SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Tropical mountains are hotspots of biodiversity, but the factors that generate this high diversity remain poorly understood. To identify possible mechanisms that influence avian species assemblages in the tropical Andes, we studied the functional and phylogenetic diversity and the structure of species assemblages of an avian feeding guild. We analysed how functional and phylogenetic diversity, structure and composition of frugivorous bird assemblages changed along a 3300 m elevational transect from the lowlands to the tree line with a novel combination of functional and phylogenetic approaches, and used null models to infer possible drivers of the observed patterns. Species richness, functional richness and phylogenetic diversity decreased almost monotonically with increasing elevation, but assemblage structure and composition changed abruptly in the Andean foothills at around 1200 m. In the lowland assemblages, species were functionally and phylogenetically less similar than expected from null models, whereas species in the highland assemblages were functionally and phylogenetically more similar than expected by chance, suggesting an abrupt reduction in the number of functionally and phylogenetically distinct species in the transition from lowlands to the highlands. Nevertheless, the functional and phylogenetic evenness of the assemblages, i.e. the regularity of the spacing of species in functional trait space and phylogeny, remained constant along the gradient, which suggests that the mechanisms that influence the co-occurrence of species within the assemblages are similar in lowlands and highlands. The observed differences between lowland and highland assemblages imply sharp distributional limits for frugivorous bird species in the Andean foothills, probably caused by environmental factors other than climate, e.g. changes in habitat types or topography, or independent species radiations in lowlands and highlands. These strong distributional limits may hinder uphill range shifts of frugivorous bird species, and the plant species they disperse, in the tropical Andes as a response to climate change.

Understanding what causes the geographic variation in species richness is one of the fundamental questions in ecology. Along elevational gradients, climatic conditions and habitats often vary greatly over small spatial extents, which makes elevational gradients excellent systems to study species richness patterns (Körner 2000, Sanders and Rahbek 2012). On humid tropical mountains, species richness generally decreases monotonically with increasing elevation (McCain 2009). Among the most plausible explanations for the decrease of species richness at higher elevations is the increasing harshness of climatic conditions towards higher elevations, i.e. decreasing temperature and increasing temperature fluctuation (McCain 2007, 2009). Species at high elevations must be adapted to more unfavourable and a wider range of climatic conditions than species at low elevations, and there seems to be a trade-off between the adaptation to a wide range of environmental conditions on the one hand, and a high specialization and competitive ability on the other hand (Janzen 1967, Ghalambor et al. 2006, Jankowski et al. 2010). Species richness patterns along elevational gradients may be additionally influenced by the evolutionary history of species assemblages (Lomolino 2001), e.g. because of different rates of speciation and degrees of endemism at different elevations (Weir 2006, Fjeldså et al. 2012).

Studying the functional trait composition and phylogeny of species assemblages may help to understand the influence of different mechanisms that generate and maintain species richness patterns along elevational gradients. The investigation of functional diversity (FD), a measure of the diversity of manifestations of ecological traits in a community (Tilman 2001), and of phylogenetic diversity (PD), the diversity of lineages in a species assemblage (Faith 1992), might help to disentangle the relative importance of abiotic and biotic factors and of evolutionary history for community assembly (Pavoine and Bonsall 2011). However, studies that compare these three measures of alpha diversity, i.e. species richness, FD, and PD, are still scarce (but see Devictor et al. 2010, Fritz and Purvis 2010, Meynard et al. 2011), especially along elevational gradients.

