**4. Discussion**

Our results show that most New World pitvipers have a narrow habitat breadth or are locally common, and around half of them are widely distributed. Twenty-two species (17.3%) are in the not rare category and fifteen (11.8%) are in the highest rarity category (Rarest). Most of the latter species are

distributed in Central America, especially in highlands and islands, and seven of these rarest species are not classified as threatened by IUCN.

Pitvipers tend to be locally common and to have narrow habitat breadth (see Tables S2 and S3). As a consequence, two of the rarity types in the Rabinowitz cross-classification (cells E and G in Figure 1) together comprise less than 4% of species and one of the rarity types (cell C in Figure 1) was not occupied by any species. This result contrasts with those observed in other groups, like plants [74], mammals [14], and frogs [22], for which empty cells were not recorded, perhaps because the number of species considered herein is smaller than those of these studies. Further rarity studies involving other taxonomically restricted snake lineages (e.g., families, subfamilies) would show whether the patterns described herein are widespread amongs<sup>t</sup> snakes.

A high frequency of habitat specialization and high local abundance was also detected in birds and plants [17,27]. Although few studies have explored patterns of habitat specialization in snakes (e.g., [79,80]), Luiselli [81], using a different approach, found that rare species (including vipers) tend to have narrow realized ecological niches, which can be related to habitat breadth (see below). Additionally, in most studies of NW snake communities, pitvipers tend to appear among the most frequently encountered species (e.g., [82–87]), suggesting a general trend to be locally common. Overall, these results are in accordance with our findings that many NW pitvipers are habitat specialists and locally common.

We found an association between geographical range and habitat breadth in NW pitvipers (see also [88] for snakes in general). Indeed, there seems to be a general trend of association between these traits in different taxa. In a review of the effects of niche breadth on range size in several groups of organisms (diatoms, algae, plants, and animals), Slatyer et al. [89] found that habitat breadth is a good predictor of range size (see also [90,91]). Indeed, many recent studies indicate that populations of specialist species tend to decline faster than populations of generalist species [89], no matter why this association occurs (see discussion in [89], see also [92]). Thus, the high degree of habitat specialization associated with small ranges in NW pitvipers (e.g., 50 species with ranges < 20,000 km<sup>2</sup> have narrow habitat breadth; see also [88]) indicates that this group may be especially vulnerable to human disturbance in their habitats. The association between geographical range and habitat breadth also indicates that there is some redundancy within the rarity aspects included in the "seven forms of rarity" framework, and that it might be possible to use one of these aspects as a proxy for the other when data are incomplete. Additionally, when redundancy within the rarity aspects is high, a simplified framework could be used (e.g., with only two aspects).

The rarest species of NW pitvipers occur in Central America and Andean and Trans-Andean South America (see Figure 2A). Actually, many of the rare pitvipers are restricted to mountains or islands, which translates into small geographical ranges (Figure S2; Table S4) and narrow habitat breadth. In fact, Böhm et al. [88] showed that snakes with a narrow habitat breadth tend to have small geographical ranges, and that snakes have smaller ranges at higher altitudes. Additionally, when the geographical distribution of rare pitvipers (Rarest category; Figure 2A) is compared to pitviper richness hotspots throughout the Americas (Figure 2D), there is a relatively high congruence between them in North and Central America, but a generally lower congruence in South America. This lack of congruence in South America perhaps reflects the fact that most hotspots of pitviper richness in this region are dominated by widespread species, whereas in Central America these hotspots are determined mainly by species with small geographical ranges (see Table S4 and Figure S2).

As predicted, our results show that NW pitvipers occurring at lower latitudes tend to be rare. Indeed, there is evidence of Rapoport's rule among these snakes (correlation between latitudinal midpoint and GR, this latter log-transformed, Pearson's *r* = 0.245, *p* < 0.05, but see [57]), indicating that GR may have contributed to this result. There is also a significant correlation between latitude and number of habitats used by species (Spearman's *r* = 0.474, *p* < 0.05), indicating that HB has also contributed to this trend. Indeed, analyzing snake communities worldwide, Luiselli [81] found that the percentage of rare species in a community is higher in the tropics than in temperate regions. Similar

results were found for other groups, such as primates [93], Australian honeyeaters [69], and Australian mammals [68]. However, latitude is just an indicator of geographical location [94] and further studies should search for the factors that are actually explaining higher levels of rarity in the tropics. Letcher and Harvey [95], for instance, tested if climatic variability could explain Rapoports's rule for mammals of the Palearctic region and their results showed that only annual temperature range could predict range size. Thus, an interesting future approach would be to test whether environmental variables, such as temperature, rainfall, and resource distribution, can predict the variation of rarity in New World pitvipers and other groups along the latitudinal gradient.

Although rarity (sensu Rabinowitz [10]) is not directly incorporated in the Red List classification using IUCN categories and criteria [96], there is some congruence between rarity and extinction risk [14,22]. This congruence may occur mainly because certain aspects of rarity, such as small geographical range and small population size, are captured in the IUCN Red List Criteria as symptoms of high extinction risk; thus, extinction risk and rarity may covary [96]. Indeed, there is a high congruence between rare and threatened New World pitvipers (Figure 2A,C, respectively; Table S4). This high congruence is somewhat surprising considering that nearly half of our Rare species are not threatened (four LC, two DD, and one NT) in the IUCN red list. This congruence probably reflects the fact that the extent of occurrence of three quarters of our Rare species falls below the threshold of IUCN Red List criterion B (20,000 km2; Table S4, [47]). This latter criterion is the most used for reptiles in general [41] and was used in the assessment of almost all threatened NW pitvipers (21 out of 22 species; [42]; this study). Furthermore, although not explicitly included as a specific criterion in the IUCN categories and criteria, local abundance is often taken into account during conservation assessments [97]. Additionally, the high congruence in the geographical distribution of rare and threatened species indicates that rarity is a useful spatial surrogate of endangerment in NW pitvipers. In fact, using a global database of fossil marine animals, Harnik et al. [19] showed that geographical range was the main factor determining extinction risk in the seven forms of rarity, followed by habitat breadth, while local abundance had a minor e ffect.

Even though we found an overall congruence between rare and threatened NW pitvipers, four species (*Cerrophidion sasai*, *C. tzotzilorum*, *Crotalus stejnegeri*, and *C. transversus*) classified in the rarest category are categorized as "least concern" on the IUCN Red List ([42]). Thus, special attention should be given to these species, as they may become threatened very easily in the near future because they may be disproportionately a ffected by habitat disturbance [4,6,68]. By the same token, both DD species that appeared in the rarest category (*Bothrops ayerbei* and *B. venezuelensis*) may prove to be threatened or NT when additional information becomes available for them.

Besides the congruence with the IUCN Red List, four of our rarest species are listed as endangered or threatened in the Mexican red list of threatened species ([78]) and additional species appear in Mexican and Central American extinction risk assessments using EVS [75–77] and with a high threat index in Maritz et al. [43] prioritization. Based on the complementary results presented herein and in the IUCN [42] and Mexican [43] red lists and EVS assessments [76–78], species that appear as high priorities in these lists and studies (see Table 2) should receive special attention in future conservation planning. Additionally, one species in the rarest category (*Mixcoatlus browni*) had a high ranking in Maritz et al.'s [43] Ecological and Evolutionary Distinctiveness index (EED). Thus, this species may also be ecologically and evolutionarily unique and therefore deserves special attention for conservation.
