Plant Ecology

, Volume 213, Issue 8, pp 1209–1220

Elevational patterns in the vascular flora of a highly diverse region in southern Mexico

Authors

    • Sociedad para el Estudio de los Recursos Bióticos de Oaxaca, A.C
    • Departamento de Ecología y Recursos Naturales, Facultad de CienciasUniversidad Nacional Autónoma de México
  • Jorge A. Meave
    • Departamento de Ecología y Recursos Naturales, Facultad de CienciasUniversidad Nacional Autónoma de México
Article

DOI: 10.1007/s11258-012-0077-6

Cite this article as:
Salas-Morales, S.H. & Meave, J.A. Plant Ecol (2012) 213: 1209. doi:10.1007/s11258-012-0077-6
  • 273 Views

Abstract

We examined general and family-specific patterns of vascular plant richness along a large elevational gradient (0–3,670 m a.s.l.), assessed the continuity of these patterns and analysed their potential underlying causes in a high diversity region of the Sierra Madre del Sur, Oaxaca, Mexico. We used a vascular plant database constructed previously. The gradient was divided into 18 200-m elevation belts. To examine elevational patterns of richness, we used both observed and estimated (interpolated) species richness, as well as genus and family observed richness, for each belt. A generalised linear model (GLM) was used to assess the effect of altitude on area-corrected species richness (standard area = 100 km2), and a numerical classification of the elevational belts based on species richness was performed. Overall, richness at the three taxonomic levels decreased with elevation, but some individual families departed from this pattern. A sharp drop in species richness was observed at 1,800 m, and the dendrogram separated two elevational floristic groups at this elevation. The GLM revealed a significant negative effect of elevation on species richness. Despite this overall decreasing pattern for vascular plants along this extensive gradient, an examination of some family-specific patterns revealed the existence of other elevation–diversity relationships, indicating taxon-specific responses to elevation. The most noticeable discontinuity in species richness, at ca. 1,800 m, is likely related to a critical temperature isocline.

Keywords:

Altitudinal gradientArea effectEnvironmental-limitation hypothesisHabitat heterogeneityOaxaca

Introduction

Tropical mountains are attractive systems for the study of elevational gradients owing to the strong environmental changes occurring over very short distances, which in turn produce considerable habitat and vegetation heterogeneity on their slopes (Körner 2000). Numerous studies of the variation of plant and animal diversity along elevational gradients indicate an inverse relationship between species richness and elevation (Kappelle and Zamora 1995; Kappelle et al. 1995; Lieberman et al. 1996; Grytnes and Vetaas 2002; Grytnes and Beaman 2006). Nonetheless, increasing evidence suggests that this inverse diversity–elevation relationship is not general (Lovett 1999; Tallens et al. 2005). This view is supported by recent studies reporting non-linear or elevation-independent trends in species richness (Kessler et al. 2001; Bachman et al. 2004; Bhattarai et al. 2004). In fact, three main elevational patterns in species richness have been described (Rahbek 1995, 1997): (1) a monotonic decrease with increasing elevation; (2) a bell-shaped or hump-shaped pattern, peaking at mid-elevations (which at present seems to be the most frequent); and (3) a constant species richness all the way from the lowlands towards higher elevations. The shapes of these elevational patterns are largely dependent on the biological group being examined or the study’s spatial scale (Kessler 2000; Lomolino 2001; Grytnes and Beaman 2006). Despite the numerous investigations on the relationship between altitudinal gradients and biological richness, it is still far from being fully understood, and other patterns may still be found (Herzog et al. 2005).

Similarly, there is no full agreement as to which are the main drivers of species spatial distributions. Attempts to explain this variation have produced more than 100 different hypotheses (Palmer 1994), among which those related to available energy, evolutionary time, area effects, geometric constraints and habitat heterogeneity or ecosystem diversity have received considerable attention (Rahbek et al. 2007). Some hypotheses are particularly relevant for the study of elevational patterns of taxonomic richness; for example, the conic form of mountains causes reductions in area with increasing elevation (Lomolino 2001; Rahbek 2005), ultimately resulting in significant reductions in species richness (Rahbek 1997; Körner 2007; Romdal and Grytnes 2007; Karger et al. 2011). Conversely, while it has been suggested that climate is a major driver of species richness globally (Francis and Currie 2003; Hawkins et al. 2003; Currie et al. 2004; Mutke and Barthlott 2005; Barthlott et al. 2007), it does not completely explain elevational patterns of species richness (Currie 1991; Rahbek and Graves 2001).

