ARTICLE IN PRESS Flora 200 (2005) 416–433 www.elsevier.de/flora Superpáramo plant species diversity and phytogeography in Ecuador Petr Sklenářa,, Henrik Balslevb a Department of Botany, Charles University, Benátská 2, 128 01 Prague, Czech Republic Department of Systematic Botany, Aarhus University, Universitetsparken Bygn. 540, DK–8000 Aarhus C., Denmark b Received 14 April 2004; accepted 21 December 2004 Abstract In 18 superpáramo sites in Ecuador we found 388 species of vascular plants belonging to 146 genera and 52 families, making the Ecuadorian superpáramo flora richer in species than that of Venezuela and that of Colombia which appears to have fewer species although the number remains uncertain. The most species rich families were Asteraceae (83 species) and Poaceae (49) which also dominate the grasspáramo that surrounds the superpáramo. Otherwise the superpáramo is dominated by families that are mostly herbaceous such as Cyperaceae, Brassicaceae, Caryophyllaceae and Valerianaceae, whereas shrubby families that dominate the subpáramo and grasspáramo, such as Melastomataceae, Ericaceae, and Solanaceae, have only few species or are absent in the superpáramo. The generic spectrum was dominated by a suite of species-rich genera (Lachemilla, Gentianella, Valeriana, Calamagrostis, Draba) with many páramo-endemic species reflecting a high level of autochthonous speciation in the (super)páramo. Species richness varied from 71 to 149 in the individual superpáramos surveyed, but species richness was only weakly correlated to their area and log-transformed area, and negatively correlated to the vertical range of each superpáramo. b-diversity was significantly correlated to the vertical range (i.e., the number of surveyed altitudinal levels) but it was not correlated to the area or log-transformed area of each superpáramo. Most of the species were narrowly distributed and 112 (29%) of them were found in a single superpáramo, while eight (2%) occurred in all 18 superpáramo sites. Floristic similarity was not correlated to the geographical distance between the sites. Redundance Analysis suggested that geological origin of the substrate (metamorphic versus volcanic bedrock) was important for the floristic composition. Occurrence of mountains built from metamorphic rocks, is however, correlated to areas with high rainfall and the amount of rainfall may be a stronger determinant for species distribution than the presence of volcanic versus metamorphic bedrocks. TWINSPAN cluster analysis divided the 18 sites into three groups, which corresponded to dry, humid and very humid superpáramos. The groups were also separated along the first ordination axis in Correspondence Analysis, while the second axis may correlate to the volcanic history of the area. r 2005 Elsevier GmbH. All rights reserved. Keywords: Andes; Habitat islands; Páramo; Phytogeography; Species distribution; Tropical alpine Introduction Corresponding author. E-mail addresses: petr@natur.cuni.cz (P. Sklenář), henrik.balslev@biology.au.dk (H. Balslev). 0367-2530/$ - see front matter r 2005 Elsevier GmbH. All rights reserved. doi:10.1016/j.flora.2004.12.006 The superpáramo is the upper vegetation belt of the páramo – an essentially aseasonal high elevation grassand shrub-vegetation of the humid tropical Andes with ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 outliers in Costa Rica and Panama. Where more seasonal climate reigns further north and south, the páramo gives way to other vegetation types above the timberline; in Guatemala and Mexico to herbaceous and tussock-grass communities called ‘‘zacatonales’’ and in central Peru and Bolivia to dry, barren grass- and bushsteppe called ‘‘puna’’ (Almeida-Lenero et al., 2004; Balslev and Luteyn, 1992; Islebe and Cleef, 1995; Luteyn, 1999; Simpson, 1983; Troll, 1959, 1968). Like other tropical mountains, the Andes are floristically rich and often have high endemism, particularly in their upper zones such as the tropical alpine belts above the timber line (Balslev and Luteyn, 1992; Hedberg, 1957; Smith, 1975). Among the tropical alpine ecosystems, the páramo is the most diverse, mainly because of its large geographic extent (Hedberg, 1992; Smith and Cleef, 1988; Vuilleumier and Monasterio, 1986). About 3500 species of vascular plants have been reported from the páramo and possibly as many as 60% of them are endemic (Luteyn, 1992, 1999). The high concentration of endemic species is usually attributed to changes of horizontal and vertical extent of the páramo during Pleistocene climatic oscillations and associated pulses of geographic isolation (Cuatrecasas, 1986; Simpson and Todzia, 1990; Van der Hammen and Cleef, 1986; Van der Hammen et al., 1973). Above 4100–4300 m elevation, the so-called superpáramo, which is the subject of this study, replaces the páramo grasslands (or the bamboo páramo in humid areas) and extends to the permanent snow-line at 74800 m. In the superpáramo, vegetation cover is broken up and patches of open soil appear, the tussock grasses that are characteristic of the grass páramo are less abundant, and the vegetation is dominated by other life-forms such as small herbs, cushion plants, and sclerophyllous dwarf shrubs. Superpáramo vegetation has been reported from Venezuela, Colombia, and Ecuador (Berg, 1998; Cleef, 1981; Cuatrecasas, 1968; Luteyn, 1992; Monasterio, 1979), whereas the summits in Panama and Costa Rica do not reach elevations that are high enough to support superpáramo. In northern Peru, several summits of the ‘‘jalca,’’ which is the local name for páramo, reach elevations at which superpáramo may have developed, but so far there are no descriptions in the literature that document the existence of superpáramo vegetation in Peru. On a few occasions the name superpáramo has been used for high-elevation vegetation as far south as Bolivia (Cuatrecasas, 1968), but since it replaces puna there, super-páramo is an inappropriate name. Since the superpáramo is confined to the highest mountains, it occurs scattered as ‘‘islands’’ separated by grass páramo, mountain forest, and deep Andean valleys (Cuatrecasas, 1958; Simpson, 1974, 1975; Van der Hammen and Cleef, 1986). In Ecuador, superpáramo is particularly well represented and occurs 417 isolated on 25 or more mountains in both the western and eastern cordilleras along a stretch of ca. 450 km from the Colombian border in the north to the province of Azuay in the south. On most of these mountains the superpáramo is developed on volcanic bedrocks but on a few mountains in the eastern cordillera it has developed on metamorphic bedrocks. Depending on local climatic conditions and orientation relative to the trade winds, the superpáramo may be dry or humid. Some areas of superpáramo are found on mountains that are high enough to have permanent glaciers whereas others do not reach such heights. This setting is an attractive one for comparative studies of species diversity, taxonomic composition, and distribution of vascular plant species, and phytogeographic relations among the superpáramo ‘‘islands’’. In this paper, we address the following questions: (1) How species rich is the Ecuadorian superpáramo compared to Colombian and Venezuelan superpáramos? (2) Is the superpáramo flora dominated by the same taxa as the grass páramo that surrounds it, and are some genera particularly diverse, indicating high levels of autochthonous speciation? (3) How does species richness and b-diversity vary among Ecuadorian superpáramos and is the variation related to their areas or their vertical ranges? (4) How are individual species distributed among the superpáramos and how is endemism structured? (5) How floristically similar are the superpáramos and are the observed differences related to ecological or other factors? Study sites We selected 16 representative areas of superpáramo in the eastern and western cordilleras of the Andes in Ecuador (Fig. 1) stretching over 430 km from Volcán Chiles (01480 N, 771570 W) on the border with Colombia to Cajas National Park (21520 S, 791180 W) in southern Ecuador. The superpáramos on Chimborazo and Antisana were studied on two opposite sides of the mountains. These data were treated as independent sets, giving a total of 18 study sites, of which 14 were located on active or extinct volcanoes, and four on mountains of non-volcanic origin with metamorphic bedrocks. Nine of the studied mountains are topped by glaciers today, but all of them were glaciated at some point during Pleistocene (Hastenrath, 1981). The basic characteristics of each site are summarized in Table 1 and further details can be