Ecography40: 1381-1394, 2017 doi: 10.1111/ecog.02567 © 2016 The Authors. Ecography © 2016 Nordic Society Oikos Subject Editor: Kenneth Feely. Editor-in-Chief: Miguel Araiijo. Accepted 4 November 2016 Latitudinal and altitudinal patterns of plant community diversity on mountain summits across the tropical Andes Francisco Cuesta, Priscilla Muriel, Luis Daniel Llambi, Stephan Halloy, Nikolay Aguirre, Stephan Beck, Julieta Carilla, Rosa Isela Meneses, Soledad Cuello, Alfredo Grau, Luis E. Gámez, Javier Irazábal, Jorge Jácome, Ricardo Jaramillo, Lirey Ramirez, Natalia Samaniego, David Suárez-Duque, Natali Thompson, Alfredo Tupayachi, Paul Viňas, Karina Yager, Maria T. Becerra, Harald Pauli and William D. Gosling F. Cuesta (francisco.cuesta@condesan.org) and W. D. Gosling (http://orcid.org/0000-0001-9903-8401), Palaeoecology and Landscape Ecology, Inst, for Biodiversity and Ecosystem Dynamics (IBED), Univ. of Amsterdam, the Netherlands. FC also at: Biodiversity Dept — Consorcio para el Desarrollo Sostenible de la Ecorregión Andina (CONDESAN). WDG also at: Dept of Environment, Earth and Ecosystems, The Open Univ., Milton Keynes, UK. — P. Muriel, J. Irazábal andR. Jaramillo, Escuela de Ciencias Biológicas, Pontificia Univ. Católica del Ecuador, Ecuador. —I. D. llambi and I. Ramirez, Inst, de Ciencias Ambientales y Ecológicas, Univ. de los Andes, Venezuela. — S. Halloy, Univ. Nacional de Chilecito, Argentina and Ministry for Primary Industries, New Zealand. — N. Aguirre and N. Samaniego, Biodiversity and Ecosystem Services Research Program, Univ. Nacional de loja, Ecuador. — S. Beck, R. I. Meneses and N. Thompson, Herbario la Paz, Univ. Mayor de San Andres, Bolivia. — J. Carilla, S. Cuello and A. Grau, Inst, de Ecología Regional, Univ. Nacional de Tucumdn, Argentina. SC also at: Inst, de Química del Noroeste, CONICET, Univ. Nacional de Tucumdn, Argentina. — I. E. Gámez, Facultad de Ciencias Forestalesy Ambientales, Univ. de los Andes, Venezuela. — J. Jácome, Pontificia Univ. Javeriana, Depto de Biológia, Unidad de Ecología y Sistemática (UNESIS), Colombia. — D. Suárez-Duque, Cooperación Técnica Alemana, GIZ, Ecuador. — P. Viňas, Naturaleza y Cultura Internacionál, Peru. — K. Yager, NASA Goddard Space Flight Center, USA. — M. T. Becerra, Earth Innovation Inst., Colombia. — H. Pauli, Inst, for Interdisciplinary Mountain Research, Austrian Academy of Sciences and Center for Global Change and Sustainability, Univ. of Natural Resources and life Sciences, Austria. The high tropical Andes host one of the richest alpine floras of the world, with exceptionally high levels of endemism and turnover rates. Yet, little is known about the patterns and processes that structure altitudinal and latitudinal variation in plant community diversity. Herein we present the first continental-scale comparative study of plant community diversity on summits of the tropical Andes. Data were obtained from 792 permanent vegetation plots (1 m2) within 50 summits, distributed along a 4200 km transect; summit elevations ranged between 3220 and 5498 m a.s.l. We analyzed the plant community data to assess: 1) differences in species abundance patterns in summits across the region, 2) the role of geographic distance in explaining floristic similarity and 3) the importance of altitudinal and latitudinal environmental gradients in explaining plant community composition and richness. On the basis of species abundance patterns, our summit communities were separated into two major groups: Puna and Paramo. Floristic similarity declined with increasing geographic distance between study-sites, the correlation being stronger in the more insular Paramo than in the Puna (corresponding to higher species turnover rates within the Paramo). Ordination analysis (CCA) showed that precipitation, maximum temperature and rock cover were the strongest predictors of community similarity across all summits. Generalized linear model (GLM) quasi-Poisson regression indicated that across all summits species richness increased with maximum air temperature and above-ground necromass and decreased on summits where scree was the dominant substrate. Our results point to different environmental variables as key factors for explaining vertical and latitudinal species turnover and species richness patterns on high Andean summits, offering a powerful tool to detect contrasting latitudinal and altitudinal effects of climate change across the tropical Andes. The alpine life zone occurs in mountain regions above the climatic upper limit of forests, and is the only life zone that can be found at all latitudes (Smith and Cleef 1988, Körner 2003). The ecosystems contained within the alpine life zone are considered highly sensitive to climate change because their distribution has been closely linked to temperature (Sala et al. 2000, Halloy and Mark 2003, Pauli et al. 2015) and precipitation patterns (Sklenář and Balslev 2005). Furthermore, the restricted altitudinal range that many species occupy within alpine ecosystems has been suggested to enhance sensitivity to climate change (Sklenář and Jorgensen 1999). Therefore, it is anticipated that climate change will force taxa to move upslope, or downslope, or from one slope to another (e.g. west - east), depending 1381 on their ecological preferences (Nagy and Grabherr 2009, Corlett and Westcott 2013, Winkler et al. 2016). However, topographic variations and the predominance of rocky substrates can also generate a mosaic of microclimatic conditions on the slopes of the mountains that could buffer the effect of warming and retard migration (Scherrer and Körner 2011). Consequently, the alpine life zone provides an ideal 'natural laboratory' in which the relationship between vegetation diversity/dynamics and climatic drivers can be studied (Körner 2007). Considerable research effort has been directed towards investigating alpine community assembly on mid- and high-latitude mountains (Körner and Spehn 2002, Körner 2003, Nagy and Grabherr 2009). However, relatively little work has been conducted at low tropical latitudes (Buytaert et al. 2011). The alpine life zone in the tropics is characterized by unique conditions including: 1) high levels of solar radiation, 2) low variability in yearly average temperatures, 3) high variability in daily temperatures, 4) inverted rainfall patterns along the elevation gradient above the treeline, and 5) the occurrence of marked precipitation gradients over short horizontal distances (Hedberg and Hedberg 1979, Halloy 1989, Rundel et al. 1994, Bader et al. 2007). Within the tropics, the largest area of alpine life zone ecosystems occurs in the high Andes. Moreover, the alpine ecosystems in the high tropical Andes are outstandingly rich in plant species with a high level of endemism (Simpson and Todzia 1990, Sklenář et al. 2014). Extending more than 4500 km in a north-south direction, the tropical Andes show a prominent precipitation and temperature amplitude gradient (Fjeldsá and Krabbe 1990) from the humid equatorial Andes to the xeric environments in the central Andes (Josse et al. 2011). The north-south precipitation/temperature gradient is paralleled by a remarkable change in topographic configuration from high fragmentation and continental insularity in the northern Andes to the large continuous alpine zone of the central Andes, including a high-elevation plateau (the Altiplano). Previous research that has encompassed the full latitudinal extent of the high tropical Andes has mainly focused on the historical and evolutionary phytogeography of the flora (van der Hammen 1974, Simpson 1983, Simpson and Todzia 1990, Young et al. 2002). Consequently, little is known about the patterns and processes that influence plant community composition, species abundance patterns and species richness, and their relation with the environmental gradients that characterize tropical alpine ecosystems (but see Peyre 2015 and Peyre et al. 2015 for the