2009 The Author DOI: 10.1111/j.1466-8238.2008.00443.x
Journal compilation  2009 Blackwell Publishing Ltd www.blackwellpublishing.com/geb 1
Global Ecology and Biogeography, (Global Ecol. Biogeogr.) (2009)
RESEARCH
PAPER
Blackwell Publishing Ltd
Global analysis of bird elevational
diversity
Christy M. McCain
ABSTRACT
Aim Elevational gradients distributed across the globe are a powerful test system
for understanding biodiversity. Here I use a comprehensive set of bird elevational
gradients to test the main drivers of diversity, including sampling, area, mid-domain
effect, temperature, temperature and water availability, and hypotheses of evolutionary
history.
Location Seventy-eight elevational gradients of bird diversity from mountains in
both hemispheres spanning 24.5° S to 48.2° N, including gradients from various
climates, biogeographical regions and habitat types.
Methods Data on bird elevational diversity were taken from the literature. Of the
150 datasets found or compiled, only those with a high, unbiased sampling effort
were used in analyses. Datasets sampled all birds, all breeding birds or all forest birds;
a few studies detailed seasonal, elevational shifts. Eighteen predictions of diversity
theory were tested, including three sets of interactions.
Results Birds display four distinct diversity patterns in nearly equal frequency on
mountains: decreasing diversity, low-elevation plateaus, low-elevation plateaus with
mid-peaks, and unimodal mid-elevational peaks. Bird elevational diversity strongly
supports current climate as the main driver of diversity, particularly combined
trends in temperature and water availability. Bird diversity on humid mountains is
either decreasing or shows a low-elevation plateau in diversity, while on dry mountains
it is unimodal or a broad, low-elevation plateau usually with a mid-elevation maximum.
The predictions of sampling, area and mid-domain effect were not consistently
supported globally. The only evolutionary hypothesis with preliminary support was
niche conservatism.
Main conclusions Both water and temperature variables are needed to comprehensively
predict elevational diversity patterns for birds. This result is consistent for
breeding and forest birds, for both hemispheres, and for local- or regional-scale
montane gradients. More analyses are needed to discern whether the mechanism
underlying these relationships is ecological, based on direct physiological limitations
or indirect food resource limitations, or historical, based on phylogenetic niche
conservation or other evolutionary trends related to climate.The species­area and middomain
effects are not supported as primary drivers of elevational diversity in birds.
Keywords
Avian, climate, environmental gradient, evolutionary history, mid-domain effect,
mountains,species­area relationship,species richness,temperature,water availability.
Correspondence: Christy M. McCain, CU
Museum of Natural History, MCOL 265 UCB,
University of Colorado, Boulder,
CO 80309-0265, USA.
E-mail: christy.mccain@colorado.edu
Department of Ecology and Evolutionary
Biology and CU Museum of Natural History,
University of Colorado, Boulder, CO 80309 USA
INTRODUCTION
The complexity of bird diversity patterns and their underlying
causes was depicted in recent global analyses (Hawkins et al.,
2003a, 2007a; Orme et al., 2005; Rahbek et al., 2007). Bird
diversity was highest in tropical montane regions, but models
of diversity mechanisms consistently underpredicted montane
diversity (e.g. Rahbek et al., 2007). One reason for this disjunct is
C. M. McCain
 2009 The Author
2 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
scale. The strong elevational shifts in climate, habitat and
topography occur at scales of 1 to 10 km, whereas the scales
of global analyses were orders of magnitude larger (110 km2
to
1° × 1° latitude­longitude). Such large scales average much of
the important variation in the potential drivers imperative for
deciphering the high richness of montane systems (e.g. Ruggiero
& Hawkins, 2008). Even the earliest biologists ­ Aristotle, von
Humboldt, Darwin, Wallace ­ were influenced by dramatic
abiotic and biotic changes occurring within short distances on
montane gradients when forming their first ideas about life on
earth and evolution. Our fundamental understanding of niche
concepts (e.g. Grinnell, 1917; Grinnell & Storer, 1924), life zones
(e.g. Merriam & Stejneger, 1890), community structure, gradient
analysis and beta diversity (e.g. Whittaker, 1952, 1960, 1967)
came directly from studies on montane gradients.The examination
of diversity gradients on mountains also offers a dynamic and
potentially powerful, complementary approach to global-scale
studies by utilizing a large number of independent, replicated
gradients at a smaller scale (Brown, 2001; Lomolino, 2001;
McCain, 2005, 2007a).
Four aspects of mountain systems make them ideally suited
for examining biodiversity drivers: globally replicated gradients,
spatial scale, power of theoretical tests and variability of the
taxonomic signal. Thousands of elevational gradients are
distributed across the globe on all continents and on most islands
in various latitudes, climates and habitats. The power to differentiate
between diversity theories along the terrestrial, latitudinal
gradient is hindered because the main diversity theories (climate,
area, spatial constraints, history) are correlated and confounded
along the two replicate gradients in the eastern and western
hemispheres (e.g.Pianka, 1966; Rohde, 1992; Brown & Lomolino,
1998; Willig et al., 2003). With the multitude of elevational
gradients distributed across the globe, the power to differentiate
between diversity theories increases since each gradient displays
a slightly different array of variables (e.g.area,rainfall,evolutionary
history), thus allowing tests to distinguish between correlated
and confounded variables by seeking globally consistent trends
for each factor. Lastly, almost all groups of organisms exhibit one
latitudinal pattern in diversity: unimodal with diversity peaking
near the equator (e.g. Hillebrand, 2004). Several diversity
patterns are exhibited on montane gradients (see Fig. 1; Rahbek,
1995, 2005; McCain, 2005, 2007a): decreasing diversity with
increasing elevation, high diversity across a plateau of lower
elevations then decreasing monotonically, a unimodal pattern
with maximum diversity at intermediate elevations, or in rare
instances increasing monotonically. These patterns appear to
differ within and among taxonomic groups and climates (McCain,
2005, 2007a). Variability in taxonomic signal, variability in
potential drivers on individual gradients, and the array of
elevational gradients sampled historically can be used to
determine the most influential drivers of diversity in powerful,
combined analyses.
Proposed drivers of biodiversity can be grouped into four
main categories: current climate, space, evolutionary history and
biotic processes (Pianka, 1966; Gaston, 2000; McCain, 2007a).
Climatic hypotheses are based on current abiotic conditions,
such as temperature, rainfall, productivity, humidity and cloud
cover. Spatial hypotheses include the classic species­area relationship
(SAR; e.g. Terborgh, 1973; Rosenzweig, 1992, 1995) and
spatial constraint hypothesis (the mid-domain effect, MDE;
e.g. Colwell et al., 2004, 2005). Speciation rates, extinction rates,
clade age and phylogenetic niche conservatism are theoretically
linked and, in some cases, empirically linked to diversity (e.g.
Rohde, 1992; Allen et al., 2002; Stephens & Wiens, 2003; Wiens
et al., 2006; Mittelbach et al., 2007). Lastly, various biological
processes have been proposed to explain patterns in species
richness. These include competition (e.g. Terborgh & Weske,
1975), source­sink dynamics and ecotone effects (e.g. Terborgh,
1985; McCain, 2004), habitat heterogeneity (e.g. MacArthur &
MacArthur, 1961) and habitat complexity (e.g. Terborgh, 1977).
Tests of the effect of these biotic processes on diversity patterns
are minimal due to the difficulty in defining critical characteristics
as well as measuring these traits for all species along a large
spatial gradient.
Many of proposed drivers of global biodiversity are testable
along elevational gradients studied over the last century of
research on montane birds around the world (Table 1).
Complete multivariate analyses are not yet possible due to data
limitations, particularly: (1) lack of global, small-scale (10s to
Figure 1 (a) The number of bird studies demonstrating the four
elevational richness patterns on the robust, informative montane
gradients (n = 78): decreasing, low plateau, low plateau with a midelevational
peak and mid-elevational peaks (see text for definitions).
(b) Comparison of the percentage of each bird diversity pattern in
the local- and regional-scale studies (n = 23 and 55,respectively); the
frequency of each elevational diversity pattern is not significantly
different between scales (X2
= 3.03, d.f. = 3, P = 0.387).
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 3
100s of metres) data on water variables like rainfall and
productivity; (2) lack of species-level, time-calibrated phylogenetic
trees encompassing all the bird diversity for the montane
gradients; and (3) lack of appropriately sampled biotic data
(e.g. species abundances, competitive interactions). Nonetheless,
the consistency of several proposed drivers can be powerfully
tested using globally distributed elevational gradients
encompassing the variability in climate, area profiles and
biogeographical history. Those factors showing a globally
consistent signal have the greatest impact on global biodiversity,
as seen for similar studies for non-flying small mammals and
bats (McCain, 2005, 2007a,b). Bird elevational gradients may
be an even more powerful test of diversity theories, since birds
are a species-rich taxon and have been studied extensively on
Table 1 Currently testable hypotheses and predictions for bird diversity on elevational gradients.
