See discussions, stats, and author profiles for this publication at: [Link]

net/publication/268076475

Tree diversity patterns in successive vegetation types along an elevation

gradient in the Mountains of Eastern Mexico

Article in Ecological Research · November 2014

DOI: 10.1007/s11284-014-1196-4

CITATIONS READS

34 2,444

2 authors:

María Toledo Guadalupe Williams-Linera

University of Toronto Institute of Ecology INECOL
6 PUBLICATIONS 178 CITATIONS 128 PUBLICATIONS 5,390 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by María Toledo on 10 February 2016.

The user has requested enhancement of the downloaded file.

Ecol Res (2014) 29: 1097–1104
DOI 10.1007/s11284-014-1196-4

O R I GI N A L A R T IC L E

Marı́a Toledo-Garibaldi • Guadalupe Williams-Linera

Tree diversity patterns in successive vegetation types along an elevation

gradient in the Mountains of Eastern Mexico

Received: 1 May 2014 / Accepted: 16 August 2014 / Published online: 26 August 2014
 The Ecological Society of Japan 2014

Abstract Tree species richness changes along elevation Keywords Cloud forest Æ Coniferous forest Æ Dry
gradients in response to underlying environmental con- forest Æ Geographical gradient Æ Vegetation structure
ditions. Our hypothesis was that richness is associated
with climatic variables and decreases with elevation. The
objective was to identify trends in species, genus and Introduction
family richness, diversity and vegetation structure in
relation to climate variables along an elevation gradient Elevation gradients have served as natural laboratories
with successive types of forest in Veracruz, Mexico. for examining how environmental factors generate and
Trees were identified and measured in 0.1 ha at 15 sites maintain distribution patterns in species richness as well
located from 140 to 4000 m a.s.l. Generalized linear as the structural characteristics of the forests growing on
models were used to fit richness, diversity, basal area and mountains (Lomolino 2001; Rahbek 2005; Nogués-
density as a function of elevation; the best model was Bravo et al. 2008). Numerous studies have demonstrated
selected using Akaike’s Information Criterion. Multi- monotonic, unimodal or multimodal patterns in species
variate analyses were used to explore climatic variables richness with elevation (Rahbek 2005). The monotonic
associated to composition of groups of sites along the pattern shows a linear decrease in species richness with
gradient. Along the entire elevation gradient, species, increasing elevation. Some studies have demonstrated a
genus and family richness decreased unimodally, and clear, linear decline in species and family richness with
diversity decreased monotonically. Richness was posi- elevation (Gentry 1988; Kitayama 1992; Vázquez and
tively correlated with temperature but not with precipi- Givinish 1998; Givinish 1999; Behera and Kushwaha
tation. Basal area increased monotonically and highest 2007; Homeier et al. 2010; Salas-Morales and Meave
basal area was associated with high humidity and certain 2012). The unimodal pattern has a hump-shaped distri-
tree species (Quercus and Abies). Ordinations indicated bution and could exhibit a peak of richness at low or
three groups of sites: lower elevation dry forest associ- high elevation. In extended gradients the unimodal
ated with temperature seasonality, mid-elevation cloud pattern has been reported for plants in regions as diverse
forest associated with precipitation-related variables, as the Barva Volcano in Costa Rica (Lieberman et al.
and coniferous forest at the top of the gradient associ- 1996), Mount Emei, China (Tang and Ohsawa 1997),
ated with elevation. Our study shows that different plant Southern Nepal (Vetaas and Grytnes 2002), Western
communities are associated with certain climatic condi- Himalaya (Oommen and Shanker 2005), Cerro Tláloc,
tions and harbour different tree species, genera and Mexico (Sánchez-González and López-Mata 2005) and
families. The results support the hypothesis that species Mount Kinabalu, Borneo (Grytnes and Beaman 2006).
richness is associated with climate, and decreases with Multimodal patterns are mainly bimodal with two peaks
elevation. of richness along the elevation gradient; bimodal pat-
terns have been reported for different life forms (e.g.,
González-Espinosa et al. 2004; Sang 2009; Acharya et al.
Electronic supplementary material The online version of this article 2011). The patterns have been associated to differences
(doi:10.1007/s11284-014-1196-4) contains supplementary material, in temperature and moisture along the elevation gradi-
which is available to authorized users.
ent (Gentry 1988; O’Brien 1993; Givinish 1999; O’Brien
M. Toledo-Garibaldi Æ G. Williams-Linera (&) et al. 2000; Bhattarai and Vetaas 2003; Hawkins et al.
Instituto de Ecologı́a, A.C., Carretera antigua a Coatepec No. 351, 2003; Sánchez-González and López-Mata 2005; Sang
91070 Xalapa, Veracruz, Mexico 2009; Acharya et al. 2011; Salas-Morales and Meave
E-mail: [Link]@[Link]
Tel.: +52 228 8421800 2012).
1098

