| |

Received: 13 September 2016    Revised: 10 November 2017    Accepted: 6 December 2017

DOI: 10.1002/ece3.3813

ORIGINAL RESEARCH

Land use change has stronger effects on functional diversity

than taxonomic diversity in tropical Andean hummingbirds

Boris A. Tinoco1,2  | Vinicio E. Santillán2 | Catherine H. Graham1,3

1
Department of Ecology and Evolution, Stony
Brook University, Stony Brook, NY, USA Abstract
2
Escuela de Biología, Ecología y Land use change modifies the environment at multiple spatial scales, and is a main
Gestión, Universidad del Azuay, Cuenca,
driver of species declines and deterioration of ecosystem services. However, most of
Ecuador
3 the research on the effects of land use change has focused on taxonomic diversity,
Swiss Federal Research Institute WSL,
Birmensdorf, Switzerland while functional diversity, an important predictor of ecosystem services, is often ne-
glected. We explored how local and landscape scale characteristics influence func-
Correspondence
Boris A. Tinoco, Escuela de Biología, Ecología tional and taxonomic diversity of hummingbirds in the Andes Mountains in southern
y Gestión, Universidad del Azuay, Cuenca,
Ecuador. Data was collected in six landscapes along a land use gradient, from an al-
Ecuador.
Email: [email protected] most intact landscape to one dominated by cattle pastures. We used point counts to
sample hummingbirds from 2011 to 2012 to assessed how local factors (i.e., vegeta-
Funding information
Secretaría de Educación Superior, Ciencia, tion structure, flowering plants richness, nectar availability) and landscape factors (i.e.,
Tecnología e Innovación; National Aeronautics
landscape heterogeneity, native vegetation cover) influenced taxonomic and func-
and Space Administration, Grant/Award
Number: NNX11AO28G; Parque Nacional tional diversity. Then, we analyzed environment – trait relationships (RLQ test) to ex-
Cajas; Decanato de Investigaciones—
plore how different hummingbird functional traits influenced species responses to
Universidad del Azuay
these factors. Taxonomic and functional diversity of hummingbirds were positively
associated with landscape heterogeneity but only functional diversity was positively
related to native vegetation coverage. We found a weak response of taxonomic and
functional diversity to land use change at the local scale. Environment-trait associa-
tions showed that body mass of hummingbirds likely influenced species sensitivity to
land use change. In conclusion, landscape heterogeneity created by land use change
can positively influence hummingbird taxonomic and functional diversity; however, a
reduction of native vegetation cover could decrease functional diversity. Given that
functional diversity can mediate ecosystem services, the conservation of native veg-
etation cover could play a key role in the maintenance of hummingbird pollination
services in the tropical Andes. Moreover, there are particular functional traits, such as
body mass, that increase a species sensitivity to land use change.

KEYWORDS
deforestation, disturbance, Ecuador, functional traits, montane forest, pollination services

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2018 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.

Ecology and Evolution. 2018;1–13. www.ecolevol.org |  1

  
 |
2       TINOCO et al.