Hypothesis: Theory Montane testable predictions
Sampling Diversity is positively related to
survey effort
A. Diversity at each 100-m elevational band will be positively correlated
to sampling effort in that elevational band
B. Local-scale studies (standardized sampling effort per elevation)
should show a different frequency of elevational diversity patterns from
non-standardized regional studies
Area Diversity is positively related to
land area
A. Diversity at each 100-m elevation band will be positively related to
area at that elevation for regional-scale studies
B. Removal of area effect results in large elevational diversity shift
C. Regional-scale studies should show more decreasing and lowelevation
plateau diversity patterns than local-scale studies (standardized
sampling area per elevation)
Mid-domain
effect (MDE)
Unimodal diversity within a
bounded domain; highest diversity
at mid-point of domain
A. Elevational diversity unimodal on all mountains
B. Empirical diversity at each 100-m band positively related to predicted
diversity at that elevation
C. Deviations of maximum diversity away from the mid-point of
mountain are randomly distributed
D. Strong, significant relationship between MDE model fit (r2
value) and
average species range size­domain size ratio
Climate I:
temperature
Diversity positively related to
temperature
A. Elevational diversity decreasing on all mountains
B. Diversity at each 100-m band positively related to average annual
temperature at that elevation
C. No statistical difference between the temperature­diversity
relationship on humid and arid mountains
Climate II:
temperature and
water (ECM*)
Diversity positively related to
productivity
A. Elevational diversity highest in warm, wet conditions: unimodal on
arid mountains; decreasing or low-plateau on humid mountains.
B. Diversity highly related to temperature on wet mountains,
relationship weaker on arid mountains
C. On arid mountains, maximum diversity occurs at wetter, cooler
conditions than the mountain base
Evolutionary
history
I = Diversity highest at elevation of
max. speciation and colonization,
min. extinction (static models)
I. Elevational diversity patterns similar on all mountains
II = Evolutionary forces different
on each mountain or region
(historical contingency)
II. Each mountain has different elevational diversity pattern; or only
regional mountains have consistent patterns
III = Diversity highest in tropical
conditions (warm, wet) within
which most taxa speciated (niche
conservatism)
III. Maximum diversity in warmest, wettest conditions; same prediction
of ECM. Further testing requires complete phylogenies; currently
unavailable
Interactions I. Area­MDE I. Area-corrected diversity improves fit to MDE
II. Area­MDE­temperature II. Multiple regressions quantify relative contribution
III. Area­MDE­ECM III. If area a primary factor in II above, then area-corrected data changes
fit to MDE and ECM
*ECM = elevational climate model, see text for definition.
Positive relationship with productivity and diversity would be the ideal montane hypothesis, but is currently not testable here given lack of sufficiently
small-scale, global data for net primary productivity or precipitation.
C. M. McCain
 2009 The Author
4 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
mountains across the globe due to their visibility and predominantly
diurnal nature. Here I examine global patterns for bird
faunas along elevational gradients in various latitudes, climates
and biogeographical regions. I detail the testable diversity
hypotheses and their specific predictions for montane gradients.
I then examine the consistency in which these diversity patterns
correspond to the predictions of climate, space (area and MDE)
and evolutionary history alone and in combination in an attempt
to delineate the primary drivers of bird diversity globally.
MATERIALS AND METHODS
Bird elevational data
I found papers on bird elevational diversity by searching BIOSIS
and Web of Science using various combinations of keywords: bird,
avian, diversity, species richness, elevation(-al), altitude(-inal).
Elevational diversity was also compiled from regional bird field
guides and faunal surveys that included elevational ranges for
each species (e.g. Inskipp et al., 1999; Grimmett et al., 2000;
Robson, 2002; Hilty, 2003). Only studies surveying all birds,
terrestrial birds, breeding birds or forest birds were used in the
quantitative analyses. Studies examining only endemics or a
particular bird guild (e.g. insectivores or thrushes) were not
included. Endemic, elevational bird diversity will be treated in a
separate paper (C.M. McCain, unpublished). If data were available
in a single study differentiating breeding, forest and oceanic
birds, summer and winter elevational ranges and/or different
mountain slopes, then each of these datasets was analysed
individually and compared. In the final analyses, breeding bird
elevational ranges excluding oceanic birds were used preferentially
if available. Only two datasets included altitudinal migrant
elevations for winter and summer (Inskipp et al., 1999;
Grimmett et al., 2000); both showed no difference in shape of
the diversity pattern, although winter diversity decreased
more rapidly with elevation than summer diversity in both
studies.
Datasets were selected based on several a priori sampling
criteria, since sampling is so important for detecting unbiased
patterns in comparative analyses (McCain, 2005, 2007a; Rahbek,
2005; see Appendix S1). First, datasets were included in quantitative
analyses only if sampling covered at least 70% of the elevational
gradient. This is a critical restriction for studies with limited
sampling at lower elevations, since discernment of pattern is
strongly contingent on diversity estimates in the lowest 600 m. In
contrast, limited sampling at the highest elevations is less critical
since diversity always decreased monotonically above some
threshold intermediate elevation. Second, sampling effort
needed to be substantial and not biased strongly elevationally,
although in practice this could be difficult to assess due to the
highly variable nature of the explanations of sampling in the text
of the studies. Third, assessments were made (or noted if
discussed by authors) of elevational correlations in species
richness with sampling intensity or wide-scale habitat disturbance
(e.g. Taiwan). In cases where the authors stated there
were elevations at the top or bottom of the mountain which were
undersampled, these elevations were not included in the analyses
(e.g. the lowest 200 m in Bhutan, Inskipp et al., 1999).
In reanalyses, I assumed that a species was present between its
highest and lowest reported elevations (range interpolation).
This improved methodological consistency, since most
published accounts assumed range interpolation. Interpolation
aids in overcoming some limitations of undersampling, but may
also inflate estimates of species richness at mid-elevations
(Grytnes & Vetaas, 2002). Species richness was then calculated
based on the number of bird ranges occurring at each 100-m
elevational band from the base to the top of the mountain (e.g.
0­99.9 m, 100­199.9 m, etc.) unless a larger resolution was used
in a particular study without raw data for reanalysis. For those
studies without species elevational range data, the bird diversity
estimates at each surveyed elevation were taken from figures or
tables in the publication.
For all gradients, species richness patterns were assigned to
four categories: decreasing, low plateau, low plateau with a
mid-peak and mid-elevation peak (see Fig. 1; more patterns were
possible but only these four were necessary with current data).
Decreasing richness patterns are those in which species numbers
decline monotonically with increasing elevation. Low-plateau
patterns have more than 300 m of consecutively high richness
at the mountain base and thereafter decreasing species
richness. Low-plateau patterns with a mid-peak have high
richness across low elevations (> 300 m) with a diversity
maximum found above 300 m from the base. Mid-elevation
peaks have a unimodal peak in diversity at intermediate
elevations (above 300 m) with 25% or more species than at the
base and top of the mountain.
Elevational gradients were generally studied at two scales: local
(alpha) or regional (gamma) species richness. Alpha richness
refers to local samples taken from field transects along single
elevational gradients. Gamma datasets are species richness
patterns compiled from sighting and capture records, specimen
records and field notes for an entire mountain or mountainous
region regardless of slope, area or standardized trapping effort
across elevations. Regional richness may be highly influenced by
area (Rahbek, 1997, 2005; Brown, 2001; Lomolino, 2001; Willig
et al., 2003; McCain, 2005, 2007b), and may have significant
sampling biases (Rickart, 2001).As the scale (grain) of the data in
alpha and gamma patterns are qualitatively and quantitatively
different, the factors producing these patterns may not necessarily
coincide (McCain, 2005; Rahbek, 2005).
Diversity hypotheses and testing methods
Many hypotheses proposed to explain large-scale patterns in
species richness apply to elevational richness patterns. Several are
currently testable, including sampling, the species­area effect,
the MDE, temperature, temperature and water (elevational
climate model,ECM),a few theoretical constructs of evolutionary
history and some combinations of these hypotheses (Table 1).
Below I will detail each hypothesis, specify its elevational
application, list its specific predictions and delineate appropriate
methods for elevational bird analyses.
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 5
Sampling
Large differences in sampling effort across the gradient may
result in an experimental bias in diversity estimation (e.g.
Colwell & Coddington, 1994). In cases where elevational bands
were sampled with unequal effort, a relationship between diversity
and elevation could simply be a result of differential sampling
effort. Two sampling predictions can be tested: (A) a positive
correlation between bird diversity and a quantification of
sampling effort (e.g.survey days,mist-net hours) across elevational
bands; (B) local-scale studies that standardized sampling
effort among elevations should show a significantly different
distribution of elevational diversity patterns than nonstandardized
regional studies. Studies with major sampling
biases and significant, positive correlations between elevational
sampling effort and diversity were excluded. Exclusion was
necessary because current analyses are conducted to delineate
non-anthropogenic influences on bird diversity. A chi-square test
of homogeneity determined if sampling methodology resulted
in different frequency distributions of elevational diversity
patterns between standardized local gradients and nonstandardized
regional gradients.