Furthermore, a comparison of the functional and phylogenetic structure of species assemblages against null models can provide insights into the mechanisms that influence species co-occurrence (Webb et al. 2002, Cavender-Bares et al. 2009, Meynard et al. 2011). For example, one can assess if the species in an assemblage are functionally more similar or less similar than expected by chance (Gotelli 2000, Webb et al. 2002). An assemblage structure where species are significantly functionally different from each other (over-dispersion) is often interpreted as a consequence of interspecific competition, based on the assumption that species that are functionally too similar cannot co-occur and exclude each other (MacArthur and Levins 1967, Fleming 1979). In contrast, an assemblage structure where species are functionally very similar (clustering) is most often attributed to environmental filtering, where specific environmental conditions allow only species with a certain set of traits to persist (Webb et al. 2002). If functional traits of species are phylogenetically conserved, competition should also lead to phylogenetic over-dispersion, and environmental filtering to phylogenetic clustering (Webb et al. 2002, Cavender-Bares et al. 2009). Along the few elevational gradients which have previously been studied, phylogenetic structure changes from over-dispersion at lower elevations towards clustering at higher elevations (Graham et al. 2009, Machac et al. 2011; but see Bryant et al. 2008). However, none of these previous studies examined functional assemblage structure, which is likely to provide additional insights into the distribution of species with different functional roles within assemblages at different elevations.

Additional insight into the mechanisms shaping species assemblages might come from the comparison of the compositional, functional and phylogenetic similarities among species assemblages (beta diversity), for instance, assessing changes in functional and phylogenetic assemblage composition along environmental gradients (Graham and Fine 2008, Devictor et al. 2010, Meynard et al. 2011). To date, functional and phylogenetic assemblage structure on the one hand and compositional, functional and phylogenetic similarity among assemblages on the other hand have rarely been studied together (but see Devictor et al. 2010, Meynard et al. 2011) and never explicitly among assemblages along elevational gradients. The comparison of assemblage structure and its turnover among assemblages might help to understand if changes in assemblage structure with elevation correspond to changes in assemblage composition.

Our study combines several methods to investigate compositional, functional, and phylogenetic alpha diversity, beta diversity, and assemblage structure in a comparable way, to gain a better understanding of the mechanisms influencing species assemblages along elevational gradients. In the present study, we investigated species assemblages of frugivorous birds. Interspecific competition should be particularly high among species that belong to the same feeding guild because these species use similar resources and have similar roles in an ecosystem (Terborgh and Robinson 1986). A feeding guild might therefore be more appropriate for testing hypotheses about how ecologically similar species co-exist in tropical communities than assemblages of all birds (Terborgh and Robinson 1986, Graham et al. 2009). We chose frugivorous birds because of their important function in tropical ecosystems, as up to 90% of tropical woody plant species produce fleshy fruit that are dispersed mostly by birds (Fleming 1979). Frugivorous birds are well-suited for studies of functional trait diversity because many morphological traits, such as body mass and beak size, are known to influence the foraging behaviour and fruit choice of avian frugivores (Moermond and Denslow 1985,Wheelwright 1985, Dehling et al. unpubl.), and these traits can be readily measured.

We studied twelve species assemblages of frugivorous birds located equidistantly along a gradient from the lowlands to the tree line in the tropical Andes in southeast Peru (Manú Biosphere Reserve). The reserve lies in one of the regions of highest avian frugivore diversity worldwide (Kissling et al. 2009), and bird distributions within it are very well known (Walker et al. 2006). First, we studied the species richness, functional diversity (FD) and phylogenetic diversity (PD) of the species assemblages (alpha diversity patterns). (1) We expected species richness to decrease with increasing elevation (Sanders and Rahbek 2012). If the loss of species went along with a loss of functional roles in the assemblages towards higher elevations, we would expect similar declines for FD and – if functional traits were phylogenetically conserved – for PD. Secondly, we studied the functional and phylogenetic structure of the assemblages with a novel combination of indices and used null models to assess if the structure differed from a random distribution of species along the gradient. (2) Given that frugivorous birds compete for the same resource, we generally expected a trend towards over-dispersed assemblage structure, with rather regular functional and phylogenetic distances between species. With increasing elevation, an increasing effect of environmental filtering should lead to a clustered assemblage structure (Graham et al. 2009, Machac et al. 2011), with generally smaller, and more variable, functional and phylogenetic distances between species. Finally, we studied the compositional, functional and phylogenetic similarity (beta diversity) between neighbouring assemblages to investigate the change in assemblage composition along the gradient. (3) Based on previous studies along the Manú elevational gradient across all avian feeding guilds (Patterson et al. 1998, Jankowski et al. 2013), we might expect a peak of species turnover at mid-elevations. However, the outcomes of these studies were largely driven by insectivorous birds (Jankowski et al. 2013) that are less mobile than frugivorous birds (Levey and Stiles 1992). We might therefore alternatively expect a more constant turn-over with elevation for the highly mobile frugivorous birds. If species turnover coincided with a turnover in functional roles and phylogenetic lineages in the assemblages, we would expect patterns in functional and phylogenetic similarity to resemble those in compositional similarity.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Dataset and phylogeny

