Catalogue of the Vascular Plants of Ecuador

By Peter M. Jørgensen and Susana León-Yánez

This section analyzes and compares the species composition of various areas within Ecuador. The attributes available for analysis and comparison are taxonomic groups, regions, elevation zones, habits, and provinces, and combinations of these variables. The number of accepted species listed in the catalogue is 16,087. This includes 595 introduced species and 186 species expected to occur in Ecuador. This leaves 15,306 documented native species. Of the introduced species, 346 were recorded as cultivated, i.e., 58% of the introduced species are crops or ornamentals, and 249 species or 42% are considered to be accidental introductions. The number of endemic species is 4,173 or 27.3% of the total number of native species; 10,988 or 71.8% were listed as native (and non-endemic); 98 or 0.6% were listed as cultivated without indication as to being native or introduced; and 47 or 0.3% were listed as of unknown origin. All further calculations were performed on the 15,306 native and endemic species, unless otherwise indicated. The numbers provided within the Catalogue include both expected as well as introduced species and may therefore differ slightly from numbers provided throughout this section.

Taxonomic groups

The number of pteridophytes listed in the catalogue is 1,298 or 8.5% of the species, the count of gymnosperms is 17 species (0.1%), and that of angiosperms 13,991 species (91.4%); the monocotyledons include 5,176 species (33.8%) and the dicotyledons 8,815 (57.6%).

The number of families reported is 273, of which 254 are native. The highest number of species is found in the Orchidaceae, followed by the Asteraceae, Melastomataceae, Rubiaceae, Poaceae, Bromeliaceae, Piperaceae, Araceae, Solanaceae, and Dryopteridaceae. Table 3 lists all native families with the number of species, the percentage of species of all species, and the number and percentage of endemic species in the family. Families with high endemicity are Nolanaceae (100%), Actinidiaceae (68.2%), Campanulaceae (61.3%), Molluginaceae (57.1%), Begoniaceae (56.9%), Gentianaceae (53.8%), Berberidaceae (53.3%), Ericaceae (50.4%), Linaceae, Loasaceae, Saxifragaceae, Thymelaeaceae, and Zygophyllaceae (last five families with 50.0%). These families are predominantly Andean, except for the Nolanaceae—with a single endemic species on Galápagos—and the Molluginaceae—with four endemic species on Galápagos and three widespread species also found in the Coastal regions. High levels of endemism are clearly associated with either the Andes or Galápagos, whereas such high levels of endemism are not observed in the lowlands of the Coastal and Amazonian regions. A total of 106 families have no endemic species in Ecuador.

The number of native genera occurring in Ecuador is 2,110. A total of 390 genera are listed in the Catalogue based only on introduced species. Additionally, a few genera are included based on material not yet identified to species or as expected genera. The highest number of species are found in Pleurothallis, followed by Epidendrum, Lepanthes, Miconia, Anthurium, Peperomia, Piper, Masdevallia, Maxillaria, and Solanum. A list of all native genera with the number and percentage of species can be found in Table 4.

It is remarkable that one in five plant species in Ecuador is an orchid, and that the Orchidaceae also contain the three most diverse genera and five of the ten largest genera. Only 85 or 4% of the genera are represented exclusively by endemic species. The number of endemic genera in Ecuador is 23, of which 12 are Andean, , seven are from the Galápagos Islands, three from the Coastal region, and only one from the Amazonian lowlands (Table 5). The highest number of endemic species is found in Pleurothallis, followed by Lepanthes, Anthurium, Masdevallia, Epidendrum, Miconia, Piper, Peperomia, Elaphoglossum, and Maxillaria. It is noteworthy that the fern genus Elaphoglossum is among the highest-ranked endemic genera; high endemism is a trait normally not seen in the pteridophytes.

Most families and genera are represented by a few species. Thirty-one families or 12% are represented by one species and 129 families or 51% have 14 or fewer species, while 125 or 49% have 15 or more species. A total of 191 families or 75% have 50 or fewer species; 1,890 genera or 90% have 14 or fewer species, while only 220 or 10% have 15 or more species (Figure 3).

Regions

The highest number of species is found in the Andean region, with 9,865 species or 64.4% of the total. The Coastal region is represented by 4,463 species or 29.2%, and 4,857 species or 31.7% are found in the Amazonian region. Only 699 species or 4.6% are found on Galápagos. It is apparent that the high diversity of plants in Ecuador is to a large degree due to the presence of the Andes.

The average number of regions recorded for a species is 1.32 with a standard deviation of 0.61, taking into account only species where at least one region was recorded. The number of species without a recorded region is 291 or 1.9% of the total. Species recorded only in one region number 11,201 (73.2%), 2,866 (18.7%) species are found in two regions, 840 (5.5%) species are found in three regions, and 108 (0.7%) species are found in all four regions.

A comparison of the species composition in the four different regions reveals low levels of similarity between them. The highest similarity is found between the Amazonian and Coastal regions, followed by the Andean and Coastal regions, and the Andean and Amazonian regions (Table 6).

To visualize how the composition of the areas is related, we used both cluster analysis and ordination. The cluster analysis used was a hierarchical agglomerative analysis with group average linking; the distance measure was Sørensen's index. The software package PC-ORD (McCune & Mefford, 1997) was used to construct the diagram. The same software package was also used to make Nonmetric Multidimensional Scaling (NMDS) ordinations, and the distance measure used was again Sørensen's index. The ordination was tested using a Monte Carlo test (20 runs) to indicate whether the axes extracted contained more information than would be expected by chance. We are aware that 20 is a small number of runs for the Monte Carlo test to be certain that the result does not deviate from random. Time constraints induced by the amount of data in the analyzed matrices forced us to be satisfied with a lower level of confidence for the time being. These methods were used for all the cluster analyses and ordinations described in the following sections.

The cluster diagram of the species composition of the regions (Figure 4) shows, as does Table 6, that the Coastal and Amazonian regions have the highest similarity; this cluster is then united with the Andean region, and all are finally connected to the Galápagos. The NMDS ordination performed on these data was not significantly different from random and is therefore not presented. The reason, presumably, is that we have only four data points; the ordination tends to place them as far away from each other as possible, and it becomes difficult to detect relationships. Another fact that influences the result is that the Galápagos region has, compared to the other regions, very few species. This does not mean that the regions have no floristic differences.

An analysis of the taxonomic groups in the regions reveals that in most cases the same families dominate the mainland regions (Table 7). The exceptions are the Fabaceae, which have a higher relative importance in the Coastal region, the Mimosaceae, which are more diverse in the Amazonian region, the Poaceae, which have high relative importance in the Andean and Coastal regions, and the Solanaceae, which are more dominant in the Andean and Amazonian regions. The Galápagos region shows an almost complementary composition of families, with diverse families being the Amaranthaceae, Convolvulaceae, Cyperaceae, Euphorbiaceae, Malvaceae, and Pteridaceae. All these families are present on the mainland but have a lower relative diversity.