Area
The SAR asserts that as survey area increases the number of
species encountered increases (Terborgh, 1973; Rosenzweig,
1992, 1995 and references therein). On mountains, the SAR
predicts that elevational bands covering more area (e.g. mountain
base) should harbour more species than elevational bands
covering a small area (e.g. mountain tops) (Rahbek, 1997; McCain,
2007b). The SAR is based on the assumption that at regional and
global scales extinction rates should decrease and speciation rates
should increase with area due to the increased likelihood of
barrier formation and increased population densities (Rosenzweig,
1995 and references therein). At small spatial scales, Rosenzweig
(1995) argued that habitat diversity and the strong ties of
particular species to habitat drive the local SAR. The elevational
diversity patterns on mountains may fall somewhere on the
continuum between these two scales and processes.
SAR prediction (A) is that studies at regional scales which
amass diversity data across many sites at each elevation should
show a positive relationship between bird diversity and elevational
band area (Rahbek, 1997; McCain, 2007b). This was
assessed by SAR regressions published in the original studies or
by calculating SAR regressions for those datasets where both
species richness data and area of elevational bands were available
at the same scale from the published paper or earlier area studies
(McCain, 2007b). In the latter case, area estimates for each
100-m elevational band were calculated using ArcGIS and
digital elevation models at 90­100 m resolution or smaller
(USGS data; see McCain, 2007b, for more details). I tested both a
linear [no. of species = constant × area (S = cA)] and a curvilinear
(log S = z log A + log c) area relationship. Regressions estimated
the strength, slope and significance of the SAR (e.g. Conner &
McCoy, 1979; Rosenzweig, 1995; McCain, 2007b).
SAR prediction (B) is that gradients with a significant, positive
relationship between area and diversity, adjusting the species
richness curve by standardizing for area should result in a large
change in the elevational diversity pattern (McCain, 2007b).
Area standardization was based on calculating area-corrected
diversity curves using the power function (S = cAz
; e.g. Conner &
McCoy, 1979; Rosenzweig, 1995) with a comparison of two zvalues:
a z-value based on the slope of log­log SAR regressions
and the canonical z-value (0.25) proposed by Preston (1962).
Area adjustment simply involves solving the power function
equation for the constant (c) after plugging in known values for
z, number of species and area estimate at each 100-m elevational
band. This is simply an area correction that calculates the
expected diversity at each elevation if the amount of area was
held constant among elevational bands, given the known logistic
relationship between diversity and area. These corrections are
not reanalyses of the area residuals from SAR regressions, so
avoid pitfalls inherent in analyses of residuals (e.g. Freckleton,
2002). Previous work compared several techniques for area
correction and estimation of parameters, and found that most
correction techniques produce highly correlated results, but
supported the technique used here as the most robust (McCain,
2007b).
SAR prediction (C) is that regional-scale studies should
display more decreasing and low-elevational plateau patterns,
since land area is generally highest toward the base of mountains,
than local-scale studies which sampled small plots of relatively
standardized area per elevation. A chi-square test of homogeneity
tested for a significantly different frequency distribution in
elevational diversity patterns between standardized local
gradients and regional gradients.
Mid-domain effect (MDE)
The MDE assumes that spatial boundaries (e.g. the base and top
of a mountain) cause more overlap of species ranges toward the
centre of an area where many large- to medium-sized ranges
must overlap but are less likely to abut an edge of the area (Colwell
et al., 2004, 2005 and references therein). On mountains, MDE
predicts (McCain, 2004) (A) a unimodal diversity curve with
maximum diversity at the mid-point of the mountain. A chisquare
goodness of fit test assessed whether there are significantly
more mid-elevation peaks than any other diversity pattern.
Additionally, the MDE predicts (B) a strong, positive association
between predicted diversity based on Monte Carlo simulations
(Mid-Domain Null; McCain, 2004) and empirical diversity at
each 100-m elevational band.Mid-Domain Null simulates species
richness between the mountain base and the summit based on
random placement of empirical range sizes sampled without
replacement (McCain, 2004). Regressions of empirical and
predicted values, based on the average of 50 000 simulations at
each 100 m elevational band give r2
estimates of MDE fit
(Colwell et al., 2004; McCain, 2004).
Two secondary predictions of MDE are testable. MDE predicts
(C) that deviations in maximum diversity away from the
mid-point of the mountain should be randomly distributed if
C. M. McCain
 2009 The Author
6 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
spatial constraints alone drive elevational diversity (e.g. area or
climate is not directionally skewing the diversity peak away from
the mid-point of the mountain; Student's t-test). Lastly, MDE
predicts (D) a strong, significant relationship between MDE fit
(r2
value) and a ratio of the average bird range size to elevational
gradient length. This is based on the premise that MDE
predictions are based mainly on the overlap of medium- and largeranged
species, and do not apply necessarily to small-ranged
species; thus gradients with more large-ranged species should
show better fits to MDE (Lees et al., 1999; Colwell et al., 2004;
Dunn et al., 2007). Linear regressions of MDE r2
values with a
ratio of average bird range size (m) divided by the length of the
elevational gradient (m) for each study will assess the strength of
MDE range size prediction (Lees et al., 1999; Dunn et al., 2007).
Climate I: temperature
Climatic tolerances put restrictions on how many species can
survive at different locations and elevations (e.g. Brown, 2001;
Hawkins et al., 2003b). A positive relationship between temperature
and diversity has been shown for many large-scale diversity
patterns, although hypotheses about the underlying mechanism
differ (Pianka, 1966; Kaspari et al., 2000; Allen et al., 2002;
Hawkins et al., 2003b; Evans et al., 2005; Sanders et al., 2007).
On mountains, temperature decreases monotonically by an
average of 0.6 °C per 100-m elevational gain (Barry, 1992).
If temperature is a main determinant of bird diversity, the
predominant elevational diversity pattern predicted is decreasing
diversity with decreasing temperature and increasing elevation
(prediction A; chi-square goodness of fit test). Specifically (B),
bird diversity in each elevational band should be significantly
and positively correlated with temperature on all montane
gradients in linear regressions. Because each study did not
publish temperature variation with elevation, the average annual
temperature for the lowest 100 m was taken from the WorldClim
database (1 km2
scale; http://www.worldclim.org/), and for
standardization assumed to decrease by 0.6 °C per 100-m elevation
gain (Barry, 1992). Lastly (C), this positive temperaturediversity
relationship should not differ among mountains in
different climatic regimes, in particular arid mountains and
wet-forested mountains should show equally strong relationships
with temperature. The equality of r2
values between arid and wet
mountains is compared with a nonparametric, Mann­Whitney
U-test.
Climate II: temperature and water
Climatic productivity has also been strongly and positively
linked to diversity (O'Brien, 1993; Gaston, 2000; Kaspari et al.,
2000; Mittelbach et al., 2001; Hawkins et al., 2003a,b; Evans et al.,
2005). Several mechanisms attempt to explain this relationship
and are nicely summarized in Evans et al. (2005); many of these
are not testable directly here due to a lack of abundance data and
small-scale productivity/water data. However, an elevational
climate model (ECM) was recently proposed and supported for
bats (McCain, 2007a), wherein separate gradients in temperature
and water availability on mountains in different climatic regimes
predict divergent elevational diversity patterns (Fig.2). Temperature
decreases with elevation on all mountains, while rainfall and
water availability follow more complex relationships with
elevation depending on the local climate. On arid mountains
(e.g. south-western US mountains), water availability is highest
at intermediate elevations where rainfall and soil water retention
are highest and evaporation lowest. Water availability drops off
dramatically towards the low elevations where high temperature
and high evaporation exceed rainfall inputs, leading to arid
habitats. Towards the upper elevations water availability also
drops toward the mountaintop, although less dramatically, as
runoff is high due to shallow soils and exposed rock, and
precipitation is seasonally inaccessible as snow and ice. On
humid mountains (e.g. the eastern Andes), water availability is
high across a broad base of lower elevations and only decreases
toward the tops of the mountains, again due to higher runoff and
decreases in rainfall.
Bird species richness is predicted to be positively related to the
warmest and wettest conditions elevationally, predicting (A)
mid-elevation peaks in bird species richness on arid mountains
and decreasing diversity on warm, wet mountains (Fig. 2). Each
elevational gradient is assigned to an arid or humid mountain
classification based on vegetation present at the base of the
mountain: arid includes hyper-arid, arid, and semi-arid vegetation
types (humidity index < 0.50; UNEP, 1997, World Atlas of
Desertification classification) and humid mountains include
various forest vegetation types (> 0.50 humidity index). A chisquare
test of homogeneity is used to assess whether particular
diversity patterns are more associated with a particular base
climate than random. ECM prediction (B) is that temperature
should show a stronger relationship with bird diversity on
Figure 2 Elevational climate model for bird species richness,
incorporating a linearly decreasing temperature gradient and a
unimodal water availability gradient. Bird species richness is
depicted in grey tones with darker tone indicating more species. The
placements of generalized wet and dry montane gradients are shown
below the x-axis. Unimodal mid-elevational peaks in bird diversity
are predicted on dry mountains and decreasing or low-plateau
patterns predicted on wet mountains.
