Iheringia Taxonomic groups with lower movement capacity may present higher beta diversity

Diversity analysis by partition is an approach employed in order to understand how communities spatially structure themselves and the factors that operate in the generation and maintenance of distribution patterns. We examined the spatial structure of species diversity of four taxonomic groups, with diff erent dispersal abilities, in 16 forest fragments in the southern region of the state of Minas Gerais, Brazil. Specifi cally, we tested: i) if the similarity in species composition would be negatively related to geographical distance between the 16 fragments; and ii) if the beta diversity of the diff erent groups could be negatively related to their dispersal abilities. Alpha diversity and the compositional similarity between localities were both low. Beta diversity was not correlated with distance for any of the groups. Primates, followed by birds, showed a higher tendency of forming similarity groupings, although in a manner that was independent from distance between fragments, as well as showed the lowest beta diversity relative values. Spermatophytes and amphibians did not defi ne groupings and presented the highest values of beta diversity. We interpreted such results as indications that the groups with higher dispersal ability (primates and birds) tend to reach, on average, farther localities and, therefore, to defi ne more similar groupings (low beta diversity). The groups with lower dispersal ability (spermatophytes and amphibians) showed the opposite tendency. Although most of the species were restricted to few localities, contributing to the low similarity, beta and gamma diversity values showed the extent which the localities are, respectively, diff erent and complementary to each other in terms of species composition. Such features reinforce and justify future conservation initiatives, both in local and regional levels.

Beta diversity thus relates the alpha to the gamma diversity (Ricklefs, 1987;Anderson et al., 2010), indicating how many species are shared between the habitats/units and, therefore, the degree of biotic heterogeneity of a region (Wilson & Shmida, 1984).Beta diversity may be positively associated with environmental heterogeneity (Soininen et al., 2007a) or be independent from it, varying only with space (Hubbel, 2001).
Several factors may influence spatial variation in beta diversity, such as geographical, environmental, historical and evolutionary processes (Soininen et al., 2007a).Those processes normally produce an inverse relationship between similarity in species composition and geographical distance among sampling units (Nekola & White, 1999;Hubbel, 2001).Such relation is partially due to the spatial autocorrelation pattern of environmental variables, with closer locations tending to be more environmentally similar to each other (Legendre, 1993), which, by itself, influences the composition of species communities (Harrison et al., 1992;Jiménez-Valverde et al., 2010).Therefore, a decrease in the similarity of environmental conditions along space may result in a correspondent decrease of the similarity in species composition (Steinitz et al., 2006).
Among the biotic factors that may influence the pattern of beta diversity there are the life history of the species (Soininen et al., 2007a) and the organisms' dispersal ability (Dobrovolski et al., 2011).Also, taxonomic groups of organisms may be broadly categorized in a gradient according to their dispersal abilities.Thus, the decay of similarity with the increase of geographical distance would be relatively higher for taxonomic groups that show lower dispersal ability (Soininen et al., 2007b;Qian 2009a;Dobrovolski et al., 2011).Spermatophytes usually have passive dispersal and depend on seed dispersal agents (e.g.Tabarelli & Peres, 2002;Almeida-Neto et al., 2008).Among terrestrial vertebrates, birds are the most vagile, followed by mammals, amphibians being the less mobile ones (Böhning-Gaese et al., 1998;Qian, 2009b;Dobrovolski et al., 2011;Qian & Ricklefs, 2012).
In disturbed landscapes, one of the factors that may influence changes in beta diversity in relation to conserved habitats also relates to different dispersal abilities.Since fragmentation generates barriers for movement, it may limit even more the dispersal ability of organisms and, thus, favor differentiation in composition among fragments (Arroyo-Rodríguez et al., 2013).Following the pattern expected for undisturbed landscapes, groups with high dispersal ability would thus have bigger chances of (re)colonizing neighbor habitats, keeping viable populations and reducing, that way, the compositional differences between fragments, the opposite occurring in groups with lower movement capacities (Soininen et al., 2007a).