It is obvious that most of the diverse genera on the mainland are epiphytic, such as Anthurium, Epidendrum, Lepanthes, Masdevallia, Maxillaria, Peperomia, Philodendron, and Pleurothallis. The only non-epiphytic genera with high relative diversity are Inga, Miconia, Paullinia, Piper, Psychotria, and Solanum (Table 8). Solanum is among the most diverse genera in all four regions, showing a high plasticity and capability to adapt to different environments. None of the diverse genera of epiphytes has reached the Galápagos Islands, and only one genus, Solanum, is among the most diverse both on Galápagos and the mainland. The nine other diverse genera on Galápagos are not among the most diverse on the mainland. An island effect is obvious here. Genera that have reached the Galápagos have been able to diversify due to geographical separation and a lack of competition.

Elevation zones

Similar to the regional analysis, we calculated the number of species found in the different 500-m elevation zones used in the Catalogue. The highest number of species is found in the lowlands, 0–500 m, followed by the 500–1000-m zone, and this pattern is continued upwards until the zone above 4,500 m is reached. It is noteworthy that the area of the different zones is not comparable to the number of species (Table 9).

A comparison of the natural logarithm of the area of the elevation zones and the number of species in the different elevation zones shows that the curves have a similar shape (Figure 5A). A plot of the logarithm of the number of species versus the logarithm of the area shows significant correlation (R² = 0.7444, p<0.01), indicating a clear relationship between total diversity and area. Two areas have fewer than expected species (their data points are placed below the trend line). They are the area above 4,500 m and the area below 500 m elevation (Figure 5B). These areas apparently have fewer species (although not significantly fewer) per unit area than the other areas. It is not surprising that the area above 4,500 m has few species, since that zone is close to the physiological limit for plant growth. The lowland areas have suffered considerably from human impact for hundreds, maybe thousands, of years on the coast, although the destruction is less drastic and more recent on the Amazonian side.

The average number of elevation zones for a species is 2.19, with a standard deviation of 1.39. Because we have recorded presence in an elevation zone, not the actual elevation, this number cannot be converted to an elevation range. It is frequently observed that a species spans 1,000 to 1,500 m in vertical elevation. The number of species without recorded elevation was 442 (2.9%); 6,303 (41.2%) species have one elevation zone recorded, 3,595 (23.5%) have two, 2,466 (16.1%) have three, 1,392 (9.1%) have four, 650 (4.2%) have five, 320 (2.1%) have six, 105 (0.7%) have seven, 29 (0.2%) have eight, and four (0.02%) have nine. Most species have a restricted elevational distribution: 64.7% occur in two or less elevation zones, and 80.8% of the species occur in three or fewer elevation zones.

The cluster analysis of the elevation zones indicates that three clusters of species can be recognized (Figure 6A). Two breaks are detected in the species composition: one at 1,500 m and the other at 3,500 m. Secondary breaks are found at 500 m, 2,500 m, and 4,500 m. Comparing the raw similarity index values in Table 9 does not indicate the same. The 2,000–2,500-m zone compared to the 2,500–3,000-m zone has a similarity index value of 0.63, and the same value is found between the 2,500–3,000-m and the 3,000–3,500-m zones. The two zones are placed in different clusters. The technique used for cluster analysis is therefore not well suited when index values are similar in more than one direction or a continuum occurs.

The NMDS configuration of the species composition of the elevation zones shows that the grouping found by the cluster analysis (surrounded by dashed lines or with gray shading) is only weakly supported (Figure 6B). The ordination is well supported by the Monte Carlo test (p£ 0.05 for the second and third axes), indicating that a significant amount of information was extracted for those axes. The distance between points within a cluster is often equal to or larger than between points belonging to different clusters, so the separation between the clusters is not well supported. It seems that each elevation zone has about the same amount of difference or similarity to the zone immediately above and below, and to zones removed one or more levels.

Another way of analyzing the number of species along the elevation gradient is to construct so-called "cohorts of species" (Stotz et al., 1996). The technique includes the counting of species according to their minimum elevation, so an elevational cohort consists of species whose minimum elevation falls within a particular 500-m interval. We counted the number of species found in the Coastal zone between 0 and 500 m and how many of those are found in the zone immediately above it. The process is then repeated for the subsequent zones. We eliminated the Galápagos region from this analysis; then we did the calculations twice, first arranging the data from a Coastal perspective and second from the Amazonian perspective. We can see that there is limited difference between the two figures (Figure 7A & B). However, it is clear that there are more species on the Amazonian side of the Andes, and that they apparently are more limited in their elevational distribution, as the curve representing 0–500 m on Figure 7B tapers off faster than the same curve on Figure 7A. The Coastal zone of 500–1,000 m has a few more species than the same zone on the Amazonian side, resulting in a less sharply marked decrease in species towards the Andes on the Coastal versus the Amazonian side. The number of species added to the flora entering the Andean region at 1,000–1,500 m elevation, regardless of whether arriving from the Amazonian (2,651 species added) or Coastal side (2,579 species added), is remarkable. The figure also shows that the documented diversity of the Andean zones remains very high until the 3,500-m level is reached, which is the lower limit of the páramo vegetation zone.

Elevation and taxonomic groups

A comparison between the elevation variable and the ten most diverse families in the different elevation zones is presented in Table 10. Families that are among the ten most diverse in all elevation zones are the Asteraceae and Poaceae. A relatively high number of families are species-rich at low elevations and extend well into the middle elevations, i.e., Araceae, Bromeliaceae, Dryopteridaceae, Melastomataceae, Orchidaceae, Piperaceae, and Rubiaceae. Families with high relative diversity exclusively at low elevations are Fabaceae, Gesneriaceae, and Mimosaceae. Families with high values exclusively at middle elevations are Campanulaceae and Ericaceae. Families that have high relative diversity in the upper elevation zones are Apiaceae, Brassicaceae, Caryophyllaceae, Cyperaceae, Gentianaceae, Lycopodiaceae, Ranunculaceae, Rosaceae, and Valerianaceae. We conclude that the diverse families in the páramo zone, above 3,500 m elevation, are different from those at middle and low elevations. The diverse families at the middle elevation show a limited turnover in comparison with the lowland zones.

The same analysis was performed for the genera. As for the families, certain genera are more diverse at certain levels (Table 11). Also, many genera maintain high levels of relative diversity from sea level to about 3,500 m elevation where another group of genera replaces them. Miconia and Solanum are remarkable, as they are among the most diverse genera in eight of the ten elevation zones. Epidendrum and Peperomia are highly diverse in seven elevation zones, and Lepanthes, Maxillaria, and Pleurothallis in six elevation zones.