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 7
mountains in humid climates, than with bird diversity on
mountains in arid climates (linear regressions and MannWhitney
U-test). Finally, on arid mountains (prediction C), the
mid-elevation climatic conditions at the diversity maximum
should be wetter and cooler than at the mountain base. Climate
conditions were estimated using annual precipitation and
temperature from the WorldClim database (1 km2
scale; http://
www.worldclim.org/). These data do not include differences in
evaporation or runoff and are at a coarser scale than most
elevational variation, so will underestimate the actual differences
in water availability. However, these data may provide additional
support to ECM if maximum bird diversity on arid mountains
occurs at mid-elevations that are statistically wetter and cooler
than conditions at the mountain base even if underestimated
(paired t-test).
Evolutionary history
Evolutionary rates hypotheses are less developed elevationally
than latitudinally, although recent analyses are forging new
ground (e.g. Wiens et al., 2007). Due to a lack of species-level
time-calibrated phylogenetic trees and speciation/extinction rate
data for each mountain, many phylogenetic trends like the
time-for-speciation effect are not testable here (Stephens &
Wiens, 2003; Wiens et al., 2007). But some general, simplistic
predictions can be examined. These tests are weaker than those
for climate and space, but will emphasize the critical characteristics
that an evolutionary hypothesis would need to be globally
supported for birds. Static evolutionary models (prediction A)
propose a region on mountains where conditions promote
speciation and dampen extinction risk, and therefore predict a
single diversity optimum. For example, if speciation is highest at
the mountain base and declines with elevation, and extinction is
highest at the mountaintop and decreases toward the base, then
the evolutionary optimum for diversity would occur at
mountain bases globally. A single, consistent diversity optimum
is the general prediction of all static evolutionary models
proposed to date (e.g. Brown, 2001, p. 107; Heaney, 2001,
hypothesis 12; Lomolino, 2001, Fig. 3). If evolutionary factors
generating montane diversity are strongly contingent on local
fauna, conditions and biogeographical history, then no consistent,
global signals in elevational diversity are predicted (prediction B).
Such a historical contingency would predict a lack of consistent
patterns globally or only regionally consistent patterns. Lastly
(prediction C), niche conservatism proposes that climatic conditions
within which a taxon evolved ­ tropical-like conditions for
most modern groups and species ­ should be conserved in the
present (Wiens & Donoghue, 2004). Thus, the basic elevation
trend that niche conservatism predicts is maximum bird diversity
in the warmest, wettest conditions on mountains; the same
general pattern as that of ECM.
Interactions
Three hypothesis interactions can be tested for variables
quantified at the same scale as bird diversity (e.g.100-m elevational
bands): area, MDE and temperature. (I) Interaction of area and
MDE, maximum diversity may be consistently skewed away from
the mountain mid-point toward the portion of the mountain
covering the most area. Such a relationship predicts that once
diversity has been standardized for area (area-corrected diversity
curves: area prediction B), then the fit to MDE should be significantly
improved. This interaction will be tested with a nonparametric,
Mann­Whitney U-test to see if MDE r2
values for areacorrected
diversity are higher than MDE r2
values for empirical
diversity. (II) Interaction of area, MDE, and temperature can all
be simultaneously tested using multiple regressions to assess the
relative contribution of each factor. Finally (III), if area is more
important to bird diversity than temperature and MDE in
multiple regressions, then the precedence of the area effect is
established. In that case, area-corrected diversity curves can be
used to reassess the fits of ECM and MDE. Lastly, spatial autocorrelation
exists both in the bird elevational diversity and
environmental data, which may influence the results of these
analyses, although not necessarily (Diniz-Filho et al., 2003; Hawkins
et al., 2007b). Due to the length and complexity of current
analyses, spatial autocorrelation analyses and influences of
multiple tests (e.g. Bonferroni corrections) need to be treated in
a separate paper.
RESULTS
I found 190 elevational gradients of bird species richness in 150
published studies. Only 78 gradients met a priori sampling
criteria and had independent robust content for quantitative
analyses (see Appendices S1 and S2 in Supporting Information).
These gradients cover mountainous regions from various
climates (tropical, subtropical, arid, temperate), latitudes (24.5° S
to 48.2° N), longitudes (119.8° W to 147.3° E), landmasses
(islands and continents), mountain heights (851 m to 8848 m)
and scales (local and regional) (Figs 3 & S1). Four elevational
diversity patterns were represented (Fig. 1a): 29% decreasing,
19% low-elevation plateaus, 26% low-elevation plateaus with
mid-peaks and 26% unimodal with mid-peaks. This distribution
of diversity patterns was the same for studies of all birds (n = 28),
terrestrial birds (n = 8), breeding birds (n = 28) or forest birds
(n = 14) (2
= 8.143, d.f. = 9, P = 0.5198). Continents and
islands generally showed the same patterns, although islands
showed fewer mid-peaks mainly since they were predominantly
tropical (see results below). There was no relationship between
elevational diversity and mountain height (r2
= 0.015, P = 0.282,
n = 78), or latitude (r2
= 0.040, P = 0.079, n = 78). Tests of the
predictions of each diversity hypothesis and interactions
(Table 1) are detailed sequentially below:
Sampling
No distinct sampling trends existed once biased datasets were
removed. No differences were detected in diversity patterns
based on sampling protocol between standardized local and nonstandardized
regional-scale studies (prediction B; Fig. 1b;
2
= 3.03, d.f. = 3, P = 0.387, n = 78).
C. M. McCain
 2009 The Author
8 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
Area
Support for the three area predictions was mixed. Positive,
significant SARs (prediction A) were detected in 16 of 22 studies
with appropriate data either in linear or curvilinear analyses:
average r2
values of 0.29 and 0.63, respectively (Fig. 4a, Table S1).
Significant shifts in diversity pattern with area-corrected
diversity curves occurred in 12 of 16 significant gradients
(prediction B): 4 showed little change from the empirical diversity
curve, 11 showed an upward shift in maximum diversity to a
mid-elevational peak, and 1 shifted to a low-elevation plateau
(Table S1). As shown in previous studies (McCain, 2007b), these
analyses are robust to methodological and parameter variation.
Lastly (prediction C), regional-scale studies with a proposed
larger species­area influence did not show a preponderance of
decreasing or low-plateau diversity patterns compared with local
studies with area standardization (Fig. 1b; 2
= 0.002, d.f. = 1,
P = 0.9643, n = 38).
Mid-domain effect
Bird diversity had low concordance with the four MDE
predictions. Elevational diversity was not uniformly unimodal
(contrary to prediction A). Fits to the null model were generally
poor: 0.192 average r2
value for those studies with available
species range data (prediction B; Fig. 4b; n = 43; Appendix S1).
Local gradients had higher average r2
values than regional gradients
(0.344 and 0.126, respectively), mainly due to slightly more
mid-peaks found in local-scale studies (although not significantly
more). Deviations in maximum bird diversity were not
randomly distributed around the mountain mid-point, but were
shifted to significantly lower elevations (prediction C; t = ­7.10,
P < 0.0001, n = 43). Lastly, there was a positive relationship
between average bird range size relative to gradient length and fit
to MDE (r2
value), although the overall relationship was not
particularly strong for all gradients (r2
= 0.274, P = 0.0003,
n = 43) or for just those with MDE predicted mid-peaks
(r2
= 0.260, P = 0.0183, n = 21).
Climate I: temperature
Support for the three temperature predictions was mixed. Bird
diversity did not exhibit predominantly decreasing and lowplateau
patterns as predicted by decreasing temperature on
mountains (prediction A). Bird elevational diversity was
positively and significantly related to temperature on 65 of 76
gradients with testable data, although the strength of the
relationship was highly variable ranging from r2
values of 0
to 0.98 (prediction B; Fig. 4c; Appendix S1). Contrary to
temperature predictions (prediction C), the strength of the
temperature­diversity relationship was contingent upon local
climate (humid average r2
= 0.807, arid average r2
= 0.496;
Mann­Whitney U-test: Z = 15.74, P < 0.001, n = 76).
Climate II: temperature and water (ECM)
The three predictions of ECM were all consistently supported.
All decreasing and most low-plateau diversity patterns occurred
on wet mountains, and the majority of mid-peak patterns
were from dry mountains (prediction A; Fig. 5a; X2
= 30.79,
d.f. = 3, P < 0.0001). The trend was strongest in the western
hemisphere (Fig. 5b; X2
= 84.70, d.f. = 3, P < 0.0001), but was
significant in the eastern hemisphere, most biogeographical
regions (Fig. 3, Fig. S1), and separately for breeding bird
Figure 3 The distribution of study sites (circles), the four main elevational richness patterns for each geographical region (bars) and the
number of patterns on wet- and dry-based mountains (dark and stippled, respectively): North America, Central America, South America,
Europe, continental Africa, Madagascar, Asia and Australasia. D = decreasing, L = low plateau, LM = low plateau with mid-elevational peak,
M = mid-elevational peak (see text for definitions).
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 9
gradients (Fig. 5c; X2
= 73.79, d.f. = 3, P < 0.0001) and local
gradients (Fig. 5d; X2
= 13.47, d.f. = 3, P = 0.0037). As predicted
(prediction B), the temperature­diversity relationship was
stronger on humid rather than dry mountains. This trend was
also supported when only considering patterns with mid-peaks
[LPMP: Z = 4.56, P < 0.001; MP: Z = 8.42, P < 0.001]. Even with
coarse climatic data (prediction C), maximum bird diversity on
arid mountains was located in significantly wetter and cooler
conditions than at the mountain base (Fig. 6; rainfall: paired
t = 2.66, P = 0.007, n = 23; temperature: paired t = ­8.55,
P < 0.0001, n = 23).