1 |  INTRODUCTION 1981), any effect of land use change on the functional diversity of
this group can have consequences for the ecological functions they
Land use change is one of the most important drivers of species loss perform. While hummingbird taxonomic diversity is often consid-
and degradation of ecosystem services (Cardinale et al., 2012; Sala ered relatively insensitive to land use change (Renjifo, 2001; Stouffer
et al., 2000). The effects of land use change on diversity are usu- & Bierregaard, 1995), a recent study by Hadley, Frey, Robinson, and
ally measured based on taxonomic diversity, while functional traits, Betts (2017), in the lowlands of Costa Rica, found that species rich-
which are more likely to influence ecosystem services, are often not ness, relative abundance, and species with specialized morphologies
considered (Cadotte, Carscadden, & Mirotchnick, 2011; Mouillot, can be sensitive to habitat loss. Nonetheless, whether hummingbird
Graham, Villéger, Mason, & Bellwood, 2013). However, taxonomic taxonomic and functional diversities respond equally to land use
and functional diversities are not always equally influenced by land change remains an open question.
use change (Luck, Carter, & Smallbone, 2013; McConkey & O’Farrill, We collected data in six landscapes in the Andes of Ecuador
2015; Villéger, Miranda, Hernández, & Mouillot, 2010). For example, along a land use gradient—from an almost intact landscape domi-
if assemblages contain species with similar functional roles, species nated by native vegetation to a landscape largely composed of cattle
loss will have a greater negative effect on taxonomic diversity than pastures to explore how local-­ and landscape-­scale characteristics
on functional diversity (Flynn et al., 2009). In contrast, if assemblages influence both taxonomic and functional diversities of humming-
contain species with unique functional traits, the loss of a species can birds. Given the morphological variation across hummingbirds, we
have greater consequences for functional diversity than for taxonomic expected that functional diversity will be at least, if not more in-
diversity because the loss of a species could eliminate a functional role fluenced by land use change than taxonomic diversity. Moreover,
(Flynn et al., 2009). Here, we evaluate how taxonomic and functional using extensive knowledge of functional traits in hummingbirds, we
diversities of hummingbirds are affected by ongoing land use change developed a set of predictions for how these traits may influence
in the tropical Andes. a hummingbird’s response to land use change (Table 1) and tested
Land use change modifies environmental characteristics at mul- those predictions by performing an analysis of environment–trait
tiple spatial scales (Tscharntke et al., 2012). At the landscape scale, relationships.
anthropogenic activities modify the type and distribution of land
cover in a landscape (Fahrig et al., 2011). A decrease in the coverage
2 | METHODS
of the original vegetation in the landscape can negatively affect the
persistence of habitat specialists (Betts, Forbes, & Diamond, 2007;
2.1 | Study region
Martensen, Ribeiro, Banks-­Leite, Prado, & Metzger, 2012). However,