Elevation and regions

To refine our analysis of species diversity by elevation, we combined the elevation parameter with the regions. As the regions are defined by elevation there should, in theory, be no noise in the data. This is, however, not always the case when we combine variables, as will be seen later. The combination of elevations and regions is used as one variable in the rest of the analysis. The highest number of species (4,303 or 28.1%) is found in the Andean region between 1,000 and 1,500 m, followed by the 1,500–2,000-m zone with 26.6% of the species. The Amazonian 0–500-m zone is ranked third with 26.1% of the species, followed by the Andean elevation zones 2,500–3,000 m with 25.6%, and 2,000–2,500 m with 25.5% species. The Coastal elevation zones 0–500 m and 500–1000 m are ranked fifth and sixth with 23.1% and 19.1%, respectively. The Amazonian elevation zone 500–1,000 m is ranked seventh with 18.6%. The rest of the Andean zones follow in the expected sequence, with some interruptions: 3,000–3,500 m with 17.8%, 3,500–4,000 m with 8.4%, Galápagos 0–500 m with 4.2%, Andean 4,000–4,500 m with 3.7%, Galápagos 500–1,000 m with 3.2%, Galápagos 1,000–1,500 m with 1.8%, and, finally, Andean above 4,500 m with 0.8% (absolute numbers can be found in Table 12). It may come as a surprise that the lowlands are not the most diverse area. The high alpha diversity seen in numerous rain forest plot-studies is more than equalled in the mountains by high beta diversity, and the total numbers for the highlands exceed the lowland numbers. It could be argued that the area between 1,000 and 1,500 m represents both the eastern and western slopes so the number of species for any given side may be only about half the number listed. It is noteworthy that the highland zone between 3,000 and 3,500 m has almost as many species recorded as the Coastal and Amazonian lowland zones between 500 and 1,000 m elevation.

The cluster analysis of the areas defined by the combination of elevations and regions shows four clusters (Figure 8A). One cluster is composed of the three elevation zones on the Galápagos Islands, a second by the Coastal and Amazonian lowlands below 1,000 m, a third by the forested part of the Andes up to 3,500 m, and fourth what could be called a páramo zone from about 3,500 m and upwards. The clustering makes sense, except perhaps for the way the Galápagos and páramo clusters are united to the other two clusters.

The NMDS ordination of the species composition of the areas defined by elevation and region has a drastic drop in stress values between 1- and 2-axes. The extracted information for all three axes is larger than what could be expected by chance (Monte Carlo test, p£ 0.05 for axes 1–3). The configuration shows clearly that the Galápagos Islands have a composition that groups the different elevation zones together and separates them from all other zones (Figure 8B). The Coastal and Amazonian lowland cluster also seems to be supported in the ordination; even a subdivision of that cluster into separate Coastal and Amazonian clusters could be justified. This is supported by the fact that only about 30% of the species are shared between the Coastal and Amazonian regions and must have evolved prior to the rise of the Andes about 5 million years ago (Table 6). It is also interesting to see that the Amazonian lowlands have a stronger association with the Andean zones than with the lowland Coastal zone. This is probably caused by a higher number of dry forest and savanna elements on the coast; such elements are generally not found in the Amazonian lowlands or in the Andean zones. The break in the Andean zones at 3,500 m is supported and probably at 4,500 m, as well. The position of the zone above 4,500 m may be affected by an outlier effect; the zone contains only 121 species, a value considerably below the other zones. The break at the 2,500-m level indicated by the cluster analysis is only very weakly supported. It should be noted in this context that the 500-m elevation zones were arbitrarily selected, and if biogeographical zones are discrete entities they apparently do not always coincide with the 500-m increments. Axis one, presented on Figure 8B, represents elevation, while axis three can be interpreted largely as an east-west gradient with some component of elevation.

Elevation, regions, and taxonomic groups

We identified the ten most diverse families (Table 13) and genera (Table 14) for the combined regions and elevation zones by splitting the lower zones in Tables 7 and 10 and Tables 8 and 11 into smaller geographical units. We have not repeated most of the Andean zones as that would be redundant. We see, when comparing Tables 7, 10, and 13, that some of the families that are among the ten most diverse families on Galápagos (Convolvulaceae, Euphorbiaceae, Malvaceae, and Pteridaceae) do not have sufficient diversity on the mainland to be included among the ten most diverse families. The families that were added to Table 10, but not present in Table 7, are Apiaceae, Brassicaceae, Campanulaceae, Caryophyllaceae, Ericaceae, Gentianaceae, Gesneriaceae, Lycopodiaceae, Ranunculaceae, Rosaceae, Scrophulariaceae, and Valerianaceae. These families are all diverse at higher elevations, but do not reach values sufficiently high to include them at the regional level. Only three families not present in Table 7 or 10 are added to Table 13: Boraginaceae, Sapindaceae, and Thelypteridaceae. These families are among the most diverse families only in a particular elevation zone within a particular region.

A comparison of Tables 8, 11, and 14 indicates that the diverse genera on Galápagos, Alternanthera, Amaranthus, Asplenium, Chamaesyce, Cyperus, Ipomoea, Paspalum, Scalesia, and Thelypteris, and a single genus Paullinia, diverse in the Amazonian region, are absent (not sufficiently diverse) when we look at diverse genera in the other elevation zones. Genera added in Table 11 are Arenaria, Azorella, Baccharis, Bartsia, Calamagrostis, Calceolaria, Centropogon, Draba, Elaphoglossum, Festuca, Gentianella, Geranium, Huperzia, Lachemilla, Poa, Tillandsia, Valeriana, and Xenophyllum. The majority of these genera are diverse only above 3,500 m, except for Calceolaria, Centropogon, and Tillandsia, which reach high levels of relative diversity at 2,500 m, 2,500 m, and 1,500 m, respectively. Four genera listed in Table 14 were not previously listed: Blechnum (with high relative diversity at 1,000–1,500 m on Galápagos), Borreria (with high diversity at 0–1,000 m on Galápagos), Diplazium (with high diversity at 500–1,000 m in the Coastal region), and Tiquilia (with high diversity at 1,000–1,500 m on Galápagos).

Habits

The habits of the species were recorded in 12 different classes: herb, subshrub, shrub, treelet, tree, vine, liana, epiphyte, parasite, hemiepiphyte, aquatic, and saprophyte. Combinations of habits for a species do occur, and in such cases all habits are counted without making a judgment of the predominant habit for the species.

The average number of habits per species is 1.27, with a standard deviation of 0.51. No species have been recorded without a habit, 11,646 (76.1%) are recorded with only one habit, 3,274 (21.4%) with two, 351 (2.3%) with three, 34 (0.2%) with four, one (0.01%) with five.

A total of 5,752 species (37.6%) are herbs, 3,953 (25.8%) are epiphytes, 3,333 (21.8%) shrubs, 2,736 (17.9%) trees, 966 (6.3%) vines, 722 (4.7%) treelets, 720 (4.7%) lianas, 670 (4.4%) subshrubs, 302 (2.0%) hemiepiphytes, 111 (0.7%) aquatics, 105 (0.7%) parasites, and 18 (0.1%) saprophytes. The number of species that exhibit more than one habit is, as expected, low, but we detected a few groupings such as herbs and epiphytes; shrubs, treelets, and trees; and maybe vines and lianas (Table 15). Between the first two groups we may place the subshrubs, and between the last two the hemiepiphytes (Table 15). We have not performed a cluster or NMDS analyses of the composition of the different habits because of the very low level of shared species between habits.

