Natives and non‐natives plants show different responses to elevation and disturbance on the tropical high Andes of Ecuador

Abstract The aim was to assess patterns of plant diversity in response to elevation and disturbance in a tropical mountain. The study area was located in north‐central portion of the Eastern Cordillera of the Ecuadorian Andes, on a road from 1,150 m a.s.l. (Osayacu) to 4,000 (Papallacta). Along a mountain road spanning a wide altitudinal gradient, at 20 elevations we sampled three plots: one at the roadside and two perpendicular to the roadside. The relationship between elevation and species richness was assessed using linear and quadratic regressions, the effect of disturbance on species richness was determined by ANCOVA and a t test with parameters obtained from quadratic equations. Similarity of species composition among the roadside and sites distant was evaluated with the Chao‐Jaccard and classic Jaccard similarity indices, the distribution of non‐native species according to their origin were analyzed with linear and quadratic regression. The native species showed a linearly monotonic decrease with elevation, whereas non‐natives showed a quadratic distribution. Disturbed areas had the greatest number of non‐native species and lower native species richness, showing also a high floristic similarity; less disturbed areas showed the opposite. The non‐native species of temperate origin were more numerous and showed unimodal elevational distribution, while species of tropical origin were few and decreased linearly with elevation. We conclude that in a tropical highland mountain range, native and non‐native plant species respond differently to elevation: native species exhibit a monotonically linear decrease, and non‐native species show a unimodal trend. Disturbance positively affects non‐native species showing higher richness and fewer species turnover. In addition, the non‐native species are located along of the elevational gradient in relation to their biogeographic origin.

including minimal effects of non-native invasive species, compared to fertile plains or urban areas, over these there is growing evidence that many non-native species have started to colonize mountain regions worldwide (e.g., Alexander et al., 2011;Becker, Dietz, Billeter, Buschmann, & Edwards, 2005;McDougall et al., 2011;Morueta-Holme et al., 2015;Seipel et al., 2012). Elevational gradients on mountain areas are usually comprised by a series of complex biotic and abiotic changes that affect species distribution (McCain & Grytnes, 2010). For example, it is well known that favorable temperatures for growth decrease with elevation acting as a filter for the richness and abundance of plant species either native and non-native (e.g., Nogués-Bravo et al. 2008, Kessler et al. 2011Marini et al., 2011;Alexander et al., 2011;Seipel et al., 2012). However, the response of native plants to elevation differ from non-natives: While the non-natives species typically decline to long elevations where the elevation acts as a filter for species that do not possess the proper adaptations to settle successfully in extreme conditions, native species tend to have their specific niches adaptations along the elevational gradient . Few studies have explicitly compared the distribution of both native and no-native species in communities that have little or a lot of disturbance, and how these communities are structured along the same elevational gradient (Marini et al., 2011).
In tropical mountains, the establishment of non-native species along an elevational gradient depends on the degree of preadaptation and plasticity inherent in species for the abiotic conditions at a given site Haider et al., 2010;Seipel et al., 2012Seipel et al., , 2015. A match between the overall climatic conditions on the native range and the elevational location on the introduced range can be expected. Thus, according to their biogeographical origin, non-native species can occupy different elevations along an elevational gradient in a tropical site, depending on the similarity between the climatic conditions of the native range and the site of introduction (Arévalo et al., 2005;Arteaga et al., 2009;Haider et al., 2010). Thus, based on biogeography, we would predict that species from tropical and subtropical origin, for example, will be present in the lower elevations of a gradient in a tropical Mountain (Arévalo et al., 2005;Arteaga et al., 2009;Daehler, 2005;Wester & Juvik, 1983). On the other hand, patterns of disturbance along an elevation gradient may uniquely shape distribution. It is well known that disturbances facilitate the arrival and dispersal of propagules of non-native species, which potentially could exhibit invasive features on these fragile ecosystems being a significant threat to the ecosystems services they provide (Alexander, Naylor, Poll, Edwards, & Dietz, 2009;Arévalo et al., 2005;Arteaga et al., 2009;Becker et al., 2005;Parendes & Jones, 2000;Pauchard & Alaback, 2004;Pauchard et al., 2009).
While it has been widely reported that anthropogenic disturbances affect the presence and abundance of non-native species (Catford et al., 2012;Pollnac, 2012;Rejmánek, 1989), little is known on disturbance gradients in tropical mountains. In the tropics, the land-use change, the human settlements, and their access to the mountains are associated with disturbances that could be influencing the distribution of non-native species. Thus, it is important to evaluate how these communities are structured.
To gain insights about the factors affecting the presence and distribution of non-native species along an elevational gradient in a tropical continental mountain, we analyzed the elevational distribution of both native and non-native species in the tropical High Andes of Ecuador.
To understand the species responses to increasing elevation and disturbance, we compared the elevational distribution of both native and non-native species away from and at the roadside as indicators of undisturbed and disturbed habitat, respectively. In particular, we assessed: 1) the distributional pattern of native and non-native species with elevation; 2) the variation in richness and composition of native and non-native species with respect to disturbance (distance to road) along an elevational gradient; and 3) the distribution of non-native species along the elevational gradient respective to its biogeographic origin.
Along the elevational gradient, we established 20 transects, located every 150-m elevation ( Figure 1b). The location of transects along the road (left or right) was determined in the field, depending on accessibility for vegetation sampling.
At each transect, three 2 × 50 m 2 plots were laid out in a T-shaped transect, so that one plot laid parallel to the road, centered on the sampling location.
The other two plots extended perpendicularly from the road (centered at the sampling location), and midpoints of the plots were 25 and 75 m from the roadside. Each plot was divided into ten 2 × 5 m 2 subplots where we recorded the presence and cover of native and non-