Vegetation structure and composition also change test quarter of the year but not with temperature (Wil-
with elevation. Some reports indicate a linear increase in liams-Linera et al. 2013).
basal area and density (Kitayama 1992; Lieberman et al. The aim of this research was to explore trends in tree
1996; Vázquez and Givinish 1998; Aiba and Kitayama species, genus and family richness and diversity, vege-
1999), but other studies show a decrease in basal area tation structure and climate along an elevation gradient
with increasing elevation (Tang and Ohsawa 1997; in central Veracruz, Mexico. We examined changes
Homeier et al. 2010). Along elevation gradients the flo- along an elevation gradient with successive types of
ristic composition of different plant communities is vegetation. The gradient runs up through seasonally dry
remarkably consistent, at least at the familial level tropical forest (SDTF), tropical montane cloud forest
(Gentry 1988). Different species belonging to certain (TMCF), and coniferous forest (CF). Our prediction
families dominate portions of the elevation gradient was that species richness was positively associated with
(Gentry 1988; Luna-Vega et al. 2001; Oommen and energy and water variables and decreases with elevation.
Shanker 2005; Zhu et al. 2005; Salas-Morales and Me-
ave 2012).
Along geographical gradients, changes in climatic
variables are relatively predictable in space and time, Methods
therefore it would be expected that species richness
gradients are essentially climate gradients (O’Brien 1993; Study area
Lomolino 2001). Other factors influencing diversity
along gradients are total forest area, nutrient availability The study area extends over an elevation gradient in
or anthropogenic disturbance (Williams-Linera and central Veracruz, Mexico (1919¢–1930¢N, 9629¢–
Lorea 2009; Homeier et al. 2010; Wassie et al. 2010). 9710¢W; 140–4000 m a.s.l.), from the lowlands to the
While they have been related to forest changes, we focus Cofre de Perote Volcano, which is 4282 m high at
on effects of elevation and climate. 100 km from the Gulf of Mexico (Rodrı́guez et al. 2009)
Climate has been hypothesized to be the most pow- (Fig. 1). This is the area where the Trans-Mexican
erful factor explaining changes in species richness and Volcanic Belt, the Gulf of Mexico Coastal Plain and the
composition along gradients, with precipitation, Mexican Central Plateau converge (Rodrı́guez et al.
humidity, temperature and radiation in particular being 2009), and the region is one of the world’s biodiversity
important variables controlling forest distribution on hotspots according to the WWF (Gillespie et al. 2012).
mountains (Grubb 1977; Gentry 1988; O’Brien 1993; In this region, we selected 15 sites along the elevation
Tang and Ohsawa 1997; O’Brien et al. 2000; Lomolino gradient, from 140 m a.s.l. near the coast to 4000 m
2001; Hawkins et al. 2003; González-Espinosa et al. a.s.l. at the tree line (Fig. 1; Table 1). Sites were dis-
2004; Sánchez-González and López-Mata 2005; Behera tributed along the entire elevation gradient and had
and Kushwaha 2007). Climatic variables associated with between 2 and 830 ha in area with relatively little dis-
water availability (amount and seasonality of precipita- turbance. Low elevation forest fragments were smaller
tion) and energy (temperature, radiation) are correlated (Williams-Linera and Lorea 2009) than mid- and upper-
with species richness (O’Brien 1993; O’Brien et al. 2000; elevation mountain forests (Williams-Linera 2002), but
Hawkins et al. 2003). Species richness should be greatest we considered that both the small and large forest pat-
where energy and moisture availability are optimal year-
round; and should decrease as they decrease or increase
beyond optimum levels (O’Brien 1993). Plant species
richness has been reported to vary positively with pre-
cipitation (Gentry 1988; O’Brien 1993; Givinish 1999;
O’Brien et al. 2000; Bhattarai and Vetaas 2003; Hawkins
et al. 2003), and parabolically with temperature because
of species richness increasing as energy moves towards
some optimum, and decreasing as it moves above or
below this threshold (O’Brien 1993; Bhattarai and Vet-
aas 2003).
In elevation gradients, depending on the type of
vegetation, species richness may be associated with dif-
ferent climatic variables. For example, tree species
richness in seasonally dry tropical forests is strongly Fig. 1 a Map of Mexico, b location of the 15 study sites along an
associated with temperature, solar radiation, and po- elevation gradient from 140 m to 4,000 m a.s.l. at the tree line in
tential evapotranspiration (Gentry 1995; Trejo and central Veracruz, Mexico (the curve was included for ease of
Dirzo 2002), but precipitation has been identified as a comparison). The X axis is horizontal distance from the summit of
Cofre de Perote Volcano toward the Gulf of Mexico. CF coniferous
poor predictor of species richness (Gillespie et al. 2000). forest, TMCF tropical montane cloud forest, SDTF seasonally dry
In tropical montane cloud forest, tree species richness tropical forest. All sites are located in the eastern windward side of
was positively correlated with precipitation in the wet- the mountains
 1099

Table 1 Characteristics of the study sites along an elevation gradient in central Veracruz, Mexico

Site N latitude W longitude Area (ha) Elevation (m a.s.l.) Tmean (C) Pp (mm) Tseasonality (SD · 100) Ppwq (mm) Vegetation