Habits, elevation zones, and regions

We counted the number of species in each habit class in the 15 combined elevation-region zones (Table 16). We have indicated the three highest values both in a vertical sense (habits) and horizontal sense (region-elevation zones) and can conclude that the most diverse habit groups in all elevation-region zones are the herbs and shrubs. The epiphytes are ranked among the most diverse in the zone from 500 to 3,500 m on the mainland. Subshrubs are a very diverse group on Galápagos and above 3,500 m elevation. Trees are ranked among the most diverse groups only between 0 and 500 m elevation in the Coastal and Amazonian regions. Mid-level maximum of diversity is found for the herbs, subshrubs, shrubs, epiphytes, and parasites between 1,000 and 3,000 m elevation. The treelets, trees, lianas, hemiepiphytes, and saprophytes show maximum diversity below 1,500 m. The vines show two maxima, one in the Coastal region below 500 m and another at mid-level elevation between 2,000 and 3,000 m. A tendency towards two maxima can also be seen in the shrubs and herbs, and several other habits have a mid-level hump at the 2,500–3,000-m zone. To some extent this may be caused by collecting bias: the highland areas of the inter-Andean valleys (2,500–3,000 m) are very well collected, leading us to believe that more species are present but not yet recorded in the elevation zones immediately below. The aquatic plants are obviously dependent on available water, so topography influences their distribution. The maximum values for aquatics are found in the lowland Amazonian and Coastal regions below 500 m elevation and in the elevation zone that includes the inter-Andean valleys where the large lakes are located. Low diversity is encountered elsewhere, where only fast-running water can be found. Several habits, e.g., parasites, subshrubs, vines, and shrubs, seem to reach a plateau at 2,500 or 3,000 m and remain somewhat constant until the lowermost zones are reached (Figure 9B & 9C). It can further be observed that the flora of the Galápagos Islands is predominantly herbaceous: more than 50% of the species are herbs, followed by high values for shrubs and subshrubs. The Coastal 0–500-m zone has a higher relative value of herbs than the same elevation on the Amazonian side where the trees dominate; the distribution is, however, more balanced among the habits in the lowlands below 1,000 m elevation than in any other elevation zone. High relative importance of trees coincides with high absolute numbers, whereas low numbers of herbs coincide with high relative importance in the highest Andean zones. Epiphytes have their maximum relative importance between 1,000 and 2,000 m, whereas the shrubs are more important with maximum relative values between 2,500 and 4,000 m (Figure 9A).

The cluster diagram of the 166 combinations of regions, elevation zones, and habits shows three super-clusters and 23 clusters (Figure 10). The super-clusters are defined by the habits. The clusters are defined by a mixture of habit and elevation. Within the first super-cluster, breaks are detected for the herbs at 2,500 m and for the epiphytes at 3,500 m elevation. The fourth cluster is composed of epiphytes from Galápagos and the Andean region between 3,500 and 4,500 m. This is a rather strange cluster and is only very loosely held together, since there are very few epiphytes in all these zones. We can also see that the different regions form sub-clusters within the four clusters. The next super-cluster is defined by woody habits. The subshrubs are undivided, or at best may be divided at 2,000 m. The woody plants found on Galápagos have a distinct composition and form a separate cluster. The shrubs break at 1,000 m elevation while the treelets and trees do not break until 2,500 m elevation. Other breaks exist for shrubs, treelets, and trees at 4,000 m. The last super-cluster is composed of the climbing habits: vines, lianas, and hemiepiphytes. The vines break at 1,500 m and the lianas at 2,000 m elevation. The hemiepiphytes are undivided, except for the splitting off of the Galápagos hemiepiphytes and lianas. The bottom five clusters are: parasites on the mainland, aquatics below 1,000 m (except for Galápagos above 1,500 m), aquatics above 1,000 m, parasites on Galápagos, and saprophytes. These three life-forms are so special that they hardly have anything in common with the other habits, and they are therefore split off "early" from the majority of species.

For the NMDS ordination of the composition of species defined by the habit, elevation, and region variables, we excluded the aquatic, parasitic, and saprophytic habits. These habits have, as shown by the cluster analysis and as indicated by Table 15, very little overlap with any of the other habits. The effect of including them would place these groups on the periphery of the diagram and, in effect, compress the other groups into the center of the diagram, resulting in lower resolution for the most interesting habits. We encountered relative large stress values, but with a drastic decrease between the 1- and 2-dimension; the Monte Carlo test indicated p£ 0.05 for all axes. The habits can be seen almost as discrete entities, and the elevation and region combination is to a large extent a continuum, so we expect a configuration with continua along an elevation gradient for each habit. That is to a large degree what we find (Figure 11) and what also was indicated by the cluster analysis. The habits are represented by elongate shapes, with the lowland areas of a habit group placed at one end and the higher elevation placed at the other end. The composition of the Galápagos is so distinct in all habits that separate groups (indicated by shaded areas) can be identified. The breaks at certain elevation levels identified by the cluster analysis are not clearly observed. It is possible that a series of ordinations performed on the individual subsets, representing the different habits, could support or refute the hypothesis raised by the cluster analysis. It is not clear what the axis in this configuration indicates; part of the problem is that habit is not an environmental variable and is therefore difficult to quantify. Axis 2 may be seen as an herbaceous (left) versus woody axis (right), and Axis 3 may show something about the "laxness" of the plant habits, where climbers and herbaceous plants are placed at the top and woody habits at the bottom.

Habits and taxonomic groups

A comparison of the ten most diverse families in the different habit classes (Table 17) shows that certain species-rich families like the Orchidaceae, Bromeliaceae, and Araceae are among the most important epiphytic, hemiepiphytic, and herbaceous families. Families that contain many shrubs, treelets, and trees are Asteraceae, Piperaceae, Rubiaceae, and Solanaceae. The families with many herbaceous plants are monocotyledons, while the families with woody habits are dicotyledons. The above-mentioned families, together with Dryopteridaceae, Gesneriaceae, and Fabaceae, are important families in three to six habit classes. The plasticity in habit expression they display may be seen as an advantage in an evolutionary context. These families were obviously able to take advantage of possibilities offered and are diverse not only in the number of species, but also in the number of habits. Because so few families and species are aquatic, parasitic, or saprophytic, familes with very low diversity also show up in this analysis.

A similar Table was elaborated for the genera (Table 18). It can be seen that only 13% of the most diverse genera are represented by more than one habit. Solanum is among the diverse genera in five habit classes; Piper, Miconia, and Anthurium are among the most diverse in three habit classes; and nine other genera are among the most diverse in two habit classes.