found in Sklenář (2000). In Ecuador, the superpáramo forms two distinct altitudinal belts. The lower superpáramo from 4100 to (4400–)4500 m is usually rich in species. It is characterized by sclerophyllous dwarf shrubs (Loricaria spp., Chuquiraga jussieui, Valeriana microphylla, ARTICLE IN PRESS 418 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 78 W 1 Ecuador South America 2 3 4 5 0 6 8 7 9 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 11 13 14 12 15 16 17 18 0 50 100 species richness Chiles 103 Cotacachi 102 Imbabura 114 Cayambe 138 Saraurcu (M) 95 Pichincha 131 Antisana - east 129 Antisana - west 143 Iliniza 143 Cotopaxi 91 Hermoso (M) 122 Tungurahua 71 Chimborazo - west 107 Chimborazo - east 131 El Altar 132 Quilimas (M) 137 Yanaurcu (M) 106 Cajas 149 beta diversity 3.2 1.8 1.9 3 1.8 2.2 2.4 2.3 3.9 3.6 1.4 1.2 1.6 2.1 1.9 1.9 0.7 1.4 150 km Fig. 1. Distribution of 18 vegetation study sites in the superpáramo in Ecuador along a 430 km long stretch of the Andes in both the western and eastern Cordilleras. Squares correspond to Group 1 and circles to Group 2 in Fig. 4. (M) after the name of site indicates metamorphic bedrock. Diplostephium rupestre, Pentacalia spp.), and cushion plants (Plantago rigida, Xenophyllum humile, Azorella spp.), but tussock grasses (Calamagrostis intermedia, Festuca asplundii) are usually also important. The upper superpáramo above (4400–)4500 m is characterized by the presence of Senecio nivalis, S. canescens, Arenaria dicranoides, and Cerastium floccosum. The vegetation is poor in species and confined to favorable habitats. Tussock grasses and sclerophyllous shrubs are mostly absent, and the most common growth-forms are shortstem grasses, prostrate subshrubs and herbs, acaulescent rosettes, and cushion plants (Sklenář, 2000). Methods Field work was carried out from May through December 1995, June through September 1997, and June through July 1999. The vegetation was sampled in a stratified random design (Mueller-Dombois and Ellenberg, 1974). This sampling was limited to the zonal vegetation, avoiding azonal patches of cushions mires, stream-sides, etc. The lower limit of the superpáramo was here defined as the transition to the grass páramo, i.e., where a continuous grass cover begins, while the upper limit was defined as where the plant cover dropped below 5%. At 100 m altitudinal (stratified) levels we randomly set three plots along a 100 m long transect running parallel with the contours. Most plots were 25 m2, but in the upper belt (generally above 4400 m) they were 100 m2 because there we found increased spatial variability in species composition. In each plot we scored all species using an 8-grade semiquantitative scale combining abundance and cover (Mueller-Dombois and Ellenberg, 1974). Data were converted to their midpoint values (r ¼ 0.05%, + ¼ 0.5%, 1 ¼ 2.5%, 2a ¼ 10%, 2b ¼ 20%, 3 ¼ 37.5%, 4 ¼ 62.5%, 5 ¼ 87.5%) for further analysis. In addition to the quantitative sampling we searched the study sites for additional species outside the sample plots but this time including also azonal habitats, such as rocky outcrops, cushion mires, and lake or stream shores, and they were added to the species lists for each site. Species planted occasionally around mountain refuges or occurring on ruderal sites around huts were excluded from the phytogeographic analyses. Introduced species that were naturalized, i.e., they reproduce and survive independently in natural plant communities, were included in all analyses. Identification was backed up with voucher specimens and usually verified by specialists for each taxonomic group. Information about species endemism in Ecuador was extracted from Jørgensen and León-Y. (1999) and Valencia et al. (2000) and was updated from our own knowledge. b-diversity was calculated from the vegetation samples for each site as: bT ¼ s=a  1, where s is the total number of species present in all samples at a site and a is the average number of species in the samples from that site (Whittaker, 1960; Wilson and Shmida, 1984). Correlation of species richness and b-diversity to environmental and spatial variables were calculated. The explanatory variables were superpáramo area (measured as plane area estimates from the IGM 1:50,000 maps; if present, the extension of glaciers was excluded), vertical range of the superpáramo (i.e., for species richness vertical distance between the base and the top of the superpáramo or the position of the permanent snow-line, but for b-diversity number of surveyed altitudinal levels), presence/absence of glacier, volcanic/metamorphic bedrock, volcanic activity yes/no, position in western or eastern cordilleras. Floristic relationship between pairs of sites was measured as the difference between observed and expected numbers of shared species (Connor and Simberloff, 1978). Expected numbers were estimated by a random pair-wise sampling from the entire species list in which each species had an adjusted probability of being sampled, for species found in the vegetation plots the expected numbers were the frequency of occurrence and for species found only outside the vegetation plots the expected numbers were the number of site Table 1. Baseline data for the 18 study sites. Maximum elevation (m) Coordinates latitude longitude Estimated superpáramo area (km2) Vertical range (m) Recent glacier Geological origin Recent volcanic activity Chiles Cotacachi Imbabura Cayambe Saraurcu Pichincha Antisana-west Antisana-east Iliniza Cotopaxi Hermoso Tungurahua Chimborazo-west Chimborazo-east El Altar Quilimas Yanaurcu Cajas Carchi Imbabura Imbabura Pichincha Pichincha Pichincha Napo 4768 4939 4609 5790 4676 4698 5704 No No No No — Yes Yes Yes Yes No Yes Yes Volcanic Volcanic Metamorphic Volcanic Volcanic No Yes — Yes No Chimborazo Chimborazo Chimborazo Azuay 5319 4730 4580 4451 564 737 330 600 576 398 500 500 600 600 539 400 700 500 500 630 380 151 Volcanic Volcanic Volcanic Volcanic Metamorphic Volcanic Volcanic 5263 5897 4639 5016 6310 6.4 6.7 1.6 43.3 12.6 6.5 29 29 26.9 65.5 7.2 11.1 62.2 62.2 44.3 47.2 15.4 17.7 No Yes No Yes Yes No Yes Cotopaxi Cotopaxi Tungurahua Tungurahua Chimborazo 01480 N 771570 W 01220 N 781200 W 01160 N 781100 W 01020 N 771590 W 01060 S 771550 W 01100 S 781330 W 01300 S 781100 W 01270 S 781070 W 01400 S 781420 W 01400 S 781260 W 11100 S 781120 W 11270 S 781260 W 11280 S 781520 W 11280 S 781460 W 11410 S 781240 W 11460 S 781240 W 21150 S 781300 W 21520 S 791180 W Yes No No No Volcanic Metamorphic Metamorphic Volcanic No — — No The vertical range of the superpáramo is the vertical distance in meters between the lowest sampling plot and the top of the mountain or the transition to the permanent snow-line, coordinates indicate the approximate location of the transects, estimated superpáramo area is a plane estimate obtained from the IGM maps (excluding glaciers where present). ARTICLE IN PRESS Province P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 Name of site 419 ARTICLE IN PRESS 420 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 occurrences (null hypothesis II of Connor and Simberloff, 1978). The size of the random sample was equal to the actual number of species for any given site. The expected numbers of shared species, independent of sample size (Connor and Simberloff, 1978; Simberloff et al., 1981), were estimated repeatedly 999 times via computer simulation (Sklenář, 2000). The chi-square (w2 ) distance between observed and expected numbers was used as a standardized measure of floristic relationships between the mountain pairs. The w2 values were tested against geographical distances between the sites by means of Mantel test, part of the R-Package (Legendre and Vaudor, 1991); the null hypothesis of the test was that there would be no linear relationship between the geographical distance and the floristic similarity. Two-way indicator species analysis (TWINSPAN; Hill, 1979) was used to group the 18 sites according to floristic and vegetation similarities with presence–absence data for all species and semi-quantitative abundance estimates for species present in the plots based on average values of each species in the plots at each site. For the abundance data we used CA (Correspondence Analysis, part of CANOCO package; Ter Braak and Šmilauer, 1998) to demonstrate the relative position of study sites in the multivariate space; data were squareroot transformed and rare species were down-weighted prior the analysis. For the presence–absence data we used RDA (Redundancy Analysis) to test the correlation between species distribution among the sites and six environmental