Páramo vegetation). While some studies have evaluated large scale patterns in plant diversity by comparing country-scale floras (Jorgensen et al. 2011), the few studies based on community-scale field plots have been restricted to specific countries or regions (Sklenář and Ramsay 2001, Sklenář and Balslev 2005, Moscol and Cleef 2009, Londono et al. 2014). Specifically, there is a lack of community data related to the vegetation on mountain summits in the tropical Andes, particularly in the subnival and nival belts, where vegetation could be more sensitive and exposed to climate change (Pauli et al. 2015). In addition, the plant communities at high elevation summits in the tropical Andes are the result of recent and little understood community assembly processes after the last glacial maximum (Hansen et al. 1984, Vuille et al. 2008, Williams et al. 2011), and, in general, they have been affected less by human land use than zonal vegetation at lower elevations, allowing the analysis of the relationship between vegetation patterns and environmental drivers (Aldenderfer 2006, Llambi and Cuesta 2013). The tropical Andes have been generally divided into a northern and a central section, in a sector where the chain is bisected by the Huancabamba depression in northern Peru at approximately 6°S (Clapperton and Clapperton 1993). This depression constitutes a barrier that divides biogeographi-cally the Andes into two major phytogeographic units (ecosystems, hereafter): the Páramo, and the Puna (Cabrera and Willink 1973, Cleef 1979, Duellman 1979, Simpson 1983, Myers et al. 2000, Weigend 2002). The proposed classifications are based primarily on physiognomic analyzes that in many cases are correlated with the major climatic provinces of the Andes (Troll 1973). However, the Puna and Páramo ecosystems share many floristic elements and show a gradual change in community composition along the latitudinal gradient (Young et al. 2002). Although, a similar spatial pattern is expected in the plant community configuration of the high Andean summits, very few studies have focused on these communities, and those available are centered in a specific country (Sklenář and Balslev 2005) or region (Peyre 2015, Peyre et al. 2015). Our general objective in the present study was to characterize patterns of change in plant community composition and structure in Puna and Páramo summits across the high tropical Andes using plant abundance data collected in permanent plots. Specifically, our aims were to: 1) assess the importance of geographic proximity and environmental gradients in explaining plant community patterns and beta diversity in all summits taken together and in Puna and Páramo summits analyzed separately and 2) to analyze the relationship between species richness, elevation and environmental drivers across all studied summits and in the Puna and Páramo summits. To address these objectives, we conducted a continental-scale comparative study of plant community composition, species abundance patterns and species richness in summits along the full extent of high tropical Andes, from Venezuela (8°N) to northwestern Argentina (26°S). We used a standardized methodology to analyze data from 50 summits, located from just above the treeline to the upper limit of vegetation in the nival belt. The base-line data was collected in the context of the GLORIA-Andes network (Cuesta et al. 2012) for long-term biodiversity and climate monitoring, following the methodology of Pauli et al. (2015) for the global GLORIA network. Although the GLORIA protocol was designed for monitoring purposes and not explicitly for studying large-scale vegetation patterns, the GLORIA-Andes network has generated the first comprehensive, continent-wide dataset from long-term vegetation study plots suitable for analyzing plant community dynamics in both Punas and Paramos. The analysis of present-day patterns of summit plant communities is a valuable step forward for understanding their possible responses to climate change across the wide environmental gradients that characterize the tropical Andes. 1382 Material and methods Study area The tropical Andes are the longest and widest high mountain region in the tropics, covering more than 1.5 million km2 (11°N to 27° S) and containing complex topographic gradients (ca 600 m up to almost 7000 m a.s.l.) (Josse et al. 2011). The tropical Andes show a predominant north-south humidity gradient with specific localities of high annual rainfall along the Eastern Andean ridge in Peru and Bolivia (Killeen et al. 2007). However, at smaller spatial scales humidity exhibits highly heterogeneous patterns, in which over short distances conditions may vary from per-humid to semi-arid (Kessler et al. 2011). The actual treeline lies around 3500 m (± 400 m) near the equator and gradually decreases southwards (Körner 1998) to as low as 2500 m in the drier western versant of the central Andes (Navarro and Maldonado 2002, Beck 2003). Previous studies divided the Paramo into four zones or altitudinal belts (Cleef 1981, van der Hammen and Cleef 1986, Ramsay and Oxley 1997): 1) the Subparamo or sub-alpine belt extends from the treeline up to 3500 (± 400) forming an ecotone between the montane forest and the alpine ecosystems, 2) the alpine or Paramo belt extends from the Subparamo up to 4200 m (± 200), 3) the subnival or Superparamo belt extends up to 4600 m (± 100), and 4) from this point onwards the nival belt occurs (~ 5000 m ± 200). The alpine belt of the central Andes covers a larger area (~ 283 865 km2) than the northern section (~ 35 000 km2), has higher elevations, and is characterized by two major vegetation types reflecting a gradual increase in the length of the dry season towards the south of the Andean chain, the Mesic and the Xeric Puna (Josse et al. 2011). As in the Paramo, both Puna regions have been previously classified in altitudinal belts (Navarro and Maldonado 2002): 1) the alpine or Altiplano belt, between 3700 and 4400 m (± 200 m), 2) the subnival or Altoandino belt up to 5000 m (± 200) and 3) the nival belt above 5000 m. Selected study sites The regional initiative 'climate change impacts on high-mountain biodiversity in the Andean region (Cuesta et al. 2012, ) promotes the establishment of long-term biodiversity-climate monitoring sites in the high tropical Andes. Here we analyze the vegetation baseline data from 13 of those sites, which comprise 50 summits along a 4200 km latitudinal gradient from 8°54'56"N (Mérida, Venezuela) to 26°4l'24"S (Tucuman, Argentina), and with an elevation range from 3220 m (Piura, Peru) to 5498 m (Cuzco, Peru; Fig. lb; Supplementary material Appendix 1, Table Al). The 50 summits selected were located across the latitudinal and elevation gradients with 26 summits in the Páramo, including 7 summits in the sub-alpine belt, 5 summits in the alpine belt, 10 in the subnival belt and 4 in the nival belt. The remaining 24 summits were located in the Puna with 4 summits in the alpine, 12 in the subnival and 8 in the nival belts. The vegetation data were collected between 2012 and 2013 during the most favorable season for plant flowering (i.e. when most plants can be identified) for each site. Sampling procedure Each site is composed of four summits, except for Podocarpus (Ecuador) and La Culata-Piedras Blancas (Venezuela), with three summits each. We selected summits with different elevations at each site following the Gloria protocol (Pauli et al. 2015), from the subalpine to the nival belt (separated by an average elevation of 178 ± 90 m), excluding summits with permanent glaciers. On each summit, we established sixteen 1 X 1 m permanent vegetation plots, four on each cardinal orientation, at the corners of a 3 X 3 m quadrat design (Pauli et al. 2015). In each plot, we visually estimated the cover of each vascular plant species and of different substrate types (rock, scree, bare soil, above-ground decomposing necro-mass