We studied assemblages of co-occurring frugivorous bird species at 12 elevational levels (‘elevations’ hereafter) placed every 300 m a.s.l. (‘m’ hereafter) from 300 to 3600 m in the Kosñipata valley in the Manú biosphere reserve in south-east Peru (‘Manú' hereafter). In a literature survey, we identified 245 bird species occurring in Manú that consume fruit as a main part of their diet (obligate and partial frugivores in the classification of Kissling et al. 2007; for details see Supplementary material Appendix 1). Studies of plant–frugivore interaction networks along the same gradient (Schleuning et al. 2014, Dehling et al. unpubl.) confirm that these species contribute the vast majority of interactions to the networks and are therefore a very good representation of the avian frugivore community. We compiled lists of co-occurring species for every elevation, using data from Walker et al. (2006), Merkord (2010), as well as data collected by DMD during field work in Manú between December 2009 and September 2011 (Dehling et al. 2013). By combining data from these different sources, we obtained comprehensive information on the elevational distribution of all 245 bird species within the Kosñipata valley (Supplementary material Appendix 1, 2, Fig. A1).

To compute the different functional indices, we selected nine morphological traits which are functionally related to the foraging and fruit-eating behaviour of frugivorous birds (Moermond and Denslow 1985, Wheelwright 1985): feeding (beak length, height and width), flight performance (wing length, tail length, Kipp's distance [i.e. the distance from the tip of the first secondary to the wing tip, which is a measure of the pointedness of the wing]) and bipedal locomotion (tarsus length, sagittal and lateral diameter of tarsus). We measured all traits on museum specimens (Eck et al. 2011) and aimed at measuring two males and two females of each bird species from local populations in Peru or adjacent Bolivia (see Supplementary material Appendix 1 for details and a list of specimens). In addition, data on body mass were compiled from Dunning (2007). For the analyses we used the species’ mean of each trait. All traits were log-transformed to approximate normality, and centred with z-transformations.

For phylogenetic assemblage indices, we constructed a phylogeny of the species in our dataset by hand, as a consensus tree of published molecular phylogenies. The family backbone followed Hackett et al. (2008) outside and Yuri and Mindell (2002) within the Passeroidea. Within families, relationships were reconstructed using a variety of sources (Supplementary material Appendix 2, Table A1). The final tree contained 244 species and was 78% resolved (Supplementary material Appendix 1, 2, Fig. A1). To assess the effects of polytomies in our phylogeny, we randomly resolved the phylogeny 100 times and calculated all phylogenetic measures for each of those fully bifurcating trees as well as for the original, unresolved tree. The results were virtually identical and are therefore not presented here. Branch lengths were not dated because the necessary sequence data were not available for all species. Because branch lengths are required by most statistical methods measuring phylogenetic assemblage structure, we imputed branch lengths on our unresolved phylogeny and on each of the randomly resolved phylogenies following Grafen (1989). To assess if results depended on branch length imputation, we computed all phylogenetic measures for each of our trees with three different settings of the scaling parameter ρ (ρ= 1. 5, ρ= 1.0, and ρ= 0.5). In general, the different branch length transformations did not affect results qualitatively, and we only report results for ρ= 1.