1 1919¢32¢¢ 9629¢7.8¢¢ 3 140 25.3 1049 224 258 SDTF

2 1921¢51.9¢¢ 9632¢37.4¢¢ 19 204 24.8 932 224 241 SDTF
3 1923¢58.7¢¢ 9639¢18.1¢¢ 2 335 24.1 945 228 253 SDTF
4 1927¢15.7¢¢ 9641¢37.2¢¢ 2 501 22.5 908 225 237 SDTF
5 1927¢53.9¢¢ 9645¢22¢¢ 7 780 20.7 1117 224 306 SDTF
6 1930¢30¢¢ 9649¢55¢¢ 4 986 19.4 1464 215 431 SDTF–TMCF
7 1930¢48¢¢ 9656¢34¢¢ 29 1250 18.5 1621 207 481 TMCF
8 1935¢ 9658¢ 54 1450 16.8 1836 202 465 TMCF
9 1930¢40.8¢¢ 970¢5.7¢¢ 14 1630 17.1 1925 199 506 TMCF
10 1929¢37¢¢ 970¢48¢¢ 52 1800 16.5 2160 195 559 TMCF
11 1929¢23¢¢ 971¢54¢¢ 65 1950 16.1 2189 197 562 TMCF
12 1931¢4.3¢¢ 973¢37¢¢ 20 2650 11.8 1189 170 284 TMCF
13 1933¢46.3¢¢ 975¢34.4¢¢ 326 3020 9.9 1384 147 344 CF
14 1931¢14¢¢ 978¢52¢¢ 57 3460 7.8 1695 126 412 CF
15 1930¢26.4¢¢ 9710¢33.4¢¢ 830 4000 8 1645 127 405 CF

Geographical coordinates, area of the forest site, elevation, mean annual temperature (Tmean), total annual precipitation (Pp), temperature
seasonality (Tseasonality), precipitation of the warmest quarter (Ppwq), and vegetation type (SDTF seasonally dry tropical forest, TMCF
tropical mountain cloud forest, CF coniferous forest)

ches maintain significant species richness (Wassie et al. Data analysis

2010).
Mean annual temperature decreases from 25 C at The response variables were species, genus and family
the lowest sites to 8 C at the highest. Mean annual richness, rarefied species, and diversity (H¢), as well as
precipitation varies from ca. 1000 mm at lower eleva- vegetation structure (basal area and density); since
tions, up to 2200 mm at mid-elevations (Williams-Li- common patterns for variation with elevation are
nera 2002). Since meteorological data were not available monotonically or unimodally related to elevation, we
for the whole elevational gradient, we decided to use the fitted linear and polynomial equations. We used gen-
WorldClim database (Hijmans et al. 2005) (Table 1). eralized linear models to test the patterns, and the best
The reliability of the WorldClim data on mean annual model was selected with Akaike’s Information Criterion
precipitation and mean annual temperature were cor- for small sample size, AICc (Burnham and Anderson
roborated with actual data from nine meteorological 2002). When the response variable was counts (number
stations located between 97 and 2426 m a.s.l. No mete- of taxa), a Poisson distribution and log link function
orological stations exist above this elevation. were used. Spearman correlation coefficients were cal-
culated for the response variables, elevation and selected
bioclimatic variables. Data were analysed in JMP ver-
Vegetation structure, richness and diversity sion 10.0.0 (SAS 2012), and R project software version
2.5.1 (R 2007).
In each site we selected 1 ha and randomly set up ten Nineteen bioclimatic variables from the WorldClim
10 · 10 m plots to measure 0.1 ha in all sites. In each database (Hijmans et al. 2005) were screened for col-
plot we measured all trees ‡5 cm diameter at 1.3 m linearity using Pearson correlation coefficients and
height (dbh), counted the number of individuals and Bonferroni’s procedure to correct for multiple compar-
identified the species. Voucher specimens were collected isons. Highly correlated bioclimatic variables were ex-
and deposited at the XAL herbarium of the Instituto de cluded from the direct ordination analysis. To analyse
Ecologia, A.C. the relationship between tree species abundance and
For each site, we calculated the basal area (m2 ha 1) bioclimatic variables, we used Canonical Correspon-
and density of trees (trees ha 1). Species, genus and dence Analysis (CCA) run in the program CANOCO,
family richness were the total number of each taxonomic version 4.5 (ter Braak and Šmilauer 1998). The species
level found in each study site (Sobs). Due to differences in matrix included 15 sites and 170 species with 5 or more
tree density among sites, we generated rarefaction curves individuals and the matrix for environmental variables
with the Mao Tau function to reduce the samples to a included elevation and seven bioclimatic variables: three
common number of individuals. Shannon’s diversity that are temperature-related (mean diurnal range, sea-
index (H¢) was estimated for each site, and the Chao- sonality and annual range) and four precipitation-re-
Jaccard index was used to measure species turnover lated (annual, seasonality, precipitation of wettest
along the elevation gradient. Rarefaction and diversity quarter and precipitation of warmest quarter). To
indices were calculated using EstimateS 8.2.0 (Colwell determine the statistical significance of the environ-
2009). mental variables we used the forward selection procedure
1100