and total vascular plant cover). At the summit level, we recorded a complete plant species lists within the area included from 10 m below the summit up to the top of the summit. The summit area sampled ranged from 0.08 ha to 1.3 ha (mean = 0.4 ha, SD ± 0.27 ha). At each site, a reference collection was compiled, and the species were identified by the site taxonomist and the plant specialists of the corresponding associated herbaria (Supplementary material Appendix 1, Table Al). Additionally, we compared and validated the taxonomic information from all sites during two taxonomic workshops, held during 2012 in Quito (QCA Herbarium-Catholic Univ., Paramo sites) and La Paz (LPB Herbarium-San Andres Univ., Puna sites). The voucher samples were contrasted among sites, and against the reference collections of both herbaria where the workshops took place. As a result of these meetings, we consolidated two regional reference collections for the central and northern Andes sites at LPB Herbarium in La Paz (Bolivia) and QCA Herbarium in Quito (Ecuador), respectively. Of the 968 morpho-species recorded, 505 taxa were identified at the species level, whereas 145 were identified at the genus level and 38 at the family level (Supplementary material Appendix 2). Currently the dataset contains 15 789 records from 920 permanent 1X1 plots, 17 681 presence/absence records from 50 summits and contains information of 17 sites, yet only 13 of those where used for this publication as the other sites were installed later, or their data sets were not ready for analysis at the time this paper was drafted. Data analysis Community composition We analyzed patterns of community composition (species abundance patterns) along altitudinal and latitudinal gradients, for which we constructed a matrix of the average cover of each vascular plant species, averaged over the sixteen 1 X 1 m quadrats in each of the 50 summits. We standardized and squared root transformed species cover, to increase the weight of low-abundance species in the analysis (i.e. 10% of the species had a mean coverage higher than 3%). We then constructed a between-summits similarity matrix using Bray Curtis as floristic similarity metric and performed 1383 (b) h 9 Caracas 1 ' VENEZUELA ilk? \___ ^ ,. • Bogota ": COLOMBIA < !r/ 4^>,Quito ; c ECUADOR 7* PERU i • Lima em BOLIVIA sm ^ % La Paz ,! Phyto region 0 Northern Paramo 0 Central Paramo 9 Southern Paramo □ Mesophytic Puna ■ Xerophytic Puna CHILE / xx Scale: 0 250 500 10D0Km ARGENTINA 40 60 -. .. .... 80-W 70- W 60" W 9mlanty(%) Figure 1. (a) Hierarchical cluster analysis (group average linkage) based on the Bray-Curtis similarity in plant species cover among the 50 studied summits. The Moraroni summit (APLMOR) was not included in the analysis since no vascular plants have been recorded, (b) Study area and distribution of the 13 monitoring sites that include the 50 studied summits. Each site comprises four summits except for Sierra de la Culata (Ve) and Podocarpus (Ec), with three summits each. The full names corresponding to each summit code are indicated in Supplementary material Appendix 1. a hierarchical agglomerative cluster analysis using a group average linkage procedure (Primer ver. 6; Clarke and Gorley 2006). The Jasasuni summit (TUCJAS) was not included in the analysis since no vascular plants have been recorded up to 2012 (only bryophytes). Based on the cluster analysis, we used a similarity percentage procedure (SIMPER, Primer ver. 6) to determine which plant species characterized the groups in each node (up to a 50% contribution to the within group cumulative similarity). Geographic distance and community turn-over We analyzed the importance of geographic distance in explaining floristic distance between sites; for this, a presence-absence species matrix for each site was built based on the complete inventory carried out on each summit (see sampling procedure). Then, we calculated a Jaccard similarity matrix (at the species and genus level) between sites, which was correlated (Spearman rank) with a geographic distance matrix using the RELATE procedure in Primer ver. 6, and evaluated the significance of these correlations using a permutation procedure (9999 permutations). In addition, plant community turnover (i.e. beta diversity) within the Paramo sites and within the Puna sites was quantified using the Whittaker index (Koleff et al. 2003) as follows: S K = - (1) oc where, S is the total number of species recorded for both sites (S = a + b + c); component a comprises the number of shared species between both sites, and components b and c comprise the number of species present exclusively in each of the two sites; « is the average number of species found within the sites. The index was calculated for the Paramo and Puna sites separately in Past ver. 3.11 (Hammer etal. 2001). Influence of habitat variables on species abundance patterns We performed a canonical correspondence analysis (CCA) to assess the relationships between plant community structure and habitat variables, following Legendre and Anderson (1999). The analysis was performed in Past ver. 3.11 (Hammer et al. 2001), for all summits and for the Paramo and Puna sites separately. 1384 The external environmental matrix used for the direct gradient analysis contained fine and broad-scale habitat variables. We collected fine-scale habitat data on average superficial substrate-type cover (rock, scree, bare-ground, above-ground decomposing necromass or litter) at each summit using the methods described above as part of the vegetation baseline (within the sixteen 1 X 1 m quadrats). The broad-scale habitat variables included climatic variables derived from WorldClim gridded datasets (Hijmans et al. 2005), extracted using the elevation and geographic location of each summit; these included: 1) total annual precipitation, 2) minimum temperature of the coldest month and 3) maximum temperature of the warmest month. To improve the characterization of the climatic conditions on each summit, the WorldClim temperature maps were downscaled from a 1 km to 90 m resolution using SRTM elevation data and a lapse rate of 5.4°C knr1 (Bush et al. 2004, Ruiz et al. 2012). Precipitation maps were kept at 1km resolution due to interpolation limitations, since rainfall patterns in the Andes do not co-vary with elevation following consistent patterns across the region (Urrutia and Vuille 2009, Buytaert et al. 2010). Patterns of species richness We explored which environmental variables were important in explaining species richness patterns for all summits and for the Puna and Paramo summits analyzed separately. Richness was calculated as the total number of species found in the sixteen quadrats established in each summit, so that the sample area was the same in all summits. Some of the predictor environmental variables were log or square root transformed to address non-normality (Shapiro-Wilk test). We applied a quasi-Poisson generalized linear model (GLM), with species richness as the dependent variable, following Quinn and Keough (2002). Poisson GLM models were not used as they showed over-dispersion of model residuals (i.e. the residual deviance was larger than the available degrees of freedom). In all cases we selected the best model in terms of explanatory power based on the Adjusted Akaike's information criterion (AICc). Given the possible lack of independence of species richness in summits within each of the 13 sites, we evaluated the existence of spatial autocorrelation of the regression residuals (Moran's I index) for the best model in each case, using SAM ver. 4.0 (Rangel et al. 2010). Quasi-Poisson GLM regression was also used to analyze the relationship between species richness and summit elevation across all summits and in the Puna and Paramo summits separately. No significant spatial autocorrelation of model residuals was found in any of the GLM analyses carried out (a = 0.05). Data