an increase in the number of land cover types in a landscape may sus- We conducted this study in the western Andes of the Azuay prov-
tain populations of species that use the resources provided by these ince in southern Ecuador between 3,000 and 3,300 m.a.s.l. (Figure 1).
novel types (Renjifo, 2001; Tscharntke et al., 2012). At the local-­habitat This region has a mean annual precipitation ranging from 1,100 to
scale, a common result of land use change is the replacement of forest 1,800 mm, and monthly mean temperatures that range from 5 to 12°C
with more structurally simple vegetation dominated by smaller trees (Celleri, Willems, Buytaert, & Feyen, 2007). Cattle pastures dominate
or pastures (Brawn, Robinson, & Thompson, 2001). In addition, there the area with remnants of Andean native forest confined to steep
are often changes in the types and abundance of resources (Feinsinger slopes and along streams (White & Maldonado, 1991).
et al., 1988; Hagen & Kraemer, 2010; Loiselle & Blake, 1994). Recent
studies have demonstrated the value of measuring the types and di-
2.2 | Study design
versity functional traits of species in a community to evaluate ecosys-
tem functioning (Flynn et al., 2009; Mouillot et al., 2013). However, Our study design was hierarchical such that local-­scale plots are
while extensive work has documented how landscape and local fac- nested within landscapes (Figure 1). We chose six inter-­Andean lin-
tors influence taxonomic diversity (e.g., Graham & Blake, 2001; Tinoco, ear valleys of approximately 300 ha in size within our study region
Astudillo, Latta, Strubbe, & Graham, 2013), few studies have evaluated (Figure 1). In each valley, we delineated an area of 2.2 km by 1 km
how these factors affect functional diversity, especially in species-­rich (200 ha) referred to hereafter as a landscape unit (LU). The dimen-
systems such as the tropical Andes (Tscharntke et al., 2008). sion of the LUs was chosen to fit the linear shape of the valleys, with
Hummingbirds are particularly suitable for the study of the effects the restriction that they were between 3,000 and 3,300 m.a.s.l. This
of land use change on biodiversity. They are among the most species-­ elevation band covers most of the suitable habitat for forest birds in
rich and abundant groups of birds in the tropical Andes (Rahbek & the study area because at higher elevations the vegetation transitions
Graves, 2000). They also vary in morphology, habitat requirements, toward páramo grasslands, and at lower elevations, the landscape
and foraging roles (Abrahamczyk & Kessler, 2014; Brown & Kodric-­ is mostly dominated by cattle pastures. The distances between LUs
Brown, 1979; Feinsinger & Colwell, 1978), resulting in high levels of ranged from 2.5 to 34 km, and all valleys were partially isolated by
local functional diversity (Graham, Parra, Tinoco, Stiles, & McGuire, steep slopes and páramo grasslands, which likely restricts most hum-
2012; Maglianesi, Blüthgen, Böhning-­Gaese, & Schleuning, 2015). As mingbird movement to each valley. Limited movement was evidenced
hummingbirds play a key role in the ecosystem as pollinators (Stiles, by the fact that there have only been three captures of hummingbirds
TINOCO et al. |
      3

