Although endangered and alien invasive plants are commonly assumed to persist under different environmental conditions, surprisingly few studies have investigated whether this is the case. We examined how endangered and alien species are distributed in relation to community biomass and N : P ratio in the above-ground community biomass in savanna vegetation in the Brazilian Cerrado.
For 60 plots, we related the occurrence of endangered (Red List) and alien invasive species to plant species richness, vegetation biomass and N : P ratio, and soil variables.
Endangered plants occurred mainly in plots with relatively low above-ground biomass and high N : P ratios, whereas alien invasive species occurred in plots with intermediate to high biomass and low N : P ratios. Occurrences of endangered or alien plants were unrelated to extractable N and P concentrations in the soil.
These contrasting distributions in the Cerrado imply that alien species only pose a threat to endangered species if they are able to invade sites occupied by these species and increase the above-ground biomass and/or decrease the N : P ratio of the vegetation. We found some evidence that alien species do increase above-ground community biomass in the Cerrado, but their possible effect on N : P stoichiometry requires further study.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Alien invasive plants are commonly stated to pose a threat to populations of native plants, especially of endangered species (Sala et al., 2000; Harrison et al., 2006), although the evidence to support this assertion is remarkably sparse. In part, this may be because few studies have attempted to establish a link between the invasion by alien plants and the local extinction of native species (Bradshaw et al., 2008; Sax & Gaines, 2008). However, a more fundamental reason may be that invasive and scarce native species usually exhibit very different functional traits (van Kleunen & Richardson, 2007; Bradshaw et al., 2008) and are therefore unlikely to persist together.
There are indications that endangered and invasive species tend to occupy different ends of environmental gradients, such as nutrient availability and/or vegetation biomass. For example, in temperate wetlands and grasslands, endangered species are found mainly in vegetation with a low biomass (Moore et al., 1989; Wheeler & Shaw, 1991; Wassen et al., 2005), and the disappearance of these species is often associated with increasing productivity caused by nutrient enrichment. In this connection, nitrogen (N) enrichment has usually been seen as the main cause of species loss (Stevens et al., 2004; Bobbink et al., 2010), though there is evidence that phosphorus (P) enrichment may also be important; for example, it has been shown that Red List species persist better at P-limited than at N-limited sites (Olde Venterink et al., 2003; Wassen et al., 2005); these Red List species have actually disappeared in at least 20% of map units in 20 years' time. By contrast, many alien invasive plant species grow quickly under nutrient-rich conditions (Baruch, 1996), and their spread may be stimulated by nutrient enrichment, especially of N (Daehler, 2003; D'Antonio & Mack, 2006). There is less evidence that P enrichment has a similar effect (cf. Daehler, 2003), which may indicate that this factor is less important for alien invasive plants. However, it may also reflect a bias in research, with most studies being conducted in areas where N limitation was more acute than P limitation, such as on young soils in Hawaii (Vitousek et al., 1993), and some areas in North America and Europe. Also, nutrient stoichiometry has been recognized as a potential controlling factor for species invasions (Harpole, 2006; González et al., 2010), although its importance in natural plant communities has scarcely been investigated.
Many of the most important hotspots of plant biodiversity are on ancient, highly weathered soils in the tropics (Phoenix et al., 2006). In these areas, P limitation may be more important than in temperate regions (D'Antonio & Mack, 2006; Lambers et al., 2008; Vitousek et al., 2010; Olde Venterink, 2011; but see Elser et al., 2007), and the processes of nutrient enrichment and vegetation change may also be different. Indeed, there is evidence that invasive plants in tropical areas are at least co-limited by N and P (Barger et al., 2003; Bustamante et al., 2012). Such information is not merely of academic interest but is also important for developing appropriate conservation strategies in cases where N- and P-limited ecosystems require different types of management (Cech et al., 2008). Hence, it is important to improve our understanding of how endangered and alien invasive plants in tropical ecosystems are distributed in relation to productivity and N : P stoichiometry.
With over 10 000 species of vascular plants, the Brazilian Cerrado is a biome of enormous floristic diversity (Myers et al., 2000; Phoenix et al., 2006; Mendonça et al., 2008), but one that is increasingly threatened by agricultural activities. There is an officially recognized list of endangered plant species in the Cerrado (cf. Supporting Information, Table S1). Cerrado soils are generally very poor in mineral available P (Goedert, 1983), although availabilities of both N and P vary widely, owing to differences in the organic N and P contents of the soil, and to contributing factors such as fire frequency and the abundance of legumes (Townsend et al., 2002; Bustamante et al., 2006).