The variable elevation conformed to a continuum, but we gained more insight when we split the lower part into Galápagos, Coastal, and Amazonian; in other words, we combined elevation with the regions. When the region and elevation zones were combined with the habits, we saw clearly that a given habit does not have the same diversity or relative diversity in all elevation zones. A disclaimer is needed here. The data were collected with relation to the species and the territory of Ecuador. A species can therefore have several habits in several elevation zones and regions. It is, however, not necessarily true that all the habits are present in all elevation zones or regional areas. It is quite logical that a species could be a tree at 2,500–3,000 m elevation, change to a shrub at 3,000–3,500 m, and maybe even become a subshrub above 3,500 m. The analyses of all combinations of variables can therefore be seen only as an analysis of the potential distribution of the habit. We believe this error to be minor, as most species are restricted to one or a few habits.

A comparison of the most diverse families and genera in the 120 combinations defined by elevation, region, and habit is presented in Tables 19–30. The number of taxa and number of combined classes is far larger than can be represented in a single Table; therefore we made individual Tables of the most diverse families and genera for each habit class with each elevation zone or 12 small Tables. We indicate the ten most diverse families or genera in each Table. In cases where the same number of species occurred for taxa ranked 8–14, only the seven highest ranked taxa are included in the Table.

All saprophytes found in Ecuador are included in Table 19. Five or 28% of all saprophytes are dicots, while the majority are monocots. They are found below 2,500 m, with a pronounced maximum diversity between 0 and 500 m on the Amazonian side. A total of 16 species can be found in the lowermost zones; they are completely absent from the Galápagos. The most diverse family is the Burmanniaceae with six genera and ten species. Two of the four families are exclusively saprophytic, whereas Wullschlaegelia (Orchidaceae) and Voyria (Gentianaceae) belong to families in which other life forms predominate.

The classic parasitic families, all dicots, are like the saprophytes few in number. Table 20 summarizes all the 105 known parasites in Ecuador, except for the Olacaceae and Santalaceae, which may all be proven to be hemiparasitic; that would add another 16 species to the Table. Eremolepidaceae are a typical Andean family, whose species are found between 1,000 and 3,500 m elevation. Balanophoraceae are present and most diverse in the lowlands, tapering off towards the higher elevation; they disappear somewhere between 3,500 and 4,000 m. Loranthaceae show a similar pattern, while the Viscaceae have a consistently high level of diversity, with a maximum between 2,500 and 3,000 m elevation. It then diminishes abruptly, disappearing between 4,000 and 4,500 m. This is also the maximum elevation the parasites attain in Ecuador. The genera Dendropthora and Phoradendron seem to be replacing each other. Dendropthora is most diverse between 2,000 and 3,500 m, whereas Phoradendron has two peaks of diversity, one at 1,000 to 2,000 m and another in the Coastal lowlands below 500 m elevation. Several other genera and families follow the same pattern, as subsequent Tables will demonstrate.

There are 111 species of aquatic plants. It is noteworthy that a considerably larger number of aquatic families and genera are listed as among the diverse taxa (Table 21) than parasitic families with a comparable number of species. The aquatic life form has evolved numerous times in many taxonomic groups, from ferns and fern allies, to dicots and monocots. The topographic features of Ecuador, with an almost complete absence of lakes between the inter-Andean valleys and the lowlands, limit the distribution of aquatic taxa. The Potamogetonaceae are among the diverse families from the lowlands to between 4,000 and 4,500 m. Other families seem to be almost restricted to the highland lakes, except for a few members in the lowlands. Their distribution may therefore be dictated by the location of lakes and standing water rather than purely by elevation. However, some genera like Eichhornea, Ludwigia, Najas, Nymphaea, and Sagittaria are found exclusively in the lowland areas between 0 and 500 m and may be considered as adapted specifically to warmer waters.

The hemiepiphytes, with 302 species, are more numerous than the aquatic plants, but have fewer very diverse families and genera (Table 22). The Table includes all but a very few families and genera that grow as hemiepiphytes, so it seems that the habit has evolved fewer times, and in several instances it has resulted in a higher number of species. Hemiepiphytes disappear along with the forest trees between 3,500 and 4,000 m and are almost absent on the Galápagos Islands. It is not surprising that families like Araceae, Araliaceae, Cecropiaceae, Clusiaceae, Dryopteridaceae, Marcgraviaceae, and Moraceae are listed among the diverse hemiepiphytic families. These families are most diverse in the lowlands, with the number of species gradually tapering off towards the highlands. Some have their upper limit between 2,000 and 2,500 m, while others reach to between 3,500 and 4,000 m. In most families hemiepiphytism is represented by one or a few genera, the exception being the Araceae represented by Anthurium, Philodendron, Monstera, Rhodospatha, Stenospermation, and Syngonium.

The number of subshrubs is 670, and the number of shared species with the herbs is 278 and with the shrubs 234. This habit is best seen as an intermediate category between herbs and shrubs, at least in the sense it has been used in the Catalogue. No fern or monocot family is represented in the most diverse taxa by subshrubs (Table 23). Families well represented by subshrubs are Asteraceae, Campanulaceae, Fabaceae, Scrophulariaceae, and Solanaceae. Their maximum diversity occurs between 2,000 and 3,500 m. Other families such as Acanthaceae, Boraginaceae, Melastomataceae, Mimosaceae, Piperaceae, and Rubiaceae have their maximum diversity of subshrubs in the lowlands. These latter families lose their diversity, at least as subshrubs, at about 2,000 m. Unlike the families, no genus is diverse through all elevation zones, except perhaps for Solanum. Some genera are diverse through several elevation zones, e.g., Piper from 0 to 2,000 m and Lupinus from 2,500 to 4,000 m. Subshrubs seem to be a very important habit on the Galápagos Islands, and genera like Acalypha, Alternanthera, Borreria, Chamaesyce, and Heliotrophe have their maximum diversity there, although the numbers are small. Subshrubs are relatively few in number, but have a mid-level diversity maximum between 2,500 and 3,000 m. Their relative importance is, however, greatest on the Galápagos Islands and at higher elevations (Figure 9).

The number of liana species is 720; all diverse taxa belong to the dicotyledons (Table 24). The pattern is the same as observed for the habit classes discussed above. The lianas almost disappear at 3,500 m, along with the forest. Like the hemiepiphytes, they depend on support from the trees. Some families are diverse in a wide variety of elevation zones, while others have a restricted distribution at varying elevations, with the majority tending towards the lower elevation zones. Some families or genera, such as Fabaceae and Hippocrateaceae, have their maximum number of species in the lowlands. Passiflora, on the other hand, reaches maximum diversity between 2,500 and 3,000 m. The majority of species are concentrated within few families. This becomes clear when a family like Ranunculaceae is included; it contains only two species of lianas belonging to the genus Clematis, but has a very wide elevational amplitude, and is included only because it grows above 3,500 m elevation. The main liana families are Apocynaceae, Asteraceae, Bignoniaceae, Malpighiaceae, Menispermaceae, Passifloraceae, and Sapindaceae. The lianas have their maximum diversity in the lowlands, below 500 m; they also have their maximum relative importance there (Figure 9).