between two adjacent valleys, Mazan and Llaviuco, during 8 years of all hummingbird individuals visually or acoustically detected in a fixed
systematic mist-­netting (Tinoco, unpublished data). Finally, the land radius of 30 m. Hummingbirds flying over the plot were excluded. All
use gradient in LUs varied from an almost intact valley, dominated by the point counts within a LU were sampled on the same day between
native vegetation, to a highly modified valley with a mosaic of pas- 06:00 hours and 10:00 hours, and on days without rain. We altered
tures for cattle ranching, native vegetation remnants and exotic for- the starting time of each point count among survey periods to ac-
ests plantations (Table 2). count for potential differences in bird detection among point counts
Within each LU, we positioned twelve 30-­m-­radius local-­scale due to time of day. We used the double-­observe method (software
plots (0.28 ha) at 200-­m intervals (Figure 1). We chose 200 m because DOBSERV; Nichols et al., 2000) to estimate abundance. Using this
it represents a distance greater than the daily movement of territorial method, detections from the two independent observers were com-
hummingbirds (Dearborn, 1998; Paton & Carpenter, 1984) and al- bined to obtain a joint detection probability, which was used to cor-
lowed us to obtain replicate plots within LUs. However, we are aware rect the abundance estimates of each species in each point count for
of potential nonindependence of data among local-­scale plots located each survey period.
within LUs, which we accounted for in our modeling procedure (see We obtained an accumulated annual species richness and a mean
Section 2.7). abundance per species in each plot per year (i.e., four survey periods).
This study was conducted in 2011 and 2012. We performed four To explore the completeness of our estimates of annual species rich-
surveys per LU each year between February and August. All valleys ness per plot, we compare our recorded values with the Chao 1 index,
were surveyed during a period of about 7 days. During each survey, a nonparametric estimator of total richness (Chao, 1984). We recorded
we gathered data on hummingbird and flower abundance. a high percentage of the total expected richness per plot according to
the Chao 1 index (mean per plot 79.9% ± 2.03). Moreover, as the same
method was applied across all plots, we can explore relative differ-
2.3 | Hummingbird sampling
ences in hummingbird diversity across plots and LUs.
We sampled hummingbirds using point counts in each of the 12 plots
in each LU during each of our eight survey periods for a total of 288
2.4 | Landscape unit characteristics
counts. Each point count lasted 10 min, a time period that has been
widely used to sample tropical bird communities (Blake & Loiselle, We used aerial photographs 1:5,000 of the study area (SIGTIERRAS
2001; O’Dea, Watson, & Whittaker, 2004), because it maximizes the -­ MAGAP, 2010) to quantify land cover types in each LU. The photo-
probability of registering most species, while decreasing the probabil- graphs were manually digitalized using ArcMap ver. 9.0 (ESRI, 2011)
ity of double-­counting individuals (Esquivel & Peris, 2008; Smith et al., and classified into the following land cover types: native vegetation,
1998). During each point count, two observers identified and counted cattle ranching pastures, and exotic forest. Native vegetation included