available from the Dryad Digital Repository: < http:// dx.doi.org/10.5061/dryad.6qfl0> (Cuesta et al. 2016). Results Floristic composition A total of 968 species, 269 genera and 76 families of vascular plants were identified in the entire summit sections surveyed (702 species were recorded within the 1 m2 quadrats). Eleven families, from a total of 76, contained 67% of all the species recorded. Asteraceae (250 spp.) and Poaceae (161 spp.) were the most diverse families (Supplementary material Appendix 2, Table A2). In the Puna 24 summits included in the study, the vegetation was composed of 45 families (133 genera, 443 species) of which 6 families contributed 72% of the species reported. The Puna families with the highest number of species were Asteraceae (128 spp.), Poaceae (95 spp.), Caryophyllaceae (30 spp.) and Brassicaceae (28 spp., Fig. 2a). In the 26 Paramo summits, a total of 63 families, 192 genera and 548 species were registered. In this case, the species were more evenly distributed between families. Eight Paramo families comprised nearly 57% of all the species recorded; the most diverse were: Asteraceae (127 spp.), Poaceae (69 spp.), and Orchidaceae (24 spp.) (Fig. 2a). At the genus level, 20 out of 268 genera contained 33% of the total vascular plant diversity. Senecio, Deyeuxia and Poa were the richest genera, comprising 104 species altogether. The Puna summits were dominated in terms of cover by Deyeuxia, Senecio, Poa and Nototriche, whereas, in the Paramo summits the most abundant genera in terms of cover were Diplostephium, Lachemilla, Senecio, and Calamagrostis (Fig. 2b; Supplementary material Appendix 2, Table A2). Cluster analysis based on Bray Curtis floristic similarity clearly separated the Paramo summits from their Puna counterparts (Fig. la). The species involved in separating the Paramo summits from the rest included Calamagrostis intermedia, Pernettya prostrata, Hypochaeris sessiliflora and Disterigma empetrifolium among others; the Puna node showed high abundance values of Deyeuxia lagurus, Festuca orthophylla, Belloa schultzii and Senecio adenophyllus (Table 1). Within the Paramo, three floristic summit groups were defined along a latitudinal gradient (Fig. la): 1) the northern Paramo (Venezuela and Colombia), 2) the central Paramo (northern and central Ecuador) and 3) the southern Paramo (southern Ecuador and northern Peru). The first group was defined by the dominance of Espeletiopsis colombiana, Castilleja fissifolia, Agrostis boyacensis, Coespeletia timotensis and Rumex acetosella (an exotic herb, which is particularly abundant in two of the Venezuelan summits). The second group was dominated by Calamagrostis intermedia, Hypochaeris sessiliflora, Baccharis caespitosa, Azorella aretioides and Xenophyllum humile. The third group showed high abundances of Hypericum lancioides, Vaccinium floribun-dum, Rhynchospora vulcani, Disterigma empetrifolium and Calamagrostis tarmensis. The Puna summits were subdivided in two sub-groups (Fig. la): 1) the xeric summits of northern Argentina and western Bolivia (Sajama) were separated from, 2) the mesic summits of eastern Bolivia and Southern Peru (Sibinacocha). The xeric summits were characterized by a predominance of Festuca orthophylla, Deyeuxia lagurus, Azorella compacta and Mulinum axilliflorum, while the mesic summits were dominated by Stipa hans-meyeri, Nototriche obcuneata, Pycnophyllum tetrastichum, Poa gymnantha and Deyeuxia minima. Geographic distance and community turn-over The matrix correlation analysis between the floristic and the geographic distance produced a Rho of 0.775 (p = 0.0001). When we analyzed the Paramo summits separately, the Rho 1385 (a) 130 120 110 100 H 90 80 70 60 -I 50 40 30 -20 10 - I Ln J j I I 2 o Q. cd C3 o £■ ra O in o OH o cd cd cd LU (b) 36 32 ■ 28 v. 24 o 0.05). Discussion Community composition At the family level, many elements were common between Puna and Páramo summits; for example, Asteraceae and Poaceae were the dominant components in both environments as previously documented (Simpson and Todzia 1990, Luteyn 1999, Sklenář and Ramsay 2001, Peyre 2015). A few genera contained most of the species present in the Puna and Páramo summits (Fig. 2). High species richness within genera has also been reported for the Páramo flora of the eastern cordillera (Simpson and Todzia, 1990) and for the superpáramo of Ecuador (Sklenář and Balslev 2005). Simpson and Todzia (1990) and Sklenář and Balslev (2005) 1387 Table 2. Plant community turnover (i.e. Beta diversity) between sites based on Whittaker's index (Koleff et al. 2003) using (a) species data, and (b) genus data. CCy = Cocuy Ang = El Angel, Ant = Antisana, Pic = Pichincha, Pnp = Podocarpus, Pac = Pacaipampa, Cpb = Piedras Biancas, Ans = Abra del Acay Cue = Cumbres Calchaqufes, Apl = Apolobamba, Saj = Sajama, Tue = Tuni Condoriri, Sib = Sibinacocha. (a) Páramo sites Ccy Ang Ant Pic Pnp Pac Cpb Ccy 0 0.87597 0.93778 0.93133 0.9542 0.89764 0.82648 Ang 0.87597 0 0.77436 0.72414 0.88793 0.85714 0.94709 Ant 0.93778 0.77436 0 0.45882 0.9799 0.95812 0.9359 Pic 0.93133 0.72414 0.45882 0 0.94203 0.88945 0.92683 Pnp 0.9542 0.88793 0.9799 0.94203 0 0.74561 0.98964 Pac 0.89764 0.85714 0.95812 0.88945 0.74561 0 0.95676 Cpb 0.82648 0.94709 0.9359 0.92683 0.98964 0.95676 0 Puna sites Ans Cue Apl Saj Tue Sib Ans 0 0.67136 0.86473 0.73563 0.82659 0.85116 Cue 0.67136 0 0.81203 0.74249 0.80172 0.81752 Apl 0.86473 0.81203 0 0.79736 0.54867 0.64179 Saj 0.73563 0.74249 0.79736 0 0.81347 0.85532 Tue 0.82659 0.80172 0.54867 0.81347 0 0.64957 Sib 0.85116 0.81752 0.64179 0.85532 0.64957 0 (b) Páramo sites Ccy Ang Ant Pic Pnp Pac Cpb Ccy 0 0.42282 0.56522 0.53642 0.6 0.51553 0.43939 Ang 0.42282 0 0.49606 0.41429 0.55556 0.53333 0.57025 Ant 0.56522 0.49606 0 0.24031 0.84962 0.69784 0.58182 Pic 0.53642 0.41429 0.24031 0 0.72603 0.60526 0.57724 Pnp 0.6 0.55556 0.84962 0.72603 0 0.39744 0.81102 Pac 0.51553 0.53333 0.69784 0.60526 0.39744 0 0.71429 Cpb 0.43939 0.57025 0.58182 0.57724 0.81102 0.71429 0 Puna sites Ans Cue Apl Saj Tue Sib Ans 0 0.40299 0.54386 0.42308 0.53398 0.57983 Cue 0.40299 0 0.47368 0.42254 0.44681 0.52866 Apl 0.54386 0.47368 0 0.44262 0.28926 0.28467 Saj 0.42308 0.42254 0.44262 0 0.38739 0.49606 Tue 0.53398 0.44681 0.28926 0.38739 0 0.36508 Sib 0.57983 0.52866 0.28467 0.49606 0.36508 0 proposed that the skewed taxonomie composition towards a few highly diverse genera indicated a high degree of a recent local speciation driven by geographic isolation between high alpine 'continental islands' or complexes. In fact, (Madrifián et al. 2013), report that the average net diversification rates of Páramo plant lineages are faster than any of the documented recent speciation processes in other ecosystems. Patterns of community similarity indicated three groups for the Páramo and two for the Puna summits (Fig. la). These groups were composed of different plant community assemblages, probably as a result of past geological-biogeo-graphical events (Sklenář and Balslev 2005, Sklenář et al. 2014), and current environmental conditions (Jorgensen et al. 2011, Kessler et al. 2011). The differences in plant community composition and abundances between high Andean summits investigated in this study supports previous findings that compared the Puna and Páramo regions (Simpson 1983, Sklenář et al. 2011, Smith and Cleef 1988). However, the high level of community dissimilarity identified between summits could also have been influenced by the limited numbers of study sites together with a relatively low surface area sampled on each summit. If a higher number of more evenly distributed summits across the Andes could be incorporated, a more gradual change in species composition and a more representative view of vegetation patterns along summits in the tropical Andes could be obtained. Incorporate Páramo summits from Colombia and Venezuela (particularly in more humid regions), as well as for the mesic