A common shortcoming in studies examining elevational biotic patterns is the exclusion of the extremes of the analysed gradients, thus limiting the possibility of drawing general conclusions (Grytnes and Beaman 2006; Nogués-Bravo et al. 2008). In a review covering >1,200 articles dealing with the altitude–diversity relationship, Rahbek (2005) found that only 92 of 204 studies involved full gradients (i.e., including elevations ≤500 and ≥2,000 m). Reasons for this bias include the destruction of the natural communities in portions of the gradients as well as practical limitations. Additionally, studies of elevational gradients are highly concentrated in humid regions (Kitayama 1992; Lieberman et al. 1996; Grytnes and Beaman 2006), which also hinders generalisations about the altitudinal patterns of the biota and the underlying factors.

In southern Mexico, the Sierra Madre del Sur located along the Pacific Ocean coastline offers an excellent opportunity to examine the variation of tropical vegetation along a full elevational gradient; in some portions of this mountain range, the vegetation cover is almost untouched from sea level to >3,700 m. Along this large gradient, there is a noteworthy transition from the seasonally dry tropical vegetation typical of the narrow Pacific Coastal Plain to plant communities characteristic of mesic habitats, increasingly subjected to stronger temperature limitations.

Using a large database of vascular plant records, we examine the elevational patterns of floristic richness along a complete and large elevational gradient. We addressed these questions: (1) What is the pattern of vascular plant richness along a large elevational gradient in a tropical region? (2) Is this pattern characterised by a smooth, continuous change or by abrupt discontinuities? and (3) What are the potential underlying causes of this elevational floristic pattern?

Considering the climatic variability encompassed along this gradient, strong temperature-related growth restrictions conceivably exist for tropical plants at upper elevations. Furthermore, in this highly seasonal region, strong moisture restrictions are expected at the lower portions of the gradient, where the high temperature enhances evaporation. Based on the potential effects of these two limiting factors, we hypothesised that growth restrictions for the tropical flora would be minimum at intermediate portions of the gradient, so that the general floristic elevational pattern would display an intermediate diversity peak, thus conforming to the hump-shaped elevational pattern.

Methods

Study region

The study was conducted in the windward (southern) slope of the Sierra Madre del Sur, Oaxaca State, southern Mexico. In this region, elevation ranges from sea level to 3,670 m a.s.l. The region belongs of the Southern Sierra Madre Morphotectonic Province (Ferrusquía-Villafranca 1993), characterised by a complex massif with steep slopes and dissected by narrow valleys, and with a narrow coastal plain at the base of the mountains. Regional soil types include Chromic Cambisols at lower elevations, Eutric Regosols and Haplic Feozems at mid-elevations, and Orthic and Humic Acrisols at high elevations. In keeping with elevational variation, climate shows steep gradients, but it is generally sub-humid across the area, changing from tropical warm in the lowlands, through semi-warm, to temperate at higher elevations. At and near the coastal plain, mean annual temperature is >26 °C and average total annual precipitation range is 800–1,000 mm. At elevations from 500 to 2,000 m, mean temperature ranges from 26 to 18 °C, and total annual precipitation from 1,000 to 1,200 mm. At higher elevations, mean temperature is 18–12 °C, and total annual precipitation ranges between 1,200 and 1,500 mm. Near the summit, mean annual temperature is <12 °C, and average total annual precipitation is 1,200–1,500 mm.

The vegetation along this elevational gradient is generally well preserved and significant human disturbance is still localised (Salas-Morales et al. 2003). The dominant plant communities in the lowlands (50–450 m a.s.l.) are classified as tropical dry forest; further up, between 500 and 1,500 m communities are classified as tropical semi-evergreen forest, and as cloud forest in some portions between 1,500 and 2,000 m. At elevations between 2,000 and 3,000 m, the prevailing vegetation is a mixed pine-oak forest, and above 3,000 m, vegetation is mostly pine forest.

Data sources and analysis

Our database contains 10,124 records of fully identified species of vascular plants (including ferns and allies) collected during more than a decade of floristic survey in the region. The extracted floristic checklist was organised in two matrices, one of presence/absence (incidence) data and the other containing number of collections (abundance) by species. To explore the relationship between altitude and plant richness, the gradient was divided into 18 bands of 200-m elevation each, and the numbers of species (SObs), genera (GObs) and families (FObs) were tallied by band. Collecting effort was assessed by elevational belt and for the entire gradient by dividing the corresponding number of collections between the number of species.