T A B L E   1   Description of functional traits and predictions of their influences on hummingbird species sensitivity to land use change

Functional trait Functional influence Prediction

Bill length Hummingbirds with long bills have a narrow diet breath Hummingbirds with long bills will be negatively
compared to hummingbirds with short bills (Maglianesi et al., affected by land use change because they are
2014; Tinoco et al., 2017) less able to respond to changes in resource
availability (Newbold et al., 2014)
Body mass Heavier birds have smaller population sizes than lighter Heavier species will be sensitive to land use
hummingbirds (Calder & Calder, 1995) change because species with a small
population size are often affected by land use
change (Hadley et al., 2017; Henle et al.,
2004)
Wing loading and width of Low wing loading and narrow wings are related to trap-­lining Species with low wing loading and narrow
wings behavior (sensu Feinsinger & Colwell, 1978) and increase the wings (i.e., trap-­liner species) will be more
efficiency with which birds fly among patches of flowers sensitive to land use change, which can result
(Stiles, 2008) in unpredictable variation in the availability of
their specialized nectar resources (Henle et al.,
2004)
Tarsus length Birds with longer tarsi tend to perch while foraging on flowers Perching while feeding is influenced by flower
(Stiles, 2008) architecture, because it requires floral
structures with landing platforms (Miller,
1985). Land use change can alter the types of
flowers available and could influence the use
of perching as a foraging option in
hummingbirds
 |
4       TINOCO et al.

both native forest and shrubs as the photographs did not permit finer cross-­checking the actual vegetation with the mapped vegetation of
classification. However, differences in vegetation structure within the each LU.
native vegetation were minor compared to the differences with the We calculated three landscape characteristics to describe land
other land cover types. In the study area, exotic forests are dominated use change for each LU: percent native vegetation cover, and two
by Eucalyptus globulus. Our classification was verified in the field by measures of landscape heterogeneity: edge density and landscape

(a) (b)

(d) (c)

F I G U R E   1   Map of the study area in the Andes of southern Ecuador where (a) is the country of Ecuador, (b) is our study region that contains
our six landscape units, (c) is an example of a landscape unit with point counts, and (d) shows the local scale where vegetation plots were
conducted

T A B L E   2   Description of landscapes where hummingbirds were sampled in the Andes of southern Ecuador

Landscape composition Landscape heterogeneity

Edge
Native vegeta- Exotic forest density (m/ Landscape Description of the type of anthropo-
Local name tion (%) Pastures (%) (%) Ha) diversity genic alterations

Mazán 89.89 7.56 0.00 235.93 0.19 Native vegetation dominated

Llaviuco 77.02 22.98 0.00 466.91 0.35 Native vegetation with pastures
Culebrillas 76.62 18.45 4.93 521.25 0.38 Native vegetation dominated mosaic
Cubilán 60.86 39.14 0.00 772.95 0.48 Native vegetation dominated mosaic
Nero 54.63 38.08 7.30 361.27 0.55 Mixed used mosaic
Aurora 34.67 65.33 0.00 601.00 0.49 Pastures dominated
TINOCO et al. |
      5