and spatial variables. Results Species richness and taxonomic composition There were 383 species, 146 genera and 52 families of vascular plants in the 18 Ecuadorian superpáramo sites surveyed. There were two subspecies in Eudema nubigena and Valeriana alypifolia, two varieties in Potentilla dombeyi, and two hybrids in Gentianella, which we, for simplicity treated in the same way as species in the analyses, so the number of taxa (subsequently referred to as ‘‘species’’) was 388. Of these, 374 were recorded within the continuous superpáramo, i.e., above the transition with the grass- or bamboo páramo at (4000–) 4100–4200 m. The remaining 14 species were found within 100 m below this limit, in isolated patches of superpáramo vegetation, such as rock crevices, surroundings of boulders, lake shores, or eroded land, and we therefore expect that they will eventually be found in the continuous superpáramo. Of the 388 species, five (Aira caryophyllea, Anthoxanthum odoratum, Poa annua, Rumex acetosella, and Sagina procumbens) are introduced to Ecuador. A full listing of the species is provided by Sklenář (2000). Species diversity was unevenly distributed among taxonomic groups (Fig. 2). Asteraceae and Poaceae entirely dominated the flora and together they comprised one-third of the total species diversity. Other families were conspicuously less diverse and there were 16 families with a single species. There were a suite of species rich genera, but the majority of the 146 genera present in the Ecuadorian superpáramo, i.e., 84 genera, were represented by a single species. Variation in species richness and b-diversity among superpáramos Species richness differed considerably among the superpáramo sites (Fig. 1). Cajas was richest in species (149 species) even if it is one of the lowest mountains (Table 1). The lowest number of species was found in Tungurahua (71) and Cotopaxi (91), which are the mountains that have the most recent volcanic activity. Species richness correlated weakly with the estimated area of superpáramo at each site (Spearman rank correlation coefficient r ¼ 0:173, p ¼ 0:49), or the logtransformed area (r ¼ 0:264, p ¼ 0:29), and correlation with the vertical range of the superpáramo was negative (r ¼ 0:191, p ¼ 0:45; n ¼ 18 in all cases). When the correlation analyses were performed independently for each of the two groups of mountains resulting from the cluster analysis (see below), the results became significant for the area and log-area of the 13 mountains of Group 2 (r ¼ 0:664, p ¼ 0:01) but not for the five mountains of Group 1 (r ¼ 0:667, p ¼ 0:22). The correlation did not change for the vertical range of the superpáramo (Group 1: r ¼ 0:395, p ¼ 0:51; Group 2: r ¼ 0:088, p ¼ 0:77). A considerable spatial pattern of b-diversity in the superpáramo vegetation (Fig. 1) was significantly correlated with the number of altitudinal levels surveyed at each site (r ¼ 0:59, p ¼ 0:01), whereas its correlation with other variables (i.e., area, vertical range of superpáramo) was weak and not significant (p40:05). This significant correlation to the altitudinal levels remained within the cluster Group 2 (r ¼ 0:605, p ¼ 0:03) but not within the Group 1 (r ¼ 0:051, p ¼ 0:93), whereas the other correlations remained not significant. Species distributions The majority of species were narrowly distributed; 112 species (29% of the total) were found only at one single site, and 274 or over two-thirds of the species occurred at fewer than seven sites (Fig. 3). Only eight species (e.g., Calamagrostis intermedia, Gentiana sedifolia, Luzula racemosa, Poa cucullata) were found at all 18 sites. Species restricted to either dry or humid sites (Groups 1 and 2 in Fig. 4, resulting from the first step of the cluster ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 421 90 Asteraceae (26 endemic) 80 Families Number of Species 70 60 50 Poaceae (12 endemic) 40 Brassicaceae (12 endemic) 30 Gentianaceae (11 endemic), Rosaceae 20 Scrophulariaceae, Caryophyllaceae, Cyperaceae 10 0 16 Genera 14 Lachemilla (2 endemic), Gentianella (10 endemic) Valeriana (5 endemic) Number of Species 12 Calamagrostis (3 endemic) Draba (8 endemic) 10 Lupinus Festuca (5 endemic), Geranium (3 endemic), S enecio 8 Cerastium,Huperzia 6 genera 63-146 4 2 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 Fig. 2. Ranked species numbers of families and genera encountered in 18 study sites in the superpáramo in Ecuador. analyses, see below) were also very narrowly distributed within these groups, and only few of them occurred at many sites (Fig. 3). The five dry sites had a total of 241 species, of which 64 were restricted to them, and only Festuca vaginalis and Plantago nubigena were present in all of them (Fig. 5A). The 13 humid sites had a total of 323 species, of which 137 were restricted to them, and there were no species occurring in all of them. The most widespread were Geranium sibbaldioides (Fig. 5H) and Luzula gigantea (Fig. 5L) which occurred in 11, and F. ARTICLE IN PRESS 422 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 120 100 All superpáramo sites Number of Species 80 Humid superpáramo sites 60 Dry superpáramo sites 40 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Number of sites (N) Fig. 3. Pattern of geographic distribution of all species among the 18 study sites in the superpáramo in Ecuador (open bars), and pattern of distribution of species restricted to the 13 humid superpáramos (hatched bars) and to the five dry superpáramos (black bars.). asplundii (Fig. 5J) and Lachemilla holosericea (Fig. 5I) which occurred in 10 humid sites, respectively. Floristic similarity among superpáramos Grouping of the superpáramo sites was similar in TWINSPAN clustering using either presence–absence data of species from the superpáramo sites (from both zonal and azonal habitats) or abundance data of the zonal vegetation, and the differences were essentially only in the final divisions. This shows that adding the species from the azonal vegetation patches (presence–absence) does not dramatically change the picture obtained on the basis of the survey that included only the zonal vegetation (abundance). The analysis based on abundance generally had higher eigenvalues and only this result is presented (Fig. 4B). The first division step separated Chimborazo-west, Antisana-west, Iliniza, Cotopaxi, and Pichincha (Group 1), due to the occurrence of Plantago nubigena and Festuca vaginalis (Fig. 5A). In the second group resulting from the first division (Group 2), Imbabura, Cajas, Cayambe, and Chimborazo-east were separated next due to occurrence of Pentacalia peruviana; Cotacachi was added to this group in the analysis using presence–ab- sence data. Tungurahua was consistently separated from the remaining nine mountains because of absence of Azorella aretioides. The rest formed two sub-groups; the southern superpáramo sites (Quilimas, Yanaurcu, Altar, Hermoso) were separated from the north-central ones (Antisana-east, Saraurcu, Chiles, Cotacachi), due to the occurrence of Calamagrostis guamanensis (Fig. 5O) in the latter. This species was recorded also on Hermoso but only outside vegetation plots and therefore this site appeared in either the southern or the north-central subgroup in the two analyses. The first CA ordination axis using the abundance data captured the gradient expressed by the first division step of TWINSPAN clustering; sites from the Group 1 occur on the right side while sites from the Group 2 on the left side of the ordination diagram (Fig. 4A). Cajas appears a very distinct site separated from the remainder by the second ordination axis. A total of 374 species formed the common species pool, which was used for estimations of the expected numbers of shared species (Appendix 1). Six sites (Chiles, Cotacachi, Imbabura, Cayambe, Saraurcu, Antisana-east) had positive w2 values between each pair of them (the only exception being Imbabura–Antisanaeast), mostly positive or low negative values with Chimborazo-east, Hermoso, Altar, Quilimas, and Ya- ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 2.5 423 Chimborazo - west Group 1 (A) Cajas (B) ! 0.473 Antisana - west 0.457 Cotopaxi 0.395 Iliniza Pichincha 0.421 0.403 Imbabura ! 0.325 Cajas 0.526 Cayambe Chimborazo - east 0.447 0.276 Hermoso (M) Cotopaxi Iliniza Altar Cotacachi Yanaurcu (M) Antisana-west Saraurcu (M) Chimborazo-east Imbabura Pichincha Chiles Quilimas (M) Cayambe Antisana-east Tungurahua Tungurahua Cotacachi Chiles Antisana - east Saraurcu (M) Hermoso (M) El Altar Yanaurcu (M) 0.495 0.353 0.304 0.398 0.311 0.445 0.423 ! 