In this study we analyzed the beta diversity variation of four groups with different dispersal abilities (spermatophytes, amphibians, birds and primates) in a region of the Atlantic Forest hotspot (Myers et al., 2000), a highly fragmented biome.Specifically, our aims were: 1-to evaluate the relative contributions of alpha and beta diversities to the gamma diversity; 2-to test if there is higher similarity between closer localities than between farther ones due to spatial autocorrelation (Legendre, 1993;Jiménez-Valverde et al., 2010), especially regarding groups with relatively higher dispersal ability (birds and primates) (Buckley & Jetz, 2008;Qian & Ricklefs, 2012); and 3-to test whether beta diversity is higher in groups with lower dispersal capacity.A better understanding of these issues will allow taking more scientifically sound decisions for the management and conservation of these fragments, especially in a biome as threatened as the Atlantic Rainforest (Pinto et al., 2006).For example, understanding the contribution of the complementarity (beta diversity) to the gamma diversity within a landscape is important to the SLOSS debate and to where we should aim our conservation efforts (Margules & Pressey, 2000).In addition, understanding the influence of the dispersal capacity on the beta diversity may help us to take management decisions fine-tuned to specific issues of the different groups.

MATERIAL AND METHODS
Study sites.We conducted rapid surveys of spermatophytes, birds, amphibians and primates in 16 forest fragments in the southern region of the state of Minas Gerais, Brazil (Fig. 1, Tab.I), covering an area of about 65,000 km 2 , during the summers of 2010 and 2011.Most of the fragments are located in areas listed as priorities for conservation (sensu Drummond et al., 2005, see green areas on the map).We employed a rapid survey approach, sampling each fragment once during two consecutive days (e.g.Herzog et al., 2002;Young et al., 2003;Penter et al., 2008).
Sampling.We recorded the occurrence of species using complementary methods, which increases the chances of sampling a greater number of species in a short period of time (Silveira et al., 2010).
For spermatophytes, we used the quadrant point's method (Cottam & Curtis, 1956), sampling 20 points on each fragment.The points were placed roughly 20 m apart from each other along a 400m transect in the central region of the fragment.In each point, we recorded the closest four individuals with diameter at breast height (DBH) ≥ 3 cm.We identified the plants through comparisons with herbarium specimens and consultations with specialists and the specialized literature.We deposited the exsiccates in the herbarium of the Universidade Federal de Alfenas (UALF).Species nomenclature followed APG III (Angiosperm Phylogeny Group, 2009).
For amphibians, we employed visual and audio surveys (Crump & Scott Jr., 1994) during a fixed time period (between 19h00 min and 00h00 min).The search for individuals was directed to breeding sites, especially water bodies (Scott Jr. & Woodward, 1994), as well as leaf litter and vegetation along trails and transects made to access the breeding sites.The sampling effort was 20 hours-  For birds, we employed the capture of understory species with mist nets (12 m x 2.5 m x 31 mm mesh) (Develey, 2003).We installed ten nets in a row along a 150 m transect inside the forest, at least 50 m distant from the edge.The nets remained opened between 07h00 min and 17h00 min, totaling 200 net-hours effort per location.Each captured individual was identified and subsequently released (IBAMA license #22020-1).
For the primate surveys, we employed a couple of different approaches.To attempt detecting the buffy-tufted-ear marmosets, Callithrix aurita (E.Geoffroy in Humboldt, 1812), the black-fronted titi monkeys, Callicebus nigrifrons (Spix, 1823), and the brown tufted capuchin monkeys, Sapajus nigritus Goldfuss, 1809, we used playbacks.We selected long-range, high amplitude calls, which are proposed to function either in inter-group communication (marmoset long calls and titi monkey duets) or to maintain contact between group members (capuchin monkey whistles).We reproduced the same calls throughout the study, both within the forest and/or at the borders.Inside the fragments we played the calls up to four times at different directions in each sampling point, so as to attempt covering a 360° radius around it.In contrast, we reproduced calls in two directions at the forest borders (each 45° from the edge line).Playbacks for a given species were ended as soon as a response was obtained.Some occurrences were also recorded through direct visual and auditory contact (Rosales-Meda, 2007).In addition, we interviewed landowners and residents near the fragments about the occurrence of species (Waters & Ulloa, 2007).This procedure was necessary to verify the occurrence of Alouatta guariba clamitans Cabrera, 1940, which usually do not respond well to playbacks.Also, marmosets and capuchins are less responsive to playbacks than titi monkeys, and are more likely not to respond, even if present.If the respondent mentioned the occurrence of marmosets, we only included these data if we could locate and identify the species, given the occurrence of an exotic species of the genus, Callithrix penicillata (E.Geoffroy, 1812), in the region.The fragment was excluded from further analysis if we could not correctly identify the species.In Pouso Alegre and Passa Quatro localities, we also obtained information regarding the presence of primates through management plans of the protected areas we surveyed.Given the lower richness of this group, we did not restrict ourselves to the two-day sampling scheme.When we had indications that a species might occur in a locality (e.g. through interviews), but were unsure about it (for example of which marmoset species it were, or due to conflicting or apparently inaccurate reports), we returned to the fragments in other occasions in a further attempt to confirm the findings.Whenever in doubt, we did not consider a species as occurring in a fragment.