diversity. Edge density, the number of edges in a landscape divided of each plant. In cases where a complete count was not possible (e.g.,
by the total landscape area, considers the spatial pattern of land cover trees, dense shrubs), we counted the number of flowers in a portion of
types irrespective of their identity (Fahrig et al., 2011). Landscape di- the plant and estimated the total number of flowers by extrapolating
versity considers the number and proportion of each land cover type the number of flowers in the sampled portion to the total area of the
in the landscape and was estimated by the Simpson index (Fahrig et al., foliage covered with flowers.
2011). These three landscape characteristics were chosen because Mean sugar production per flower of each species in our plots was
they are important predictors of biodiversity (Fahrig et al., 2011) and obtained from recently opened flowers that were bagged for 24 hr to
describe the gradient of land use in our study area (Table 2). LUs with prevent access by hummingbirds, after which nectar was extracted
low human influence had more native vegetation, low edge density, with capillary tubes and sugar concentration measured with a hand-
and low values of landscape diversity, while those with greater human held refractometer. Flowers were depleted of nectar before bagging
influence had lower coverage of native vegetation, higher edge den- (the number of flowers sampled and nectar production per species
sity, and higher values of landscape diversity. are provided in Appendix Table S2). Sugar production was calculated
as the product of nectar volume in milliliters multiplied by sugar con-
centration (mg/ml) following the table provided by Kearns and Inouye
2.5 | Local plot habitat characteristics
(1993). Sugar production rates per plot and sampling period were ob-
From the center point of each plot located within LUs, we established tained by multiplying the mean sugar production over 24 hr per flower
four 20-­m-­long transects in each of the four cardinal directions. We with the total number of open flowers of the respective plant species.
quantified foliage height diversity, canopy cover, abundance of shrubs, Flower abundance and sugar production rates were significantly cor-
and number of trees in different diameter at breast height (DBH) related (r = .80, p < .01); therefore, we only included sugar production
classes following a protocol commonly used in bird studies (James & in further analysis because it is a direct measure of energy availabil-
Shugart, 1970). To measure foliage height diversity, we placed a 3-­m ity for pollinators (Potts, Vulliamy, Dafni, Ne’eman, & Willmer, 2003).
pole at 4-­m intervals along each transect and recorded the presence Richness of flowering species and sugar production rates were aver-
or absence of vegetation touching the pole at 0.5-­m intervals from aged across survey periods within each year to obtain annual mean
0 to 3 m. Above 3 m, we visually estimated the presence or absence estimates of resource availability per plot.
of vegetation at 1-­m intervals to the top of the canopy which gener-
ally was less than 10 m tall. We calculated a Shannon diversity index
2.6 | Taxonomic and functional diversities
to quantify foliage height diversity. We visually estimated the canopy
cover using a scale of 1 to 5 (1 = 0%–19%, 2 = 20%–39%, 3 = 40%– We calculated taxonomic and functional diversities of humming-
59%, 4 = 60%–79%, 5 = 80%–100%). Foliage height diversity and birds at each plot using the annual mean abundance of each species.
canopy cover were averaged across all readings inside a vegetation We excluded species with less than three records across the study.
plot (n = 20). The abundance of shrubs (plants with <3 cm of DBH) Taxonomic diversity was calculated by the Simpson index. We used
was obtained by counting the shrubs that contacted the extended the following hummingbird functional traits to calculate functional di-
arms of a person walking along each transect. Finally, all trees present versity: body mass (weight of a live individual), bill length (length of
inside the plot were counted and assigned to one of four different the bill from base to tip), tarsus length (length from the outer bend of
DBH categories: 3–8, 9–15, 16–23, 24–38, and >38 cm. In most cases the tibiotarsal articulation to the base of the toes), wing loading (the
(68.2%), trees of the largest DBH categories (24–38 and >38 cm) were ratio of body mass to wing area), and wing aspect ratio (the quotient
the exotic species E. globulus. of twice the square of the wing length divided by wing area). High
To obtain a composite description of the vegetation structure in wing loading represents a high body mass to wing area ratio, and a
each plot, we used principal components analysis (PCA) with all the high aspect ratio denotes narrow wings. We only used morphological
vegetation structure variables measured in each plot. PCI accounted data for males because of their greater sample size. While some spe-
for 54.2% of the variation and represented a gradient of plots dom- cies in the studied community are sexually dimorphic, the standard
inated by open vegetation, to plots with closed vegetation (See deviation of a trait value within a species including measures of both
Appendix Table S1). PCII depicted a gradient from plots with abundant sexes is much lower than the standard deviation of a trait across males
large trees (mainly Eucalyptus trees) to plots dominated by shrubs and of different species (Tinoco et al., 2017). Intraspecific variation related
accounted for 20.9% of the variation (Appendix Table S1). to sex is, therefore, unlikely to influence our results. Moreover, over
During each survey period in each plot, we sampled the richness 48% of our records in point counts were acoustic, from which it was
of flowering plants and abundance of flowers. We included plant spe- impossible to sex individuals. However, we are aware that there could
cies that were observed to be used by hummingbirds during 405 hr be behavioral differences between sexes in hummingbirds, and future
of observation (Tinoco, Graham, Aguilar, & Schleuning, 2017). Plant research should explore if this variation influences functional diversity
species were identified by local experts from the Herbarium Azuay— by studying behavior, movements, and landscape use at the individual
University of Azuay in Cuenca, Ecuador. Flowering plant richness was level (Volpe, Robinson, Frey, Hadley, & Betts, 2016).
the total number of plant species with open flowers. We measured We calculated functional diversity using the Rao quadratic diver-
the abundance of flowers by counting the number of open flowers sity index (Botta-­Dukát, 2005). The Rao index is equivalent to the
 |
6       TINOCO et al.