There are 722 species of treelets. Diverse representatives are found both in the ferns as well as in the monocots, but the overwhelming majority are dicotyledons (Table 25). The patterns found in other habit classes are repeated. There are, however, several genera that have carried this habit to high levels of diversity in several elevation zones: Miconia, Piper, Saurauia, Siparuna, and Solanum. Most treelets disappear at 3,500 m, a few reach 4,500 m, and only one species grows above 4,500 m. Except for the genus Scalesia they are almost absent from the Galápagos Islands. Like the subshrubs this habit may be seen as intermediate between the shrubs and trees; that is further indicated by the number of shared species, 460 with the shrubs, and 343 with the trees. Treelets have their maximum diversity below 500 m, but nearly reach their maximum value between 2,500 and 3,000 m. Their relative importance is almost constant except at elevations above 3,500 m (Figure 9).

There are 966 species of vines. Table 26 lists the most diverse taxa in each elevation zone and may be compared to Table 24 on the lianas. The families Apocynaceae, Asteraceae, Bignoniaceae, Campanulaceae, Convolvulaceae, Fabaceae, Menispermaceae, Passifloraceae, Sapindaceae, and Solanaceae are shared among the most diverse taxa for both habits. The most diverse families are Asclepiadaceae, Asteraceae, Convolvulaceae, Cucurbitaceae, Fabaceae, and Passifloraceae. The most diverse genus is Passiflora. All other families have their diversity spread among several genera. Of the genera only Mikania, Passiflora, and Solanum appear on both Tables 23 and 25. The number of species that exhibit both habits is only 113, so it seems, contrary to what could be expected, that the two habits are well separated. Vines seem to have their maximum diversity either in the lowlands below 500 m, like the lianas, or at 2,000 to 3,000 m elevation. This bimodal diversity does not make sense from an evolutionary perspective. It may be an artifact of collection intensity. The destruction of natural vegetation makes more edges and open areas available, a habitat type well suited for vines as well as for plant collectors. The relative diversity of vines seems to be highest in three distinct zones: Galápagos and Coastal below 500 m, and 2,000 to 3,000 m in the Andes (Figure 9).

The total number of trees species is 2,736. The majority of families have high diversity at lower elevations, and the number of species in some families is large (Table 27); see Annonaceae, Arecaceae, Bombacaceae, Chrysobalanaceae, Clusiaceae, Euphorbiaceae, Fabaceae, Flacourtiaceae, Lauraceae, Melastomataceae, Mimosaceae, Moraceae, and Rubiaceae. Families like Euphorbiaceae, Lauraceae, Melastomataceae, and Rubiaceae continue with a high level of diversity up the Andean slopes to the tree line at about 3,500 m. Other families are well represented at higher elevations from 2,000 or 2,500 m and upward; they are Araliaceae, Asteraceae, Chloranthaceae, Cunoniaceae, Rosaceae, and Theaceae. The Solanaceae seem to be most diverse in the highlands and stretching down to the Coastal and the Amazonian lowland. Most genera are not diverse over a wide elevation span. They have a "normal" distribution of diversity as seen in, e.g., Blakea. The genera that are among the most diverse in various regions and elevation zones are Clusia, Cyathea, Inga, Miconia, Ocotea, and Solanum. The diversity of trees shows two dramatic increases, one from the treeline from 3,500 m to 2,500 m elevation, and a second reaching the lowlands below 500 m. Between the two dramatic jumps a relatively sTable increase in species occurs. The increase of tree species from premontane forests above 500 m to true lowland forest below 500 m is even more dramatic than going from no forest to forest around 3,500 m (Figure 9). The relative importance of trees shows increasing values with decreasing elevation, and a clear maximum diversity on the Amazonian side of the Andes below 500 m (Figure 9).

A total of 3,333 species are shrubs. The families Asteraceae and Solanaceae are diverse in all region-elevation zones, reaching their highest diversity at 2,500–3,000 m (Table 28). The Ericaceae are likewise a very diverse family, but only on the mainland, and their maximum diversity lies at 1,500–2,000 m. The Melastomataceae conform to that pattern, but have two peaks of diversity, one at 1,000–1,500 m and another at 2,500–3,000 m. Some families are diverse in the lowlands and extend to higher elevations, but are not ranked among the most diverse; examples include Verbenaceae and Malvaceae. There is a clear tendency towards many shrubby species in the Andean zones. This may be seen as an adaptation to the mountainous terrain. Creating several stems may increase the survival rate when branches or trunks fall on top, as well as during the frequent landslides. Most genera are very diverse in only one to three region-elevation zones; a few exceptions are Baccharis, Cordia, Miconia, Palicourea, Piper, Psychotria, and Solanum. The diversity of shrubs reaches an absolute maximum between 2,500 and 3,000 m elevation, but a second maximum can be observed below the 500-m elevation line (Figure 9). The relative importance of shrubs is greatest from 2,500 to 4,000 m (Figure 9).

The epiphytes include 3,953 species, but surprisingly few families and genera are represented among the most diverse epiphytic taxa (Table 29). Most families and genera have a wide elevational amplitude, except for the Lycopodiaceae and Brassicaceae, which are among the most diverse only at higher elevations. Only three families are monocotyledons, four are dicotyledons, and nine families are ferns. It is also characteristic that the diversity of epiphytes decreases dramatically at 3,500 m elevation, again explained by the forest line. As with the hemiepiphytes, lianas, and vines, the epiphytes all need support, so they depend on the trees and continuous forest cover to survive. The genera that are among the most diverse in one to three region-elevation zones are Adiantum, Aechmaea, Ceradenia, Columnea, Dichaea, Draba, Elaphoglossum, Huperzia, Melpomene, Nephrolepis, Odontoglossum, Pleopeltis, Polypodium, Stelis, and Terpsichore. Contrary to all other habits, the epiphytes decrease in diversity below 500 m elevation. They have a well-marked mid-level diversity maximum between 1,000 and 1,500 m elevation, and perhaps up to 2,000 m. The highest relative diversity values are found at the same elevations (Figure 9).

There are 5,752 herbaceous species. The number of families listed among the most diverse is 26, of which 12 have an elevational amplitude of more than three zones (Table 30). The fern and fern allies are represented by five families, the monocotyledons by six, and the dicotyledons by 15. Herbs are found in all regions and elevation zones, and are the most diverse group except in the Amazonian lowlands below 500 m, where they are surpassed by the trees (Figure 9 & Table 16). Asteraceae and Poaceae are diverse groups in all elevation zones. Some of the families listed are generally considered temperate groups and are well adapted to higher elevations; these include Apiaceae, Brassicaceae, Caryophyllaceae, Gentianaceae, Geraniaceae, Ranunculaceae, Rosaceae, and Scrophulariaceae. The ecologically dominant family in the páramo areas, the Poaceae, has its highest diversity at lower elevations. The highest diversity of grasses, 133 species, is found between 2,500 and 3,000 m, followed by 125 species in the Coastal lowlands below 500 m, and 119 species at 1,000–1,500 m. This reminds us that diversity does not necessarily correlate with ecological importance. The genera follow the same pattern; some have wide elevation amplitudes, while others are more diversified at a certain elevation. Their relative importance varies with increasing numbers of the other habits (Figure 9). The highest relative values are reached above 4,500 m, but it should also be noted that the herbs are very important on the Galápagos Islands (Figure 9).