In this paper we follow the approach of Wassen et al. (2005) to investigate the distribution of two ecological plant groups in the Cerrado – endangered herbaceous species and alien invasive species – in relation to above-ground biomass and N : P stoichiometry of the vegetation. We hypothesized that the two groups would persist at opposite ends of these environmental gradients, with endangered species occurring at sites with a low biomass and high N : P ratios, and alien invasive species at sites with a high biomass and low N : P ratios.
Materials and Methods
The study was conducted in pristine Cerrado fragments in five nature reserves located in two regions c. 200 km apart. One of the regions was in Brasília-DF (Central Plateau), where three protected areas together occupy a total of some 10 000 ha of Cerrado habitat (at 15°55′S, 47°55′W). The other two reserves were located in Alto Paraíso-GO (Veadeiros Plateau) and cover a total area of nearly 400 ha (Água Fria Reserve at 14°05′S, 47°30′W and Pândavas Reserve at 14°09′S, 47°40′W). The two regions differ in their major soil types, with Brasília-DF having predominantly clayey Oxisols and the Alto Paraíso-GO sites having Cambisols. In both regions, the annual precipitation ranges between 1100 and 1600 mm, and the average annual temperature is c. 20°C (Munhoz & Felfili, 2006; Nardoto et al., 2006).
The study sites were located in vegetation types known as campo limpo (open grassland), campo sujo (open shrub savanna), and cerrado sensu stricto (shrub-dominated savanna). We selected a total of 60 plots, of size 2 × 2 m2 (35 in Brasília-DF and 25 in Alto Paraíso-GO), containing only herbaceous vegetation. Plots located in the more woody vegetation types were selected in patches containing only herbaceous plants. Twenty-six of these 60 plots were selected because they contained Red List species (endangered or E plots), which, based on decrease in abundance, are those classified as ‘vulnerable’, ‘threatened’ or ‘critically threatened’ to extinction according to the International Union for Conservation of Nature and Natural Resources (IUCN) category groups for threatened species (IUCN, 2012). A further 15 plots were selected because they contained one or more alien invasive species (alien or A plots). Additionally, each E plot and A plot was matched with a reference plot 100–500 m away containing no alien or endangered species but with the same vegetation type and a similar community structure. There were only 19 reference plots, because some of them were paired with two or three A or E plots.
Vegetation relevées were made between 15 January and 1 February 2008, at the peak of the growing season, and the location of each plot was recorded using GPS. With the aid of local botanists, all herbaceous species were identified and their cover in each plot estimated using the Braun–Blanquet scale (cf. van der Maarel, 1979). For species determination, we used the Angiosperm Phylogeny Group (APG) (2003), The Plant Names Project (http://www.ipni.org, July 2012), and the website W3 Tropicos (http://www.tropicos.org, July 2012). Species were considered as endangered if they were listed in the Brazilian Red List, elaborated following the IUCN Red List criteria (Dias et al., 2008), and as alien invasives if they were recorded as such in the lists of Mendonça et al. (2008) and Felfili et al. (2004). In this study we define alien invasive species as those species that immigrated to an area, became naturalized and expanded in abundance over this new habitat.
The total above-ground biomass of each plot was estimated by clipping the vegetation at 2–3 cm in three randomly sampled 0.25 m2 subplots, and the material was then pooled. Three soil samples (upper 10 cm) were collected from each plot, using a 3-cm-diameter auger, and the material pooled. The samples were either immediately brought to the laboratory or frozen for a maximum of 48 h before analysis.
The vegetation samples were dried at 60°C for 48 h, weighed and ground. N and P concentrations in the living plant material were determined after Kjeldahl digestion (heating at 420°C for 60 min in a mixture of concentrated sulfuric acid and CuSO4), by means of a continuous flow injection analyzer (FIA Star, Foss Tecator Höganäs, Sweden). The soil samples were extracted by shaking 2 g fresh soil with 1 M KCl for 1 h. Extractable ammonium was determined colorimetrically using Nessler reagent, while nitrate was determined by UV absorption according to Meier (1991). Soil pH was measured after shaking 10 g fresh soil with 25 ml distilled water. Extractable phosphorus was determined from soil samples dried at 70°C, using Mehlich extraction and continuous FIA.