Simpson diversity index when species completely differ in functional of flowering plants, sugar production, edge density, and landscape di-
traits, a property that facilitates comparison between taxonomic and versity were log-­transformed to improve normality. We evaluated the
functional diversities. Given that we were interested in measuring goodness of fit of the global model by estimating the marginal R2 (the
functional diversity beyond that which is explained by body size, we variance explained by the fixed factors alone) and conditional R2 (the
used the residuals from linear regressions of total bill length and tar- variance explained by both the fixed and random factors, following
sus length against body mass as uncorrelated functional traits (Reist, Nakagawa and Schielzeth (2013)). We used multimodel inference to
1985). The Rao index was obtained using the package FD (Laliberte & compare and ranked the models by Akaike’s information criterion cor-
Legendre, 2010) in R (R Development Core Team, 2013). rected for small sample size (AICc) (Burnham & Anderson, 2002) using
the R package MuMin (Barton, 2011). We obtained model average
coefficients from the top selected models with ΔAICc <2. Statistical
2.7 | Data analysis
significance of a factor was inferred when its 95% confidence intervals
To consider the hierarchical structure of our study design (plots (CI) excluded zero values (Burnham & Anderson, 2002). Further, as the
nested within LUs), we used two approaches. First, we evaluate the pairwise spatial distances among LUs were not equal, we checked for
importance of cross-­scale correlations between local and landscape spatial autocorrelation in the residuals of the global models by assess-
scales in hummingbird diversity responses using variance partition- ing the significance of Morans’ I values using the package spdep in R
ing analysis (Whittaker, 1984). Next, we explored the environmental (Bivand, 2013). None of the models revealed spatial autocorrelation.
correlates of diversity using linear mixed models, which incorporated Variation in functional diversity of a community can be influenced
the hierarchical structure of the design in the modeling (Zuur, Ieno, by changes in species composition (i.e., beta diversity) or differences
Walker, Saveliev, & Smith, 2009). in functional traits. We used RLQ analysis to explore associations be-
An important issue in studies that have a hierarchical spatial de- tween hummingbird functional traits and environmental characteris-
sign, such as ours, is the potential cross-­scale correlations between tics, accounting for differences in species composition and abundance
scales (i.e., correlations between local and landscape scales). This issue (Doledec, Chessel, terBraak, & Champely, 1996). RLQ analysis is a
could produce biological responses that are not independent between constrained ordination that maximizes the covariance between sites
scales; therefore, we used variance partitioning analysis to test the ef- and species on the basis of environmental variables and species’ traits.
fects of cross-­scale correlations (Whittaker, 1984). Variance partition- RLQ analysis uses three matrices: environment by sites (R); species
ing analysis estimates the amount of the response variable that can abundances by sites (L); and trait values by species (Q). The analysis
be attributed independently to one scale, once the effect of the other measures the strength of correlation between environmental variables
scale is accounted for using partial multiple regressions (Cushman & (R) and traits (Q) mediated by species abundances (L). The significance
McGarigal, 2002). The relative size of independent and shared vari- of the costructure between the R and Q matrices was obtained by
2
ation measured by adjusted R of partial multiple regressions is used 999 Monte Carlo permutations where permutations on the rows of
to determine the relative importance of cross-­scale correlations the R and Q matrices were compared to the observed total inertia. A
(Cushman & McGarigal, 2002). Variance partitioning was implemented probability of less than 0.05 of the observed inertia was considered
using the R package vegan (Oksanen et al., 2015). significant costructure between R and Q matrices. See details of this
The variance partitioning of species richness, taxonomic diversity, analysis in Doledec et al. (1996).
and functional diversity revealed that the independent contribution to We performed RLQ analysis separately for the landscape and local
variance of local and landscape scales was greater than their shared scales. At the landscape scale, we used native vegetation cover, land-
contribution (Appendix Table S3). This result indicates that is possible scape diversity, and edge density in R, annual mean abundance per
to explore the associations of landscape and local factors with hum- plot in L, and our five hummingbird functional traits in Q matrix. At the
mingbird diversity without considering cross-­scale correlations, for local scale, we used PCI, PCII, sugar production and richness of flow-
which we used linear mixed models. ering plants per plot in R, and the same L and Q matrices as used in the
To examine the influence of landscape-­scale factors (landscape landscape analysis. RLQ was implemented in the package ade4 (Dray
diversity, edge density, native vegetation cover) and local-­scale fac- & Dufour, 2007) in R (R Development Core team 2013).
tors (PCI, PCII, flowering plant richness, sugar production) on species
richness, taxonomic diversity, and functional diversity of humming-
birds, we used likelihood-­based linear mixed models. We included the 3 | RESULTS
identity of the LU as random factor to accommodate the hierarchi-
cal structure of our study design because local plots within a given We recorded 15 hummingbird species during our 576 point counts
landscape may be more similar to each other than plots in different (Appendix Table S4). The most abundant species were Metallura
LUs. We also included year as random factor to account for potential tyrianthina, Eriocnemis luciani, and Coeligena iris. Mean hummingbird
temporal correlation in surveys. All local-­ and landscape-­scale factors species richness per point count among LUs across years varied from
were used as fixed factors. We constructed models with all possible 3.66 in the less disturbed LU (Mazan) to 2.54 in the most disturbed
combinations of local-­ and landscape-­level factors, but did not in- LU (Aurora) (Table 3). Species turnover among LUs, measured by
clude interactions because of our low degrees of freedom. Richness Bray–Curtis distance, varied from 0.17 to 0.50 (Table 3); the two least
TINOCO et al. |
      7