Total species richness by belt (SEst) was estimated through interpolation; to this end it was assumed that each species occurred in all belts between its observed minimum and maximum altitudinal limits. No interpolation was made for higher rank taxa (genera and families), because in some cases, their potential continuity was suggested not by the same but by different species occurring above and below the belt where the higher taxon was absent; thus, assuming that one or both species were also present in the focal belt would represent a case of extrapolation instead of interpolation. We examined general elevational patterns by drawing scatterplots that included both observed (SObs) and estimated (SEst) species richness, as well as observed genus (GObs) and family (FObs) richness.

Area is an important factor affecting species richness, and in altitudinal gradients, it widely varies. We wanted to solely look at the relationship between altitude and SEst, so to completely remove the area effect, we performed a correction by area of SEst, procedure used in other studies (Rahbek 1997; Bachman et al. 2004). This required calculating each elevational belt’s area within a rectangular geographical window covering approximately 750 km2. Measurement of area included a slope correction based on a digital elevation model with a resolution <10 m and was performed on a ArcGis 9.2 platform (ESRI, Redlands, CA, USA). To perform the correction by area of SEst, we calculated the number of species in each elevational belt for a constant area of 100 km2 by using the power model proposed by Arrhenius in 1921 (Karger et al. 2011):
$$ S \, = \, cA^{Z} , $$
where S is species richness, c represents species richness in the smallest possible sampling area, A is sampling area (the predictor variable) and z is the slope of the linearised version of this equation when both S and A are plotted on logarithmic scales (Brown and Lomolino 1998). Although the calculation of z has resulted in a variable figure, we decided to use a constant value of z = 0.25, as this value is often obtained in theoretical and empirical studies (Crawley and Harral 2001). The corrected values were expressed as S100.

Thereby, to examine the relationships between S100 as response variable and mid-point of elevation of elevational belts as explanatory variable, we performed a regression with a generalised linear model (GLM). Owing to data overdispersion, we used a negative binomial distribution. This analysis of deviance was performed with the GLM function in R (R Development Core Team, Vienna, Austria).

Elevational patterns for selected individual families were described by fitting functions to their richness/elevation curves with TableCurve2D (AISN Software Inc., San Jose, CA, USA); the criterion to select the best-fit model was the largest adjusted R2.

We assessed floristic similarities between elevational belts by performing a hierarchical classification analysis of the elevational belts based on incidence interpolated data with Wards’s (minimum variance) method and Euclidean distances as dissimilarity measure; this analysis was performed with Statistica ver. 8 (StatSoft, Inc., Tulsa, OK, USA).

Results

Baseline information for the full gradient included 10,124 records, among which were represented 2,321 species (917 genera, 161 families). The prevailing families by species richness were Leguminosae (302), Asteraceae (235), Orchidaceae (125), Rubiaceae (118) and Poaceae (87). Table 1 lists the most speciose families by elevational belt for elevations of up to 2,000 m (above that elevation, nearly all families are listed, with the exception of a few very rare families). Notably, 36 families were represented by a single species and 17 families were exclusively recorded in the lowest elevational belt on the coastal plain (<200 m).
Table 1

The ten most speciose families by elevational belt for elevations up to 2,000 m a.s.l. and families occurring in those elevational belts above this elevation (some very rare families were excluded)

Families

Elevational belts (m, midpoint)

     