We performed simple linear and second-order polynomial regressions, with biomass or species richness as dependent variables, and vegetation (biomass, species richness, N and P concentrations, N : P ratio) and soil variables (, , : ratio, P-PO43−, pH) as predictors. Based on the same vegetation and soil variables, we used logistic regressions (based on presence/absence) to calculate the probabilities of finding either one or more endangered species; one or more alien invasive species; or only common species. Using multiple logistic regression analyses, we determined the minimal regression models of the effect of the above-mentioned vegetation and soil variables on the presence of endangered and alien invasive species with elimination of nonsignificant terms based on likelihood ratio. We sequentially entered the remaining explanatory variables from this model in another multiple logistic model to estimate their individual and shared effects on the presence of endangered and alien invasive species. A t-test was used to test the significance of differences in species richness between A and E plots. Differences in species richness, above-ground biomass and vegetation N : P ratio among endangered and alien invasive plots and their respective reference plots were tested by means of paired t-tests. Because above-ground biomass and N : P ratios could be determined by the dominance of alien plants in the plots, we tested correlations (Pearson's) between these two variables and the cover of alien species in the plots (Braun–Blanquet scale). If assumptions of normality and homogeneity of variance were not fulfilled, data were log-transformed. All significance levels were set to P <0.05 and all statistical analyses were performed with SPSS 19.
The 60 plots contained a total of 682 vascular plant species, with individual 4 m2 plots containing from one to 47 species (Fig. 1). A total of 13 Red List species were present in the 26 E plots (i.e. plots containing endangered species), but these were never abundant, accounting for < 5% of the vegetation cover in any plot. The 15 A plots contained a least one of four widely distributed alien invasive species (Table S1), namely two African grasses, Melinis minutiflora and Brachiaria decumbens, a North American-European fern Pteridium aquilinum, and a parasitic vine native to the Brazilian Atlantic coast, Cassytha filiformis. Paired t-tests showed that species richness was not different in the endangered plots (means ± SE = 26.4 ± 2.1 species) and their respective reference plots (23.8 ± 2.4 species), but the alien plots had significantly fewer species (15.8 ± 3.6 species) than their reference plots (22.2 ± 2.0 species; P <0.05). Furthermore, endangered plots had significantly more species than alien plots and both sets of reference plots had the same species richness (t-test; P <0.05). In 2008, when the survey was conducted, none of the plots contained both alien invasive and endangered species, but in 2010 we observed that one of the E plots (containing Paepalanthus extremensis) had been invaded by M. minutiflora.
Species richness in plots was negatively related to above-ground biomass, which ranged from 160 to 4000 g m−2 (Fig. 1a), and positively related to tissue N : P ratio (mass-based), which ranged from 7 to 21 (Fig. 1b), although with a very low R2 in the latter case. The N : P ratio decreased significantly with increasing biomass (Fig. 2a), mainly as a consequence of declining N concentrations (Fig. 2b), with P concentrations remaining relatively constant (Fig. 2c). As a result, there were no plots combining a high N : P ratio and high biomass.
Endangered and alien invasive species differed significantly in their relationships to both above-ground biomass and N : P ratio of vegetation (Fig. 1c–f). Thus, endangered species occurred only in sites with an above-ground biomass < 1200 g m−2 and a N : P ratio > 11, whereas alien invasive species mainly occurred where the biomass was > 1000 g m−2 and the N : P ratio was < 13. However, the association of endangered species with high tissue N : P ratio was no longer significant when only sites of low biomass were considered (241–1166 g m−2), while the relationship for alien species was significant across the entire biomass range in which they occurred (553–3874 g m−2). The presence of endangered plants was positively related to the nitrogen concentration in the vegetation, whereas the presence of alien invasive species was related negatively to nitrogen and positively to phosphorus concentration in the above-ground biomass (Fig. S1). The presence of endangered plants was negatively related to N and P stocks in the vegetation, while the opposite pattern was observed for the presence of alien invasive plants (Fig. S1). Compared with their corresponding reference plots, the E plots had a significantly (P <0.05) lower biomass and a higher N : P ratio, whereas the A plots had a significantly (P <0.01) higher biomass and a lower N : P ratio. In the A plots, the cover of alien species was not significantly correlated with either above-ground biomass or N : P ratio of the vegetation, but we note that a moderate or high cover of alien plants was never associated with a low biomass (< 1000 g m−2) or a high N : P ratio (> 15; Fig. 3).