Evolutionary history
Of the three general predictions of historical factors underlying
elevational diversity, only niche conservatism was supported.
There appears to be no single optimum in species diversification
on mountains (theory I: static models), since there are four
distinct trends in elevational diversity. Similarly, diversity
patterns were neither ubiquitously different nor regionally
consistent (theory II: historical contingency), but varied within
regions due to local climatic conditions. The warm, wet climates
under which birds are purported to have diversified consistently
do harbour the most species on mountains, thus supporting the
simplest prediction of niche conservatism (theory III).
Interactions
An area­MDE interaction was not supported (interaction I).
There was no significant increase in MDE fit when calculated
with area-corrected bird diversity with either z-value, as
mean r2
values were nearly identical: empirical = 0.134,
mountain z = 0.131, z of 0.25 = 0.139 (Mann-Whitney U-test
(mountain z): Z = 0.624, P = 0.268, n = 11; (z = 0.25) Z = 0.295,
P = 0.386, n = 11). Multiple regression analyses of area,
temperature and MDE with bird elevational diversity tested with
each 100 m elevational band (interaction II; all variables logtransformed
for normality) found all three factors significant in
the final model [r2
= 0.7211, F ratio = 441.72, P < 0.001, n = 542
(elevation bands)] with temperature contributing the most to
the model (r2
= 0.2343), followed by MDE (r2
= 0.1176), then
area (r2
= 0.0554). Because area in the multiple regressions was
shown to be the factor with the least importance, there is no
statistical or biological foundation for testing interaction III.
DISCUSSION
In the not so distant past it was thought that all diversity patterns
of birds, in fact all faunal and floral groups, decreased with
increasing elevation (e.g. MacArthur, 1972; Terborgh, 1977;
Brown & Gibson, 1983; Brown, 1988; Rohde, 1992; Stevens
1992). Rahbek (1995) falsified this assumption by showing that
many elevational gradients have mid-elevational peaks in
diversity. This bird analysis shows conclusively using all available
and well-sampled datasets that neither decreasing nor midelevational
peaks are the sole predominant pattern for birds on
mountains. Bird diversity follows four general patterns in nearly
equal frequency: decreasing, low plateau, low plateau with a
mid-elevation peak and unimodal with a mid-elevational peak
(Fig. 1). These patterns are seen across all comprehensive bird
assemblages (all, terrestrial, breeding or forest birds), across both
theeasternandwesternhemispheres,andwithineachbiogeographical
region (Figs 3 & S1).This elevational variability may be attributable
Figure 4 Regression analyses of the species­area, mid-domain and
temperature­diversity relationships using elevational patterns in
bird diversity: (a) Fits to the log­log species­area relationship
(n = 22) with an overall mean r2
of 0.634.(b) Fits to the mid-domain
model (n = 43) with an overall mean r2
of 0.192. (c) Fits to the
temperature­bird diversity relationship with an overall mean r2
of
0.710, contrasting humid and dry mountain relationships
(r2
= 0.807, n = 51; r2
= 0.496, n = 24, respectively).
C. M. McCain
 2009 The Author
10 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
to particular proposed drivers of global biodiversity, including
sampling, spatial, climatic or evolutionary factors (Table 1), but
is the support for any of these consistent enough to be globally
persuasive?
Sampling
Robust patterns in bird elevational diversity are not due to
sampling bias. After the removal of studies with obvious
sampling biases, there was no sampling signal in diversity
patterns between standardized and non-standardized studies.
Additionally, sampling simulations conducted on elevational
diversity patterns found that unrealistically large changes in a
significant proportion of the species ranges are necessary to
change overall diversity patterns (e.g. an average error of 2000­
3000 m in range sizes; McCain, 2007a). Simulations showed
undersampling insufficient to change most decreasing patterns
to a mid-peak, unless 25­50% of species occurring on a
mountain are currently unknown and with ranges that occur
only at intermediate elevations (McCain, 2007a). Such vast
numbers of unknown bird species and huge range underestimations
for large percentages of the birds on each mountain are
highly unlikely with our current distributional knowledge of
birds. Additionally, it would be very difficult to associate
differential sampling biases to the dichotomous diversity trends
seen on humid and arid mountains.
Area
Some predictions of the SAR were supported for montane birds:
many studies showed a significant, positive relationship between
elevational bird diversity and elevational area, and more than
half of the significant SAR studies showed substantial change in
the elevational diversity pattern with area corrections (Fig. 4a,
Table S1). Whereas other SAR predictions were not supported,
regional-scale studies, many with large area effects, did not show
significantly more decreasing or low plateaus in diversity than
area-standardized, local-scale studies (Fig. 1), and area had the
least influence on bird diversity after temperature and MDE in
combined multivariate analyses (interaction II).
Mixed support for the area effect on mountains is consistent
with similar analyses of non-flying mammals and bats (McCain,
2007b). Like in birds, strong area effects were detailed for only
half of the mammalian gradients. Other taxonomic groups also
show mixed SAR effects on mountains: supportive (plants: Jones
et al., 2003; Bachman et al., 2004; ants: Sanders, 2002; fish: Fu
et al., 2004) and not supportive (plants: Odland & Birks, 1999;
Vetaas & Grytnes, 2002; Bhattarai et al., 2004; Oommen &
Shanker, 2005; ants: Sanders et al., 2003). Generally, strong
correlations with area occur on mountains with decreasing or
low plateau diversity, but infrequently with low plateaus
with mid-peaks and unimodal patterns in diversity (Table S1;
McCain, 2007b). These mixed results warrant two main
Figure 5 The elevational climate model
(ECM), based on temperature and water
availability,predicts maximum bird richness at
mid-elevations on dry mountains, and
decreasing and low plateaus in bird richness of
wet mountains. ECM was statistically
supported across the robust elevational
gradients in bird species richness: (a) all
localities (X2
= 30.8, d.f. = 3, P < 0.0001),
(b) western hemisphere (X2
= 84.7, d.f. = 3,
P < 0.0001), (c) breeding birds (X2
= 73.8,
d.f. = 3, P < 0.0001), (d) local-scale gradients
(X2
= 13.47, d.f. = 3, P = 0.0037). Low Plateau
(MP) refers to low plateaus with mid peaks.
Figure 6 A contrast of climate conditions (average annual
precipitation and temperature) at the mountain base and at the
mid-elevation with maximum bird richness on arid mountains. Bird
diversity occurs at significantly wetter (paired t = 2.66, P = 0.007)
and cooler (paired t = ­8.55, P < 0.0001) conditions at higher
elevations on dry mountains (WorldClim data; pixel scale = 1 km2
).
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 11
conclusions. First that area, which on most mountains decreases
with increasing elevation,tends to be correlated with the decreasing
temperature effect on mountains, and for montane birds,
temperature was shown to be the stronger of the two effects in
multiple regressions. Second, the fact that many bird gradients
showed a negative or no relationship with area, and no large
differences existed between regional- and local-scale studies
shows conclusively that area cannot be the main driver of
bird elevational diversity, only a factor of secondary importance.
For that reason, I caution against methodologies promoting
the initial removal of area effects before examining other
potential drivers of diversity (e.g. Rahbek, 1997; see Freckleton,
2002).
Mid-domain effect (MDE)
Spatial constraints on species ranges are theoretically implicated
and, in some cases, cannot be rejected as a contributing factor to
mid-elevational peaks in species richness (e.g. Colwell et al.,
2004, 2005; but see Hawkins et al., 2005; Zapata et al., 2005). The
MDE predicts unimodal species richness with a maximum at the
mid-point of all elevational gradients, but montane bird
assemblages were not predominantly unimodal (Fig. 1). In fact,
none of the four MDE predictions were strongly supported. The
vast majority of MDE regressions found low fits to MDE
predictions (average r2
= 0.192; Fig. 4b, Appendix S1). And
deviations of maximum diversity away from the mountain
mid-point were significantly skewed toward lower elevations, not
randomly distributed as expected.There was a positive relationship
between average bird range size and the MDE fit, but the
relationship was weak (r2
= 0.27), nearly the strength of the
average MDE fit. Lastly, MDE fits were not improved when area
and MDE were tested together. Similar trends in low MDE fits,
skewed deviations and lack of an area­MDE interaction were
documented along elevational gradients globally for non-flying
small mammals (r2
= 0.238; McCain, 2005, 2007b) and bats
(r2
= 0.156; McCain, 2007a,b). For most vertebrate groups and
other taxa examined so far (e.g. Dunn et al., 2007), elevational
patterns are not directly consistent with MDE. Thus, the main
drivers of elevational bird diversity do not include either spatial
variable considered here, SAR or MDE.