for ranking and evaluated significance using a Monte with elevation (Fig. 2c; Table 2). Basal area increased
Carlo permutation test. monotonically (Fig. 2f) whereas tree density showed no
Nonmetric multidimensional scaling (NMDS) was significant trend with elevation (Table 2; Table S2).
run on abundance data using the Sørensen distance The Chao-Jaccard index values ranged from 0.01 to
measure to detect groups of sites. Differences in tree 0.83. Species turnover was high along the elevation
species composition among groups of sites were tested gradient. Sites 13 and 14 were the most similar
with a multiresponse permutation procedure (MRPP), (85–87 %), followed by 7, 8, 9 (37–39 %), and 1, 2 and 3
which is a nonparametric method for testing multivari- (34 %). Site 15 was dissimilar to all the other sites. Also,
ate differences among pre-defined groups. The test sta- sites 5 and 6 were different from each other and from all
tistic (T) describes the separation between groups and other sites (Fig. 3).
the chance-corrected within-group agreement Mean annual temperature and temperature season-
(A) describes within-group homogeneity compared to ality were negatively correlated with elevation. Total
random expectation (McCune and Grace 2002). Fol- annual precipitation was positively correlated with
lowing the MRPP we ran the indicator species analysis elevation even though precipitation was the highest at
(ISA) to identify characteristic species within each of the mid-elevation sites. Richness and the diversity vari-
groups detected by the NMDS. ISA yields an indicator ables were negatively correlated with elevation, and
value and a statistical significance for this value using a positively correlated with temperature variables but
Monte Carlo technique. NMDS, MRPP and ISA anal- not correlated with precipitation. Basal area was pos-
yses were done using PC-ORD software (McCune and itively correlated with elevation and annual precipita-
Grace 2002). tion, but negatively correlated with temperature
variables (Table 3).
The CCA results retained two significant climate
Results variables (temperature seasonality and precipitation of
the warmest quarter). Axis 1 (eigenvalue = 0.99) and
A total of 2011 trees belonging to 170 species, 109 gen- axis 2 (eigenvalue = 0.97) represented 16.8 and 16.3 %
era, 58 families and 10 morphospecies was recorded of the variance of the species-environment relationship
across the entire elevation gradient (Table S1). Species (Monte Carlo test of first axis 1: F = 0.75, P = 0.002;
richness, rarefied species, genus and family richness de- test for all canonical axes: F = 1.47, P = 0.002). CCA
creased unimodally with elevation, but family richness results reveal one group of sites (1–5) in the lower left
exhibited a peak at intermediate elevation (Fig. 2a, b, d, quadrant associated with temperature seasonality, a
e). Shannon’s diversity index decreased monotonically second group of sites (7–12) in the upper right quadrant

Fig. 2 Relationship between

species richness (a), rarefied
richness (b), Shannon’s
diversity index (c), genus
richness (d), family richness (e),
and basal area (f) with elevation
along a gradient in central
Veracruz, Mexico. The fitted
line represents the best model
(DAICc within 0–2), see Table 2
for summary statistics
 1101

Table 2 Summary of model fitting of tree species richness, rarefied associated with precipitation of the wettest quarter, and
richness, Shannon’s diversity index (H¢), genus richness, family a third group of sites (13–15) associated with elevation.
richness, basal area and density in relation to elevation in central
Veracruz, Mexico The first group was associated with tree species such as
Bernardia mexicana, Bursera simaruba, Ceiba aesculifo-
Residual Deviance v2 P AICc DAICc lia, Ipomoea wolcottiana, Leucaena lanceolata, and
deviance explained (%) Lysiloma acapulcense. The second group was associated
with Carpinus tropicalis, Cinnamomum effusum, Clethra
Species richness
1 51.92 50.39 52.73 <0.0001 123.8 23.8 macrophylla, Liquidambar styraciflua, Magnolia schi-
2 29.15 72.14 75.50 <0.0001 104.2 4.2 edeana, Oreomunnea mexicana, Oreopanax xalapensis,
3 21.12 79.81 83.52 <0.0001 100.0 0 Prunus rhamnoides, Quercus corrugata, Q. delgadoana,
Rarefied species richness Q. lancifolia, Q. xalapensis, Styrax glabrescens and
1 31.11 57.08 41.4 <0.0001 99.6 11.7
2 17.41 75.98 55.1 <0.0001 89.1 1.2 Symplocos limoncillo. The third group was associated
3 12.37 82.93 60.1 <0.0001 87.9 0 with Abies religiosa and Pinus hartwegii. Site 6 was
H¢ associated with Q. sapotifolia (Fig. 4).
1 4.14 70.65 18.39 <0.0001 31.45 0 The presence of groups was confirmed by the indirect
2 3.22 77.19 22.17 <0.0001 31.49 0.04
3 3.22 77.20 22.17 <0.0001 36.16 4.71
gradient analysis. NMDS ordination detected five
Genus richness groups of communities. Those groups represented dry
1 37.72 54.28 44.78 <0.0001 106.4 14.2 forest (sites 1–5), lower (7–9) and upper (10–12) mon-
2 20.68 74.94 61.83 <0.0001 92.5 0.3 tane forest, coniferous forest (13–15) and an ecotone
3 16.58 79.90 65.9 <0.0001 92.2 0 (site 6) between dry and montane forests. Tree species
Family richness
1 49.24 29.10 20.21 <0.0001 114.2 27.2 composition varied significantly among groups of sites
2 20.43 70.58 49.02 <0.0001 88.6 1.6 and the heterogeneity within groups tends to equal what
3 14.95 78.48 54.50 <0.0001 87.0 0 one would expect by chance (MRPP, T = 6.84,
Basal area A = 0.167, P < 0.0001). The ISA identified nine spe-
1 6029 43.66 8.6 0.003 140.7 0
2 4841 54.76 11.9 0.003 141.2 0.5 cies as strong indicators for dry forest, six and seven
3 3868 63.85 15.3 0.002 142.5 1.8 species for lower and upper montane cloud forest,
Density respectively, and two species for coniferous forest
1 2172953 18.79 3.1 0.077 229.0 0 (Table 4).
2 1922630 28.14 5.0 0.084 231.0 2.0
3 1723853 35.57 6.6 0.086 234.0 5.0