| Data analysis
The relationship between species richness of native and non-native species with elevation was assessed by fitting linear and quadratic regressions models and according to the R square obtained we selected that with the higher R square. To assess the relationship between the distribution of non-native species along the elevational gradient with their biogeographic status, we also used linear and quadratic regression models for tropical and temperate origin.
Analysis of covariance (ANCOVA) was used to assess if distance from the road (a proxy of disturbance) affect native species richness across elevations, where native species richness was the dependent variable, distance from the road (roadside, medium-distance, and distant from the road plots) was the independent variable and elevation as covariate. As the elevational distribution of non-native species followed a hump-shaped model (see results), to assess if disturbance affects the elevational distribution of non-native species, for each distance to the road we obtained the values and standard deviations of the parameters of a quadratic model (ax² + bx + c) and were then compared among them with Student's t tests. To be rigorous, we considered that there were significant differences among distance to roads when all the quadratic model parameters (a, b, and c) showed significant differences.
To compare the similarity between the composition of native and non-native species among the different plots, we used Chao-Jaccard (based on abundance) and classic Jaccard (presence-absence species) indices of similarity.
All statistical analyzes were performed at 95% confidence level, using the softwares EstimateS 8.20 and SPSS 15.0.

| Total diversity
We found a total of 771 species, where 728 were native species and 43 non-native species (see Appendix 1). These species belonged to 120 families where the most diverse were as follows: Asteraceae (11%) and Poaceae (7%). The number of genera found was 420, where the most diverse were Solanum (21 species), Miconia (3%) for native species, and Trifolium (7%) for non-native species. The most abundant native species across the gradient was Baccharis latifolia (2%) and the non-native species Pennisetum clandestinum (12%). The non-native species present in a higher number of transects were as follows: Conyza bonariensis (18 transects), Cerastium glomeratum (14 transects), Holcus lanatus (14 transects), and Taraxacum officinale (13 transects).
The herbaceous habit was the most frequent for both native (365 species) and non-native species (40 species), followed by the shrubs (196 native and three non-native species).

| Elevational distribution patterns
Compared with non-native species, the number of native species was always higher at all elevations. Two clearly defined distribution patterns were found along the gradient for native and non-native species, respectively (Table 2). For native species, we found a linear monotonic decline (R² = 0.62, p < .001). In contrast, for non-native species, we found a unimodal or hump-shaped pattern (quadratic), with an increase in the number of species in the middle areas of the gradient (R² = 0.58, p < .001) (Figure 2a). In addition, the proportion of nonnative over native species also showed a quadratic trend (R² = 0.46, p = .002) ( Figure 2b, Table 2).

| Disturbance and species distribution
All plots located at the roadside contained non-native species and always showed the highest level of disturbance (80% very disturbed),  Table 2). Student's t test for fit parameters to quadratic model indicated significant differences on species richness-elevation relationship among the three group of plots (roadside, near, and far) (Table 3), with non-native species decreasing in number with increasing distance from the road across elevations.
Similarity analysis with Chao-Jaccard and Jaccard indices showed the same consistent results, the variation in floristic composition between roadside and far from the road communities across the elevational gradient. The roadside plots always showed greater similarity in the composition of native and non-native species than in the plots far from road, where the similarity was lower (Figure 4).