Climate
Support for the temperature model was mixed, reflecting that
temperature alone is not responsible for the distribution of
elevational diversity patterns in birds. I did not find a predominance
of decreasing or low plateaus in bird diversity on mountains. and
the temperature­diversity relationship was significantly stronger
on humid mountains than dry mountains (Figs 1 & 4c,
Appendix S1). Temperature had the strongest effect in multivariate
analyses, but still a substantial portion of variation in diversity
remained unexplained. Both water and temperature variables are
needed to explain montane bird diversity.
Temperature­water predictions of the elevational climate
model (ECM; Fig. 2) were the only set of predictions that were
uniformly and strongly supported (Figs 3, 4c, 5 & 6). Maximum
bird diversity occurred at mid-elevation on dry mountains and
at low elevations on humid mountains as predicted (Figs 3 & 5).
Temperature was more strongly linked to diversity on wet
mountains than on dry mountains (Fig. 4c). On arid mountains,
maximum bird diversity occurred at significantly wetter and
cooler climatic conditions at mid-elevations than climatic
conditions at the base of the mountain (Fig. 6), even with
underestimates of water availability. These results emphasize that
water in a necessary factor modifying the temperature effect on
elevational diversity.
Montane bats also strongly supported ECM: bat diversity was
always unimodal on dry mountains and always decreasing on
humid mountains expect for a single exception which simulations
show may be undersampled at low elevations (McCain, 2007a).
Additionally, water and energy/temperature are correlated with
elevational species richness of epiphytes (Krömer et al., 2005)
and ferns (Bhattarai et al., 2004; Kluge et al., 2006). A positive
relationship between species richness and water and energy
regimes has previously been demonstrated for woody plants
across southern Africa (O'Brien, 1993) and across multiple
large-scale geographical studies for plants and animals
(Hawkins et al., 2003b, and references therein). Clearly,
water and energy regimes are essential to the physiology of
plants and directly influence distribution and species richness
(O'Brien, 1993; Hawkins et al., 2003b; Bhattarai et al., 2004;
Krömer et al., 2005; Kluge et al., 2006). For animal richness
patterns, the focus becomes the relative importance of direct
(physiological limitations) and indirect (food resource) effects of
water and temperature on diversity (Hawkins et al., 2003b;
McCain, 2007a).
Endothermic physiological constraints of birds may directly
limit their ability to sustain populations at cold temperatures or
high aridity (e.g. Root, 1988a,b; Weiner, 1992, and references
therein). These energetic constraints are thought to act as a`filter
barrier' to upward or downward elevational range expansion,
and result in the hard limits of bird distributions (e.g. Heaney,
2001). Cold temperatures clearly limit montane bird distributions,
as fewer than 20 species were found at the highest elevations
(> 75% mountain height). Temperature and water availability
could also influence bird species richness indirectly through food
resources and availability (e.g. Terborgh & Weske, 1975;
Terborgh, 1977; Root, 1988b; Loiselle & Blake, 1991). Insect
abundance, fruit and nectar production, and abundances of
many small vertebrates are all predicted to be low in cold
temperature regimes (e.g. Janzen, 1973; Janzen et al., 1976;
Fauth et al., 1989; Loiselle & Blake, 1991; Navas, 2003), and low
or patchy in arid conditions (e.g. Noy-Meir, 1973; Janzen,
1973; MacMahon, 1997; Perrin & Boyer, 2000). The optimal
conditions for bird species richness ­ warm and wet ­ may be the
most optimal conditions for production of their food resources.
More studies addressing food resource distributions, bird
resource use and physiological adaptations of bird species along
elevational gradients are needed to tease apart the physiological
and resource contributions to elevational diversity and distributional
patterns.
C. M. McCain
 2009 The Author
12 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
Evolutionary history
The four distinct patterns in bird elevational diversity negated
the possibility of a single evolutionary optimal from static
models or ubiquitously divergent patterns in montane diversity
from local historical contingencies. The latter was not due to
high species overlap among mountains. Very few species were
shared among mountains in the various biogeographical regions
(0­7.7% species overlap), and predictably the amount of overlap
decreases with distance (Table S2). Most evolutionary hypotheses
on mountains are static, meaning that the factors thought to
enhance speciation and immigration, and decrease extinction,
are proposed to be the same for all mountain systems (e.g.
Brown, 2001; Heaney, 2001; Lomolino, 2001). For example,
Lomolino (2001) proposed that diversity should be higher at the
base of mountains where immigration is higher. Diversity should
then decrease with elevation as immigration rate declines with
increasing isolation of high-elevation communities. This
hypothesis would predict decreasing diversity on all mountains;
hence it is a static diversity prediction. This analysis for birds
shows unequivocally that static evolutionary models will not be
supported for birds. Future evolutionary models, to be globally
persuasive, will need to be tied to evolutionary trends that are
different on humid and arid mountain systems. This is the case
for niche conservatism (Wiens & Donoghue, 2004) and why it is
the only evolutionary prediction that has initial support for birds.
Niche conservatism posits that most large-scale richness
patterns result from taxonomic groups diversifying when the
majority of the earth was dominated by a tropical-like climate
(Wiens & Donoghue, 2004). If most birds evolved niches in
warm, wet conditions then bird diversity should be concentrated
at warm, wet elevations on mountains. This is supported here
elevationally (see ECM) and previously latitudinally (e.g.
Hawkins et al., 2003a, 2007a). But so far, this result just
emphasizes that montane bird diversity patterns are consistent
with the basic tenet of niche conservatism. To directly test niche
conservatism in future studies, time-calibrated phylogenies are
needed to examine whether the oldest species are concentrated in
the areas of highest diversity and whether niches of those species
are more constrained to warm, wet conditions than are younger
species across all of these mountain systems (e.g. Stephens &
Wiens, 2003; Wiens et al., 2006, 2007).
In summary, both water and temperature variables are needed
to comprehensively predict the various elevational diversity
patterns seen for birds. More analyses are needed to discern
whether the mechanism underlying these relationships is ecological,
based on physiological or food resource limitations, or historical,
based on phylogenetic niche conservation or other evolutionary
trends tied to climate. Two issues remain unresolved: why do
some humid mountains show mid-elevational peaks in diversity
and what explains the existence of low-elevation plateaus in dry
climates? Most of the mid-peak outliers occurred in humid
regions with high human disturbance over long time periods, for
instance Taiwan, China and Madagascar. In fact, almost all of the
Madagascan bird gradients are decreasing or low plateau if only
sampled within intact habitats (e.g. Goodman et al., 1997;
Hawkins, 1999; Goodman et al., 2000). Similarly, studies in
Taiwan and China found strong impacts of disturbance and
human habitation at low elevations (Lan & Dunbar, 2000; Shiu
& Lee, 2003; Lee et al., 2004). But in general, the low elevations
on arid mountain gradients were relatively undisturbed anthropogenically
due to the low productivity and hence low human
populations in these areas. Mid-elevational peaks also occur on
elevational gradients in highland valleys of wet mountainous
regions (e.g. the Prattigau Valley, Switzerland). In these cases, the
typical low-elevation bird fauna of the regional elevational
gradient may be missing or reduced, thus making the occurrence
of a mid-elevational peak more likely. Finally, low-elevation
plateaus in dry climates (e.g. Tanzania; Corsica, France) have a
large portion of species that range across most of the lower and
mid-elevations. In fact, the length of the plateau is longer in dry
climates than wet climates globally (1373 m vs. 997 m; 47% vs.
29% of mountain height, respectively). Teasing apart the causes
of low plateaus in bird diversity further requires more information
of water availability curves and area profiles, which is currently
unavailable, but will be a fruitful future direction. Lastly, high
biodiversity on mountains underlines the need for current
protection plans in montane regions to encompass large intact
gradients in elevation and climate. Those montane bird faunas
most at risk from global climate change are those where
maximum diversity is at mid-elevations, including mountains in
arid environments and those already devastated by low-elevation
habitat destruction.
ACKNOWLEDGEMENTS
I am indebted to the researchers whose work is reanalysed herein.
Carsten Rahbek initially helped in suggesting potential papers to
include, made scientific suggestions and provided encouragement
for pursing a meta-analysis of bird elevational patterns in diversity.
This work was begun while I was a post-doctoral associate at the
National Center for Ecological Analysis and Synthesis, a Center
funded by NSF (grant no. DEB-0072909), the University of
California, and the Santa Barbara campus.
REFERENCES
Allen, A.P., Brown, J.H. & Gillooly, J.F. (2002) Global biodiversity,
biochemical kinetics, and the energetic-equivalence rule.
Science, 297, 1545­1548.
Bachman, S., Baker, W.J., Brummitt, N., Dransfield, J. & Moat, J.
(2004) Elevational gradients, area and tropical island diversity:
an example from the palms of New Guinea. Ecography, 27,
299­310.
Barry, R.G. (1992) Mountain weather and climate. Routledge,
New York.
Bhattarai, K.R., Vetaas, O.R. & Grytnes, J.A. (2004) Fern species
richness along a central Himalayan elevational gradient,
Nepal. Journal of Biogeography, 31, 389­400.
Brown, J.H. (1988) Species diversity. Analytical biogeography (ed.
by A. Myers and R.S. Giller), pp. 57­89. Chapman & Hall,
Chicago.