Provinces

To analyze plant diversity and composition by political areas such as provinces may at first seem illogical, but they are the only recorded parameters that give us any information on the distribution of the plant species in a north-south direction. The numbers of species found in the different provinces are listed in Table 31. The location of the provinces is found on the political map on the inside front cover.

The average number of provinces in which a species is found is 3.23, with a standard deviation of 2.84. For 224 species no province was recorded. A total of 5,194 species (33.9%) is found in only one province, 2,966 (19.4%) in two provinces, 2,003 (13.1%) in three, 1,333 (8.7%) in four, 1,013 (6.6%) in five, 662 (4.3%) in six, 499 (3.3%) in seven, 366 (2.4%) in eight, 292 (1.9%) in nine, 228 (1.5%) in ten, 526 (3.4%) in 11 to 20 provinces. The majority (54%) are found in only one or two provinces; 76% of the species are found in four or fewer provinces.

The highest number of species (5,886 or 38.5%) is recorded in the province of Napo, followed by Pichincha with 4,759 species (31.1%), Morona-Santiago 3,353 species (21.9%), Pastaza 3,151 species (20.6%), Loja 3,039 species (19.9%), Carchi 2,911 species (19.1%), Zamora-Chinchipe 2,715 species (17.7%), Azuay 2,518 species (16.5%), Esmeraldas 2,333 species (15.2%), Imbabura 2,231 species (14.6%), Tungurahua 2,052 species (13.4%), Chimborazo 2,038 species (13.3%), Sucumbíos 1,888 species (12.3%), Cotopaxi 1,850 (12.1%), Los Ríos 1,711 (11.2%), Guayas 1,621 species (10.6%), El Oro 1,294 species (8.5%), Bolívar 1,271 species (8.3%), Cañar 1,084 species (7.1%), Manabí 1,001 species (6.5%), and Galápagos 694 species (4.5%) (Table 31). We have tried to correlate the number of species per province with the area of the provinces, but the correlation is not significant (R² = 0.3755). The lack of correlation may be explained by three factors. Other environmental factors such as humidity and precipitation play an important role in determining diversity. The isolation of the Galápagos has obviously placed that province in a different category where diversity is concerned. The difference in knowledge, due to uneven collecting intensity, most certainly plays an important role as well.

The Sørensen's similarity index varies greatly between provinces, but a pattern emerges upon careful examination. Any given province has highest similarity with the provinces in the same region and often with adjacent provinces (Table 31). For example, Azuay has the highest similarity with Loja, then Chimborazo and Cañar; Bolívar shares the most species with Chimborazo, Cotopaxi, and Cañar; Cañar shares most species with Azuay, Chimborazo, and Bolívar. It is interesting that Loja shares many species with Azuay, and with Chimborazo, not its neighbors to the east (Zamora-Chinchipe), west (El Oro) or the next province to the north (Cañar). A reason may be the large areas of dry inter-Andean valleys found in both Loja and Chimborazo. The provinces of the Andean region play a role in the Coastal region because large parts of some of these provinces, such as Carchi and Pichincha, include coastal lowlands. Galápagos shows the highest similarity with provinces like Guayas, Manabí, and Los Ríos, but the similarity values are about 50% smaller than between the mainland provinces. The highest value of similarity was found between the Amazonian provinces Sucumbíos and Zamora-Chinchipe with a value of 0.58. The maximum similarity values for typical Coastal provinces range between 0.38 and 0.52, for Andean provinces between 0.41 to 0.48, and for Amazonian provinces between 0.47 and 0.58.

The cluster analysis of the species composition of the provinces shows three main clusters or three distinct floristic regions (Figure 12A). It confirms the regions used in the Catalogue. The Andean cluster can be subdivided in three parts following a north-south gradient. The Coastal cluster includes Galápagos, but the attachment is very loose. This cluster can be split in two using humidity as a parameter. The humid cluster is composed of Esmeraldas and Los Ríos, versus the drier provinces of Manabí, Guayas, and El Oro. The way the regions are connected may be justified. It is, however, difficult to evaluate what is more important, the Coastal elements included in the Andean provinces or the Andean elements included in the Amazonian provinces.

The NMDS configuration of the species composition by provinces (Figure 12B) encountered relatively large stress values, but with a drastic decrease between the 1- and 2-dimension. The Monte Carlo test indicated that axes Two and Three extracted more information than expected by chance. Gray shaded areas represent the clusters discussed above, while the larger clusters are circled. The triangular shape of the Coastal cluster can be explained between affinities by the humid provinces Esmeraldas and Los Ríos and the humid Amazonian provinces. El Oro is situated closer to the Andean provinces because of its combination of areas of high elevation as well as lowland areas. The same can be said for provinces like Pichincha and Cotopaxi. Galápagos is so distinct that it is placed far from all other provinces. The Andean subclusters are not well supported in the NMDS configuration. A northern group would include Cotopaxi and exclude Tungurahua. Tungurahua would possibly be grouped together with Bolívar and Chimborazo. The southern group of provinces would then include Cañar, Azuay, and Loja. Loja is placed quite far from Azuay and Cañar, a fact also documented by Jørgensen and Ulloa Ulloa (1994). The Amazonian provinces form a distinct group, but there are some relationships between Napo and Pichincha; and Zamora-Chinchipe is not placed close to Loja. Axis One on the configuration represents a north-south gradient. Axis Three can be seen as an elevational gradient, considering that the Amazonian provinces contain many Andean elements, whereas the Andean provinces include fewer Coastal elements.

Provinces and taxonomic groups

An analysis of the most diverse families and genera in each province may seem a strange thing to do, because the political borders only to a very limited degree reflect natural limits for the plants and vegetation. Nevertheless, there are a few interesting observations to be made (Tables 32 & 33). The number of Amaranthaceae species on the Galápagos Islands is only surpassed by the number in the province of Guayas. The Araceae are most diverse in the lowlands up to 2,000 m elevation (Table 10), and are poorly represented in provinces that lack that range of elevation. We would therefore have expected more Araceae in the provinces Azuay, Bolívar, Cañar, and Zamora-Chinchipe, but we are not surprised that there are few Araceae in Tungurahua and Chimborazo. The collection effort may explain this result. Likewise, it cannot be explained why so few Asteraceae are found in Esmeraldas and Sucumbíos. Although the family does not do well in such superhumid areas, for Sucumbíos we would have expected values similar to Napo and for Esmeraldas similar to Los Ríos. It seems that similar explanations, of either an ecological or adaptation preference, or a collecting artifact, can be made for several families and genera, which would lead to a better focused collection effort or an increased understanding of the plants' requirements.