Results for the models are residual deviance, deviance explained Discussion
(%), v2 and P. Boldface indicates the best model that was selected
as DAICc = 0; however, where DAICc is less than 2, the simplest Unimodal decreased of tree taxa richness along an entire
model has substantial support and receives consideration according
to the principle of parsimony (Burnham and Anderson 2002). elevation gradient in central Veracruz, Mexico, is con-
According to variation with elevation, Model 1 describes a sistent with previous studies and support interpretations
monotonic pattern (a + bx) whereas Model 2 (a + bx + cx2) and concerning the complex role of climatic factors associ-
Model 3 (a + bx + cx2 + dx3) describe a unimodal pattern; x is ated with peaks in richness. Although a unimodal rela-
elevation (m). Regression coefficients of each model are in Table S2
AICc Akaike Information Criterion for small size samples, DAICc tionship between species richness and elevation has been
measure of each model relative to the best model reported by several authors (Lieberman et al. 1996; Tang
and Ohsawa 1997; Vetaas and Grytnes 2002; Oommen
and Shanker 2005; Grytnes and Beaman 2006), linear
decrease or increase patterns are also frequently re-
ported. In the Sierra Madre del Sur mountain range,
Oaxaca, Mexico, a clear decrease in species, genus and
family richness with increasing elevation was revealed
(Salas-Morales and Meave 2012). For Oaxaca, Salas-
Morales and Meave (2012) proposed that the decrease in
species richness is likely the result of a critical temper-
ature isocline, where the thermophilous lowland flora
seems to be replaced by a floristic array adapted to
thermal restrictions. On the Cerro Tláloc mountain in
Mexico, a unimodal richness pattern was explained as
resulting from differences in temperature and moisture
along the elevation gradient: the lowest values of species
richness were seen in the upper half of the gradient
Fig. 3 Chao-Jaccard index for tree species comparisons between where temperatures are lower, while the highest rainfall
adjacent pair of sites along an elevation gradient in central values coincided with the elevation range where species
Veracruz, Mexico. Jaccard values vary from 0 which represents
complete dissimilarity to 1 which represents complete similarity in richness was highest (Sánchez-González and López-
tree species composition Mata 2005).
1102

Table 3 Spearman q correlation coefficients between elevation, climate variables and response variables

Elevation Tmean TSD Pptotal Ppwq BA Den Sspecies S Rarefied Sgenus Sfamily

Tmean 0.99***
TSD 0.95*** 0.95***
Pptotal 0.62* 0.63* 0.63*
Ppwq 0.51 0.50 0.50 0.97***
BA 0.80*** 0.82*** 0.78*** 0.65** 0.51
Den 0.34 0.33 0.20 0.12 0.04 0.24
Sspecies 0.59* 0.59* 0.61* 0.18 0.15 0.28 0.10
SRarefied 0.65** 0.64* 0.64* 0.16 0.13 0.31 0.28 0.92***
Sgenus 0.60* 0.58* 0.59* 0.12 0.10 0.23 0.13 0.94*** 0.91***
Sfamily 0.31 0.31 0.28 0.28 0.33 0.06 0.19 0.77*** 0.75** 0.82***
H¢ 0.66** 0.65** 0.68** 0.23 0.18 0.36 0.29 0.91*** 0.98*** 0.88*** 0.67**

Climate variables are mean annual temperature (Tmean), temperature seasonality or standard deviation (TSD), total annual precipitation
(Pptotal), and precipitation of the warmest quarter (Ppwq). Response variables are basal area (BA), density of trees (Den), species richness
(Sspecies), rarefied richness (SRarefied), genus richness (Sgenus), family richness (Sfamily) and Shannon’s diversity index (H¢)
* P < 0.05, ** P < 0.01, *** P < 0.001

dient where precipitation and energy are relatively low.