Total species richness was negatively correlated with soil extractable P (Fig. S2). However, the probability of occurrence of endangered or alien invasive species was not significantly influenced by either soil inorganic N or P concentrations, soil N : P ratio of available forms or NH4 : NO3 ratio. Species richness showed a significant quadratic relationship with soil pH, and the probability of occurrence of endangered species decreased with increasing soil pH, whereas that of alien invasive species increased (Fig. S2). Biomass was weakly positively correlated with soil ammonium concentrations (P =0.031) but not with either soil nitrate or phosphate.
Multiple regressions confirmed that the two most important factors explaining the presence of either endangered and alien invasive species were vegetation biomass and N : P ratio (Table 1). In the case of endangered species, these factors explained 54% of the variation, while for invasive species they explained 75% of the variation. Biomass was a stronger explanatory variable for endangered species, and N : P ratio was a stronger explanatory variable for alien invasive species (Table 1). In addition, for endangered species soil, pH was also a significant factor (P <0.05; Table 1).
Table 1. Minimal binary logistic regression models of the effects of vegetation (biomass, species richness, N and P concentrations in the biomass, and biomass N : P ratios) and soil variables (ammonium, nitrate, ammonium : nitrate ratio, available phosphorus, and pH) on the presence/absence of endangered (a) and alien invasive species (b), using elimination of nonsignificant terms based on likelihood ratio at P <0.05
Change in R2 (%)
Analyses were automatically terminated after eight steps with biomass and N : P ratio remaining as explanatory variables for both dependent variables, and additionally pH for the endangered species; slopes and levels of significance are shown. Cumulative explained variances (Nagelkerke's R2) attributed to biomass, N : P ratio and pH on presence of endangered and alien invasive species at the Cerrado, as well as the increments in R2 are shown. Note the different orders of the model components in (a) and (b). *, P <0.05; ***, P <0.001.
As far as we know, this study is the first to investigate patterns of both endangered and alien invasive species in relation to important ecological characteristics in a natural plant community. As hypothesized, we found that endangered and alien plants in the Cerrado occupied opposite ends of the ranges in above-ground biomass and vegetation N : P ratios, with endangered plants occurring in plots with low biomass (< 1200 g m−2) and high N : P ratios, while alien invasive species occurred in sites with high biomass and low N : P ratios (Fig. 1). In the following, we discuss these findings in light of the response of these plant groups to nutrient availability, limitation and enrichment. We have assumed that N and P are the main limiting nutrients for growth, as they are in most terrestrial plant communities (cf. Elser et al., 2007; Harpole et al., 2011), and hence that enrichment of N would induce relative limitation of P, and vice versa. The limited data available on nutrient limitation in the Cerrado show that growth of the vegetation, including alien species, was limited by P and, to a lesser extent, by N, but we cannot rule out that growth of the native vegetation might have been co-limited by a nutrient other than N or P (Bustamante et al., 2012; Lannes, 2012).
The endangered species of this study have all decreased in their occurrence in the Cerrado over time as a result of environmental change (Dias et al., 2008). Besides direct habitat destruction, this environmental change could include factors such as altered fire frequencies (Mistry, 1998), nutrient enrichment and eutrophication, or alien plant invasions; our study was focused on the latter two factors.
Endangered species in relation to biomass and vegetation N : P
Our finding that endangered Cerrado plants only occur in plots with a low above-ground biomass is consistent with similar patterns observed in herbaceous wetlands and grasslands in Eurasia and Canada (Moore et al., 1989; Wheeler & Shaw, 1991; Wassen et al., 2005). In those cases, it was suggested that the endangered species were poor competitors for light and therefore not able to persist under nutrient-rich, highly productive conditions. The same mechanism probably applies in the Cerrado, where most of the endangered species (nine out of the 13 species sampled) were also relatively small and could be outcompeted by shading of taller plants, though the maximum biomass at which endangered species occurred was twice as high as in temperate vegetation (600 g m−2; Fig. 1), perhaps because of greater insolation compared with temperate regions (Landsberg, 1961).