Provinces, elevation zones, and regions

The cluster analysis was made of the species composition defined by the combination of the provinces, regions, and elevation variables (Figure 13). The first division observed in the cluster diagram is found to represent a highland/lowland boundary, but the boundary fluctuates between 1,000 and 3,000 m elevation depending on which province or side of the Andes is considered. The lowland cluster, at the top of Figure 13, is divided in three subclusters, following the regional limits to a certain extent. The Galápagos is the first to be isolated. The Andean elevation zones included in the otherwise largely Coastal and Amazonian clusters is consistent with adjacent areas. The elevation range reached for certain provinces is higher than expected, but may be partly explained by two factors. The "noise" factor may be playing a significant role for certain areas of the provinces Azuay, Loja, El Oro, Bolívar, Cotopaxi, and Cañar. This is particularly the case in certain elevation ranges of these provinces that are undercollected, and they are therefore assigned a biased species composition. Second, collectors often record very wide elevational ranges for their localities, frequently spanning several hundred meters. This influences the data in a way that keeps elevation zones together within a province, thus creating more similarity upwards and downwards than laterally to the adjacent province. The Andean cluster, at the bottom, is divided at 3,500 to 4,000 m elevation, coinciding with the border between forest and non-forest vegetation types. The lower cluster is divided at 2,000 to 2,500 m and the upper cluster at 4,500 m elevation. We again observe the tendency of smaller clusters by province spanning about three elevation zones. The collecting practice may explain part of this or it could indicate a significant turnover in species composition from province to province or from north to south.

The NMDS configuration (Figure 14) encountered relative large stress values, but with a drastic decrease between the 2- and 3-dimensions; the Monte Carlo test indicated p£ 0.05 for all three axes. In Figure 14, we have chosen not to superimpose the cluster analysis and we find that groups can be identified almost exclusively using elevation as the main criterion. It is noteworthy that the groups are easier to separate at higher elevations, whereas below 1,000 m the grouping breaks down for certain provinces and fails to separate the Coastal elevation zones from the Amazonian zones; this indicates that many species are shared between certain Coastal areas and the Amazonian lowlands. The "true" Coastal provinces below 1,000 m are, however, clearly separated at the top of the configuration. As in the cluster analysis, the province Sucumbíos has a deviating position, but we can only attribute that to a lack of recorded species from that province. Data points for several elevation zones in El Oro seem to be "wrongly" placed; they are placed closer to points from elevation zones 500 m below. This may be attributed to lack of collecting in the zones from 1,500 to 3,000 m elevation in the province of El Oro. There is furthermore a tendency to place the provinces Loja, Zamora-Chinchipe, and Azuay together (to the right side of the figure) between 500 and 4,000 m elevation. The Galápagos and the zone above 4,500 m are the groups most clearly distinguished. The axis identified by the analysis can largely be defined by an elevational gradient (Axis 3) and a north-south gradient (Axis 2).

We have not performed a comparison of the province variable with habit, elevation, and region as individual variables. We do not believe that the habit will show any north-south variation, so we have omitted this analysis.

Summary and conclusions

We have documented the occurrence of 16,087 species growing in Ecuador. A total of 15,306 are native to the country, and 4,173 of these are endemic. The number of new species described with distribution in Ecuador has averaged 165 per year since 1975, with 91 new endemic species being described each year. There are no signs that this trend will change. If we compare the number of species found in Ecuador with neighboring countries, we see that Ecuador is extremely rich. Brako and Zarucchi (1993) documented 17,143 species in Peru; the corresponding number for Ecuador (subtracting the ferns and fern-allies) is 13,991. Estimates for Colombia, Venezuela, and Bolivia are, respectively, 50,000, 16,500, and 16,500 (including all species). It is interesting to compare the number of species with the area that the countries occupy. An index of diversity can be made by taking the logarithm of the number of species divided by the logarithm of the area. The values obtained, listed in descending order, are Colombia 0.78, Ecuador 0.76, Venezuela 0.71, Bolivia 0.70, and Peru 0.69. The estimated number of species for Colombia may be an exaggeration, so Ecuador may have the highest number of species relative to area in South America. Another indication of Ecuador's extreme diversity can be found by comparing the 1,298 ferns and fern-allies with the 1,358 species included in Flora Mesoamericana, which covers an area 6–7 times larger (Davidse et al., 1995).

The number of families registered for Ecuador is 273; of these 254 are native to the country. The largest families are Orchidaceae (2,999 species), Asterceae (863), Melastomataceae (553), Rubiaceae (493), and Poaceae (451). Ecuador has 3,013 orchids (Orchidaceae plus Cypripediaceae), compared to the 1,587 species in Peru (Brako & Zarucchi, 1993); this indicates not only that Ecuador is diverse, but may also reveal a disparity in knowledge. Calaway Dodson, an orchid specialist, has lived and collected for the last 20 years in Ecuador, which has certainly increased our knowledge of orchids in Ecuador and at the same time pointed out our lack of knowledge in Peru. The opposite effect may occur in the Asteraceae: 1,432 species were found in Peru (Brako & Zarucchi, 1993), which may be the result of extensive collecting by the Asteraceae specialist Michael Dillon.

Species are restricted in their elevational distribution. We found that the species composition at any given elevation zone has the highest number of species in common with the adjacent elevation zone, but more importantly we can see that this relationship goes in both directions, making it difficult to detect discrete zones or breaks in species composition. We have found correlations between area and number of species in the different elevation zones, but no correlation between the area of the provinces and number of species. One clearly marked elevation break coincides with the upper limit of continuous forest at about 3,500 m. The transition from forest to páramo becomes apparent when the data are viewed from several perspectives. The fact that the diversity of the Andean forest surpasses the lowland rain forest may have surprised Humboldt, but it has been known for some time (see, for example, papers in Churchill et al., 1995). The forests on the outer slopes of the Andes are very species-rich, yet still very poorly known, and their extent is very limited. Ecuador would lose a very significant portion of its biodiversity should these areas be converted for agriculture. Currently, a new species is being described from Ecuador every other day, clearly indicating that the era of exploration and discovery is far from over in the tropics.

It would have been very difficult to gather the data presented in this catalogue without computers, and an attempt to analyze the dataset contained in a matrix with about 4.3 million cells would have been impossible. We believe that the use of analytical tools such as cluster analysis and ordination has helped us to detect and confirm patterns that may have been detected previously, some already by Humboldt, but nonetheless have never before been documented with the dataset of a complete flora. We see this analysis as only the first step in the development of methods to analyze biodiversity and plant geographic patterns. In our analysis we have not gone into detail with the endemic species, but it would be interesting to analyze that subset in greater detail in order to provide more pragmatic advice for conservation purposes.