Puna in central Peru would be particularly important, as they are underrepresented in the currently available data set. 1388 Table 3. Correlation coefficients between environmental variables and axis scores from a canonical correspondence analysis (CCA) based on all plant species recorded on 49 high Andean summits. Significant correlations (p<0.05) are represented in bold with an asterisk. Signs reflect arbitrary selection of gradient direction by PAST. All summits Puna summits Páramo summits Variable 1st axis 2nd axis 1 st axis 2nd axis 1 st axis 2nd axis LogPrec -0.845* 0.678* 0.862* 0.258 -0.460 0.476 Tmax 0.646* 0.384 -0.749* 0.186 -0.944* 0.102 Rock 0.565* -0.629* -0.373 -0.438 0.397 -0.626* Bsoil 0.323 -0.636* 0.036 -0.510* 0.491 -0.380 Log(DNec) -0.132 -0.070 -0.518* -0.070 0.317 -0.016 Scree 0.287 0.233 0.178 0.169 0.132 -0.601 * Lat -0.335 0.133 0.242 -0.150 0.098 -0.189 Long 0.174 -0.156 0.010 0.043 0.156 0.135 At a continental scale, floristic similarity between summits declined with geographical distance, as previously documented in the Paramos (Londono et al. 2014, Sklenář et al. 2014). Moreover, there was a higher correlation between floristic and geographic distance in the Paramos than in the Punas at species and genus level. Our results support the idea of continental insularity (i.e. the fragmentation of alpine areas separated from each other by forests) as an important driver of high levels of species turnover, particularly in the northern Andes (Anthelme et al. 2014). The studied Paramo sites showed higher values of community turnover, something that could be linked with the differences in landscape connectivity of high elevation areas between the more insular Paramos and the more continuous Punas. The observed pattern could have important consequences for long-term biodiversity dynamics in high Andean regions, as land use practices and climate anomalies are expected to increase the effects of continental insularity in the coming years, (a) 6 -0.8 AngCpen CcyClnADPIcLdp CcyMolA - - AntCbu PIcChuA* m APIcIng AAntChp •AntChg CDucAlz □SaJPac Cuclsa •SibYur SajSum» «AnsAfi •SibOra »AnsAna Saj Jas ASajHui AnsAbr (b) 2.0 1.5 1J> CucAbDnJCucSh SaJPaCA^ua 0.5 _,_,_^5"*-— •SlbPum •TucPal ATucCop -1.8 -1.2 —TEBLbflriNej/ AnsAnaA fiook ASaJSum -1.0 -1.5 SaJHulftSaJJas •AnsAbr •AplMIt ATucWat 'bsoII •SlbOrq Figure 3. Biplots from canonical correspondence analysis (CCA) based on plant community similarity (Chi-quare distance) in plant species cover among the 49 studied summits and its relation with normalized habitat variables: (a) all summits; (b) Puna summits; (c) Paramo summits. Summits are classified into vegetation belts based on their recorded elevation. Dot-darkblue = subalpine, square-light blue = alpine, triangle-black = subnival, dot-olive = nival. 1389 All summits 4500 5000 Puna summits 3800 4000 4200 4400 4600 4800 5000 5200 5400 5600 Elevation (m) Páramo summits 3000 3500 4000 4500 Elevation (m) 5000 Figure 4. Change in quadrat species richness with elevation in high Andean summits: (a) all summits; (b) Puna summits; (c) Paramo summits. GLM quasi-Poisson regression model fitted to the data in each case. There was no significant spatial autocorrelation of model residuals. especially by reducing the permeability of the lowland matrix that encloses tropical alpine summits (Anthelme et al. 2014). Nevertheless, Sklenář and Balslev (2005) found environmental similarity (i.e. humidity) between superpáramo areas were more important for species distribution than other factors, such as geographic distance. Hence, at smaller geographic scales climatic factors could be more relevant for explaining community assemblages than biogeographic processes. Our results also indicate that processes linked with environmental factors (e.g. correspondences in precipitation regimes) also influence patterns of community similarity at continental scales, particularly across the Puna summits; for example, the xeric summits of Sajama in Bolivia, showed higher community similarity with the xerophytic Puna sites in northern Argentina than with the other, geographically closer, but more mesic sites in Bolivia. Plant communities and the influence of habitat variables The importance of environmental variables in explaining community similarity patterns was supported by results from the CCA. Annual precipitation, maximum temperatures and substrate types, mostly rock and scree, were the most important variables explaining plant community structure across the summits (Table 3), as has been reported for vascular plants at global, regional and local scales (Dufour et al. 2006, Kreft and Jetz 2007 Crous et al. 2013). A strong latitudinal precipitation gradient separated the Paramo summits from their Puna counterparts. This was associated with a stronger thermal seasonality in the Punas, which have a less equatorial distribution than the Paramos. A second, temperature gradient discriminated the summits according to their elevation belt (from the subalpine to the nival), this being particularly evident across the Paramo summits (Fig. 3a). The ordination performed only with the Puna summits confirmed the existence of a marked precipitation gradient that divided the xeric summits from the mesic ones. Further, a maximum temperature gradient separated the alpine summits (all xeric Puna summits) from the subnival and nival counterparts. The CCA analysis performed only for the Paramo summits, diverged from the analysis based purely on floristic information (cluster analysis). In this case, the summits were ordered according to a strong temperature gradient and to a lesser degree following a precipitation gradient (Fig. 3c). Hence, the summits were organized according to their elevation belt, with the subalpine summits (southern Paramos), followed by the alpine, then subnival and then nival summits (irrespective of their geographic distribution along the central and northern Paramos); the Piedras Blancas summits of Venezuela were separated from the rest due to their lower annual precipitation and the predominance of scree and rock as the dominant substrates; the presence of very extensive cattle grazing could also be a factor linked with this separation, which could in turn be linked with the high cover of the exotic species Rumex acetosella in the two lower Venezuelan summits. It is important to take into account the geographic specificities of our summit based data-set for the purpose of analyzing environment—floristic relationships; on the one hand, the lowest (mean = 3416 m) and more humid (mean = 3089 mm yi-1) summits in our Paramo data were geographically clumped, as they were all located in southern Ecuador and northern Peru, where the cordillera is at one of its lowest elevations throughout the Andes. On the other hand, within the xeric Puna we had a lower frequency of nival summits than in the mesic Puna. This obviously 1390 Table 4. Generalized linear model - GLM (quasi-Poisson) - of plant species richness as a function of environmental variables in the summits studied along the tropical Andes and in the Puna and Paramo summits analyzed separately. The parameters selected on the basis of the ACIc criteria for the best model are presented in each case. Significant relations are represented with an asterisk: ***highly significant, "significant, *marginally significant. Summits Parameter Estimate SD error t value P(>|t|) All Intercept 3.371 0.223 15.09 <0.001*** Tmax 0.053 0.016 3.295 0.002** Log(DNec) 0.259 0.125 2.073 0.045* Log(Scree) -0.130 0.030 -4.341 <0.001*** Puna Intercept -1.665 1.524 1.092 0.288 Log(Prec) 1.218 0.379 3.213 0.005** Log(Scree) -0.125 0.054 -2.310 0.032* Tmax 0.205 0.045 4.551 <0.001*** Páramo Intercept 1.683 1.073 1.568 0.133 Log(Scree) -1.147 0.059 -2.503 0.0216* Log(Prec) 0.740 0.324 2.284 0.0341* increased the association between precipitation and temperature in our data, suggesting special care is needed when