Species richness, functional diversity and phylogenetic diversity

For every assemblage, we calculated three types of alpha diversity: species richness, functional diversity and phylogenetic diversity. Species richness was the number of species present at an elevation. Functional diversity (FD) was measured as functional richness, FRic (Villéger et al. 2008). FRic measures the volume of a convex hull around all species of an assemblage projected in a multidimensional trait space using principal coordinate analysis. Species are arranged in trait space according to the Euclidean distances between them as calculated from the morphological traits. FRic of each assemblage was standardized against the total FRic calculated from all species of the regional species pool. Since FRic is determined by the extreme values of trait combinations, we also calculated the functional evenness of an assemblage (FEve; Villéger et al. 2008) which measures the regularity of distances between species in trait space along a minimum spanning tree. FEve ranges between 0 and 1, with values close to 1 indicating very constant and values close to 0 indicating very irregular functional distances between species in an assemblage (Villéger et al. 2008).

Following Faith (1992), we calculated phylogenetic diversity (PD) as the sum of all branch lengths contained by the minimum spanning tree that links all species in an assemblage on the regional pool phylogeny, including the link to the root. As an equivalent to functional evenness, we calculated the phylogenetic evenness (PhyloEve) of the assemblages. For comparability between the evenness measures, we constructed a minimum spanning tree from the pairwise phylogenetic distances of the species in an assemblage and then calculated PhyloEve using the same formula as for FEve. Accordingly, PhyloEve ranges from 0 to 1, with values close to 1 indicating similar phylogenetic distances and values close to 0 indicating irregular phylogenetic distances between species in an assemblage.

To assess the extent to which measures of FD and PD could be expected to reflect each other given our choice of morphological traits, we measured the phylogenetic signal in each trait with Blomberg's K (Blomberg et al. 2003, Kembel et al. 2010). This statistic has only been implemented for fully resolved phylogenetic trees, so we calculated K for each of the 100 randomly resolved trees after applying the different branch length transformations as described above. All morphological traits showed significant phylogenetic signal according to Blomberg's K, even when accounting for polytomies and different branch length transformations (Supplementary material Appendix 2, Table A2).

Functional and phylogenetic assemblage structure

We investigated phylogenetic assemblage structure with the net relatedness index (NRI; Webb et al. 2002), which is calculated as the inverse of the standardized effect size of the mean phylogenetic distance (MPD) between all taxon pairs in the assemblage phylogeny (Webb et al. 2002). To obtain NRI, the observed MPD value within each assemblage was compared against the values from 999 sets of randomized assemblages created with the independent swap algorithm (Gotelli 2000), with all species present along the elevational gradient as the source pool. This algorithm keeps the species numbers of the assemblages constant but randomizes the species occurrences across the assemblages, according to the occurrence frequencies of species in the original dataset. NRI was calculated as NRI =−1 × (observed MPD – mean of MPD values from randomizations)/standard deviation of MPD values from randomizations (Webb et al. 2002). Therefore, NRI values > 0 indicate phylogenetic clustering, NRI values < 0 indicate phylogenetic evenness or over-dispersion, and NRI = 0 denotes random phylogenetic structure (Webb et al. 2002). For every assemblage, we also calculated the nearest taxon distance (NTD, Webb et al. 2002) which is the mean distance of every species to its most closely related species in the phylogeny. We did not compute standard effect sizes but compared observed values of NTD with expectations from the null-model.

As a functional equivalent to NRI, we calculated the net functional relatedness index (NFRI), which is the mean pairwise functional distance (MPFD) standardized against null-model expectation as described above for the standardization of MPD to NRI. MPFD is the mean Euclidean distance between all species of an assemblage that are projected in functional trait space according to their functional traits with principal coordinate analysis (see description of FRic above). As a functional equivalent to NTD, we also calculated the nearest taxon functional distance (NTFD) as the mean distance of every species to its most similar species in trait space. We compared observed values of NTFD to null-model expectations as for NTD.