| Origin of non-native species in the gradient
Thirty-two non-native species (74%) came from temperate regions, whereas eight species (19%) were from tropical regions and only three species (7%) had a cosmopolitan distribution. The results of the regression models for these species showed a quadratic trend for the group from temperate regions, with an increase in the number of species at middle elevations. In contrast, species of tropical origin showed linear monotonic decrease with elevations, whereas the cosmopolitan showed no trends (Figure 5a, Table 2).
Non-native species of temperate origin showed wider elevational ranges than tropical ones, most located between 2,050 and 3,250 m a.s.l.; Poa annua had the broadest elevational range 1,600-4,000 m a.s.l.,

| DISCUSSION
We found that along a wide elevational gradient located in a tropical continental area of South America, native and non-native plant species showed different distributional patterns that also differ from what has been reported for temperate areas. Disturbance, assessed as the distance to road, and the temperate or tropical origin of non-native species also play important roles in determining the species richness and the shape of the elevational distribution of non-native species.

| Elevational distribution patterns
The diversity of native species found in this study is remarkably high .
T A B L E 2 Results of regression models with respect species richness, native, and non-native, with the elevation Some studies show that in tropical regions, the elevational distribution of species, seemed to follows a hump-shaped curve (Balslev 1988, Arévalo et al., 2005Pauchard et al., 2009;Kessler et al., 2011). For the Eastern Cordillera of Ecuador, the highest number of species has been found at middle elevation (between 1,000-1,500 m a.s.l.), declines from 1,000 m a.s.l. to lowlands and from 1,500 m a.s.l. to higher elevations (Gentry, 1995;Jørgensen & León-Yánez, 1999). Balslev (1988) when conducting a regional assessment of plant species richness in Ecuador found that the largest number of species is found between 900 and 3,000 m a.s.l. However, in the present study, we found a linear monotonic decrease of native species, and this pattern was generated due to the elevational extent that comprised our gradient. The lowest elevation we studied was 1,150 m a.s.l., located mostly in the Humid Temperate Region, where is the highest diversity of Ecuador, with intermediate temperatures and high humidity (Cañadas, 1983).
Nogués-Bravo et al. (2008) reported that the form of the elevational F I G U R E 2 (a) Distribution patterns of native and non-native species in the T transects along the elevational gradient.
(b) Proportion of non-native and native species along elevational gradient (regression results in Table 2) gradient of species richness depends not only on the scale, but also on the length of the gradient (see also Kessler et al., 2011). Thus, the inclusion of sites from lower elevations might produce that expected hump-shaped pattern for native species. Notwithstanding, the fact that native species richness showed a monotonic decline with elevation points to the elevational increase in climate harshness as an important filter as it has been widely discussed in other studies (Grytnes & Vetaas, 2002;Grytnes et al., 2006;Kluge et al., 2006;Nogués-Bravo et al. 2008, Kessler et al., 2011. The unimodal (hump-shaped) distributional pattern found for nonnative species suggest that the best environmental conditions for their successful establishment are located in middle elevations. On the one hand, the decrease in the number of non-natives toward higher elevations can be easily understood by the climate filter as occurs for natives (see also Alexander et al., 2011); further higher elevations in our study sites corresponded to a protected area with a very sparse human presence, and as determined by Traux, Vankat, and Schaefer (1992) and Pauchard et al. (2009), elevation is associated with a gradient of land use, which is also an important determinant of the distribution of non-native species. On the other hand, the areas with high anthropogenic disturbance due to the presence of human settlements in Ecuador are commonly found around 2,900 m a.s.l.
The decrease in number of non-native toward lower elevation could be related to the biotic resistance theory proposed by Elton F I G U R E 3 (a) Linear regression for native species on the subtransects (roadside, near, far), along of the elevational gradient. (b) Quadratic regression for nonnative species on subtransects (roadside, near and far), along of the elevational gradient (regression results in Table 2) 1958), who determined that the native species-rich communities have more resistance to invasions of exotic species (Rejmánek, 1989).
At low elevations, communities contain more native species, having higher competitive ability and lacking available empty niches for the recruitment of new species; therefore, the number of non-native species that can be established at these low elevation sites will be low (Herben, Mandák, Bímová, & Munzbergová, 2004). Further, the humpshaped pattern could also be related to the fact that most of the nonnative species come from temperate regions, finding their optimum of distribution at medium elevations.