In Chiapas, Mexico, González-Espinosa et al. (2004)
found relatively high tree diversity associated with the
least productive dry end of the rainfall gradient, lower
diversity under intermediate rainfall conditions and then
increased diversity where precipitation was highest. In
contrast, Sang (2009) reported that in Xinjiang, China,
species richness showed a unimodal pattern, but water
was more important at low elevations, and temperature
at high elevations.
Some studies have reported increases or decreases in
biomass (as approximated by basal area) along entire
elevation gradients (e.g., Lieberman et al. 1996; Homeier
et al. 2010), but we observed the highest values in basal
Fig. 4 Canonical correspondence analysis triplot for tree species in
area corresponding to the upper part of the TMCF
15 sites located along an elevation gradient in central Veracruz, distribution and a CF. This peak can be explained by the
Mexico. Symbols and numbers are sites. Open circles are seasonally high levels of humidity at those sites since basal area was
dry tropical forest, squares are tropical montane cloud forest, filled positively associated with precipitation and negatively
circles are coniferous forest and the triangle is a site between dry associated with mean temperature and temperature
and cloud forest. Vectors are significant explanatory variables,
temperature seasonality or standard deviation (TSD), precipitation seasonality. Also, the trend of increasing basal area
of the warmest quarter (Ppwq) and elevation (m). Acronyms are along an elevation gradient can be attributed to other
species with higher weights in the ordination; abi, Abies religiosa; factors besides climatic or soil variables, such as tree
ber, Bernardia mexicana; bus, Bursera simaruba; car, Carpinus species composition. The highest basal area in upper
tropicalis; cei, Ceiba aesculifolia; cin, Cinnamomum effusum; clm,
Clethra macrophylla; ipo, Ipomoea wolcottiana; leu, Leucaena TMCF corresponds to a Quercus corrugata-dominated
lanceolata; liq, Liquidambar styraciflua; lys, Lysiloma acapulcense; forest, and an Oreomunnea mexicana-dominated forest.
mag, Magnolia schiedeana; orm, Oreomunnea mexicana; ore, However, in a study carried out in the same region,
Oreopanax xalapensis; pih, Pinus hartwegii; pru, Prunus rhamno- when the TMCF part was analysed as a function of
ides; qco, Quercus corrugata; qde, Q. delgadoana; qla, Q. lancifolia;
qsa, Q. sapotifolia; qxa, Q. xalapensis; sty, Styrax glabrescens; sym,
elevation there was no significant trend (Williams-Linera
Symplocos limoncillo et al. 2013). In the CF the highest basal area corresponds
to a high elevation site where Abies religiosa is the
dominant species. These coniferous forests are tall,
In our study, the highest tree species richness in the dense communities reaching 40 m height, and the pre-
low elevation SDTF is associated with energy variables sence of fir is associated with high humidity (Rzedowski
(Gentry 1995; Trejo and Dirzo 2002). At mid-elevations, 1978). Therefore, the peak indicates the presence of large
the TMCF richness is associated with summer precipi- trees with high wood volume, suggesting that biomass
tation. In another study dealing with only the TMCF production may depend on water availability rather
belt, Williams-Linera et al. (2013) also found that tax- than on any measure of energy.
onomic richness was not correlated with temperature, The ordination results strongly suggest that different
but that its correlation with precipitation seasonality climatic conditions were associated with portions of the
was very strong. The lowest species richness was re- elevation gradient. The changes with increasing eleva-
corded for the CF sites in the upper section of the gra- tion form vegetation zones defined by elevation belts
 1103

Table 4 Indicator species’ values calculated for each group of sites of species richness and a positive correlation with the
with their corresponding family name, observed maximum indica- precipitation of the wettest quarter. The distribution of
tor value >0.50 and the Monte Carlo test of significance (P values)
with 1000 permutations TMCF has been extensively associated with persistent
clouds and fog during most of the year (e.g., Grubb
Species Family Maximum P 1977; Rzedowski 1978; Gentry 1988). The Lauraceae,
indicator Fagaceae and Pentaphylacaceae (Theaceae) families
value
were dominant along our gradient, as reported by
1 Seasonally dry tropical forest Gentry (1988) for Neotropical forests and by Zhu et al.
Bernardia mexicana Euphorbiaceae 0.80 0.029 (2005) for Chinese tropical montane evergreen broad-