Another clear pattern in our data was that endangered species only occurred at vegetation N : P ratios > 11, and their probability of occurrence increased with increasing values of N : P. This result is consistent with the study of Wassen et al. (2005), who found that endangered plants in Eurasian wetlands were associated with relatively high N : P ratios, which they interpreted as reflecting P limitation. In temperate regions, the critical value of N : P below which N limitation occurs is thought to be c. 14 (Olde Venterink et al., 2003; Güsewell, 2004), but in savanna vegetation dominated by C4 grasses this critical value is probably lower, perhaps around 9; (Cech et al., 2008). If this lower value applies for Cerrado vegetation, the endangered plants in our study were probably growing under P-limiting conditions, though co-limitation by N and P cannot be excluded from N : P ratios alone (Olde Venterink et al., 2003; Cech et al., 2008).
The restriction of endangered species to sites with a low biomass and high vegetation N : P ratio could have arisen in various ways. For example, it could have resulted from a general increase in vegetation biomass produced by the widespread use of N fertilizers in the Cerrado in recent years (Filoso et al., 2006). According to this explanation, endangered species have reduced in their occurrence, mainly because of increasing vegetation biomass in these fertilized sites (or sites indirectly affected by N enrichment through atmospheric deposition or pollution), and only persist in nature reserves and other sites where P is limiting or co-limiting and so the biomass remains low (which is consistent with the pattern in Fig 2a). Alternatively, the pattern could have arisen if most of the formerly P-limited and unproductive vegetation had been altered by P enrichment or destroyed, so that only a few fragments remain. A problem with both explanations, however, is that we found no evidence of a link between the presence of endangered species and soil extractable pools of either N or P. This could imply either that the extractable pools did not reflect N or P availabilities for plants (i.e. plant and microbial N and P uptake might have been so fast that it had removed N or P derived from deposition or mineralization very quickly without having an effect on extractable nutrient pool sizes), or that above-ground biomass increments were induced not by increased soil N or P availabilities but, for instance, by an altered species composition (such as alien plant invasion, see the section ‘Do alien invasive plants pose a threat to the endangered native species?’).
We did find a link between the presence of endangered species and soil pH. Unlike in Europe, however, where rare and endangered species are more frequently associated with soils of a relatively high pH (> 5) (Hodgson, 1986; Roem & Berendse, 2000; Pärtel et al., 2004), we found they were more likely in the Cerrado to occur on acidic soils. We speculate that this biogeographical difference could be related to large-scale patterns in the distribution of soil pH and the size of the associated species pools, which are themselves the result of soil age and weathering processes, on the one hand, and evolutionary processes on the other (e.g. Pärtel, 2002; Lambers et al., 2010). Theoretically, the correlation between soil pH and endangered species could also be affected by the relationship between pH and P availability (Hinsinger, 2001), if P availability were positively correlated to pH, but this was not the case in our sites.
Alien invasive species in relation to biomass and vegetation N : P
The four alien species all occurred in plots of medium to high above-ground biomass, which contained more biomass than their corresponding reference plots. This pattern is consistent not only with our hypothesis, but also with other studies showing that alien invasive species plants flourish under conditions of high productivity, and high above-ground biomass. These include both studies using the same species as the ones we investigated – M. minutiflora, B. decumbens, and P. aquilinum (Baruch et al., 1989; Klink, 1994; Baruch, 1996; Whitehead et al., 1997; Hoffmann & Haridasan, 2008; Martins et al., 2011) – and studies in other regions with other species (de Gruchy et al., 2005; Wilsey et al., 2009).