analyzing the interactions between these two climatic drivers, particularly when the Páramo and Puna summits are analyzed separately. Moreover, climatic variables and substrate types did not account for all of the variability in community composition. This suggests that other factors such as biogeo-graphical isolation, geological/edaphic heterogeneity and landscape history could also be important determinants of changes in vegetation structure (resulting from processes such as different histories of volcanism, glaciation or land use). For example, Londono et al. (2014) reported high levels of flora dissimilarity in Páramo summits disconnected by less than 100 km in Colombia as previously reported for the eastern Andes realm by Cleef (1981). This is also the case in northern Argentina, where the Pampean mountains, of which Cumbres Calchaquies are part, conform entirely distinct floristic compositions on each summit, separated by deep valleys but only a few dozen kilometers apart (Aagesen et al. 2012). Patterns of species richness We found a clear linear decline in species richness with elevation across all summits and in the Puna and Páramo analyzed separately. Moreover, richness was positively associated with maximum temperature in the GLM using environmental variables for all summits (Table 4). A monotonie decrease in species richness with increasing elevation above the treeline has been well documented in many mountain regions, including the Paramos of Colombia (Cleef 1981) and Ecuador (Sklenář and Ramsay 2001). In contrast, a hump-shaped pattern of richness with elevation has also been reported for humid Páramo areas (Sklenář and Balslev 2005). The upper condensation belt in the lower subnival belt (~ 4200 m) creates a humid climate enhancing species richness in these humid regions (Sklenář et al. 2008). As suggested by Körner (2003) the inverse relationship between temperature/elevation and species richness may be partly due to the conical shape of high mountains, which result in an overall decrease in the size of the species pool as the surface area decreases with altitude (other relevant factors include increased isolation of high elevation areas). Even so, the GLM model for species richness in all summits indicated a significant positive relationship with above-ground decomposing necromass (plant litter), which could be interpreted as a result of higher site productivity (Al-Mufti et al. 1977). Moreover, species richness in the summits increased mono-tonically with total plant cover in all summits and in the Puna and Paramo analyzed separately (Table 4). Al of this suggests that environmental factors limiting productivity at high altitudes could also play a role in influencing species richness, as predicted for the low-end of productivity gradients in the classical humpback diversity-productivity model (Grime 1973, see Peyre 2015 for the Paramo region). As for the relationship between precipitation and richness, this variable was not included in the best GLM model for all summits, but it was included as a positive effect in the models for the Puna and Paramo analyzed separately. In the case of the Paramo this could have been influenced by the lower elevation summits (Podocarpus, Ecuador) which show the highest observed species richness and annual rainfall in our data; at the other end, the driest summits in Venezuela, showed some of the lowest species richness values. In the case of the Puna summits, the positive relationship between richness and precipitation could be linked with higher overall richness in the more mesic summits than the more xeric ones (after accounting for differences in temperature/elevation in the GLM). In addition, the regression models for all summits, and for the Puna and Paramo analyzed separately, indicated a significant negative relation between richness and scree cover. Scree and rocks were in fact the dominant substrates in nival and subnival summits, and appear to be indicative of limiting conditions for soil and vegetation development. The predominance of scree should have a direct link with the occurrence of periglacial conditions and frequent freezing temperatures inducing gelifraction processes (Monasterio 1986, Perez 1986). Finally, the decline in species richness in high altitude summits can also be linked with the relatively recent retreat of glaciers and snow covered areas (i.e. after the Little Ice Age, Vuille et al. 2008). These dynamic historical changes in climatic conditions need to also to be considered for explaining species richness patterns on high Andean summits. 1391 Conclusions and implications for climate change research Overall, our results indicate that community structure along tropical high Andean summits is influenced by the complex interaction of biogeographic processes, reflected in the patterns of association of floristic similarity and geographic distribution, and latitudinal/elevation gradients, which are in turn reflected in a strong association of climatic and substrate characteristics with community composition and species richness. The large proportion of species belonging to a few highly diversified genera in combination with a high replacement of species composition between our study summits, is consistent with the general finding that vegetation in the high tropical Andes is characterized by many species with narrow distributions and high species turn-over (Sklenář and Balslev 2005, Sklenář et al. 2011, Anthelme et al. 2014). Consequently, vegetation in our summits might be highly sensitive to the effects of climate change, especially considering that tropical alpine environments are among the terrestrial ecosystems that are likely to experience higher warming rates during this century (Bradley et al. 2006), and there is growing evidence that the rate of warming is amplified with elevation (Mountain Research Initiative 2015). Moreover, species which occupy high summits will have limited space for vertical migration (Ramirez-Villegas et al. 2014), and high Andean plants appear to be shifting rapidly towards their upper elevation limit, as recently documented on the Chimborazo volcano, two centuries after Humboldt's visit (Morueta-Holme et al. 2015). Nevertheless, microclimatic refugia are common in alpine ecosystems and this might help to attenuate the expected trend in species vertical migration (Scherrer and Körner 2011, Valencia et al. 2016). Our results also support the idea of continental insularity as driver of high levels of species turnover, particularly in the northern Andes (Anthelme et al. 2014). Even so, geographic proximity did not account for all the variation in community structure between summits. A general latitudinal gradient of precipitation and minimum air temperatures was strongly associated with differences in community composition between the Paramos and Punas. Variation in species richness across all summits was mainly explained by altitu-dinal changes in maximum air temperature, above-ground necromass, and scree cover. Hence, our results point to different climatic drivers as key factors for explaining both vertical and latitudinal species turnover and species richness patterns in high Andean summits. This could offer a powerful tool to detect contrasting latitudinal and altitudinal effects of climate change across the tropical Andes. In particular, it raises interesting questions regarding the differential effect of temperature vs precipitation on summit community diversity at continental scales (Pauli et al. 2012). For instance, on European GLORIA summits, the documented recent increase in air temperature has been related to a shift towards a more ther-mophilous species composition (Gottfried et al. 2012) and to an increase of species richness on temperate and boreal summits, whereas on the drier Mediterranean summits a reduction in annual rainfall has resulted in stasis or decrease in richness (Pauli et al. 2012). Given that