Compositional, functional and phylogenetic similarity

We computed three types of beta diversity between neighbouring assemblages along the elevational gradient: compositional, functional and phylogenetic similarity. We calculated compositional similarity between two assemblages using the Sørensen index (Oksanen et al. 2011). As a corresponding measure for functional similarity, we calculated the mean of the Euclidean distances from all species in one assemblage to their functionally most similar species in the other assemblage in multidimensional trait space. For a species that is present in both assemblages, this distance is zero. For the phylogenetic similarity between assemblages we used the PhyloSor index (Bryant et al. 2008, Kembel et al. 2010), which produces values between zero (no branches in the phylogenetic tree occur in both assemblages) and one (all branches occur in both assemblages). We tested if each of the beta diversity indices differed from random expectations by comparing the similarity observed between neighbouring assemblages with the similarity between neighbouring assemblages in the 999 randomized sets. The indices were then transformed to standardised effect sizes as described above for NFRI and NRI.

For all statistical analyses, we used R ver. 2.12 (R Development Core Team) and the packages picante (Kembel et al. 2010), FD (Laliberté and Legendre 2010) and vegan (Oksanen et al. 2011).

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Species richness, functional diversity and phylogenetic diversity

Species richness, functional richness (FRic) and phylogenetic diversity (PD) of frugivorous birds decreased nearly monotonically with increasing elevation (Fig. 1). PD decreased more slowly than species richness and FRic until 3000 m, but then decreased rapidly between 3000 and 3600 m (Fig. 1). Functional evenness (FEve) did not show a relationship with elevation, whereas phylogenetic evenness (PhyloEve) increased slightly with increasing elevation (Fig. 1).

image

Figure 1. Patterns of three types of alpha diversity, functional evenness and phylogenetic evenness for frugivorous birds along the Manú elevational gradient. (a) Species richness. (b) Functional richness (FRic, Villéger et al. 2008). (c) Faith's (1992) phylogenetic diversity. (d) Functional evenness (FEve, Villéger et al. 2008). (e) Phylogenetic evenness (PhyloEve) calculated from a minimum spanning tree derived from the pairwise phylogenetic distances among the species in the assemblages.

Download figure to PowerPoint

Functional and phylogenetic assemblage structure

Net functional relatedness index (NFRI) and net relatedness index (NRI), indicated significant functional and phylogenetic over-dispersion in the three lowland assemblages (300–900 m), random assemblage structure at 1200 m, and a tendency towards clustering in the highland assemblages (1500–3600 m; Fig. 2a, b). Functional clustering was significantly higher than expected by chance from 2100 to 2700 m (Fig. 2a). These patterns arise because mean pairwise functional distance (MPFD) and mean phylogenetic distance (MPD) decreased with increasing elevation, whereas null-model expectations showed no trend along the elevational gradient (Fig. 2c, d). Phylogenetic and functional distances between the most functionally similar or the most closely related species pairs, NTFD and NTD, were constant until about 1200 m and then increased with increasing elevation (Fig. 2e, f). All values of NFTD and NTD lay within the range of values expected under the null model along the elevational gradient (Fig. 2e, f).

image

Figure 2. Functional and phylogenetic assemblage structure for frugivorous birds along the Manú elevational gradient. (a) Net functional relatedness index, NFRI, and (b) net relatedness index, NRI, are standard effect sizes of observed values of mean pairwise functional distance and mean pairwise phylogenetic distance against 999 sets of simulated assemblages. They reflect functional and phylogenetic over-dispersion (values < 0), random structure (values = 0) or clustering (values > 0) of assemblages. Asterisks denote significant clustering or over-dispersion compared to the randomizations. (c) Mean pairwise functional distance, MPFD, and (d), MPD, are the average distances among species pairs in the functional trait space and the phylogeny, respectively. Solid lines indicate the mean, and the dotted lines indicate the 95% confidence interval of values obtained from 999 sets of randomized assemblages along the elevational gradient. Asterisks denote significant deviations from null model expectations. (e) Nearest taxon functional distance, NTFD, and (f) nearest taxon distance, NTD, are the mean minimum distances among species pairs in the functional trait space and the phylogeny, respectively. Solid lines indicate the mean, and the dotted lines indicate the 95% confidence interval of values obtained from 999 sets of randomized assemblages along the elevational gradient.