Bursera simaruba Burseraceae 1.00 0.001 leaved forests.
Cnidoscolus spinosus Euphorbiaceae 0.60 0.037 At the upper part of the gradient, coniferous forest was
Comocladia engleriana Anacardiaceae 0.80 0.031
Croton reflexifolius Euphorbiaceae 0.80 0.029 not associated with any particular climatic variables, but
Ipomoea wolcottiana Convolvulaceae 0.60 0.128 rather with elevation. Pinaceae was monodominant from
Leucaena lanceolata Fabaceae 0.80 0.024 above 3000 m a.s.l. to the treeline, with dominance
Lysiloma acapulcense Fabaceae 0.60 0.197 changing from Pinus patula forest to Abies religiosa forest
2 Tropical montane cloud forest
Lower montane
to Pinus hartwegii forest. O’Brien (1993) suggested that
Carpinus tropicalis Betulaceae 1.00 0.005 suboptimal levels of moisture and energy are associated
Clethra macrophylla Clethraceae 1.00 0.005 with a decrease in species richness. Our coniferous forests
Ocotea psychotrioides Lauraceae 1.00 0.005 did have the lowest richness, which coincided with low
Quercus lancifolia Fagaceae 0.67 0.082 precipitation and temperature on the gradient.
Quercus sartorii Fagaceae 0.61 0.082
Turpinia insignis Staphylaeceae 1.00 0.005
Upper montane
Alchornea latifolia Euphorbiaceae 0.55 0.124 Conclusions
Clethra schlechtendalii Clethraceae 0.67 0.091
Oreopanax xalapensis Araliaceae 0.57 0.089
Phyllonoma laticuspis Phyllonomaceae 0.67 0.091 Along the entire elevation gradient, a decreasing pattern
Prunus rhamnoides Rosaceae 1.00 0.005 in richness was apparent in which temperature season-
Quercus corrugata Fagaceae 0.67 0.108 ality was associated with SDTF at the lower end of the
Quercus delgadoana Fagaceae 0.67 0.108 gradient, and precipitation-related variables influenced
3 Coniferous forest
Abies religiosa Pinaceae 0.67 0.083 the limits of the TMCF at mid-elevations; the coniferous
Pinus hartwegii Pinaceae 1.00 0.007 forests in the upper part of the gradient were associated
with a geographic factor (elevation). Vegetation struc-
Maximum indicator values varied between 0 and 1. Boldface ture is poorly explained by richness or climate; however,
indicates significant P values
basal area peaked in the upper TMCF and in a fir forest,
suggesting that basal area may be associated with both
(Tang and Ohsawa 1997; Bach and Gradstein 2011). high humidity and tree species composition. As Hawkins
Groups of sites that were geographically close had more et al. (2003) demonstrated, there is a link between water,
similar species composition (Fig. 3). energy and richness, but it is difficult to tease apart the
SDTF sites were associated with temperature sea- direct and indirect effects of climate on richness and
sonality at the lower end of the elevation gradient where underlying mechanisms. Our study shows that different
environmental conditions such as topographic variation, plant communities are associated with certain climatic
elevation, slope and aspect were reported as important conditions and harbour different tree species, genera and
factors for the distribution of woody plants in diverse families. The results support the hypothesis that species
plant communities (Williams-Linera and Lorea 2009). richness is associated with climate, and decreases with
For dry Neotropical forest communities, Gentry (1995) elevation.
found that there is no significant change in diversity with
precipitation. Similarly, results for Mexican SDTFs Acknowledgments We thank Claudia Gallardo Hernández for
indicate that there is no association between the species assistance with species identification, and Javier Tolome for help
during field work. Two anonymous reviewers provided valuable
diversity of woody plants and total precipitation, but suggestions that greatly improved the manuscript. This work was
species richness does vary with potential evapotranspi- supported by the European Community under INCO Project
ration (Trejo and Dirzo 2002). The Leguminosae, Eu- Reforlan (CT2006-032132), CONACyT (through a Master’s
phorbiaceae and Malvaceae were dominant at the lower scholarship 239909 to MTG, Biological Sciences program,
UNAM), and the Instituto de Ecologı́a, A.C. (Project 2003010145).
end of our elevation gradient (<1000 m a.s.l.). The
distribution of families with elevation coincides with
Gentry’s (1988) observation for the lowlands of Neo-
tropical forests where the Leguminosae dominate up to References
about 1000 m a.s.l.
At mid-elevations on the gradient, TMCF sites were Acharya KP, Vetaas OR, Birks HJB (2011) Orchid species richness
associated with precipitation. In this elevation band, along Himalayan elevational gradients. J Biogeogr
Williams-Linera et al. (2013) found a unimodal pattern 38:1821–1833
1104