1

1

1

1

1

2

2

2

2

2

3

3

3

1

3

5

7

9

1

3

5

7

9

1

3

5

7

9

1

3

5

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

Leguminosae

1

1

1

1

1

1

3

3

3

4

2

3

4

3

4

1

Asteraceae

2

2

2

2

2

2

2

1

3

2

1

3

1

3

Euphorbiaceae

3

4

9

6

6

6

6

Poaceae

4

5

4

4

7

6

10

3

Convolvulaceae

5

5

5

Rubiaceae

6

3

3

3

4

4

4

3

2

3

1

Malvaceae

7

10

6

10

2

Boraginaceae

8

2

Solanaceae

9

7

6

10

8

5

5

5

2

3

4

1

Acanthaceae

10

10

6

5

5

5

7

10

7

Malpighiaceae

10

7

Cyperaceae

6

Apocynaceae

10

10

Cucurbitaceae

Asclepiadaceae

6

10

Bignoniaceae

7

6

Sapindaceae

10

Sterculiaceae

2

3

3

Tiliaceae

2

Cactaceae

10

Amaranthaceae

2

1

Lamiaceae

10

10

2

1

Bromeliaceae

10

6

7

10

6

4

4

Rhamnaceae

10

2

Moraceae

6

6

7

6

Scrophulariaceae

3

3

Agavaceae

2

3

4

Orchidaceae

8

3

3

1

2

1

1

2

1

4

1

2

1

1

3

Adiantaceae

6

Piperaceae

10

9

Melastomataceae

9

6

9

Campanulaceae

3

3

Aspleniaceae

10

1

Iridaceae

2

3

Fagaceae

2

3

2

3

Ericaceae

3

2

3

Rosaceae

3

3

4

Pinaceae

3

3

3

3

2

3

Crassulaceae

3

4

3

Numbers are family rankings according to species richness by family, figures in italics typeface indicate species ranking first, second and third in each elevational belt

Mean collecting effort by elevational belt was 1.7 specimens per species. This ratio, however, was unequal among belts: while the ratio for the lowest belt was 4.7 specimens per species, in the remaining ones this ratio had a mean value of 1.5 specimens per species (Table 2).
Table 2

Area and floristic information for 18 elevational belts along the windward slope of the Sierra Madre del Sur, Mexico

Elevation (m a.s.l.)

Area (km2)

No. of coll.

Coll. effort

SObs

SEst

S100

GObs

FObs

200

92

4,719

4.7

1,009

1,009

1,029

485

103

400

146

1,129

1.7

663

798

726

403

96

600

130

657

1.5

444

668

625

295

86

800

70

662

1.5

434

642

702

293

85

1,000

49

1,025

1.8

572

709

849

358

103

1,200

47

525

1.4

367

500

604

262

86

1,400

52

296

1.3

223

360

425

176

59

1,600

54

309

1.3

240

321

375

177

68

1,800

51

492

1.9

265

284

337

187

71

2,000

52

70

1.4

49

68

80

47

25

2,200

53

35

1.7

21

33

39

20

16

2,400

52

24

1.5

16

27

32

13

11

2,600

44

35

1.3

27

36

44

23

15

2,800

41

27

1.4

20

27

34

15

6

3,000

44

36

1.3

27

34

42

22

15

3,200

41

46

1.3

35

38

47

29

22

3,400

29

22

1.6

14

14

19

14

10

3,600

17

15

1.4

11

11

17

9

7

Floristic information includes number of collections (No. of coll.), collecting effort (Coll. effort), raw observed richness (SObs), total estimated (interpolated) richness (SEst), area-corrected richness (standard area = 100 km2, S100), raw observed genus richness (GObs), and raw observed family richness (FObs)

Elevational patterns of floristic richness

The elevational patterns for total and observed species richness (SEst and SObs) and observed genera (GObs) showed overall decreasing trends with increasing elevation (Fig. 1). Both SEst and SObs had their maxima at the lowest elevation (1,009), in strong contrast with the total richness of 11 species at the highest elevation. The reduction in species richness was not constant along the gradient. For example, we noted a second peak of species richness at 800–1,000 m altitudinal belt (SEst = 709). More outstanding, however, was the abrupt reduction observed at 1,800 m, equivalent to ca. 80 % of both SEst (from 284 to 68 species) and SObs (from 265 to 49). When the variation in floristic richness is separately examined for those elevational bands below and above the 1,800 m elevation, the decreasing pattern is much less marked and could also be described as a very slow reduction (for lower belts) or a fluctuating richness pattern (for higher belts). The elevational pattern for observed family richness (FObs) included much less variation from sea level up to 2,000 m, but at this elevation, FObs abruptly dropped from 71 to 25 families (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs11258-012-0077-6/MediaObjects/11258_2012_77_Fig1_HTML.gif
Fig. 1

Elevational changes of total estimated species richness (SEst), raw observed richness (SObs), raw observed genus richness (GObs), and raw observed family richness (FObs), along the studied elevational gradient