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 13
Brown, J.H. (2001) Mammals on mountainsides: elevational
patterns of diversity. Global Ecology and Biogeography, 10,
101­109.
Brown, J.H. & Gibson, A.C. (1983) Biogeography. Mosby, St Louis.
Brown, J.H. & Lomolino, M.V. (1998) Biogeography. Sinauer
Assoc., Inc., Sunderland, MA.
Colwell, R.K. & Coddington, J.A. (1994) Estimating terrestrial
biodiversity through extrapolation. Philosophical Transactions
of the Royal Society Series B: Biological Sciences, 345, 101­118.
Colwell, R.K. & Hurtt, G.C. (1994) Nonbiological gradients in
species richness and a spurious Rapoport effect. The American
Naturalist, 144, 570­595.
Colwell, R.K., Rahbek, C. & Gotelli, N.J. (2004) The mid-domain
effect and species richness patterns: what we have learned so
far? The American Naturalist, 163, E1­E23.
Colwell, R.K., Rahbek, C. & Gotelli, N.J. (2005) The mid-domain
effect: there's a baby in the bathwater. The American Naturalist,
166, E149­E154.
Conner, E.F. & McCoy, E.D. (1979) The statistics and biology of
the species­area relationship. The American Naturalist, 113,
791­833.
Diniz-Filho, J.A.F., Bini, L.M. & Hawkins, B.A. (2003) Spatial
autocorrelation and red herrings in geographical ecology.
Global Ecology and Biogeography, 12, 53­64.
Dunn, R.R., McCain, C.M. & Sanders, N.J. (2007) When does
diversity fit null model predictions? Scale and range size
mediate the mid-domain effect. Global Ecology and Biogeography,
16, 305­312.
Evans, K.L., Warren, P.H. & Gaston, K.J. (2005) Species-energy
relationships at the macroecological scale: a review of the
mechanisms. Biological Reviews, 80, 1­25.
Fauth, J.E., Crother, B.I. & Slowinski, J.B. (1989) Elevational
patterns of species richness, evenness, and abundance of the
Costa Rican leaf-litter herpetofauna. Biotropica, 21, 178­185.
Freckleton, R.P. (2002) On the misuse of residuals in ecology:
regression of residuals vs. multiple regression. Journal of
Animal Ecology, 71, 542­545.
Fu, C., Wu, J., Wang, X., Lei, G. & Chen, J. (2004) Patterns of
diversity, altitudinal range and body size among freshwater
fishes in the Yangtze River basin, China. Global Ecology and
Biogeography, 13, 543­552.
Gaston, K.J. (2000) Global patterns in biodiversity. Nature, 405,
220­227.
Goodman, S.M., Pidgeon, M., Hawkins, A.F.A. & Schulenberg,
T.S. (1997) Birds of southeastern Madagascar. Fieldiana:
Zoology (New Series), 87, 1­125.
Goodman, S.M., Hawkins, A.F.A. & Razafimahaimodison, J.-C.
(2000) Birds of the Parc National de Marojejy, Madagascar:
with reference to elevational distribution. Fieldiana: Zoology
(New Series), 97, 175­200.
Grimmett, R., Inskipp, C. & Inskipp, T. (2000) Birds of Nepal.
Princeton University Press, Princeton.
Grinnell, J. (1917) The niche-relationships of the California
thrasher. The Auk, 34, 427­433.
Grinnell, J. & Storer, T.I. (1924) Animal life in the Yosemite.
University of California Press, Berkeley.
Grytnes, J.A. & Vetaas, O.R. (2002) Species richness and altitude:
a comparison between null models and interpolated plant
species richness along the Himalayan altitudinal gradient,
Nepal. The American Naturalist, 159, 294­304.
Hawkins, A.F.A. (1999) Altitudinal and latitudinal distribution
of eastMalagasyforestbirdcommunities. Journal of Biogeography,
26, 447­458.
Hawkins, B.A., Porter, E.E. & Diniz-Filho, J.A.F. (2003a) Productivity
and history as predictors of the latitudinal diversity
gradient of terrestrial birds. Ecology, 84, 1608­1623.
Hawkins, B.A., Field, R., Cornell, H.V., Currie, D.J., Guégan, J.-F.,
Kaufman, D.M., Kerr, J.T., Mittelbach, G.G., Oberdorff, T.,
O'Brien, E.M., Porter, E.E. & Turner, J.R.G. (2003b) Energy,
water, and broad-scale geographic patterns of species richness.
Ecology, 84, 3105­3117.
Hawkins, B.A., Diniz-Filho, J.A.F. & Weis, A.E. (2005) The
mid-domain effect and diversity gradients: is there anything to
learn? The American Naturalist, 166, E140­E143.
Hawkins, B.A., Diniz-Filho, J.A.F., Jaramillo, C.A. & Soeller, S.A.
(2007a) Climate, niche conservatism, and the global bird
diversity gradient. The American Naturalist, 170, S16­S27.
Hawkins, B.A., Diniz-Filho, J.A.F., Bini, L.M., De Marco, P. &
Blackburn, T.M. (2007b) Red herrings revisited: spatial
autocorrelation and parameter estimation in geographical
ecology. Ecography, 30, 375­384.
Heaney, L.R. (2001) Small mammal diversity along elevational
gradients in the Philippines: an assessment of patterns and
hypotheses. Global Ecology and Biogeography, 10, 15­39.
Hillebrand, H. (2004) On the generality of the latitudinal
diversity gradient. The American Naturalist, 163, 192­211.
Hilty, S.L. (2003) Birds of Venezuela, 2nd edn. Princeton University
Press, Princeton.
Inskipp, C., Inskipp, T. & Grimmett, R. (1999) Birds of Bhutan.
Christopher Helm Ltd, London.
Janzen, D.H. (1973) Sweep samples of tropical foliage insects:
effects of seasons, vegetation types, elevation, time of day, and
insularity. Ecology, 54, 687­708.
Janzen, D.H., Ataroff, M., Farinas, M., Reyes, S., Rincon, N.,
Soler,A.,Soriano,P. & Vera,M. (1976) Changes in the arthropod
community along an elevational transect in the Venezuelan
Andes. Biotropica, 8, 193­203.
Jones, J.I., Li, W. & Maberly, S.C. (2003) Area, altitude and
aquatic plant diversity. Ecography, 26, 411­420.
Kaspari, M., O'Donnell, S. & Kercher, J.R. (2000) Energy, density,
and constraints to species richness: ant assemblages along
a productivity gradient. The American Naturalist, 155, 280­
293.
Kluge, J., Kessler, M. & Dunn, R.R. (2006) What drives elevational
patterns in diversity? A test of geometric constraints, climate
and species pool effects for pteridophytes on an elevational
gradient in Costa Rica. Global Ecology and Biogeography, 15,
358­371.
Krömer, T., Kessler, M., Gradstein, S.R. & Acebey, A. (2005)
Diversity patterns of vascular epiphytes along an elevational
gradient in the Andes. Journal of Biogeography, 32, 1799­1809.
Lan, D. & Dunbar, R. (2000) Bird and mammal conservation in
C. M. McCain
 2009 The Author
14 Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd
Gaoligongshan Region and Jingdong County, Yunnan, China:
patterns of species richness and nature reserves. Oryx, 34,
275­286.
Lee, P.-F., Ding, T.-S., Hsu, F.-H. & Geng, S. (2004) Breeding bird
species richness in Taiwan: distribution on gradients of
elevation, primary productivity and urbanization. Journal of
Biogeography, 31, 307­314.
Lees, D.C., Kremen, C. & Andriamampianina, L. (1999) A null
model for species richness gradients: bounded range overlap of
butterflies and other rainforest endemics in Madagascar.
Biological Journal of the Linnean Society, 67, 529­584.
Loiselle, B.A. & Blake, J.G. (1991) Temporal variation in birds
and fruits along an elevational gradient in Costa Rica. Ecology,
72, 180­193.
Lomolino, M.V. (2001) Elevation gradients of species-density:
Historical and prospective views. Global Ecology and Biogeography,
10, 3­13.
MacArthur, R.H. (1972) Geographical ecology. Harper & Rowe,
New York.
MacArthur, R.H. & MacArthur, J.W. (1961) On bird species
diversity. Ecology, 42, 594­598.
McCain, C.M. (2004) The mid-domain effect applied to
elevational gradients: species richness of small mammals in
Costa Rica. Journal of Biogeography, 31, 19­31.
McCain, C.M. (2005) Elevational gradients in diversity of small
mammals. Ecology, 86, 366­372.
McCain, C.M. (2007a) Could temperature and water availability
drive elevational species richness patterns? A global case study
for bats. Global Ecology and Biogeography, 16, 1­13.
McCain, C.M. (2007b) Area and mammalian elevational
diversity. Ecology, 88, 76­86.
MacMahon, J.A. (1997) Deserts. Chanticleer Press, New York.
Merriam, C.H. & Stejneger, L. (1890) Results of a biological
survey of the San Francisco Mountain region and desert
of the Little Colorado, Arizona. North American Fauna, 3, 1­
136.
Mittelbach, G.G., Steiner, C.F., Scheiner, S.M., Gross, K.L.,
Reynolds, H.L., Waide, R.B., Willig, M.R., Dodson, S.I. &
Gough, L. (2001) What is the observed relationship between
species richness and productivity? Ecology, 82, 2381­2396.