0.328 Quilimas (M) -1.0 -1.0 Group 2 Chimborazo-west 0 1.5 Fig. 4. Correspondence Analysis of floristic abundance data from 18 superpáramo sites in Ecuador (l1 ¼ 0:316, l2 ¼ 0:138, total inertia 1.537); squares (&) indicate sites of Group 1 and circles (J) indicate sites of Group 2 resulting from the cluster analysis (A) and cluster (TWINSPAN) analysis of abundance data from the 18 sites with the first division separating two groups based on the presence (Group 1) or absence (Group 2) of Festuca vaginalis and Plantago nubigena; eigenvalues for each division step provided, ! indicate divisions with higher residual than the tolerance given by the algorithm (B); (M) indicates site with metamorphic bedrock. naurcu, and high negative values with Chimborazowest, Cotopaxi, Antisana-west, and Iliniza (Appendix 2). The latter four sites had positive values between each pair of them. Three sites (Pichincha, Tungurahua, Cajas), did not appear to demonstrate a clear pattern. Pichincha showed mostly small values (negative or positive), the highest in combination with Cotacachi, Cayambe, and Iliniza, and the lowest with Hermoso and Cajas. Tungurahua had small negative values with most of the mountains, only with Chimborazo-west the value is remarkably low (w2 ¼ 24:56). Cajas was unusual since all but one of its values were negative, the lowest one with Chimborazo-west. The overall spatial pattern of floristic similarities between the mountains was not correlated with the geographic distance between the sites; a Mantel test between the two matrices (Appendices 1 and 2) gives r ¼ 0:034, p ¼ 0:38, using 999 permutations. The RDA showed that of the six variables tested (estimated area, vertical range of superpáramo, presence/absence of a recent glacier, position in western/eastern cordillera, presence/absence of recent volcanic activity, and volcanic/metamorphic bedrocks), only the latter (bedrock volcanic/metamorphic) was significantly correlated to the observed species composition of the sites (forward manual selection, po0:05, 199 permutations). However, the bedrock variable accounted for only a small portion (10.7%) of total data variability. Discussion Species richness Most of the sites were visited only once, but Iliniza and Chimborazo were explored repeatedly. The first 5days survey of Iliniza yielded 127 species, but 16 additional species (13% of the total) were found during the second 7-days intensive research of the mountain three months later. In 1999, the transect area on Chimborazo-west was revisited during a 1-day trip. Four additional species were found (3.5% of the total), but three of them occurred above the limit surveyed in 1995. Some ‘‘gaps’’ in our species distributions may therefore result from sampling error. However, if our experience at Iliniza is representative, we have recorded 80–90% of all species at each superpáramo site. The remarkable diversity of neotropical páramos, previously highlighted by Smith and Cleef (1988) and Luteyn (1999), is underlined by our record of 388 species from the superpáramo in Ecuador. This number ARTICLE IN PRESS 424 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 Fig. 5. Examples of distributions of species among 18 study sites in the superpáramo in Ecuador ( species not found at study site; J species found at study site); A–F species characteristic of dry sites, G–L species characteristic of humid sites, M–R species characteristic of very humid sites, S–T species negatively characterizing very humid sites, U–V narrowly distributed species with trans-Andean distribution, W–X narrowly distributed species with and cis-Andean distribution. ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 includes some species that typically occur in the grass páramo at lower elevations (e.g., Hieracium frigidum, Stachys elliptica) or bamboo páramo (e.g., Gentianella splendens), and only marginally enter the lower parts of the superpáramo belt. On the other hand, we are aware of several species, such as Azorella ecuadorensis, Eudema rupestris, and Geranium chimborazense, previously collected in the superpáramo in Ecuador, that were not encountered in this survey. Those species may either be very rare, or restricted to sites that were not covered by our study (e.g., Xenophyllum roseum from the central part of the Cajas National Park), or possibly they were not recognized in the field (e.g., sterile specimens of Carex are very hard to identify to species level). With 388 species of vascular plants the Ecuadorian superpáramo flora seems to be richer in species than other superpáramo floras. In the superpáramo in Venezuela Ricardi et al. (1997) reported 234 species in 119 genera and Berg (1998) found 168 species in 147 vegetation samples. In the Colombian eastern cordillera Van der Hammen and Cleef (1986) found 112 species in 69 genera, mostly in the lower superpáramo belt (Cleef, 1981). A transect study in the Parque Los Nevados in the central cordillera of Colombia, revealed 170 species growing above 4000 m (Cleef et al., 1983; Rangel et al., 1983). In this study we surveyed representative superpáramo sites throughout Ecuador on 16 mountains, but at least ten distinct superpáramo areas were not explored (e.g., Carihuairazo, Quilindaña, Sincholagua). In Venezuela, superpáramo is confined to only three areas in the surroundings of Mérida (Monasterio, 1979), and the low number of superpáramo species reported from that country may therefore reflect the actual situation well. In Colombia, in contrast, there are many areas that reach the superpáramo belt (Cleef, 1981), but their total area may not surpass that of the Ecuadorian superpáramo (Hofstede et al., 2003). Different total areas of superpáramo in Ecuador and Colombia may determine the differences in species richness between their floras. Nevertheless the number of known Colombian superpáramo species will certainly grow when more sites are explored (A. M. Cleef, pers. comm.; P. Pedraza, pers. comm.) so definitive conclusions concerning the relative number of species in the three countries cannot be made until the species lists are more complete. Taxonomic composition It is well known that Asteraceae and Poaceae dominate the páramo flora (Luteyn, 1999; Luteyn et al., 1992; Ramsay, 1992; Ricardi et al., 1997) and this study confirms that these two families are the most species rich also in the superpáramo flora. Together they account for one third of all superpáramo species in Ecuador, which is roughly the same proportion as in the páramo in general 425 (Luteyn, 1999). Their importance is further underlined by the fact that nearly half of the 40 most important (combining frequency and abundance) species of the Ecuadorian superpáramo belong to these two families (Sklenář, 2000). Orchidaceae, especially epiphytic ones, and ‘‘shrubby’’ families, such as Melastomataceae, Ericaceae, and Solanaceae rank among the most important páramo groups (Luteyn, 1999), but they are poor in species or even absent in the superpáramo flora. In contrast, several ‘‘herbaceous’’ families (e.g., Cyperaceae, Brassicaceae, Caryophyllaceae, Valerianaceae) are among the most species-rich groups in the superpáramo flora whereas they are (relatively) less important when the whole páramo belt is considered. The most species-rich genera, usually with numerous country-endemic species, are Lachemilla, Gentianella, Valeriana, and Draba (Fig. 2), and they are diverse also in the páramos of Colombia and Venezuela (Gaviria, 1997; Santana Castañeda, 1994; van der Hammen and Cleef, 1986; Xena de Enrech, 1992). In Ecuador, especially Gentianella and Draba are very conspicuous superpáramo genera, contributing probably with the highest number of endemic species (Jørgensen and León-Y, 1999; León-Y, 2000; Pringle, 1995). However, if Senecio were considered in a broad sense, including Dorobaea and Pentacalia, its total number of species would be 17 and the genus would then be the richest in the superpáramo. At the other extreme, 57% of the genera have only one species. Such unequal distribution of species richness among genera was also found in the páramo flora of the eastern cordillera in Colombia by Simpson and Todzia (1990), who argued that presence in the páramo of many species-rich genera reflected a high degree of autochthonous speciation. The skewed taxonomic composition towards a few speciose genera documented here for the Ecuadorian (super)páramo would agree with that scenario. Variation in species richness among superpáramos We expected to find a strong positive correlation between the area of the surveyed superpáramo and its species richness (Simpson, 1974), but instead we found