Elevational patterns based on observed richness for some individual families departed notably from the general decreasing familial pattern. For Leguminosae and Asteraceae (Fig. 2a, b), two very species-rich groups, elevational patterns were similar to the general one described earlier for the entire flora. The reduction in species richness for Rubiaceae was best fitted to the same negative exponential model, but members of this family were completely absent above 2,400 m (Fig. 2c). We also observed a strict elevational limit for Acanthaceae, a family also absent at elevations ≥2,200 m (Fig. 2d); however, its general pattern was fitted to the same model that best described the elevational changes of Bromeliaceae (Fig. 2e), a family whose members reached much higher elevations. The most striking trait in the elevational patterns of Acanthaceae and Bromeliaceae was a second peak at ca. 1,000 m (for Bromeliaceae, the largest richness along the gradient). The best-fit model for the moisture-loving Orchidaceae differed from all others: orchids had a hump-shaped pattern, with its species richness peaking at around 800–1,000 m (Fig. 2f).
https://static-content.springer.com/image/art%3A10.1007%2Fs11258-012-0077-6/MediaObjects/11258_2012_77_Fig2_HTML.gif
Fig. 2

Elevational changes of observed species richness (SObs) for selected families along the studied elevational gradient. The best-fit simple models according to adjusted R2 describing the elevational trend are shown (P < 0.05 for all models)

As most elevational belts had areas smaller than 100 km2 (Table 2), area-corrected richness values (S100) derived from the Arrhenius equation were generally larger than estimated total richness (SEst). Although the parameter c (intercept) in the Arrhenius equation is considered to be constant for a given biological group and biogeographical area, it is interesting that we observed two sets of c values (i.e., richness in the smallest possible sampling area) along the gradient, namely, a set of higher values (mean value of 199.3) at elevations from sea level to 1,800 m, and a set of much lower values (mean value of 12.4) at elevations higher than 1,800 m.

The analyses of deviance showed a highly significant negative effect of elevation on area-corrected (S100) species richness (Table 3). We fitted a model in which the factor elevation was excluded and performed an ANOVA to compare the best-fit models with and without elevation; this yielded a significant difference (P = 0.0000), thus confirming the explanatory power of elevation on species richness along the gradient.
Table 3

Results of the GLM-based analysis of deviance performed to evaluate the effect of elevation on area-corrected species (S100) by elevational belt

Factor

Estimate

df

Deviance

P

Species

    

Intercept

7.470

1

  

Elevation

−0.00131

1

39.75

<0.001

Classification of elevational belts

The dendrogram obtained from the hierarchical classification analysis showed a clear distinction between two major groups of elevational belts at a linkage distance of 60 (Fig. 3). These two major groups were closely related to elevation, as all belts between 0 and 1,800 m were included in one group, whilst all remaining belts were comprised in the other. The low-elevation group could be further subdivided at a linkage distance of 45, forming two almost equal-sized subgroups of belts above and below 800 m. The dendrogram also showed that the high elevation group was relatively more homogeneous floristically than its low-elevation counterpart.
https://static-content.springer.com/image/art%3A10.1007%2Fs11258-012-0077-6/MediaObjects/11258_2012_77_Fig3_HTML.gif
Fig. 3

Hierarchical classification of the elevational belts based on their floristic composition (incidence data), obtained using Ward’s method and Euclidean distances as measures of between-belt dissimilarity

Discussion

Elevational floristic patterns and underlying causes

Based on the environmental-limitation hypothesis, we expected a hump-shaped pattern in species richness along this large elevational gradient. This expectation was not supported by our results, as these revealed a clear elevational decrease in species richness, which was also true for genus and family richness. Although for some time the general decreasing pattern was repeatedly pointed out in the literature as being prevalent (Gentry 1988; Kappelle et al. 1995), newer evidence has shifted to some extent this viewpoint so that presently it is generally recognised that the hump-shaped pattern represents the most common responses of different biotas to the heterogeneous environments of mountains (Rahbek 1995). There are several potential explanations for this discrepancy. For one, it could be argued that it is a methodological consequence derived from the dichotomy between analysing full or partial gradients. The consequences of excluding sections of elevational gradients were elegantly examined by Nogués-Bravo et al. (2008); based on the analysis of 400,000 elevational records of vascular plants, they suggested that the recognition of monotonic decreasing patterns was largely related to the exclusion of the lowest sections of the studied gradients. Our study challenges this view, as the records of plant collections used by us for the analysis covered the entire gradient.