| The effect of disturbance: distance from the road
Throughout the elevational gradient, we found that the roadsides were highly disturbed habitats that favored the establishment of non-native species and the decline of native diversity. This is mainly because in the roadside, there are more disturbances and removal of vegetation due to maintenance of the road and the traffic, with a concomitant increase propagule pressure, generate advantages for invasion of non-natives (Seipel et al., 2012). While increasing distance from the road disturbances decreased promoting communities with fewer non-native species and more native species across elevation. The roadside plots have fewer species turnover between them; therefore, they have a more similar floristic composition, with non-native species that are often present across the entire altitudinal gradient. Among far plots, the floristic similarity is lower, because the identity of the species varies with a greater turnover along the gradient. The roadside habitats are more homogeneous in structure because species adapted to disturbed environments are sustained within its composition. It is well known that roads are promoters for the entry of non-native species (Arteaga et al., 2009;Becker et al., 2005;Otto et al., 2014) and act as biological corridors and dispersal agents of their propagules, providing suitable habitat for their establishment (Pauchard & Alaback, 2004), while in distant habitats the species composition responds to local heterogeneity patterns by low intensity of disturbance. Similar to McDougall et al. (2011), in this study most of the non-native species richness belong to Poaceae and Asteraceae families. Probably F I G U R E 4 Similarity analysis of Chao-Jaccard and classic Jaccard indices, between assemblages of species, native and non-native, in subtransects roadside and far of road, along the elevational gradient F I G U R E 5 (a) Distribution of non-native species respect their bioclimatic origin, located in the elevational gradient (regression results in Table 2). (b) Elevational distribution range of 43 non-native species located in the 20 transects, with their maximum, minimum, and median

| Origin of non-native species in the gradient
this is related to fact that these families contain annuals or perennials herbaceous species with broad range of climate distribution. Our results on the distribution patterns of non-native species along the elevational gradient in relation to their biogeographical origin concurs with other studies (Arévalo et al., 2005;Arteaga et al., 2009;Becker et al., 2005;Daehler, 2005;Pauchard & Alaback, Pauchard et al., 2009). The distribution of non-native species across an elevational gradient is determined by the similarity of climatic conditions between the native range of a non-native species and the area of introduction (Seipel et al., 2012). Accordingly, we found that species from tropical and subtropical areas were located in the lower elevations, while species of temperate origin showed their optimum at medium and high elevations (Becker et al., 2005;Pauchard & Alaback, 2004;Pauchard et al., 2009). When the elevation increases, the species of tropical origin are replaced by species of temperate affinities (Wester & Juvik, 1983  . In addition, the dominance of species from temperate origin could be due to intentional and accidental introductions, mostly after the European colonization (McDougall et al., 2011). Cosmopolitan species were few (3 species) and preferably located in the middle elevational zones, suggesting that there is low propagule pressure for this species in our studied region.
According to Alexander et al. (2011), the climatic niche of some nonnative species may have changed in the introduced range, as seen in species that have expanded their range. In our case, Poa annua that comes from a temperate climate was found from 1,600 to 4,000 m a.s.l., and Conyza bonariensis of tropical origin was found among 1,150 to 3,850 m a.s.l., indicating the ability to colonize zones with different climate conditions than those occurring in their native ranges.

| CONCLUSIONS
Although the diversity of native species was higher than that of nonnative species along the entire tropical elevational gradient studied, non-native species were present in all sites, especially along roadside habitats. This type of disturbance determined plant species distribution across elevations, generating a unique plant assemblage in the roadsides with a higher proportion of non-native plants than less disturbed areas.
While native species have a monotonic linear decline of richness with increasing elevation, non-native species showed a hump-shaped pattern, with increases in richness in middle elevations. These contrasting patterns suggest that factors involved in shaping the distribution of native plants differ from that of the non-natives. This is particularly noticeable at lower elevations where either lower density of human settlements and disturbances, or higher resistance of a highly diverse native flora decrease the presence of non-native species. In the higher end of the gradient, patterns of non-native and native species distributions tend to equate to those found in temperate regions of the world. Both the presence of disturbance and the biogeographic origin of the nonnative species have important influence in their elevational distribution, with European species becoming more relevant at mid elevations and reducing their abundance in the highest areas. While disturbance is a promoter of the success of non-native species, the biogeographic origins appear to play a larger role in determining the elevational range that non-native species can occupy in the mountains of Ecuador.
Our study highlights the importance of having more data on how native and non-native species are distributed in tropical regions of the world. Although there are some commonalities with temperate regions, which have been profusely studied, there are striking differences in how species are distributed in the lower, warmer zones of these mountains. Understanding the interaction between non-native and native plants and the role of elevation and disturbance in shaping their distribution will be critical to better conserve biodiversity in tropical areas under an increasing threat by human development.