Data analysis.Herein we considered gamma diversity (γ) as the total number of registered species in the 16 fragments sampled (regional richness, but see Tuomisto, 2010).We considered alpha diversity (α) to be the number of species in each fragment (local richness).Last, we defined beta diversity (β) as the non-directional variation on species composition between the fragments (sensu Anderson et al., 2010).We calculated beta diversity in two ways, using three different measures (β W ; β add ; β C ): (1) Variation on the number of species among the localities: here we used both multiplicative beta diversity (β W ) (Whittaker, 1960) and additive beta diversity (β add ) (Lande, 1996;Crist & Veech, 2006) measures.The multiplicative beta diversity is given by the formula [β W = γ / α mean ], where γ is the total number of species for the region and α mean is the average number of species of the 16 fragments.β W thus indicates "the number of times by which the richness in a region is greater than the average richness in the smallerscale units" (Anderson et al., 2010).On the other hand, the additive beta diversity is given by the formula [β add = γ -α mean ] and it informs the average number of species that are not shared among all the sampling units (Anderson et al., 2010).
(2) Variation on the species composition between localities: beta diversity as a measure of complementarity (β C ).Through paired comparison of species between localities, the proportion of species that occur in only one of those localities is evaluated in relation to the total number species of both localities (Colwell & Coddington, 1994).The beta diversity values, in this case, are represented by the inverse of the similarity indexes of species' values (see next paragraph), that is, [β C = 1 -Cj], where Cj is the similarity index value (Krebs, 1999).Thereby, pairs of locations with low similarity in species composition show high beta diversity, and vice versa.Complementarity values vary from zero (identical species composition between two localities) to 1 (completely different species composition between two localities) (Colwell & Coddington, 1994).We considered the values of average complementarity as significant if β C was ≥ 0.5 (50%) (Vasconcelos et al., 2011).
Similarity in species composition was quantified both for each taxonomic group and for the four groups combined through grouping or cluster analysis (UPGMA) and computation of the Jaccard's similarity index (Cj) (Magurran, 1988), which determines the proportion of species shared between each pair of localities.We considered as a grouping every pair or group of localities showing Cj ≥ 0.5.To verify if there was any correlation between species composition similarity and geodesic geographical distance between the localities, we applied the Mantel test (Legendre & Legendre, 2012), whose r values may vary from -1 (strong negative correlation) to +1 (strong positive correlation), zero meaning absence of correlation.The tests were carried out in the program R, version 3.0.1.(R Development Core Team, 2013).
Last, to classify the species according to the frequency of occurrence in the 16 localities, we used the following categories (adapted from Dajoz, 1983): "frequent" (species with registered presence in at least nine locations); "common" (occurrence in five to eight locations); and "rare" (occurrence in four locations or less).Both the exotic species and the ones recorded by chance (i.e.outside the standardized sampling methods) were not computed on the data analysis.However, we included them in the general relation of species described in the supplementary material (Appendixes 1 to 4).

RESULTS
We found 259 species of spermatophytes (Appendix 1), 45 of amphibians (Appendix 2), 66 of birds (Appendix 3) and four of primates (Appendix 4).However, local richness was usually much smaller than that (Tab.II).We also registered one exotic anuran species Lithobates catesbeianus (Shaw, 1802) in Delfim Moreira and one introduced primate species Callithrix penicillata in Guaxupé, Caxambu, Aiuruoca, Passa Quatro and Bocaina de Minas.The list of all species we recorded, as well as their distributions along the 16 fragments and respective frequency of occurrence (FO) can be found in the supplementary material (Appendixes 1 to 4).

Locality
As for the frequency of occurrence within the set of fragments, most of the species (84.4%) were "rare" when we consider the four groups combined, occurring in only one, two, three or four localities.The same pattern was observed for spermatophytes (93.8%), amphibians (77.3%) and birds (57.6%), but not for primates (0%) (Fig. 2).