A further methodological explanation may be related to an unbalanced collecting effort across elevational belts. Collecting effort is often responsible of substantial biases in assessing spatial richness patterns (Lomolino 2001; Grytnes and Beaman 2006; Karger et al. 2011). Although in our study region a much larger collecting effort was done in the lowest elevational belt than in all others, collecting effort was clearly homogeneous across all remaining belts, so that the general decreasing pattern in species richness cannot be interpreted as a methodological artefact. In fact, species richness of individual families such as Acanthaceae and Bromeliaceae peaked at intermediate points of the gradient, and those peaks did not correspond to belts with larger collecting efforts. In particular, the hump-shaped pattern observed for Orchidaceae cannot be attributed to a deficient collecting effort at lower elevations; only one orchid species among more than 4,000 records was obtained from the lowest elevational belt. Thus, explanations for this pattern beyond the methodological strategy of the study are needed.

The potential causes underlying hump-shaped elevational patterns of species richness have been discussed extensively (e.g., Vetaas and Grytnes 2002; Bhattarai et al. 2004; Grytnes and Beaman 2006; Grau et al. 2007). Interestingly, fewer explanations have been put forward for the general decreasing pattern, and these always involve unidirectional changes in resource limitations, thermic restrictions, habitat diversity or available space (Grytnes 2003; Fiedler and Beck 2008). The relatively scarce evidence for the decreasing pattern led Colwell and Lees (2000) to state that the monotonic decreasing pattern was actually more of a dogma than a reality. Our results demonstrate, however, that the overall decreasing trend is a real response of a group of organisms (vascular plants) to elevational changes, but with the observation that in the gradient studied by us an abrupt decrease in floristic richness around 1,800–2,000 m (depending on the taxonomic level) is the most striking feature.

Our working hypothesis established the existence of two major gradients of growth limitations for tropical plants acting in opposite directions along the elevational gradient: a temperature-related gradient from the lowlands upward and a seasonal water limitation gradient increasing from the higher elevations downward. Implicit in this hypothesis is the notion that these two factors have symmetrical effects on a tropical flora. However, this may not be the case. In discussing the timing of the origin of the tropical dry forest of Western Mexico based on a time-calibrated phylogeny of Bursera, Becerra (2005) suggested that during the long evolutionary history of this biome (ca. 30 million years), its floristic elements developed numerous adaptations to hot, dry conditions. A critical consequence of these adaptations is a low tolerance to cold temperatures and a total intolerance to freezing (Becerra 2005). It is thus likely that thermal restrictions have a much stronger effect on this flora than water stress, which many species can avoid through various functional strategies (Holbrook et al. 1995; Poorter and Markesteijn 2008; Markesteijn and Poorter 2009; Lebrija-Trejos et al. 2010).

Floristic discontinuities along the elevational gradient

Aside from the obvious decrease in floristic richness towards higher elevations, this study also revealed an inconstant rate of change in species richness with elevation, given the large discontinuity observed at 1,800 m for species and genus richness and at 2,000 m for family richness. There are several indications that temperature may also be the critical factor involved here.

Temperature is widely associated with elevational changes. Several studies on elevational floristic patterns have attributed their results to critical variations in this factor (Grubb 1977; Hamilton and Perrot 1981; Sang 2009). For example, it is widely accepted that most tropical plants face physiological restrictions that prevent their occurrence at higher elevations, such as the short growing season, direct temperature restrictions on metabolic processes or an unfavourable energy balance (Körner 1999; Francis and Currie 2003). Unfortunately, the lack of specific climatic information in our study region implies that any discussion on the role of particular environmental factors is necessarily speculative. Even so, the discontinuity in species richness observed at 1,800 m is likely related to a critical temperature isocline (possibly frost line) that could be associated with a major division between two large physiological groups of plants: a cold-resistant guild above 1,800 m and a thermophilous flora occurring below it (Holdridge 1978).