Aiba S, Kitayama K (1999) Structure, composition and species O’Brien EM (1993) Climatic gradients in woody plant species
diversity in an altitude-substrate matrix of rain forest tree richness: towards an explanation based on an analysis of
communities on Mount Kinabalu, Borneo. Plant Ecol Southern Africa’s woody flora. J Biogeogr 20:181–198
140:139–157 O’Brien EM, Field R, Whittaker RJ (2000) Climatic gradients in
Bach K, Gradstein R (2011) A comparison of six methods to detect woody plant (tree and shrub) diversity: water-energy dynamics,
altitudinal belts of vegetation in tropical mountains. Ecotropica residual variation, and topography. Oikos 89:588–600
17:1–13 Oommen MA, Shanker K (2005) Elevational species richness pat-
Behera MD, Kushwaha SPS (2007) An analysis of altitudinal terns emerge from multiple local mechanisms in Himalayan
behavior of tree species in Subansiri district, Eastern Himalaya. woody plants. Ecology 86:3039–3047
Biodivers Conserv 16:1851–1865 R (2007) R, a language and environment for statistical computing.
Bhattarai KR, Vetaas OR (2003) Variation in plant species richness R Foundation for Statistical Computing, Vienna.
of different life forms along a subtropical elevation gradient in [Link]
the Himalayas, east Nepal. Glob Ecol Biogeogr 12:327–340 Rahbek C (2005) The role of spatial scale and the perception of
Burnham KP, Anderson DR (2002) Model selection and multi- large-scale species richness patterns. Ecol Lett 8:224–239
model inference: a practical information-theoretic approach. Rodrı́guez SR, Morales-Barrera W, Layer O, González-Mercado E
Springer, New York (2009) A quaternary monogenetic volcanic field in the Xalapa
Colwell RK (2009) Statistical estimation of species richness and region, Eastern Trans-Mexican volcanic belt: geology, distri-
shared species from samples, version 8.2. University of Con- bution and morphology of the volcanic vents. J Volcanol
necticut, Connecticut. [Link] Geotherm Res 197:149–166
Gentry AH (1988) Changes in plant community diversity and flo- Rzedowski J (1978) Vegetación de México. Editorial Limusa,
ristic composition on environmental and geographical gradi- Mexico
ents. Ann Mo Bot Gard 75:1–34 Salas-Morales SH, Meave JA (2012) Elevational patterns in the
Gentry AH (1995) Diversity and floristic composition of Neo- vascular flora of a highly diverse region in southern Mexico.
tropical dry forest. In: Bullock SH, Mooney HA, Medina E Plant Ecol 213:1209–1220
(eds) Seasonally dry tropical forests. Cambridge University Sánchez-González A, López-Mata L (2005) Plant species richness
Press, Cambridge, pp 146–194 and diversity along an altitudinal gradient in the Sierra Nevada,
Gillespie TW, Grijalva H, Farris CN (2000) Diversity, composi- Mexico. Divers Distrib 11:567–575
tion, and structure of tropical dry forest in Central America. Sang W (2009) Plant diversity patterns and their relationships with
Plant Ecol 147:37–47 soil and climatic factors along an altitudinal gradient in the
Gillespie TW, Lipkin B, Sullivan L, Benowitz DR, Pau S, Keppel middle Tianshan Mountain area, Xinjiang, China. Ecol Res
G (2012) The rarest and least protected forests in biodiversity 24:303–314
hotspots. Biodivers Conserv 21:3597–3611 SAS (2012) JMP user’s guide. SAS Institute, Cary
Givinish TJ (1999) On the causes of gradients in tropical tree Tang CQ, Ohsawa M (1997) Zonal transition of evergreen, decid-
diversity. J Ecol 87:193–210 uous, and coniferous forests along the altitudinal gradient on a
González-Espinosa M, Rey-Benayas JM, Ramı́rez-Marcial N, humid subtropical mountain, Mt. Emei, Sichuan, China. Plant
Huston MA, Golicher D (2004) Tree diversity in the northern Ecol 133:63–78
Neotropics: regional patterns in highly diverse Chiapas, Mex- ter Braak CJF, Šmilauer P (1998) CANOCO reference manual and
ico. Ecography 27:741–756 user’s guide to Canoco for windows: software for canonical
Grubb PJ (1977) Control of forest growth and distribution on wet community ordination (version 4). Microcomputer Power,
tropical mountains: with special reference to mineral nutrition. Ithaca
Annu Rev Ecol Evol Syst 8:83–107 Trejo I, Dirzo R (2002) Floristic diversity of Mexican seasonally
Grytnes JA, Beaman JH (2006) Elevational species richness pat- dry tropical forests. Biodivers Conserv 11:2048–2063
terns for vascular plants on Mount Kinabalu, Borneo. J Bio- Vázquez JA, Givinish TJ (1998) Altitudinal gradients in tropical
geogr 33:1838–1849 forest composition, structure, and diversity in the Sierra de
Hawkins BA, Field R, Cornell HV, Currie DJ, Guégan JF, Ka- Manantlán. J Ecol 86:999–1020
ufman DM, Kerr JT, Mittelbach GG, Oberdorff T, O’Brien Vetaas OR, Grytnes JA (2002) Distribution of vascular plants
EM, Porter EE, Turner JRG (2003) Energy, water and broad- species richness and endemic richness along the Himalayan
scale geographic patterns of species richness. Ecology elevation gradient in Nepal. Glob Ecol Biogeogr 11:291–301
84:3105–3117 Wassie A, Sterck FJ, Bongers F (2010) Species and structural
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) diversity of church forests in a fragmented Ethiopian Highland
Very high resolution interpolated climate surfaces for global landscape. J Veg Sci 21:938–948
land areas. Int J Climatol 25:1965–1978 Williams-Linera G (2002) Tree species richness complementarity,
Homeier J, Breckle SW, Günter S, Rollenbeck RT, Leuschner C disturbance and fragmentation in a Mexican tropical montane
(2010) Tree diversity, forest structure and productivity along cloud forest. Biodivers Conserv 11:1825–1843
altitudinal and topographical gradients in a species-rich Ecu- Williams-Linera G, Lorea F (2009) Tree species diversity driven by
adorian montane rain forest. Biotropica 42:140–148 environmental and anthropogenic factors in tropical dry forest
Kitayama K (1992) An altitudinal transect study of the vegetation fragments of central Veracruz, Mexico. Biodivers Conserv
on Mount Kinabalu, Borneo. Plant Ecol 102:149–171 18:3269–3293
Lieberman D, Lieberman M, Peralta R, Hartshorn GS (1996) Williams-Linera G, Toledo-Garibaldi M, Gallardo Hernández C
Tropical forest structure and composition on a large-scale (2013) How heterogeneous are the cloud forest communities in
altitudinal gradient in Costa Rica. J Ecol 84:137–152 the mountains of central Veracruz, Mexico? Plant Ecol
Lomolino MV (2001) Elevation gradients of species-density: his- 214:685–701
torical and prospective views. Glob Ecol Biogeogr 10:3–13 Zhu H, Shi JP, Zhao CJ (2005) Species composition, physiognomy and
Luna-Vega I, Morrone JJ, Alcántara Ayala O, Espinosa Organista plant diversity of the tropical montane evergreen broad-leaved
D (2001) Biogeographical affinities among Neotropical cloud forest in southern Yunnan. Biodivers Conserv 14:2855–2870
forest. Plant Syst Evol 228:229–239
McCune B, Grace JB (2002) Analysis of ecological communities.
MjM Software Design, Gleneden Beach
Nogués-Bravo D, Araújo MB, Romdal T, Rahbek C (2008) Scale
effects and human impact on the elevational species richness
gradients. Nature 453:216–220

View publication stats