The occurrence of alien invasive plants in the Cerrado at the lower range of N : P ratios (Fig. 1f) suggests that these species perform well under conditions of (relative) N limitation. This conclusion is supported by various pieces of evidence from other studies. First, fertilization experiments in Cerrado vegetation show that two of the invasive species studied, M. minutiflora (Bustamante et al., 2012; L. S. Lannes et al., unpublished) and P. aquilinum, respond positively to P addition (Werkman et al., 1996) (i.e. to simulated N limitation, assuming that N and P are the main limiting nutrients and that P addition alleviated P limitation). Secondly, the grasses M. minutiflora and B. decumbens are known to have high N-use efficiencies (NUEs) – higher at least than those of native South American C4 grasses (Baruch et al., 1985; Baruch, 1996). Thirdly, the same two species often dominate in burned sites (Hughes & Vitousek, 1993; D'Antonio et al., 2001; Silva & Batalha, 2008; personal observation), which tend to be N-limited because N is lost to the atmosphere when plant material burns (cf. Cech et al., 2008). And in Hawaii, M. minutiflora also invades young, N-limited volcanic soils (Vitousek et al., 1993; D'Antonio & Mack, 2006). Fourthly, it may be significant that the families Fabaceae (90% from the frequently N-fixing subfamilies Papilionoideae and Mimosoideae) and Poaceae are strongly represented in the alien flora of the Cerrado, since both taxa possess adaptations to cope with low N availability, for example, N fixation and a high NUE (Table S2). Taken together, these various lines of evidence suggest that alien invasive plants in the Cerrado are strong competitors under N-limiting conditions, which may be because they benefit more than native species from an increased P supply (Bustamante et al., 2012; L. S. Lannes et al., unpublished) and/or because they produce more biomass per unit of acquired N (Baruch et al., 1985; Asner & Beatty, 1996; Whitehead et al., 1997).
Do alien invasive plants pose a threat to the endangered native species?
Does the spread of nonnative plants in Cerrado as such (i.e. just the invasion, and not invasion as a result of, for instance, nutrient enrichment) pose a threat to native species? One could argue that if the alien species are able to lower the N : P below the range where endangered species persist (N : P = 11, Fig. 1d), they could pose a threat. However, we know of no direct mechanism through which a change in the vegetation N : P ratio could cause competitive exclusion, and hence we don't have good arguments for alien plant invasions causing species extinction through only altering foliar N : P stoichiometry. This is different for the amount of above-ground biomass. If alien species could increase above-ground biomass above 1 kg m−2 (see Fig 1c), endangered species would no longer occur because they would probably be outcompeted for light (Hughes & Vitousek, 1993; Hautier et al., 2009; D'Antonio et al., 2011). There is also some evidence that alien plants can indeed increase the above-ground biomass of a site, even without nutrient enrichment. For example, by comparing biomass in an invaded site with that in the same site where M. minutiflora was systematically removed at very young stages, Martins et al. (2011) showed that the alien grass M. minutiflora increased the above-ground biomass of Cerrado vegetation by 40% over a 2 yr period. Furthermore, a global meta-analysis showed that alien plant species often increase the productivity of invaded communities (Vilà et al., 2011), while an experimental study in a North American prairie showed that, under the same nutrient conditions, most of the 20 alien species investigated produced more biomass than native species (Wilsey et al., 2009). If alien plants can increase biomass because of a more efficient use of N (higher NUE), rather than a higher P efficiency (PUE), it could explain our observed patterns between foliar N : P and occurrence of alien species. Furthermore, since alien plants likely produce litter with a low N : P ratio, this might feed back on a low N : P ratio of availabilities in soil, and on their invasion success. The speculations about NUE, PUE and plant invasions, however, require further study.
In conclusion, our results show that endangered and alien invasive grasses exhibit contrasting distributions in relation to the biomass of vegetation and its N : P status. They also suggest that there is a threshold amount of above-ground biomass (1 kg m−2) beyond which endangered native species can no longer persist. Depending on the type of nutrient limitation at a particular site, this threshold could be exceeded if the soil is enriched with either N or P. Alien species perform well under conditions of relative N limitation, which could be a result of both an efficient use of N and an ability to respond well to increased P supply. In addition, there is some evidence that even without nutrient enrichment, the alien species M. minutiflora may increase above-ground biomass to an amount at which endangered species can no longer persist. More information is needed on this point, however, as well as on whether alien species can alter N : P stoichiometry under unfertilized natural conditions.
We greatly acknowledge the help of Prof. Cássia Munhoz, Maria Aparecida da Silva and Renata Martins with finding the endangered plants and selecting the sites in Brasília. Sabine Güsewell is gratefully acknowledged for her advice concerning data analysis, and we also thank her, Jake Alexander, Deborah Scharfy, Richard Norby and the reviewers for their helpful comments on the manuscript. We wish to thank Chesterton Eugênio and Ana Carolina Corrêa for their valuable help with plant identification. We thank Betânia Góes and Gustavo Dauster for giving us permission to work at the nature reserves. This study was funded by the ETH Zürich North-South Centre, University of Brasília, and by the Swiss National Science Foundation (grant 31003A_122563).