, medial teeth protruding well beyond the lateral teeth (Figure 4 I), when compared with that of the S. cubensis specimens from Matanzas, Cuba, whose anterior tooth-plate border is only slightly curved (Figure 4 C). When viewed ventrally, there is some noticeable intra-island variation of the tooth-plates within close proximity to one another. A good example of this is observed in the tooth-plate of the S. alternans specimen from Planisance, Haiti, when compared with that of the S. alternans specimen from Trou Caiman, Haiti (Figures 4 E, F, respectively). Notice the lateral tooth-plate edges of the specimen from Planisance consistently slope laterad moving posteriorly, where as that of the Trou Caiman’s has a region near its base that is incurved. Preliminary data for coxosternal tooth-plates hold promise for morphological characters such as the ratio of the maximum length of median embayment to leading edge of teeth to the distance between coxosternal condyles; as well as the overall structure, sclerotization, and fusing of coxosternal teeth but large specimen series are needed to do a statistical analysis.
In S. alternans from Puerto Rico the posterior edge of the ultimate tergite has heavier sclerotization (Figure 5 C) than that for S. longipes or S. alternans from Haiti. Sternites 2–20 in the Haitian S. alternans have complete, rather pronounced, paramedian sutures. Sternites 2–10 in the Puerto Rican S. alternans have incomplete paramedian sutures, 11–20 complete with the intensity of the sutures slightly increasing posteriorly. As it is in S. longipes, the posterior edge of the ultimate sternite in S. alternans from Haiti is gently rounded to rounded (Figure 6 C). The ultimate sternite of S. alternans from Puerto Rico has a very gently rounded to straight posterior edge, and also exhibits a dark colored border (Figure 6 D), which does not appear in S. longipes or in the specimens from Haiti.
The boxplot in figure 8 shows the median length to width ratio of the ultimate prefemur for S. alternans from Hispaniola at 3.33, S. alternans from Puerto Rico at 3.28 and S. alternans from Saba Island at 3.63. Two small specimens from San Juan, Puerto Rico have ultimate prefemora whose length to width ratios are close to the low range of S. longipes but they have no dorsomedial spines on the penultimate prefemora. The maximum width of the ultimate prefemur occurs distally in S. longipes and the Puerto Rican S. alternans, but in the Haitian specimens it was located medially where it is crassate and shows more pronounced spination (Figures 2 A–F).
It should be noted that a series of specimens from Trou Caiman, Haiti always exhibited at least 1 dorsal spine on the left and right, with one specimen having 2 on one left penultimate prefemur but the length to width ratio was always less than that of S. longipes, was overall stockier looking and had similar coloration like that of the live Haitian specimen in figure 1 B. The Etang Saumâtre specimen exhibited a mottled pattern on the head and tergites, which was unique.
Lewis (1989) studied 9 specimens of what he identified as S. alternans from St. John (U.S. Virgin Islands). Three things from his study provide further evidence for inter-island variation: 1) all 9 specimens ranged from 20–69 mm in length, which is a rather small body length range considering these specimens were collected in two different years in the months of February, March, May and October; 2) the color description ranging from olive to dark brown is dissimilar to the rusty-brown color of S. alternans 3) in his figure 2, the paramedian sutures are incomplete on the cephalic plate and absent from tergite 1. Although this suggests that the animal from St. John may be distinct from S. alternans, it is possible that these were all juveniles, and if so, this would further indicate that there is a need for detailed studies of character variation in different growth stages. Furthermore, Lewis (1989) observed variations in the margination of the tergites from the St. John S. alternans specimens and correlated it to body length. Perhaps there is also some variation in tergite widths and patterns with age and gender.
Shelley (2002) synonymized Scolopendra hirsutipes Bollman, 1893 with S. alternans by process of elimination, based on the lack of an anterior transverse suture on tergite 1 and spur counts of the ultimate legs. According to Shelley, the holotype of S. hirsutipes cannot be found at the USNM, but the description alone, with a lack of cephalic plate sulci and 25–27 antennal segments suggests it is not S. alternans. Bollman (1893) did cite the habitat of S. hirsutipes as “West Indian fauna”; however, because he thought all other unlabeled material in this particular collection seemed to be from Surinam that the S. hirsutipes specimen may have also been from Surinam.