that the correlation was weak and not significant. However, when the correlations were calculated for the ecologically more homogeneous groups of mountains separately, the expected positive correlation pattern did appear, at least for the group of humid mountain sites (Group 2 in Fig. 4). We also expected to find a positive correlation between the vertical range of the superpáramo and its species richness (Simpson, 1974), but instead we found a negative correlation. These unexpected correlation patterns reflect the great ecological variability among the superpáramo sites. Small mountains with only species-rich lower superpáramo, such as Cajas, are more diverse than ARTICLE IN PRESS 426 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 some of the bigger mountains with an overall impoverished flora, such as Cotopaxi, Cotacachi, or Chimborazowest, despite their larger superpáramo area and/or larger vertical range. Apparently, the exceptional superpáramos of Cotopaxi and Chimborazo-west influenced the overall correlation pattern of species richness with area and also produced the strong negative albeit not significant correlation among the five dry sites (Group 1 in Fig. 4B). Variation in b-diversity among superpáramos We expected, and found a positive correlation between b-diversity and the vertical range of the superpáramo (measured as number of surveyed altitudinal levels), i.e., the longer the gradient of the superpáramo, the greater its b-diversity. As previously demonstrated, species composition and vegetation structure change considerably from the lower to the upper superpáramo belts (Cleef, 1981; Sklenář, 2000; Sklenář and Ramsay, 2001), which would give greater bdiversity on higher mountains (e.g., Cayambe and Iliniza mts.). The Cotopaxi superpáramo is among the most b-diverse in our sample despite the fact that it is among the least species rich (Fig. 1). This paradox may be related to the occurrence of a pioneer vegetation in the Cotopaxi superpáramo in addition to grass páramo as documented by Stern and Guerrero (1997) and Sklenář (2000). Lower mountains, where only lower superpáramo is developed (e.g., Yanaurcu, Cajas mts.), have lower b-diversity values, even though the overall species richness may be high. Among the highest Ecuadorian mountains (Antisana, Cayambe, Cotopaxi, Iliniza, and Chimborazo) and consequently those with the largest superpáramos (Table 1), the Chimborazo-west has rather low bdiversity. This superpáramo is among the least diverse in Ecuador, but it is conspicuous because of its vegetation structure and presence of several central/ southern Andean floristic elements (e.g., Werneria apiculata, Stipa spp., Viola polycephala) and due to this it is sometimes compared to ‘‘puna’’ which is the dry páramo-equivalent in Peru (Acosta-Solı́s, 1985). Through the combined effects of climate (rain-shadow of the mountain) and human activities (burning and grazing), this area has been altered into patchy, desertlike superpáramo (Sklenář and Lægaard, 2003) with reduced species richness and b-diversity. Smaller mountains seem to exhibit consistently lower b-diversity, with the exception of Chiles. Plant communities on Chiles in general have lower a-diversity (less species per samples) than communities at other sites of similar size but there is a high species turnover along the altitudinal gradient in the rocky habitats (Sklenář, 2001a) which produce high b-diversity at this particular site. Species distributions The superpáramo flora is distinct from the grass páramo and subpáramo belts below (Cleef, 1981; Cuatrecasas, 1968; Ramsay and Oxley, 1997). Species endemism is generally high in the superpáramo (Berg, 1998; Luteyn, 1992) although it may be even higher in the montane forest (Balslev, 1988; Sklenář and Jørgensen, 1999; but cf. Young et al., 2002). The proportion of species endemic to Ecuador in our sample was estimated at 23%, which is consistent with previous reports for the Ecuadorian páramo (León-Y, 2000; Sklenář and Jørgensen, 1999). Two genera, Floscaldasia and Raouliopsis (Asteraceae), seem to be restricted to superpáramo vegetation, but only the former reaches to Ecuador (Cuatrecasas, 1979; Luteyn, 1999; Sklenář and Robinson, 2000; Van der Hammen and Cleef, 1986). Almost one-third (29%) of the Ecuadorian superpáramo species are restricted to a single mountain and almost one-half of them (47%) are distributed in less than four superpáramo sites. This is consistent with findings of Ramsay (1992) and Sklenář and Jørgensen (1999), who reported a high number of narrowly distributed species in the Ecuadorian páramo. The number of very narrowly distributed superpáramo species may be partly inflated by the occasional occurrence of grass páramo and bamboo páramo species in the superpáramo belt at one or a few mountains. In addition some species reach Ecuador only marginally. That is true for Valeriana henrici and Lysipomia multiflora that are found in the Cajas region but are otherwise distributed in Peru (Ayers, 1999; Eriksen, 1989). Nevertheless, many Ecuadorian superpáramo species have a truly limited geographical distribution, e.g., Gentianella sulphurea (Fig. 5W) and Loricaria antisanensis (Fig. 5X), and some are restricted to a single mountain. Other examples of such stenoendemic species are Senecio ferrugineus at Chimborazoeast (Sklenář, 2001b), Aphanactis antisanensis from Antisana (Robinson, 1997), Oritrophium llanganatense from Hermoso (Sklenář and Robinson, 2000), and several species from the Cajas region, e.g., Gentianella longibarbata, G. hirculus (Pringle, 1995), Valeriana secunda (Eriksen, 1989), and Draba steyermarkii. Mountain ridges and deep valleys acted as major barriers to migrations of high-Andean plants during glacial and interglacial periods (Jørgensen and Ulloa, 1994; Jørgensen et al., 1995). Many páramo species are wind-dispersed or epizoochorous (Frantzen and Bouman, 1989; Melcher et al., 2000, 2004; Simpson and Todzia, 1990), which would make them capable of dispersal over long distances. Superpáramo species do tend to occur over broad vertical ranges which generally correlate with the species’ geographic distribution (Sklenář and Jørgensen, 1999). In our data, about 60% of the species have a trans-Andean distribution, ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 i.e., they occur in both cordilleras and are distributed across the deep inter-Andean valley. This is somewhat less than 72% of trans-Andean species previously reported for the highlands of Ecuador (Balslev, 1988), but our number is certainly affected by the singlemountain species. Trans-Andean distributions are found in many species, e.g., Senecio nivalis, Nototriche spp., Pentacalia microdon (Fig. 5V), Saxifraga magellanica, Arenaria dicranoides, and Xenophyllum rigidum (Fig. 5U), from the highest elevations, that should be the most isolated from each other by the inter-Andean valley. This may be a ‘‘relictual’’ occurrence of species which were widespread during glacial periods or it may have resulted from a post-glacial over-land dispersal of propagules along and across the Andes. Nevertheless, the pattern suggests that the distribution of (super)páramo plants is less affected by topographical barriers than are the distributions of montane forest species where only about half of species have trans-Andean distribution (Balslev, 1988). Although the species richness of separate superpáramo areas seems to be higher in Ecuador, onemountain endemics are much more common in the eastern cordillera of Colombia. In the Sierra Nevada del Cocuy, for instance, 20% of the 107 superpáramo species are endemic (Van der Hammen and Cleef, 1986). We found that only about 3% and 2% of the species found at Cajas and Chimborazo, respectively, were endemic, and most other mountains have no stenoendemic species at all. Unlike the metamorphic eastern cordillera of Colombia, the major Ecuadorian superpáramos occur on volcanoes, several of which are still active. Repeated destruction of vegetation through volcanic events followed by recolonization of the new substrate may not provide sufficient time between successive eruptions for a speciation. It is interesting to note in this context that the páramo flora of the volcanic central cordillera of Colombia is poorer in species than that of metamorphic eastern cordillera (Salamanca, 1992; Van der Hammen and Cleef, 1986). An additional factor may be the size of individual superpáramos; possibly the larger area of the Sierra Nevada del Cocuy compared to Ecuadorian páramo areas (Hofstede et al., 2003) also contributed to the high proportion of endemic species found there. An additional contributing factor could be the varying degree of isolation of the areas today and in the past. While downward shifts of vegetation belts by 1000–1500 m during glacial periods (Simpson, 1974; Van der Hammen, 1974; Van der Hammen and Cleef, 1986) would merge most areas of Ecuadorian superpáramos and facilitate species migrations between the now separated superpáramos, the Colombian Cocuy superpáramo area would remain isolated from other areas of superpáramo allowing longer time for speciation. Phylogeographic studies of wide-spread species could test this hypothesis. 