When we analyzed the species composition similarity, there was no grouping between localities regarding spermatophytes (Fig. 3).As for the amphibians (Fig. 4), there was only the MAR-DEL grouping, with both localities showing the same type of phytophysiognomy (ombrophilous forest), but located 117 km apart from each other (Fig. 1).In birds, four groupings of similarity were formed (Fig. 5).The localities from the groupings POU-MAR and MON-GUA showed the same phytophysiognomy (seasonal semideciduous forest) and are relatively close to each other (61 and 40 km, respectively).However, the localities of the groupings EXT-SRJ and MVE-DEL, though presenting the same phytophysiognomy of ombrophilous forest, are distant from each other 235 and 90 km, respectively.The primates grouping (Fig. 6) was the only one with high composition similarity, forming six locality groupings, with all of them showing the same phytophysiognomy between the fragments Fig. 2. Frequency of occurrence of spermatophytes (n=259), amphibians (n=44), birds (n=66) and primates (n=4) species in the 16 localities sampled in Minas Gerais, Brazil.Frequent (species with registered presence between nine and 16 locations); common (between five and eight localities); and rare (between one and four localities). of each grouping, independently from geographical distance.The only grouping common to birds and primates was MON-GUA.We did not find any grouping when we analyzed the four groups simultaneously (Fig. 7).Finally, we did not find any relation between species composition and geographical distance between localities for any of the groups through the Mantel test: spermatophytes (r = 0.03, p = 0.35), amphibians (r = -0.01,p = 0.5), birds (r = -0.006,p = 0.49), primates (r = 0.14, p = 0.072), and all the four groups together (r = 0.01, p = 0.45).

DISCUSSION
Beta diversity showed no correlation with geographic distance.Primates, followed by birds, presented a greater tendency to form location groups with species compositions more alike between themselves, though in a way that was independent from geographic distance, as well as showed the lower beta diversity values.On the other hand, spermatophytes and amphibians did not define such groupings and showed the highest values of beta diversity.We interpreted those    results as indications that the groups with higher dispersal ability (primates and birds) reached, in average, more distant locations and tend, therefore, to define locality groups with more similar compositions (i.e.low beta diversity).In less vagile groups (spermatophytes and amphibians), the low dispersal ability does not favor them in occupying nearest locations, even if the ecological conditions allow; such groups, thus, present the reverse tendency of not forming locality groups with similar compositions (i.e.high beta diversity).
Primates, moreover, were the only group in which there were no species with "rare" frequency of occurrence, probably due to the low regional richness of only four species.This may have increased the probability of generating locality groupings with similar compositions by chance alone.By excluding this group, because of its exceedingly low species number, it is possible to conclude that beta diversity, be it of spermatophytes, amphibians, birds or all these groups together, exerted a greater influence on regional diversity (gamma) than local species richness (alpha) (Pineda & Halffter, 2004).
The absence of correlation between the similarity in species composition and the geographical distance and groupings of localities with similar compositions, independent from geographical distance, in birds and primates, suggests that other factors might be necessary to explain the variation on the species composition of each group along the space.Accordingly, several studies have demonstrated a host of different factors that may influence variation of species composition through space: (1) spatial and environmental gradients (e.g.Clark et al., 1999;Nekola & White, 1999;Oliveira-Filho & Fontes, 2000;Carneiro & Valeriano, 2003;Steinitz et al., 2006Steinitz et al., , 2007b;;Legendre et al., 2009); (2) differences in the physiology, in the degree of biological interactions and in the dispersal ability of the species (Nekola & White, 1999;Tuomisto et al., 2003;Gilbert & Lechowicz, 2004); (3) barriers imposed by the configuration of the landscape and the influence of weather on species' dispersion (Nekola & White, 1999;Hubbel, 2001); (4) stochastic processes generated randomly and independently of environmental dissimilarities (Neutral Theory sensu Hubbel, 2001;Soininen et al., 2007b;Steinbauer et al., 2012); (5) species' tolerance to fragmentation (Arroyo-Rodríguez et al., 2013); (6) spatial scale (extension, resolution; Nekola & White, 1999;Steinitz et al., 2006;Soininen et al., 2007b;Arroyo-Rodríguez et al., 2013).Therefore, more accurate analysis involving other variables could yield further explanations regarding the spatial distribution observed in this study.
Although our results support the relationship between dispersal ability and beta diversity, there are some controversial results in the literature (McKnight et al., 2007).Soininen et al. (2007a), for example, when comparing different trophic levels, showed that autotrophs have smaller beta diversity than omnivores and carnivores.This particular result was very different from ours, since the spermatophytes from our study showed a higher average beta diversity value than all other groups, constituted by animals.A possible explanation for these differences can be found in the suggestions by the same authors (Soininen et al., 2007a): beta diversity is something very complex and influenced by extrinsic (e.g.landscape structure and environmental variations) and intrinsic factors (e.g.peculiar features of the organisms).