The original description of Koch’s (1847) Scolopendra crudelis from St. Barthelemy was based on two specimens he said were very different in color from each other. One was said to be pale yellow ochre and the other a rusty-red with yellow legs. Due to this fact, there is a good chance this species was described from two different species and should not be considered a synonym of S. alternans until the types and/or fresh material can be studied. He stated that the ultimate prefemur had 24–26 spines in 7 uninterrupted rows on the ventral and medial surfaces. This is slightly less than the 28 observed on a few specimens in this study. Although Meinert (1886) listed Florida as the locality for a specimen of S. crudelis he described from Double Headed Shot Key, this island is now considered part of the Bahamas and is about 60 nautical miles southeast of the Florida Keys. His description of S. crudelis is much closer to matching the population of S. alternans from Haiti than it is to S. longipes. Perhaps it is conspecific with a population from Cuba, but because the medial part of the ultimate prefemur is described as rounded and the total body length as 150 mm, it is unlikely to be S. longipes.
There is certainly no overlap of the length to width ratio of the ultimate prefemur with S. longipes and S. cubensis, or the S. alternans of Saba Island with S. cubensis, but the populations of Hispaniola and Puerto Rico certainly need further resolution. It is obvious that general morphological characters of S. alternans do not make it easy to differentiate the species present in the Caribbean region. Nonetheless, minute details of some of the morphological evidence from this preliminary investigation of the S. alternans -complex point to a highly diverse fauna throughout the Caribbean region. Although further study of the S. alternans species-group is beyond the scope of this paper, an overview of factors presented below convey the possible origins of the Caribbean centipede fauna and why Scolopendra alternans is a species-group rather than a single species.
Origin and distribution. Pereira et al. (1997) mentioned that “…the most puzzling element of the Neotropical fauna of Geophilomorpha is constituted by the large non-endemic genera whose distribution cannot be explained within the usual framework of old Gonwanian elements or recent Northern immigrants.” Foddai et al. (2004) reiterated that our knowledge of the Neotropical Geophilomorpha is limited. The entire centipede diversity found in the Caribbean region remains relatively understudied. The fauna’s origin is enigmatic because it is not known how the islands were colonized nor is it clear how the S. alternans species-group radiated over such an enormous geographical area. We do know that most Old World Scolopendra lack an anterior transverse suture on their first tergite. Therefore, the presence of at least 3 Scolopendra species, S. longipes, S. cubensis and S. alternans, in the Neotropics without a transverse suture suggests this closely related group evolved from a Gondwanan relict. The existence of such relicts is supported by the work of Moran and Smith (2001) on phytogeographic relationships between Neotropical and African-Madagascan pteridophytes. In addition, the presence of the predominantly African genus Ballophilus Cook (1896), with two species in South America and one in Puerto Rico, is also an indicator of a probable Gondwanan component.
By simply analyzing known distributions of some geophilid genera (Ityphilus Cook, 1899; Polycricus Saussure and Humbert, 1872; Telocricus Chamberlin, 1915 and Titanophilus Chamberlin, 1915), the scolopendrid genus Newportia Gervais, 1847 and two scutigerid genera (Dendrothereua Verhoeff, 1944 and Sphendononema Verhoeff, 1904) in the southern Nearctic and Neotropical region, we observe that the Greater and Lesser Antillean chilopod fauna consists of Central and South American components. The expansive distribution and diversity of these various chilopod representatives in southern Florida, Central and South America, as well as throughout the Greater and Lesser Antilles suggests that they most likely colonized these areas through a land-bridge.