Mittelbach, G.G., Schemske, D.W., Cornell, H.V., Allen, A.P.,
Brown, J.M., Bush, M.B., Harrison, S.P., Hurlbert, A.H.,
Knowlton, N., Lessios, H.A., McCain, C.M., McCune, A.R.,
McDade, L.A., McPeek, M.A., Near, T.J., Price, T.D., Ricklefs,
R.E., Roy, K., Sax, D.F., Schluter, D., Sobel, J.M. & Turelli, M.
(2007) Evolution and the latitudinal diversity gradient: speciation,
extinction, and biogeography. Ecology Letters, 10, 315­
331.
Navas, C.A. (2003) Herpetological diversity along Andean
elevational gradients: links with physiological ecology and
evolutionary physiology. Comparative Biochemistry and
Physiology Part A, 133, 469­485.
Noy-Meir, I. (1973) Desert ecosystems: environment and
producers. Annual Review of Ecology and Systematics, 4, 25­51.
O'Brien, E.M. (1993) Climatic gradients in woody plant species
richness: towards an explanation based on an analysis of
southern Africa's woody flora. Journal of Biogeography, 20,
181­198.
Odland, A. & Birks, H.J.B. (1999) The altitudinal gradient of
vascular plant species richness in Aurland, western Norway.
Ecography, 22, 548­566.
Oommen, M.A. & Shanker, K. (2005) Elevational species
richness patterns emerge from multiple local mechanisms in
Himalayan woody plants. Ecology, 86, 3039­3047.
Orme, C.D.L., Davies, R.G., Burgess, M., Eigenbrod, F., Pickup,
N., Olson, V.A., Webster, A.J. et al. (2005) Global hotspots of
species richness are not congruent with endemism or threat.
Nature, 436, 1016­1019.
Perrin, M.R. & Boyer, D.C. (2000) Seasonal changes in the
population dynamics of hairy-footed gerbils in the Namib
Desert. South African Journal of Wildlife Research, 30, 73­84.
Pianka, E.R. (1966) Latitudinal gradients in species diversity: a
review of concepts. The American Naturalist, 100, 33­46.
Preston, F.W. (1962) The canonical distribution of commonness
and rarity. Parts I, II. Ecology, 43, 185­215, 410­432.
Rahbek, C. (1995) The elevational gradient of species richness: a
uniform pattern? Ecography, 18, 200­205.
Rahbek, C. (1997) The relationship among area, elevation, and
regional species richness in Neotropical birds. The American
Naturalist, 149, 875­902.
Rahbek, C. (2005) The role of spatial scale and the perception of
large-scale species-richness patterns. Ecology Letters, 8, 224­
239.
Rahbek, C., Gotelli, N.J., Colwell, R.K., Entsminger, G.L., Rangel,
T.F.L.B.V. & Graves, G.R. (2007) Predicting continental-scale
patterns of bird species richness with spatially explicit models.
Proceedings of the Royal Society Series B: Biological Sciences,
274, 165­174.
Rickart, E.A. (2001) Elevational diversity gradients, biogeography
and the structure of montane mammal communities in the
intermountain region of North America. Global Ecology and
Biogeography, 10, 77­100.
Robson, C. (2002) Birds of Thailand. Princeton University Press,
Princeton.
Rohde, K. (1992) Latitudinal gradients in species diversity: the
search for the primary cause. Oikos, 65, 514­527.
Root, T. (1988a) Energy constraints on avian distributions and
abundances. Ecology, 69, 330­339.
Root, T. (1988b) Environmental factors associated with avian
distributional boundaries. Journal of Biogeography, 15, 489­
505.
Rosenzweig, M.L. (1992) Species diversity gradients: we know
more and less than we thought. Journal of Mammalogy, 73,
715­730.
Rosenzweig, M.L. (1995) Species diversity in space and time.
Cambridge University Press, Cambridge.
Ruggiero, A. & Hawkins, B.A. (2008) Why do mountains support
so many species of birds? Ecography, 31, 306­315.
Sanders, N.J. (2002) Elevational gradients in ant species richness:
area, geometry, and Rapoport's rule. Ecography, 25, 25­32.
Sanders, N.J., Lessard, J.-P., Fitzpatrick, M.C. & Dunn, R.R.
(2007) Temperature, but not productivity or geometry, predicts
Global analysis of bird elevational diversity
 2009 The Author
Global Ecology and Biogeography, Journal compilation  2009 Blackwell Publishing Ltd 15
elevational diversity gradients in ants across spatial grains.
Global Ecology and Biogeography, 16, 640­649.
Sanders, N.J., Moss, J. & Wagner, D. (2003) Patterns of ant
species richness along elevational gradients in an arid ecosystem.
Global Ecology and Biogeography, 12, 93­102.
Shiu, H.-J. & Lee, P.-F. (2003) Seasonal variation in bird species
richness along elevational gradients in Taiwan. Acta Zoologica
Taiwanica, 14, 1­21.
Stephens, P.R. & Wiens, J.J. (2003) Explaining species richness
from continents to communities: the time-for-speciation effect
in emydid turtles. The American Naturalist, 161, 112­128.
Stevens, G.C. (1992) The elevational gradient in altitudinal
range: an extension of Rapoport's latitudinal rule to altitude.
The American Naturalist, 140, 893­911.
Terborgh, J. (1973) On the notion of favorableness in plant
ecology. The American Naturalist, 107, 481­501.
Terborgh, J. (1977) Bird species diversity on an Andean elevational
gradient. Ecology, 58, 1007­1019.
Terborgh, J. (1985) The role of ecotones in the distribution of
Andean birds. Ecology, 66, 1237­1246.
Terborgh, J. & Weske, J.S. (1975) The role of competition in the
distribution of Andean birds. Ecology, 56, 562­576.
UNEP (United Nations Environment Programme) (1997) World
atlas of desertification, 2nd edn. UNEP, London.
Vetaas, O.R. & Grytnes, J.A. (2002) Distribution of vascular plant
species richness and endemic richness along the Himalayan
elevation gradient in Nepal. Global Ecology and Biogeography,
11, 291­301.
Weiner, J. (1992) Physiological limits to sustainable energy
budgets in birds and mammals: ecological implications. Trends
in Ecology and Evolution, 7, 384­388.
Whittaker, R.H. (1952) A study of summer foliage insect
communities in the Great Smoky Mountains. Ecological
Monographs, 22, 1­44.
Whittaker, R.H. (1960) Vegetation of the Siskiyou Mountains,
Oregon and California. Ecological Monographs, 30, 279­338.
Whittaker, R.H. (1967) Gradient analysis of vegetation.
Biological Reviews, 42, 207­264.
Wiens, J.J. & Donoghue, M.J. (2004) Historical biogeography,
ecology and species richness. Trends in Ecology and Evolution,
19, 639­644.
Wiens, J.J., Graham, C.H., Moen, D.S., Smith, S.A. & Reeder, T.W.
(2006) Evolutionary and ecological causes of the latitudinal
diversity gradient in hylid frogs: treefrog trees unearth the
roots of high tropical diversity. The American Naturalist, 168,
579­596.
Wiens, J.J., Parra-Olea, G., García-París, M. & Wake, D.B. (2007)
Phylogenetic history underlies elevational biodiversity
patterns in tropical salamanders. Proceedings of the Royal
Society Series B: Biological Sciences, 274, 919­928.
Willig, M.R., Kaufman, D.M. & Stevens, R.D. (2003) Latitudinal
gradients of biodiversity: pattern, process, scale, and synthesis.
Annual Review of Ecology and Systematics, 34, 273­309.
Zapata, F.A., Gaston, K.J. & Chown, S.L. (2005) The mid-domain
effect revisited. The American Naturalist, 166, E144­E148.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article:
Figure S1 Colour version of Fig. 3.Global map of the distribution
of study sites and four elevational richness patterns of birds.
Table S1 Area analyses of bird elevational gradients.
Table S2 Number and percentage of bird species shared among
mountains in different biogeographical regions.
Appendix S1 List, characteristics and citations of robust and
informative elevational gradients used in analyses of bird species
richness.
Appendix S2 Literature cited for gradients not used in elevational
bird analyses.
Please note: Wiley-Blackwell is not responsible for the content or
functionality of any supporting materials supplied by the authors.
Any queries (other than missing material) should be directed to
the corresponding author for the article.
Editor: Mikko Mönkkönen
BIOSKETCH
Christy McCain is an Assistant Professor in the
Department of Ecology and Evolutionary Biology and
Curator of Vertebrates in the Museum of Natural History
of the University of Colorado at Boulder.She did her postgraduate
work in mammalian ecology at the University of
Kansas and post-doctoral work on ecology at the National
Center for Ecological Analysis and Synthesis. Christy is an
ecologist focusing on large-scale ecological patterns with
particular emphasis on species richness, abundance and
distribution patterns along ecological gradients. She is
also interested in null models, ecological modelling of
species ranges, montane biogeography, climate change
and theoretical and empirical aspects of the causes and
maintenance of species diversity.