many models predict a reduction in precipitation in the dry central Andes in the next few decades, but not in the northern Andes -except for the drier Venezuelan Paramos (Buytaert et al. 2010, Tovar et al. 2013), the monitoring system established in high Andean summits should provide a sensitive system of permanent sites to evaluate the possibility of diverging impacts of climatic changes on plant diversity and vegetation composition across the region. Acknowledgements — This paper has been developed thanks to the financial support of the Swiss Agency for Development and Cooperation (SDC, < www.eda.admin.ch/sdc>) that supported the GLORIA-Andes network and the baseline establishment of 6 monitoring sites through the CIMA Project conducted by CONDESAN. The German International Cooperation (GIZ, < www.giz.de >) through its Proyecto Tri-Nacional financed the establishment of the Complejo Volcánico Pichincha Site and the taxonomie workshops where the idea of this study was conceived. PM acknowledges the Andean Community (CAN) and its former Proyecto de Adaptación al Impacto del Retroceso Acelerado de Glaciares en los Andes Tropicales, PRAA (contract no. 078-2012-SGCAN) and the Pontificia Univ. Católica del Ecuador (PUCE) for the financial support to implement the Reserva Ecológica Antisana Site (EC-ANT). JC and RG thank the Consejo Nacionál de Ciencias de Investigaciones Científicas y Técnicas of Argentina (CONICET, ) and recognize the financial support of the ALARM Project of the European Commission. JJ recognizes the Alexander von Humboldt Biological Resources Inst, for its support on the establishment of the Parque Natural Cocuy Site. We are grateful for the support of the General Secretariat of the Andean Community and its former Project Programa Regional Andino, funded by the Spanish International Cooperation (AECID, www.aecid.es) together with Conservation International, which helped to create the Gloria-Andes network. All the authors acknowledge the following government agencies for granting permissions to implement the study sites in the Andean Countries and for their support to the GLORIA-Andes Network: Ministerio de Ambiente (Venezuela), Parques Nacionales Natu-rales de Colombia, Ministerio del Ambiente (Ecuador), Servicio Nacionál de Areas Naturales Protegidas por el Estado (SERNANP, Peru), Servicio Nacionál de Areas Protegidas (SERNAP, Bolivia), Dirección de Flora, Fauna Silvestře y Suelos, Ministerio de Desarrollo Productivo, Provincia de Tucumán and Secretaria de Medio Ambiente y Desarrollo Sustentable, Provincia de Salta (Argentina). LDLL is grateful to Nelson Marquez y Teresa Schwarzkopf from the Inst, de Ciencias Ambientales y Ecológicas, Univ. de los Andes (ICAE), and acknowledges the financial support of the Secretaria de Educación Superior, Ciencia, Tecnología e Innovación de la República del Ecuador (SENESCYT), through its Proyecto Prometeo to support his stay in Ecuador. FC has also received additional funding to complete this study from the EcoAndes Project conducted by CONDESAN and UNEP, funded by the Global Environmental Fund (GEF) and from the Andean Forest Program conducted by CONDESAN and Helvetas Swiss Intercooperation, and funded by COSUDE. We gratefully thank Joost Duivenvoorden and Carolina Tovar for their useful comments on early versions of the manuscript. We are grateful with Wouter Buytaert for his support in downscaling the Worldclim gridded data sets. References Aagesen, L. et al. 2012. Areas of endemism in the southern central Andes. - Darwiniana 50: 218-251. 1392 Aldenderfer, M. 2006. Modelling plateau peoples: the early human use of the world's high plateaux. — World Archeol. 38: 357-370. AI-Mufti, M. M. et al. 1977. A quantitative analysis of shoot phenology and dominance in herbaceous vegetation. — J. Ecol. 65: 759-791. Anthelme, F. et al. 2014. Biodiversity patterns and continental insularity in the tropical high Andes. - Arct. Antarct. Alp. Res. 46: 811-828. Bader, M. Y. et al. 2007. High solar radiation hinders tree regeneration above the alpine treeline in northern Ecuador. — Plant Ecol. 191: 33-45. Beck, S. G. 2003. The Paramo of the Bolivian Yungas — a landscape formed by man. - In: Körner, C. and Spehn, E. M. (eds), Linking diversity with fire, grazing and erosion. Unpublished abstracts, p. 8. Bradley, R. et al. 2006. Threats to water supplies in the tropical Andes. - Science 312: 1755-1756. Bush, M. B. et al. 2004. 48 000 years of climate and forest change in a biodiversity hot spot. - Science 303: 827-829. Buytaert, W. et al. 2010. Uncertainties in climate change projections and regional downscaling in the tropical Andes: implications for water resources management. — Hydrol. Earth Syst. Sei. 14: 1247-1258. Buytaert, W. et al. 2011. Potential impacts of climate change on the environmental services of humid tropical alpine regions. - Global Ecol. Biogeogr. 20: 19-33. Cabrera, A. L. and Willink, A. 1973. Biogeografia de America Latina. — Programa Regional de Desarrollo Cienufico y Technolögico, WA. Clapperton, C. M. and Clapperton, C. 1993. Quaternary geology and geomorphology of South America. — Elsevier. Clarke, K. and Gorley, R. 2006. PRIMER ver. 6 user manual and program. - PRIMER-E, UK. Cleef, A. M. 1979. The phytogeographical position of the Neotropical vascular Paramo flora with special reference to the Colombian Cordillera Oriental. - In: Cleef, A. M. et al. (eds), Tropical botany. Academic Press, pp. 175—184. Cleef, A. M. 1981. The vegetation of the Paramos of the Colombian Cordillera Oriental. - Meded. Bot. Mus. Herb. Rijksuniv. Utrecht 481: 1-320. Corlett, R. T. and Westcott, D. A. 2013. Will plant movements keep up with climate change? — Trends Ecol. Evol. 28: 482-488. Crous, C. J. et al. 2013. Associations between plant growth forms and surface rockiness explain plant diversity patterns across an Afro-montane grassland landscape. — S. Afr. J. Bot. 88: 90-95. Cuesta, F. et al. 2012. Biodiversidad y cambio climatico en los Andes tropicales — conformaeiön de una red de investigaeiön para monitorear sus impactos y delinear acciones de adaptation. - Red Gloria-Andes, p. 180. Cuesta, F. et al. 2016. Data from: Latitudinal and altitudinal patterns of plant community diversity on mountain summits across the tropical Andes. — Dryad Digital Repository, . Duellman, W. E. 1979. The South American herpetofauna: its origin, evolution, and dispersal. — Museum of Natural History, Univ. of Kansas. Dufour, A. et al. 2006. Plant species richness and environmental heterogeneity in a mountain landscape: effects of variability and spatial configuration. — Ecography 29: 573—584. Fjeldsä, J. and Krabbe, N. 1990. Birds of the high Andes: a manual to the birds of the temperate zone of the Andes and Patagonia, South America. — Zoological Museum, Univ. of Copenhagen. Gottfried, M. et al. 2012. Continent-wide response of mountain vegetation to climate change. - Nat. Clim. Change 2: 111-115. Grime, J. P. 1973. Competitive exclusion in herbaceous vegetation. - Nature 242: 344-347. Halloy, S. 1989. Altitudinal limits of life in subtropical mountains: what do we know? - Pac. Sei. 43: 170-184. Halloy, S. and Mark, A. 2003. Climate-change effects on alpine plant biodiversity: a New Zealand perspective on quantifying the threat. - Arct. Antarct. Alp. Res. 35: 248-254. Hammer, 0. et al. 2001. Past: paleontological statistics software package for education and data analysis. — Palaeontol. Electronica 4: 1-9. Hansen, B. et al. 1984. Pollen studies in the Junin area, central Peruvian Andes. - Geol. Soc. Am. Bull. 12: 1454-1465. Hedberg, I. and Hedberg, O. 1979. Tropical-alpine life-forms of vascular plants. - Oikos 33: 297-307. Hijmans, R. J. et al. 2005. Very high resolution interpolated climate surfaces for global land areas. — Int. J. Climatol. 