427 Due to its scattered occurrence along the Andes, the páramo has often been treated as an insular system (Simpson, 1974, 1975; Vuilleumier, 1970). In one classical study the extent of the páramo during glacial periods was found to be more important for explaining the patterns of páramo plant species diversity than the current extent of the páramo (Simpson, 1974). In Ecuador, species richness is only weakly correlated to the present size of the individual superpáramos. On the other side the extent of superpáramo during glacial periods is not meaningful given the fact that the assumed downward movement of vegetation belts would merge almost all Ecuadorian areas of superpáramo to one large confluent area. Although geographic distance at various scales is (negatively) correlated to floristic similarity in the Andes (e.g., Luteyn, 2002; Ricardi et al., 1997; Simpson and Todzia, 1990) it does not explain the observed general pattern of floristic similarity in the Ecuadorian superpáramos. This finding of poor explanatory power of geographic distance confirms previous results obtained from a sample of the high-páramo flora (Sklenář and Jørgensen, 1999). The presence of volcanic versus metamorphic bedrock is the only significant variable in the RDA, but it accounts for only a small portion of the variability. Nevertheless, in three of the sites (Hermoso, Saraurcu, Quilimas), we thoroughly examined the boundary between volcanic and metamorphic basements, and we failed to discern any obvious changes in species composition or vegetation structure when crossing the boundary, although Vargas et al. (2000) reported distinct changes in species composition for the Llanganatis region where Hermoso is located. The cluster and indirect ordination analyses (Fig. 4) give a similar result; the superpáramos of mountains built from metamorphic rocks are not clustered together but are grouped with superpáramos of nearby volcanoes. The four superpáramo sites developed on metamorphic bedrocks all occur in the eastern cordillera where they face the Amazon lowlands and therefore receive the highest amounts of precipitation among the studied superpáramo sites (Bendix and Lauer, 1992; Hastenrath, 1981). We believe that the significance of geology (volcanic versus metamorphic bedrocks) for the species distributions (Fig. 5) in the RDA is an artifact and that this variable acts as a substitute for climate, especially humidity on the sites. Unfortunately, there are no climatic data of the necessary spatial resolution available for the highest elevations of the Ecuadorian Andes for a rigorous test of this hypothesis. Some support for it, however, comes from data concerning the effect of a rain-shadow on species composition of two Ecuadorian superpáramos, Antisana and Chimborazo (Sklenář and Lægaard, 2003), that was based on published precipitation measurements. Other circumstantial evidence for the effect of humidity on the ARTICLE IN PRESS 428 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 distribution of superpáramo species comes from the observation that some species that are mostly restricted to the ‘‘humid’’ superpáramos from the eastern slopes (e.g., Distichia muscoides, Caltha sagittata, Cuatrecasasiella isernii), can be found in wet sites along springs or small depressions in the terrain that occasionally occur in the ‘‘dry’’ superpáramos (Sklenář, 2000; Cerón, 1994). Floristic similarity among superpáramos Based on the distribution of species from the montane forests and páramos, two floristic regions have been proposed for the high Andes of northern and central Ecuador, with the divide being the inter-Andean valley (Jørgensen and Ulloa, 1994). Our data suggest two somewhat different regions. Although the climatic data is very limited, we attribute this pattern to differences in humidity based on our field observations of vegetation physiognomy. Most mountains of the eastern cordillera, northern Ecuador (Chiles, Cotacachi, Imbabura), Chimborazo-east (in the western cordillera), and Cajas in the south belong to humid superpáramos (Group 2 in Fig. 4B), which are characterized by the presence of Jamesonia spp., Geranium sibbaldioides, Lachemilla holosericea, F. asplundii, Carex sect. Acicularis, and Luzula gigantea among others (Figs. 5G–L). The dry superpáramos (Group 1 in Fig. 4B), are distinguished by the occurrence of Festuca vaginalis, Plantago nubigena, Astragalus geminiflorus, Bidens andicola, Conyza cardaminifolia, Calamagrostis mollis, Cerastium imbricatum, and Silene thysanodes (Figs. 5A–F; although the latter two were also found in some more humid superpáramos), and comprise Iliniza, Pichincha, Cotopaxi, and the Chimborazo-west and Antisana-west. Among the dry superpáramo sites, Chimborazo-west stands out from the other sites (see above). Among the humid superpáramo sites, two groups are apparent. Sites from the eastern cordillera (except Cayambe), Chiles and possibly also Cotacachi from the western cordillera represent very humid superpáramos that receive high amounts of precipitation originating in the Amazon basin. Among them, Tungurahua is exceptional, being an active volcano on which the vegetation is constantly recovering from the influence of volcanic events (Sklenář, 2000). These very humid superpáramo sites are characterized by the occurrence or abundance of species, such as Huperzia rufescens, Nertera granadensis, Loricaria complanata, Calamagrostis guamanensis, C. ecuadoriensis, Draba spruceana, and Xenophyllum sotarense (Figs. 5M–R). These species often have a strictly cis-Andean distribution, but usually do not occur on all of those mountains. These very humid sites are also characterized by the absence of a group of species that occur only in the dry or moderately humid superpáramo sites, such as Perezia pungens and Trisetum spicatum (Figs. 5S and T). The remaining four mountains (Cayambe, Imbabura, Chimborazo-east, and Cajas) basically differ from the very humid superpáramo by the absence of the above species rather than by forming a homogeneous group among themselves. However, in agreement with previous studies that pointed to the peculiarity of the Cajas region (e.g., Sklenář and Jørgensen, 1999) this site occurred well separated from the other sites in the Correspondence Analysis. Volcanic activity terminated during the Tertiary in the southern Ecuadorian Andes, including Cajas (Hall, 1977), whereas the northern páramos all experienced volcanic activity up till present time. The different volcanic histories may be reflected in the second CA ordination axis. Simpson (1983) pointed to the importance of the ecological similarity between areas to explain the composition of different Andean floras. At the geographic scale studied here of tens to hundreds of kilometers, we suggest that ecological similarity, considered mainly as humidity, is more important for distribution of superpáramo species than other factors, such as geographic distance or topographic separation by valleys. Our findings support previous works that stressed the importance of precipitation for vegetation composition by distinguishing humid and dry superpáramos in the Parque Los Nevados, Cordillera Central of Colombia (Pérez and Van der Hammen, 1983), and others who recognized rain-shadow desert páramo on the western side of Chimborazo (Ramsay, 1992). Several interesting questions remain for further exploration in the high páramo. Some of these are: How can we explain the smaller number of one-mountain endemic species in Ecuador compared to Colombia? At what geographic scale do island biogeographical factors become more important than ecological factors? Do grass páramo and subpáramo species conform to the pattern as observed in the superpáramo flora? Acknowledgements Petr Sklenář thanks the Danish Research Academy and the Grant Agency of the Czech Republic (Grant No. 206/97/1198) and Henrik Balslev thanks the Danish Natural Science Research Council (Grant No. 21-010617) for support. Two anonymous reviewers are thanked