Whether the Greater and/or Lesser Antilles were ever connected to North, Central and/or South America, and how, has been and remains very controversial (e.g. MacFadden 1980, Iturralde-Vinent and MacPhee 1999, Graham 2003, Hedges 2006, Ali 2012, Alonso et al. 2012). Nevertheless, there are three main theories that could explain how the terrestrial fauna of the Greater and Lesser Antilles arrived:
1) over-water dispersal and rafting have been suggested methods of island colonization for centipedes in the Caribbean region (Shelley, 2002; Shelley & Sikes, 2012);
2) the Greater Antilles + Aves Ridge land-bridge hypothesis dubbed GAARlandia by Iturralde-Vinent and MacPhee (1999) where the Aves Ridge connected northern South America with the Greater Antilles somewhere between 35 – 33 Ma;
3) the tectonic reconstruction model of the Caribbean region by Pindell and coauthors (e.g. Pindell and Barrett 1990, Pindell 1994, Pindell and Kennan 2009, Pindell et al. 2011), which suggests that the proto-Antillean arc was connected from Mexico to South America starting ~ 130 Ma through ~ 59 Ma.
In regards to over-water dispersal and rafting, there is no reason to believe that this can’t happen; however, it most likely consisted of limited events. For example, Heatwole and Levins (1972) found a dead centipede on a piece of flotsam within 16 km of Puerto Rico, but all of the flotsam they found about 120 km from the nearest land lacked terrestrial animals. Iturralde-Vinent and MacPhee (1999) stated that: “…surface-current dispersal of propagules is inadequate as an explanation of observed distribution patterns of terrestrial faunas in the Greater Antilles.” Furthermore, Fritsch and McDowell (2003) also concluded that more than one biogeographical scenario is required to account for the current distribution and biology of the Antillean flora.
The main difference between the two remaining theories is when the land emerged and became colonizable. Although both of these scenarios may have occurred and are possible explanations, the latter tectonic model provides the most credible explanation for the initial origin of the centipede fauna seen in the Greater/Lesser Antilles and southern Florida for the following three reasons:
1) Near the end of the existence of the inter-American land bridge created by the proto-Antillean arc as suggested by Pindell et al., 2011, around 65.5 Ma, the Chicxulub asteroid impacted the Yucatan Peninsula and caused a world-wide mass extinction (Schulte et al. 2010). Although Graham (2003) suggested that the initial period during which the Greater Antilles became available for colonization by terrestrial flora and fauna was ~ 49 Ma in the Middle Eocene, this time-period was after the Antillean arc had separated from the continents based on Pindell’s model. It seems reasonable that biota would have radiated sooner into the proto-Antillean arc, immediately after the Chicxulub asteroid impact, while it was still connected to Central and South America.
2) The diverse presence of the predominant Central American Polycricus (Geophilomorpha: Geophilidae) throughout the Greater and Lesser Antilles suggests that these species or their ancestors were most likely in existence on the Antillean arc before it separated from Central America, and that at least some parts of the Greater Antilles remained emergent during glacial minima for their persistence.
3) According to Pindell (1994) there was a Bahamian-Antillean collision termination in the middle Eocene, which could explain the presence of S. longipes in southern Florida and the Bahamas.
Since the effective colonization of the Caribbean region by centipedes, ongoing cyclical selective pressures have been created through glacial maxima and minima affecting sea levels. Fleming et al. (1998) postulated that somewhere between 16,000 and 7,000 years ago the ocean levels rose from about 100 meters less than they are today to within 3–5 meters of today’s. Yokoyama et al. (2000) concluded that the Last Glacial Maximum (LGM) was between 22,000 and 19,000 years before present. According to Abe-Ouchi et al. (2013) these glacial cycles are primarily driven by insolation cycles that last ~ 100,000 years, with a saw-toothed pattern of gradual growth and fast termination. Their data indicates that the global sea-levels may have dropped by as much as 120 m during the LGM. Even with the current sea floor depths in the Caribbean region, if the ocean level were dropped by 120 m it would seemingly not create any land-bridges connecting North, Central or South America to the Greater Antilles. Although glacial maxima and minima don’t seem to explain the biogeography of centipedes in the Caribbean or even how S. longipes arrived in southern Florida, it may have assisted the S. alternans species-group to radiate via land-bridges throughout the Greater and Lesser Antilles during times of glacial maxima.