On the other hand, Qian (2009a) and Arroyo-Rodríguez et al. (2013), both working with plants, showed results that were similar to those we found here, that is, high negative correlation between the dispersal ability and the beta diversity.According to Arroyo-Rodríguez et al. (2013), increasing the distance between forest fragments leads to communities with very specific compositions (i.e.high beta diversity), given seed dispersal is limited.Such parallel may indicate a potential problem for the landscapes in southern Minas Gerais, given its high degree of fragmentation, which may lead in the future to local extinctions.
Amphibians showed a very similar result to the spermatophytes, only with slightly lower beta diversity values and much smaller gamma diversity (regional).The high beta diversity of the group can be explained by two factors.First we have amphibians strict environmental requirements (Duellman & Trueb, 1994;Werner et al., 2007;Buckley & Jetz, 2008), especially their need for both water (for reproduction), and higher temperatures, due to ectothermy (Buckley & Jetz, 2008;Qian, 2009b).Since the studied fragments vary in terms of water availability and altitude (thus temperature), the environmental requirements for amphibians vary accordingly.Besides, amphibians are usually considered as animals with low dispersal ability (Qian, 2009b;Dobrovolski et al., 2011;Qian & Ricklefs, 2012).When compared to mammals and birds, amphibians always show higher beta diversity (see Buckley & Jetz, 2008;Qian, 2009b;Dobrovolski et al., 2011;Qian & Ricklefs, 2012).The same was observed for reptiles, an equally ectothermic and little vagile group (see Qian, 2009b;Qian & Ricklefs, 2012).Amphibians may even show turnover values four times higher than birds (Buckley & Jetz, 2008).The opposite happens to the median size of the geographic distributions of these two groups.In general, birds have occurrence areas four times bigger than amphibians, a clear sign of the greater dispersal ability made possible by flight.The relationship between these patterns is so outstanding that the amphibians' turnover has been proved to be a better predictor for the birds' turnover than the environmental variables (Buckley & Jetz, 2008).
The high percentage of spermatophytes, amphibians and birds species with "rare" occurrence in this study is a sign of the low number of species shared between the local communities, especially on the first two groups.Some of these rarities actually represent less abundant species under some degree of threat (see International Union for Conservation of Nature, 2013), and the register of their occurrence is important for conservation purposes on the respective sampled municipalities.The registers of geographical distribution expansions are equally important, as, for example, in the case of the amphibian Ischnocnema holti (Cochran, 1948) in the locality of Extrema, which extended its range to about 160 km southwest of its typical locality (see Da Costa et al., 2008;Targino & Carvalhoe-Silva, 2008).
The exotic species we found deserve mention due to the negative impacts they cause.The anuran Lithobates catesbeianus, originally from North America, is commercially farmed as a food source.Due to negligence in their containment, they commonly end up escaping captivity and settling up populations in natural environments, where they may cause problems to the native communities (Both et al., 2011;Silva et al., 2011).The locality in which this species was collected constitutes a new register for the state of Minas Gerais.The primate Callithrix penicillata, a typical species of the Cerrado biome (Miranda & Faria, 2001), was deliberately introduced in Atlantic Forest environments.Worse, it is a species with high adaptability and dispersion power, causing several impacts and possibly even hybridizing with native species of this biome (Stevenson & Rylands, 1988;Auricchio, 1995).
A potential criticism to our results relates to the rapid survey sampling scheme we have adopted here.Such sampling scheme may be regarded insufficient (Lawton et al., 1998;Steinbauer et al., 2012), a problem compounded by the different levels of detectability of each species (Boulinier et al., 1998), given that the surveys were relatively fast.However, the overall conclusion would not be affected by a longer sampling on each location.Although alpha diversity values might increase on each fragment with higher sampling efforts, beta diversity would remain high.The reason is that longer sampling allows registering species that are rarer, which, however, naturally show a patchy distribution in the landscape, particularly in a fragmented one.Therefore, our sampling scheme was sufficient to gather the data necessary to provide the basis for our conclusions.
In summary, although alpha diversity within fragments was low, we found low similarity in species composition between localities (i.e.high beta diversity values), thus resulting in a high gamma diversity.Primates were the only group with average complementarity below 50%.In addition, they presented a greater tendency (followed by birds) to form location groups with species compositions more alike between themselves, though in a way that was independent from geographic distance.Thus, the decreasing gradient of beta diversity observed [spermatophytes (92%) > amphibians (83%) > birds (68%) > primates (48%)] coupled with the results of the groupings, indicate that the taxa with higher dispersal ability (primates and birds) may have reached, in average, more distant locations and tend, therefore, to define locality groups with more similar compositions (i.e.low beta diversity).