for suggestions to improve the quality of the paper. Both authors acknowledge Renato Valencia, director of QCA herbarium, P.U.C.E., Quito, for research facilities, INEFAN, Quito, for research and export permits. Veronika Kostečková is thanked for assistance in the field and Flemming Nørgaard for producing figures. Appendix 1 Matrix of OBSERVED (above diagonal) and EXPECTED numbers of species held in common (below diagonal) between pairs of sites, diagonal indicates the number of species for each site used in the analysis. Cota Imba Caya Sara EAnti WAnti Pich Ilin Coto WChim EChim Herm Tung Altar Quil 56 64 34 69 67 55 64 62 43 72 67 59 71 59 42 68 64 54 93 72 46 82 77 63 60 74 39 76 76 63 77 84 48 84 84 66 87 53 39 62 72 49 83 54 45 68 67 55 83 48 42 67 67 53 46 30 29 34 34 30 53 29 27 44 41 35 129 58 45 73 76 61 71.6*** 117 48 74 83 65 47.1 44.3 71 45 46 43 77 71.4 47.1 130 92 87 80 73.9* 48.5 80*** 135 80 66.1 61.9 41.5 66*** 68.3** 106 85 78.6** 51*** 84.7 87.7 72.4* 61 71 69 83 60 73 77 71 79 47 53 78 68 39 82 83 65 147 Species pool ¼ 374, significant differences marked with asterisks (0.05Xp40.01*, 0.01Xp40.001**, pp0.001***). ARTICLE IN PRESS Chiles 100 62 60 67 67 70 44 51 49 26 30 Cotacachi 53.8* 102 77 78 59 69 61 76 69 39 41 Imbabura 57.9 59.4*** 114 92 55 66 66 74 72 42 43 Cayambe 66 67.4** 73.4** 137 69 84 85 92 91 50 53 Saraurcu 49.6*** 50.8* 54.6 61.7* 93 77 45 49 47 22 26 Antisana-east 63.2* 64.5 70.1 80.6 59.4*** 129 67 66 62 35 33 Antisana-west 66.7*** 68.4* 74.4* 85.8 63.1*** 82.1*** 140 87 102 66 69 Pichincha 63.5*** 65.1** 70.7 81.1** 59.7** 77.8** 82.4 130 95 62 61 Iliniza 68.5*** 70.2 76.2 87.9 64.5*** 84.1*** 89.4** 84.4** 143 74 73 Cotopaxi 48.7*** 49.6*** 53.6** 60.7** 46*** 58.5*** 62 58.8 63.3** 91 54 Chimborazo-west 53.6*** 55*** 59.3*** 67.4*** 50.8*** 64.9*** 68.5 65.1 69.8 49.9 103 Chimborazo-east 63.3* 64.9 70.2 80.8** 59.6 77.3 81.9 77.7 84.1 58.6*** 64.6** Hermoso 59.1 60.5 65.2 74.8 55.6*** 71.5*** 76*** 72.1*** 77.7*** 54.7*** 60.3*** Tungurahua 39.6* 40.6 43.7 49.1 37.6 47 49.8*** 47.5 50.6** 37.1** 40.6*** El Altar 63.2 64.7* 70.1 80.7 59.6*** 77.3 82.2** 77.7** 83.9*** 58.6*** 64.6*** Quilimas 65.2 66.9 72.5 83.5 61.3*** 79.9 84.8** 80.2*** 87.1*** 60.3*** 66.8*** Yanaurcu 54.9 56.2 60.6 69.2 51.8*** 66.3 70.2*** 66.5*** 71.9*** 50.8*** 56*** Cajas 69* 70.7 77* 88.8 64.9 84.8** 90.4** 85.6*** 92.8*** 63.6*** 70.9*** Yana Cajas P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 Chil 429 430 Appendix 2 Matrix of geographic distances between the sites in kilometers (above diagonal) and matrix of floristic dissimilarities measured by w2 distance (below diagonal) which were used in the Mantel test. Imba Caya Sara E Anti WAnti Pich Ilin Coto WChim EChim Herm Tung Altar Quil Yana Cajas 0 1.25 0.076 0.015 6.104 0.732 7.73 2.46 5.55 10.6 10.4 0.84 .406 0.79 0.532 0.05 0.0002 0.93 64 0 5.215 1.667 1.324 0.314 0.8 1.825 0.02 2.27 3.56 0.01 0.037 0.142 0.824 0.0001 0.14 0.001 64 22 0 4.713 0.003 0.24 0.95 0.154 0.23 2.51 4.48 0.009 0.59 0.07 0.06 1 0.72 0.83 140 94 80 56 45 0 2.78 1.79 5.81 9.44 15.7 0.001 2.185 0.021 0.581 0.21 0.001 1.64 183 122 120 111 107 69 62 58 0 1.809 0.147 0.01 11.4 1.46 3.4 4.64 4.97 2.05 172 115 108 93 85 43 35 57 30 0 0.337 2.71 11.2 1.77 10.3 11.5 8.52 4.33 272 212 208 193 185 140 133 149 91 101 0 2.08 16.2 4.56 6.57 9.96 7.88 4.52 345 292 282 260 248 205 198 232 177 176 96 92 125 89 64 55 0 0.76 85 54 33 0 0.864 0.143 0.01 1.465 0.109 1.89 3.08 1.842 0.1 0.2 0.021 0.51 0.56 0.38 100 70 49 17 0 5.215 5.19 1.92 4.75 12.5 12.1 0.003 6.089 0.052 4.513 3.525 2.422 0.37 147 98 85 63 52 8 0 0.257 1.776 0.258 0.004 0.318 6.96 2.34 4.96 1.93 6.4 1.99 126 64 64 67 71 58 56 0 1.331 0.174 0.26 0.362 4.54 0.13 1.21 2.17 1.99 2.49 268 209 204 188 179 134 126 147 89 96 11 0 2.58 0.09 0.21 0.2 0.39 0.58 220 171 159 136 123 80 74 118 79 61 81 71 0 0.309 0.095 1.121 0.155 1.43 256 202 193 172 161 117 110 143 92 87 48 37 41 0 0.09 0.13 0.054 2.82 281 228 218 196 184 141 134 169 118 113 57 47 62 26 0 1.8 6.682 0.09 290 237 228 205 193 150 143 179 127 122 62 53 70 35 9 0 2.004 0.25 The positive or negative signs with the w2 values indicate respectively whether or not the observed number of shared species is higher or lower than the expected number. 434 375 370 354 344 299 292 312 254 263 163 167 225 185 165 158 112 0 ARTICLE IN PRESS Cota P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 Chiles Cotacachi Imbabura Cayambe Saraurcu Antisana-east Antisana-west Pichincha Iliniza Cotopaxi Chimborazo-west Chimborazo-east Hermoso Tungurahua El Altar Quilimas Yanaurcu Cajas Chil ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 References Acosta-Solı́s, M., 1985. El Arenal del Chimborazo, ejemplo de puna en el Ecuador. Rev. Geográfica (Quito) 22, 115–122. Almeida-Lenero, L., de Azcarate, J.G., Cleef, A.M., Trapaga, A.G., 2004. Plant communities of the zacatonal alpino area of the Popocatepetl and Nevado de Toluca volcanoes in Central Mexico. Phytocoenologia 34, 91–132. Ayers, T., 1999. Biogeography of Lysipomia (Campanulaceae), a high elevation endemic: an illustration of species richness at the Huancabamba Depression, Peru. Arnaldoa 6, 13–27. Balslev, H., 1988. Distribution patterns of Ecuadorian plant species. Taxon 37, 567–577. Balslev, H., Luteyn, J.L., 1992. Páramo: an Andean Ecosystem under Human Influence. Academic Press, London. Bendix, J., Lauer, W., 1992. Die Niederschlagsjahreszeiten in Ecuador und ihre klimadynamische Interpretation. Erdkunde 46, 118–134. Berg, A., 1998. Pflanzengesellschaften und Lebensformen des Superpáramo des Parque Nacional Sierra Nevada de Mérida in Venezuela. Phytocoenologia 28, 157–203. Cerón, C.E.M., 1994. Vegetación y diversidad en la Reserva de Producción Faunı́stica del Chimborazo–Ecuador. Rev. Geográfica (Quito) 33, 19–42. Cleef, A.M., 1981. The vegetation of the páramos of the Colombian Cordillera Oriental. Diss. Bot. 61, 1–320 J. Cramer, Vaduz. Cleef, A.M., Rangel, C.O.J., Salamanca, S.V., 1983. Reconocimiento de la vegetación de la parte alta del transecto Parque Los Nevados. In: Van der Hammen, T., Pérez, P.A., Pinto, P.E. (Eds.), La Cordillera Central Colombiana. Transecto Parque Los Nevados. Introducción y datos iniciales, vol. 1. J. Cramer, Vaduz, pp. 150–173. Connor, E.F., Simberloff, D., 1978. Species number and compositional similarity of the Galápagos flora and avifauna. Ecol. Monogr. 48, 219–248. Cuatrecasas, J., 1958. Aspectos de la vegetación natural de Colombia. Rev. Acad. Colomb. Ci. Exact. 10, 221–268. Cuatrecasas, J., 1968. Páramo vegetation and its life forms. In: Troll, C. (Ed.), Geo-ecology of the tropical mountainous regions of the tropical Americas. Colloq. Geograph., vol. 9. Bonn, pp. 163–186. Cuatrecasas, J., 1979. Comparación fitogeográfica de páramos entre varias cordilleras. In: Salgado-Labouriau, M.L. (Ed.), El Medio Ambiente Páramo. Centro de Estudios Avanzados, Caracas, pp. 89–99. Cuatrecasas, J., 1986. Speciation and radiation of the Espeletiinae in the Andes. In: Vuilleumier, F., Monasterio, M. (Eds.), High altitude Tropical Biogeography. Oxford University Press, New York, pp. 267–303. Eriksen, B., 1989. 186. Valerianaceae. In: Harling, G., Andersson, L. (Eds.), Flora of Ecuador, vol. 34. Department of Systematic Botany, University of Göteborg, and the Section for Botany, Riksmuseum, Stockholm, pp. 1–60. Frantzen, N.M.L.H.F., Bouman, F., 1989. Dispersal and growth form pattern of some zonal páramo vegetation types. Acta Bot. Neerl. 38, 449–465. Gaviria, J., 1997. Sinopsis del género Lachemilla (Focke) Rydberg (Rosaceae) para Venezuela. Plantula 1, 189–212. 431 Hall, M., 1977. El Volcanismo en el Ecuador. Biblioteca Ecuador, Quito. Hastenrath, S., 1981. The Glaciation of the Ecuadorian Andes. Balkema, Rotterdam. Hedberg, O., 1957. Afroalpine vascular plants. A taxonomic revision. Symb. Bot. Upsal. 