Download figure to PowerPoint

Compositional, functional and phylogenetic similarity

Similarity values between neighbouring assemblages along the elevational gradient were nearly always significantly higher than expected under the null model, which randomizes species occurrences across the whole elevational gradient (Fig. 3). Compositional, functional and phylogenetic similarity was very low between the assemblages at 900 and 1200 m (Fig. 3), with phylogenetic similarity as low as expected under the null model. Above 3000 m, all types of similarity decreased considerably, with phylogenetic similarity between neighbouring assemblages being not significantly different from the null model expectation (Fig. 3). Assemblages above 3000 m were all nested within the assemblage at 3000 m, i.e. the low similarity between assemblages above 3000 m was solely due to species loss.

image

Figure 3. Compositional, functional and phylogenetic similarity between neighbouring assemblages along the Manú gradient. All indices are standardized effect sizes based on 999 sets of randomized assemblages. Compositional similarity reflects Sørensen dissimilarity (Oksanen et al. 2011), functional similarity is the mean of the Euclidean distance of every species in one assemblage to its functionally most similar species in the neighbouring assemblage in functional trait space, and phylogenetic similarity was calculated with PhyloSor (Bryant et al. 2008, Kembel et al. 2010). Points are connected for visualization purposes only. All similarity values differed significantly from randomizations, except those marked with ‘ns’.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

All types of alpha diversity (species richness, FRic, Faith's PD) decreased with increasing elevation, whereas functional and phylogenetic evenness were rather constant along the elevational gradient. Lowland frugivorous bird assemblages were characterized by functionally and phylogenetically over-dispersed assemblage structure, whereas highland assemblages above 1200 m had a clustered assemblage structure. The similarity between neighbouring assemblages (beta diversity) indicated a break in assemblage composition in the Andean foothills between 900 and 1200 m.

The decrease in FRic along the elevational gradient resembles the pattern for species richness and reflects a gradual loss of species with extreme trait combinations (i.e. species that lie on the margins of the functional trait space; Supplementary material Appendix 2, Fig. A2), suggesting that species that fulfil specialized functional roles disappear successively towards higher elevations. By contrast, the relatively slow decrease of PD until 3000 m indicates that the decrease of PD is not caused by the disappearance of whole lineages but that the number of species within many lineages decreases gradually with increasing elevation. In fact, the number of bird orders in Manú is nearly constant until 3000 m (the lowland assemblages contain two additional orders with only one species each), but above 3000 m there is a strong decline across all taxonomic levels and only five species in two Passeriformes families occur at 3600 m (Supplementary material Appendix 2, Fig. A1). Above 3000 m and up to the tree line at around 3700 m, an increasing inter-digitation of forest and grasslands probably leads to a rapid decline in the availability of fleshy fruit above 3000 m and thus the decline in species richness, FRic and PD.

The patterns of over-dispersed functional and phylogenetic assemblage structure in lowland assemblages and clustered assemblage structure in the highland assemblages in Manú correspond largely to patterns of phylogenetic assemblage structure reported from other elevational gradients (Bryant et al. 2008, Graham et al. 2009, Machac et al. 2011). These patterns are often attributed to a prevalence of competition in lowlands assemblages and a prevalence of environmental filtering in highland assemblages (Graham et al. 2009, Machac et al. 2011). However, this interpretation might be over-simplified and is therefore contentious (HilleRisLambers et al. 2012). Competition among functionally similar species should lead to regular distances between species in trait space and, if traits are conserved, across the phylogeny (MacArthur and Levins 1967, Fleming 1979). However, the standard indices NFRI and NRI are calculated from the mean functional (MPFD) and mean phylogenetic (MPD) distances between all species pairs in an assemblage, and the patterns of MPFD and MPD were very similar to those for FRic and PD, i.e. the total volume of trait space and total length of phylogenetic branch lengths. Hence, NFRI and NRI do not assess the regularity of functional and phylogenetic distances between species in an assemblage but rather show whether an assemblage has an above-average or below-average range and diversity of functional trait combinations (or phylogenetic lineages) relative to a random assemblage.