In tropical Mexico, the upper elevational limit for tropical forest vegetation generally coincides with the 0 °C isotherm of extreme low temperature, above which tropical communities are commonly replaced by plant associations of temperate affinity (Rzedowski 1978). In our elevational transect, well-known tropical families displayed upper elevational limits between 1,800 and 2,200 m, probably concurring with the 0 °C isotherm, thus contributing to the abrupt reduction in species richness. Similar results have been reported for other tropical mountains in Malesia (Van Steenis 1984), Sumatra (Ohsawa et al. 1985) and Tanzania (Hemp 2006). A further potential factor related to abrupt elevational upper limits of tropical vegetation, which is likely to operate in our study area, is a sharp discontinuity in temperature, humidity or both, resulting from a thermal inversion (Martin et al. 2011). The formation of two clearly distinct groups in the hierarchical classification of elevational belts further supports the existence of an altitudinal discrimination of the floristic components typical of tropical and temperate environments.

Although temperature seems to be the cause of the abrupt drop in species richness at 1,800 m, species richness also decreased markedly in other sections of the gradient, apparently regardless of temperature. For example, the large reduction observed for both SObs (346 species, 34.3 % loss) and SEst (211 species, 20.9 %) between the lowest (0–200 m) and the adjacent elevational belt (200–400 m) cannot be attributed to a change in temperature, as the high-temperature life zone extends much higher than 200 m, even in seasonally dry tropical conditions (Holdridge 1978; Murphy and Lugo 1995). Thus, this decline in species richness may rather result from a reduction in the variety of habitats between these two elevations. In addition to tropical dry forest, at the bottom of the gradient wetlands and coastal vegetation such as mangrove forests, flooded savannahs and sand dune vegetation are found (Lorence and García-Mendoza 1989; Salas-Morales et al. 2003); these systems host many exclusive species and families (e.g., families such as Rhizophoraceae, Bataceae, Aizoaceae, Juncaceae and Molluginaceae, and species such as Pistia stratiotes L., Batis maritima L. and Ipomoea pes-caprae (L.) R. Br., among many others). This explanation is consistent with the habitat diversity hypothesis as a mechanism contributing to species richness (Rosenzweig 1995; Guégan et al. 1998; Wohlgemuth 1998; Rahbek and Graves 2001; Rahbek et al. 2007; Fahr and Kalko 2011).

Elevational patterns of individual taxa

Patterns of species richness may be understood more thoroughly by dissecting the flora into different functional or taxonomic groups (Pausas and Austin 2001). The different elevational patterns observed for the examined families suggest that these are taxon-specific, because different taxa possess ecophysiological traits associated with their general richness patterns (Linden 1991; Kessler 2002; Grytnes and Beaman 2006). These traits seem to explain the presence of plant groups with higher water requirements at elevations away from the lowlands, such as Bromeliaceae and Orchidaceae, whose richness peaked at intermediate elevations; the strong response of these groups can also explain the second peak in overall species richness found at 1,000 m on our gradient. Our results agree with various studies on the elevational distribution of vascular epiphytes that have also reported hump-shaped patterns for these families (Nieder et al. 2001; Kreft et al. 2004; Küper et al. 2004; Cardelús et al. 2006).

Conclusions

Our results show general decreasing patterns of richness for species, genera and families of vascular plants along the large elevational gradient on this mountain range. This pattern does not result from an incomplete knowledge of the flora along the gradient as it holds when total species richness by belt is estimated through interpolation. Yet this elevational pattern is not necessarily true for all individual taxa, as demonstrated by the individual families we analysed, which revealed the existence of other taxon-specific elevation–diversity relationships, including hump-shaped and bimodal distributions. The strong discontinuity in species richness observed as a sharp decrease at ca. 1,800 m is likely related to a critical temperature isocline, where the thermophilous lowland flora seems to be replaced by a floristic array adapted to thermal restrictions. Future studies on the environmental heterogeneity along this gradient, particularly on the thermal regime, will shed further light on the underlying causes of this floristic pattern.

Acknowledgments

We are indebted to all the people who participated in the botanical exploration and the taxonomic determination of thousands of vouchers collected in the region. The senior author thanks the Graduate Programme in Biological Sciences of the Universidad Nacional Autónoma de México and CONACyT for a doctoral scholarship. Funding was provided by SERBO A.C. and CONACyT (Grant no. CB-2009-01-128136). The insightful comments of Robert Colwell, Alberto Gallardo-Cruz, Trudy Kavanagh, Michael Kessler, Emily Lott, Eduardo Pérez-García and two anonymous reviewers improved earlier versions of this manuscript. Edgar J. González, Gilberto Hernández and Marco A. Romero provided assistance in model fitting, GIS use and figure preparation, respectively.

Copyright information

© Springer Science+Business Media B.V. 2012