15, 1–411. Hedberg, O., 1992. Afroalpine vegetation compared to páramo: convergent adaptations and divergent differentiation. In: Balslev, H., Luteyn, J.L. (Eds.), Páramo: an Andean Ecosystem under Human Influence. Academic Press, London, pp. 15–29. Hill, M.O., 1979. TWINSPAN: a FORTRAN Program for Arranging Multivariate Data in an Ordered Two-Way Table by Classification of the Individuals and Attributes. Cornell University, Ithaca. Hofstede, R., Segarra, P., Mena, P.V. (Eds.), 2003. Los Páramos del Mundo. Proyecto Atlas Mundial de los Páramos. Global Peatland Initiative/NC-IUCN/EcoCiencia, Quito. Islebe, G.A., Cleef, A.M., 1995. Alpine plant communities of Guatemala. Flora 190, 79–87. Jørgensen, P.M., León-Y, S. (Eds.), 1999. Catalogue of the vascular plants of Ecuador, Monogr. Syst. Bot. Missouri Bot. Gard. 75, 1–1181. Jørgensen, P.M., Ulloa, C.U., 1994. Seed plants of the high Andes of Ecuador – a checklist. AAU Reports 34, 1–443. Jørgensen, P.M., Ulloa, C.U., Madsen, J.E., Valencia, R.R., 1995. A floristic analysis of the high Andes of Ecuador. In: Churchill, S.P., Balslev, H., Forero, E., Luteyn, J.L. (Eds.), Biodiversity and Conservation of Neotropical Montane Forests. The New York Botanical Garden, Bronx, pp. 221–237. Legendre, P., Vaudor, A., 1991. The R Package: Multidimensional Analysis, Spatial Analysis. Département de Sciences Biologiques, Université de Montréal. León-Y, S., 2000. La flora de los páramos ecuatorianos. In: Josse, C., Mena, P.A., Medina, G. (Eds.), La Biodiversidad de los Páramos. Serie Páramo 7, GTP/Abya Yala, Quito, pp. 5–21. Luteyn, J.L., 1992. Páramos: why study them. In: Balslev, H., Luteyn, J.L. (Eds.), Páramo: an Andean Ecosystem Under Human Influence. Academic Press, London, pp. 1–14. Luteyn, J.L., 1999. Páramos: a checklist of plant diversity, geographical distribution, and botanical literature. Mem. New York Bot. Gard. 84, 1–278. Luteyn, J.L., 2002. Diversity, adaptations, and endemism in neotropical Ericaceae: biogeographical patterns in the Vaccinieae. Bot. Rev. 68, 55–87. Luteyn, J.L., Cleef, A.M., Rangel, C.J.O., 1992. Plant diversity in páramo: towards a checklist of páramo plants and a generic flora. In: Balslev, H., Luteyn, J.L. (Eds.), Páramo: an Andean Ecosystem Under Human Influence. Academic Press, London, pp. 71–84. Melcher, I.M., Bouman, F., Cleef, A.M., 2000. Seed dispersal in páramo plants: epizoochorous and hydrochorous taxa. Plant Biol. 2, 40–52. Melcher, I.M., Bouman, F., Cleef, A.M., 2004. Seed atlas of the monocotyledonous genera of the páramo. Flora 199, 286–308. ARTICLE IN PRESS 432 P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 Monasterio, M., 1979. El páramo desertico en el altiandino de Venezuela. In: Salgado-Labouriau, M.L. (Ed.), El Medio Ambiente Páramo. Centro de Estudios Avanzados, Caracas, pp. 117–146. Mueller-Dombois, D., Ellenberg, H., 1974. Aims and Methods of Vegetation Ecology. Wiley, New York. Pérez, P.A., Van der Hammen, T., 1983. Unidades ecogeográficas y ecosistemas en el Parque Natural Los Nevados: una sintesis inicial. In: Van der Hammen, T., Pérez, P.A., Pinto, E.P. (Eds.), La Cordillera Central Colombiana. Transecto Parque Los Nevados. Introducción y datos iniciales, vol. 1. J. Cramer, Vaduz, pp. 277–300. Pringle, J., 1995. 159A. Gentianaceae. In: Harling, G., Andersson, L. (Eds.), Flora of Ecuador, vol. 53. Department of Systematic Botany, University of Göteborg, and the Section for Botany, Riksmuseum, Stockholm, pp. 1–131. Ramsay, P.M., 1992. The páramo vegetation of Ecuador: The community ecology dynamics and productivity of tropical grasslands in the Andes. Ph.D. thesis, University of Wales, Bangor. Ramsay, P.M., Oxley, E.R.B., 1997. The growth form composition of plant communities in the Ecuadorian páramos. Plant Ecol. 131, 173–192. Rangel, C.O.J., Diaz, S.P., Jaramillo, R.M., Salamanca, S.V., 1983. Lista del material herborizado en el transecto del Parque Los Nevados (Pteridophyta–Spermatophyta). In: Van der Hammen, T., Pérez, P.A., Pinto, P.E. (Eds.), La Cordillera Central Colombiana. Transecto Parque Los Nevados. Introducción y Datos Iniciales, vol. 1. J. Cramer, Vaduz, pp. 174–205. Ricardi, H.M.S., Gaviria, J., Estrada, J., 1997. La flora del superpáramo venezolano y sus relaciones fitogeográficas a lo largo de los Andes. Plantula 1, 171–187. Robinson, H., 1997. New species of Aphanactis in Ecuador and Bolivia and new combinations in Selloa (Heliantheae: Asteraceae). Brittonia 49, 71–78. Salamanca, 1992. La vegetación del páramo y su dinámica en el macizo volcanico Ruiz-Tolima (Cordillera Central, Colombia). Análisis Geográficos (Bogotá) 21, 1–155. Santana Castañeda, E., 1994. Las especies de Draba de la Cordillera Oriental, Colombiana. Trabajo de grado, Universidad Nacional de Colombia, Bogotá. Simberloff, D., Heck, K.L., McCoy, E.D., Connor, E.F., 1981. There have been no statistical tests of cladistic biogeographical hypotheses. In: Nelson, G., Rosen, D.E. (Eds.), Vicariance Biogeography. A Critique. Columbia University Press, New York, pp. 40–93. Simpson, B.B., 1974. Glacial migrations of plants: island biogeographic evidence. Science 185, 698–700. Simpson, B.B., 1975. Pleistocene changes in the flora of the high tropical Andes. Paleobiol 1, 273–294. Simpson, B.B., 1983. An historical phytogeography of the high Andean flora. Rev. Chilena Hist. Nat. 56, 109–122. Simpson, B.B., Todzia, C.A., 1990. Patterns and processes in the development of the high Andean flora. Am. J. Bot. 77, 1419–1432. Sklenář, P., 2000. Vegetation ecology and phytogeography of Ecuadorian superpáramos. Ph.D. thesis, Charles University, Prague. Sklenář,, P., 2001a. Superpáramo flora and vegetation of Volcán Chiles. In: Ramsay, P.M. (Ed.), The Ecology of Volcán Chiles: High Altitude Ecosystems on the Ecuador–Colombia Border. Pebble & Shell, Plymouth, pp. 39–45. Sklenář, P., 2001b. Notes on taxonomy, distribution and ecology of Ecuadorian high páramo Asteraceae, tribe Senecioneae. Compositae Newsletter 36, 1–8. Sklenář, P., Jørgensen, P.M., 1999. Distribution patterns of páramo plants in Ecuador. J. Biogeogr. 26, 681–691. Sklenář, P., Lægaard, S., 2003. Rain-shadow in the high Andes of Ecuador evidenced by páramo vegetation. Arctic Antarctic Alpine Res. 35, 8–17. Sklenář, P., Ramsay, P.M., 2001. Diversity of páramo plant communities in Ecuador. Diversity Distribution 7, 113–124. Sklenář, P., Robinson, H., 2000. Two new species of Oritrophium and Floscaldasia (Asteraceae, tribe Astereae) from the superpáramos of Ecuador. Novon 10, 144–148. Smith, J.M.B., 1975. Notes on the distribution of herbaceous angiosperm species in the mountains of New Guinea. J. Biogeogr. 2, 87–101. Smith, J.M.B., Cleef, A.M., 1988. Composition and origins of the world’s tropicalpine floras. J. Biogeogr. 15, 631–645. Stern, M.J., Guerrero, M.G., 1997. Sucesión primaria en el Volcán Cotopaxi y sugerencias para el manejo de hábitats frágiles dentro del Parque Nacional. In: Valencia, R., Balslev, H. (Eds.), Estudios Sobre Diversidad y Ecologı́a de Plantas. Memorias del II Congreso Ecuatoriano de Botánica, Pontificı́a Universidad Católica del Ecuador, Quito. Centro de Publicaciones, P.U.C.E., Quito, pp. 217–229. Ter Braak, C.J.F., Šmilauer, P., 1998. CANOCO Reference Manual and User’s Guide to Canoco for Windows. Wageningen, Centre of Biometry. Troll, C., 1959. Die tropische Gebirge. Bonner Geogr. Abhandl. 25, 1–93. Troll, C., 1968. The cordilleras of tropical Americas. Aspects of climatic, phytogeographical and agrarian ecology. In: Troll, C. (Ed.), Geo-ecology of the Tropical Mountainous Regions of the Tropical Americas. Colloq. Geogr, vol. 9. Bonn, pp. 15–56. Valencia, R., Pitman, N., León-Y., S., Jørgensen, P.M. (Eds.), 2000. Libro Rojo de las Plantas Endémicas del Ecuador 2000. Herbario QCA, P.U.C.E., Quito. 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., Cleef, A.M., 1986. Development of the high Andean páramo flora and vegetation. In: Vuilleumier, F., Monasterio, M. (Eds.), High Altitude Tropical Biogeography. Oxford University Press, New York, pp. 153–201. Van der Hammen, T., Werner, J.H., van Dommelen, H., 1973. Palynological record of the upheaval of the Northern Andes: a study of the Pliocene and Lower Quaternary of the Colombian Eastern Cordillera and the early evolution of its high-Andean biota. Palaeogeogr. Palaeoclim. Palaeoecol. 16, 1–24. ARTICLE IN PRESS P. Sklenář, H. Balslev / Flora 200 (2005) 416–433 Vargas, H., Neill, D., Asanza, M., Freire-Fierro, A., Narvaez, E., 2000. Vegetación y flora del Parque Nacional Llanganates. In: Vazquez, M.A., Larrea, M., Suarez, L. (Eds.), Biodiversidad en el Parque Nacional Llanganates. EcoCiencia, Quito, pp. 13–66. Vuilleumier, B.S., 1970. Insular biogeography in continental regions. I. The northern Andes of South America. Am. Naturalist 104, 373–388. Vuilleumier, F., Monasterio, M., 1986. High Altitude Tropical Biogeography. Oxford University Press, New York. 433 Whittaker, R.H., 1960. Vegetation of the Siskiyou Mountains, Oregon and California. Ecol. Monogr. 30, 279–338. Wilson, M.V., Shmida, A., 1984. Measuring beta diversity with presence–absence data. J. Ecol. 72, 1055–1064. Xena de Enrech, N., 1992. Valerianaceae. In: Morillo, G. (Ed.), Flora de Venezuela, vol. 5(1) Acta Cientı́fica Venezolana, Caracas. pp. 221–267. Young, K.R., Ulloa, C.U., Luteyn, J.L., Knapp, S., 2002. Plant evolution and endemism in Andean South America: an introduction. Bot. Rev. 68, 4–21.