25: 1965-1978. Jorgensen, P. M. et al. 2011. Regional patterns of vascular plant diversity and endemism. — In: Herzog, S. et al. (eds), Climate change and biodiversity in the tropical Andes. Inter-American Inst, for Global Change Research (IAI) and scientific committee on problems of the environment (SCOPE), p. 192. Josse, C. et al. 2011. Physical geography and ecosystems in the tropical Andes. — In: Herzog, S. et al. (eds), Climate change and biodiversity in the tropical Andes. Inter-American Inst, for Global Change Research (IAI) and scientific committee on problems of the environment (SCOPE), pp. 152-169. Kessler, M. et al. 2011. Gradients of plant diversity: local patterns and processes. — In: Herzog, S. et al. (eds), Climate change and biodiversity in the tropical Andes. Inter-American Inst, for Global Change Research (IAI) and scientific committee on problems of the environment (SCOPE), pp. 204-219. Killeen, T. J. et al. 2007. Dry spots and wet spots in the Andean hotspot. - J. Biogeogr. 34: 1357-1373. KolefF, P. et al. 2003. Measuring beta diversity for presence—absence data. - J. Anim. Ecol. 72: 367-382. Körner, C. 1998. A re-assessment of high elevation treeline positions and their explanation. — Oecologia 115: 445-459. Körner, C. 2003. Alpine plant life: functional plant ecology of high mountain ecosystems; with 47 tables. — Springer. Körner, C. 2007. The use of altitude' in ecological research. -Trends Ecol. Evol. 22: 569-574. Körner, C. and Spehn, E. 2002. Mountain biodiversity: a global assessment. — CRC Press. Kreft, H. and Jetz, W. 2007. Global patterns and determinants of vascular plant diversity. - Proc. Natl Acad. Sei. USA 104: 5925-5930. Legendre, P. and Anderson, M. J. 1999. Distance-based redundancy analysis: testing multispecies responses in multifactorial ecological experiments. — Ecol. Monogr. 69: 1—24. Llambi, L. D. and Cuesta, F. 2013. La diversidad de los paramos andinos en el espacio y en el tiempo. — In: Cuesta, F. et al. (eds), Avances en investigaeiön para la conservation en los Paramos Andinos. CONDESAN, pp. 7-40. Londono, C. et al. 2014. Angiosperm flora and biogeography of the Paramo region of Colombia, northern Andes. — Flora 209: 81-87. Luteyn, J. 1999. Introduction to the paramo ecosystem. — In: Luteyn, J. (ed.), Paramos: a checklist of plant diversity, geographical distribution and botanical literature. New York Botanical Garden, Brooklyn, pp. 1-39. Madrinan et al. 2013. Paramo is the world's fastest evolving and coolest biodiversity hotspot. — Front. Genet. 4: 192. Monasterio, M. 1986. Adaptive strategies of Espeletia in the Andean desert Paramo. — In: Vuilleumier, F. and Monasterio, M. (eds), High altitude tropical biogeography. Oxford Univ. Press, pp. 49-80. 1393 Morueta-Holme, N. et al. 2015. Strong upslope shifts in Chimborazo's vegetation over two centuries since Humboldt. - Proc. Natl Acad. Sei. USA 112: 12741-12745. Moscol, M. C. and Cleef, A. M. 2009. A phytosociological study of the Paramo along two altitudinal transects in El Carchi province, northern Ecuador. — Phytocoenologia 39: 79-107. Mountain Research Initiative EDW Working Group 2015. Elevation-dependent warming in mountain regions of the world. — Nat. Clim. Change 5: 424^30. Myers, N. et al. 2000. Biodiversity hotspots for conservation priorities. - Nature 403: 853-858. Nagy, L. and Grabherr, G. 2009. The biology of alpine habitats. — Oxford Univ. Press. Navarro, G. and Maldonado, M. 2002. Geografia ecolögica de Bolivia: vegetaeiön y ambientes acuaticos. — Centro de Ecologia Simon I. Patino, Depto de Difusiön. Pauli, H. et al. 2012. Recent plant diversity changes on Europe's mountain summits. - Science 336: 353-355. Pauli, H. et al. 2015. The GLORIA field manual - standard multi-summit approach, supplementary methods and extra approaches. — GLORIA-Coordination, Austrian Academy of Sciences and Univ. of Natural Resources and Life Sciences. Perez, F. L. 1986. Talus texture and particle morphology in a north Andean Paramo. - Z. Geomorphol. 30: 15-34. Peyre, G. 2015. Plant diversity and vegetation of the Andean Paramo. — PhD thesis, Univ. of Barcelona. Peyre, G. et al. 2015. VegParamo, a flora and vegetation database for the Andean paramo. — Phytocoenologia 45: 195-201. Quinn, G. P. and Keough, M. J. 2002. Experimental design and data analysis for biologists. — Cambridge Univ. Press. Ramirez-Villegas, J. et al. 2014. Using species distributions models for designing conservation strategies of tropical Andean biodiversity under climate change. — J. Nat. Conserv. 22: 391-404. Ramsay, P. M. and Oxley, E. R. 1997. The growth form composition of plant communities in the Ecuadorian paramos. — Plant Ecol. 131: 173-192. Rangel, T. F. et al. 2010. SAM: a comprehensive application for spatial analysis in macroecology. — Ecography 33: 46—50. Ruiz, D. et al. 2012. Trends, stability and stress in the Colombian central Andes. - Clim. Change 112: 717-732. Rundel, P. W. et al. 1994. Tropical alpine environments: plant form and function. — Cambridge Univ. Press. Sala, O. E. et al. 2000. Global biodiversity scenarios for the year 2100. - Science 287: 1770-1774. Scherrer, D. and Körner, C. 2011. Topographically controlled thermal-habitat differentiation buffers alpine plant diversity against climate warming. — J. Biogeogr. 38: 406—416. Simpson, B. B. 1983. An historical phytogeography of the high Andean flora. - Rev. Chil. Hist. Nat. 56: 109-122. Simpson, B. B. and Todzia, C. A. 1990. Patterns and processes in the development of the high Andean flora. - Am. J. Bot. 77: 1419-1432. Sklenář, P. and Jorgensen, P. M. 1999. Distribution patterns of Páramo plants in Ecuador. — J. Biogeogr. 26: 681—691. Sklenář, P. and Ramsay, P. M. 2001. Diversity of zonal Páramo plant communities in Ecuador. — Divers. Distrib. 7: 113-124. Sklenář, P. and Balslev, H. 2005. Superpáramo plant species diversity and phytogeography in Ecuador. — Flora 200: 416—433. Sklenář, P. et al. 2008. Cloud frequency correlates to plant species composition in the high Andes of Ecuador. — Basic Appl. Ecol. 9: 504-513 Sklenář, P. et al. 2011. Tropical and temperate: evolutionary history of Páramo flora. - Bot. Rev. 77: 71-108. Sklenář, P. et al. 2014. Island biogeography of tropical alpine floras. -J. Biogeogr. 41: 297. Smith, J. M. B. and Cleef, A. M. 1988. Composition and origins of the world's tropicalpine floras. — J. Biogeogr. 15: 631—645. Tovar, C. et al. 2013. Diverging responses of tropical Andean biomes under future climate conditions. — PLoS One 8: e63634. Troll, C. 1973. High mountain belts between the polar caps and the equator: their definition and lower limit. — Arct. Alp. Res. 5: A19-A27. Urrutia, R. and Vuille, M. 2009. Climate change projections for the tropical Andes using a regional climate model: temperature and precipitation simulations for the end of the 21st century. -J. Geophys. Res. 114: 1-15. Valencia, B. G. et al. 2016. Andean microrefugia: testing the Holocene to predict the Anthropocene. — New Phytol. 212: 510-522. van der Hammen, T. 1974. The Pleistocene changes of vegetation and climate in tropical South America. — J. Biogeogr. 1: 3-26. van der Hammen, T. and Cleef, A. M. 1986. Development of the high Andean Páramo flora and vegetation. — In: Vuilleumier, F and Monasterio, M. (eds), High altitude tropical biogeography. Oxford Univ. Press, pp. 153-201. Vuille, M. et al. 2008. Climate change and tropical Andean glaciers: past, present and future. — Earth-Sci. Rev. 89: 79-96. Weigend, M. 2002. Observations on the biogeography of the Amotape-Huancabamba zone in northern Peru. — Bot. Rev. 68: 38-54. Williams, J. J. et al. 2011. Vegetation, climate and fire in the eastern Andes (Bolivia) during the last 18 000 years. — Palaeogeogr. Palaeoclimatol. Palaeoecol. 312: 115—126. Winkler, M. et al. 2016. The rich sides of mountain summits - a pan-European view on aspect preferences of alpine plants. — J. Biogeogr. doi: 10.111 l/jbi.12835 Young, K. et al. 2002. Plant evolution and endemism in Andean South America: an introduction. - Bot. Rev. 68: 4-21. Supplementary material (Appendix ECOG-02567 at ). Appendix 1-2. 1394