Assessing the regularity of distances among species in an assemblage, however, is crucial for an understanding of the co-occurrence of functionally and phylogenetically similar species, and can be measured with functional and phylogenetic evenness (Villéger et al. 2008). If there were a prevalence of competition in the lowlands and a prevalence of environmental filtering in the highlands, then the regularity of distances between species in an assemblage should decrease with increasing elevation. By contrast, the constant values for FEve and the slight increase of PhyloEve with increasing elevation argue against a higher influence of competition in the lowlands compared to the highlands. Provided that competition between species (or rather the avoidance thereof) is indeed manifested in regular functional and phylogenetic distances between species in functional trait space and phylogeny, then competition appears to have played a similar role within all assemblages along the elevational gradient.

The increasing minimum distances between species pairs in the functional trait space and the phylogeny with increasing elevation further contradict the assumption of higher competition in the lowlands than in the highlands, because such a mechanism should lead to larger-than-expected distances among lowland species and smaller-than-expected distances among highland species. The increasing minimum distances in the phylogeny and in trait space are caused by the decline in species numbers across all major phylogenetic lineages, and show that in the lowland assemblages species are more densely packed in the functional trait space, i.e. a higher number of relatively similar species co-occurs in the lowlands than at higher elevations. This suggests a higher redundancy in the functional roles of species in the lowlands which may increase the robustness of lowland assemblages to species loss (Schleuning et al. 2012). However, the fact that species from all major lineages (i.e. species from all orders except Caprimulgiformes [Steatornithidae] and Gruiformes [Psophidae]; Supplementary material Appendix 2, Fig. A1, A2) are present in all assemblages up to 3000 m suggests that basic functional roles of frugivorous birds might be maintained in all these assemblages. For instance, although Ramphastos toucans in the lowlands can probably handle larger fruits than Andigena toucans in the highlands, both eat and disperse fruits of plant species that are primarily dispersed by toucans and other large avian frugivores.

Together with the constant decline in alpha diversity, the rapid change from over-dispersed towards clustered assemblage structure between 900 and 1500 m indicates that the assemblages up to 900 m have a higher number of species from functionally and phylogenetically distinct lineages. In fact, between 900 and 1200 m there is a reduction in the number of species from basal lineages of non-Passeriformes (notably of tinamous, parrots, and toucans; Supplementary material Appendix 2, Fig. A1) that lie close to the periphery in the functional trait space (Supplementary material Appendix 2, Fig. A2).

The patterns in beta diversity further confirmed these abrupt changes. Beta diversity between neighbouring assemblages was nearly always significantly higher than expected from null models, because the randomization did not consider continuity in species’ elevational ranges. It is therefore particularly striking that the similarity between the 900 and 1200 m assemblages was so low that it did not significantly differ from the null model, indicating that many species’ elevational ranges ended or started here. In fact, the species turnover in the Andean foothills between 900 and 1200 m was the highest along the elevational gradient, both in number and proportion of species exchanged (42 spp. with elevational maximum at 900 m and 26 spp. with minimum at 1200 m; Supplementary material Appendix 2, Fig. A1). The turnover in species composition of frugivorous birds thus peaked slightly lower than reported for the whole avian community of Manú (1400 m; Patterson et al. 1998, Jankowski et al. 2013) that corresponds largely to changes in forest types (Patterson et al. 1998). Patterns in the whole avian community might be strongly driven by the dominance of insectivorous birds in the assemblages (Jankowski et al. 2013). Although the species turnover of frugivorous birds is also influenced by compositional changes in tree species (Jankowski et al. 2013), it might further be associated with changes in other habitat characteristics and topography. In Manú, forests up to around 1000 m are characterized by relatively plain terrain and open understory, contrasting with the steep slopes and thicker understory found at higher elevations. This should affect large ground-dwelling frugivorous bird species (Mitu, Tinamidae and Psophidae) whose numbers decline abruptly between 900 to 1200 m (Supplementary material Appendix 2, Fig. A1). Moreover, the increasingly rugged topography towards higher elevations is probably less favourable for frugivorous bird species that forage in the canopy and over large distances, i.e. large parrots and toucans, whose numbers also decline abruptly between 900 to 1200 m (Supplementary material Appendix 2, Fig. A1).

The differences in assemblage structure and composition between over-dispersed lowland and clustered highland bird assemblages might also be influenced by different evolutionary histories of lowland and highland assemblages because speciation within regions and low dispersal between regions can result in phylogenetically clustered assemblage structure (Emerson and Gillespie 2008, Cavender-Bares et al. 2009). There is some evidence that some highland and lowland taxa in Amazonia underwent separate radiations, with a break at about 1000 m (Weir 2006, Fjeldså et al. 2012). Lowland avifaunas in the Neotropics are phylogenetically older than highland avifaunas, with many highland clades originating as recently as the Pleistocene (Fjeldså and Rahbek 2006, Weir 2006). The clustered assemblage structure in the highlands might thus to some extent also be influenced by relatively younger speciations in the highlands.

Conclusions

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Our study illustrates that the investigation of functional groups of species, such as avian frugivores, is well-suited to detect patterns in assemblage structure, and that the inclusion of functional and phylogenetic data can provide deeper insights into the mechanisms that influence geographic patterns in the species richness and composition of species assemblages. In contrast to previous studies, we show that the way species fill the available trait space is very similar along the elevational gradient, suggesting that the mechanisms that influence the assemblages are similar across the elevational gradient. We also reveal an abrupt change in assemblage composition and structure of frugivorous birds in the Andean foothills. This rapid change is probably due to factors unrelated to climate, e.g. changes in habitat structure and topography, or independent species radiations in lowlands and highlands and suggests that there are strong distributional limits for many species along the elevational gradient, especially in the foothill region. Temperature and precipitation in the Andes are expected to change drastically in the near future (Bush 2002) and species are either expected to adapt locally or to shift their ranges to track their preferred climatic conditions (Hof et al. 2011). Our results imply that frugivorous bird species in the tropical Andes, and the plant species that they disperse, may have difficulties shifting their ranges uphill as a response to global change.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Mathias Templin helped with the compilation of bird diet data. R. van den Elzen (ZFMK Bonn), R. Prŷs-Jones and M. P. Adams (NHM Tring), G. Mayr (SMF Frankfurt/M.) and R. Winkler (NMB Basel) provided access to the bird collections kept in their charge. M. Hennen, J. Bates and D. Willard (FMNH Chicago) sent specimens, and J. V. Remsen and S. W. Cardiff (LSUZM Baton Rouge) and D. Willard (FMNH Chicago) provided additional measurements. We thank R. Diesener, S. Frahnert, C. Bracker, P.-R. Becker, J. Fjeldså, N. Krabbe and J. Mlíkovsky for information about collection holdings. This study was supported by the research funding programme ‘LOEWE – Landes-Offensive zur Entwicklung Wissenschaftlich-ökonomischer Exzellenz’ of Hesse's Ministry of Higher Education, Research, and the Arts. Field work in Manú was also supported by a grant from the German Academic Exchange Service (DAAD) to DMD and was conducted under the permits 041-2010-AG-DGFFS-DGEFFS, 008-2011-AG-DGFFS-DGEFFS, 01-C/C-2010-SERNANP-JPNM, and 01-2011-SERNANP-PNM-JEF.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Supplementary material (Appendix ECOG-00623 at <www.oikosoffice.lu.se/appendix>). Appendix 1–2.