altered LUs, Mazan and Llaviuco, had the most similar species com- which had negative loadings, depicting a gradient from high to low
position, while Cubilán, an intermediately altered LU, and Aurora, the levels of human land use change (Figure 2). The positive loadings of
most altered LU, were most different from the other LUs (Table 3). body mass, wing aspect ratio, and relative tarsus length indicated
There was a large variation in the functional traits of hummingbird an association of these traits with less altered LUs (Figure 2). When
species (Appendix Table S5). Chaetocercus mulsant and Ramphomicron species were plotted in functional traits space, we found that spe-
microrhynchum had body mass and bill lengths several orders smaller cies positively associated with characteristics of less altered LUs were
than Pterophanes cyanopterus and Ensifera ensifera. Chaetocercus mul- E. ensifera, P. cyanopterus, C. iris, and A. cupripennis (Figure 2), while
sant and R. microrhynchum also had high wing loading, while species species that likely preferred characteristics of more altered LUs in-
with low wing loading included M. tyrianthina, Aglaeactis cupripennis, cluded M. tyrianthina and L. nuna.
and P. cyanopterus. Species with narrow wings, which have a higher as- In the local-­scale RLQ analysis, there was a significant association
pect ratio, were E. luciani and M. baroni, while M. tyrianthina and Lesbia between local factors and species traits (Permutation test; p = .001)
nuna had broad wings (Appendix Table S5). with 82.4% and 10.1% of the variation was accounted for by Axis I
and Axis II, respectively. We only interpreted Axis I, which was mainly
described by PCII with a negative loading, and sugar production with
3.1 | Hummingbird diversity
a positive loading (Figure 3), thus representing a gradient from plots
The global linear mixed model of hummingbirds’ species richness with large trees (mainly Eucalyptus), and less sugar availability on the
explained 18% of the variation in the data (marginal R2 = .15, con- left side of the ordination to plots characterized by the presences of
ditional R2 = .18). Hummingbirds’ species richness was significantly shrubs and greater sugar availability in the right side. Body mass, wings
positively associated with the local-­scale factor flowering plant rich- aspect ratio, and bill length were associated with plots dominated by
ness (Table 4). Around 26% of the variation in taxonomic diversity shrubs and high sugar availability (Figure 3). Plotting species within
was explained by the global linear mixed model (marginal R2 = .18, functional traits space revealed that species associated with plots with
2
conditional R  = .26), and it varied positively with richness of flower- abundant shrubs and high resource availability included P. cyanopterus,
ing plants and the landscape factor, edge density (Table 4). The global A. cupripennis, and C. iris, while species associated with plots with the
linear mixed model of functional diversity explained 23% of the vari- presence of large trees and low sugar availability included L. nuna and
ation in the data (marginal R2 = .21, conditional R2 = .23). At the local M. tyrianthina (Figure 3).
scale, functional diversity increased with richness of flowering plants
and was negatively associated with open plots (PCI) (Table 4); at the
landscape scale, edge density, landscape diversity, and coverage of 4 | DISCUSSION
native vegetation were all positively associated with functional diver-
sity (Table 4). Land use change modifies the environment at multiple spatial scales,
which could differentially affect facets of biodiversity (Frishkoff et al.,
2014; Luck et al., 2013). Here, we used a hierarchical study design
3.2 | Trait–environment relationships
to evaluate the effects of land use change at the landscape and local
RLQ analysis identified a significant relationship between landscape scales of both taxonomic and functional diversities of hummingbirds
variables and species traits (Permutation test; p = .002). We only con- in the tropical Andes. At the landscape scale, both facets of diver-
sidered Axis I because it accounted for 95.3% of the variance in the sity of hummingbirds were positively associated with an increase in
RLQ analysis. Important variables included coverage of native vegeta- landscape heterogeneity. However, unlike taxonomic diversity, func-
tion with a positive loading, and landscape diversity and edge density tional diversity was positively associated with native vegetation cover,

T A B L E   3   Mean bird species richness

Low level of land use High level of land
per point count and Bray–Curtis pairwise
change use change
dissimilarity distance across six landscapes
Mean species richness Mazan Llaviuco Culebrillas Cubilán Nero Aurora with different levels of land use change in
the south central Andes of Ecuador
Mazán (3.66 ± 1.65) — 0.17 0.26 0.49 0.29 0.27
Llaviuco (3.13 ± 1.14) 0.3 0.46 0.23 0.35
Culebrillas 0.46 0.28 0.24
(3.20 ± 1.38)
Cubilán (2.91 ± 0.92) 0.38 0.50
Aurora (2.54 ± 0.97) 0.38
Nero (3.45 ± 1.25) —

Mean values and species composition for dissimilarity calculations were obtained by averaging data
from 12 point counts and eight sampling periods per valley.
 |
8       TINOCO et al.

T A B L E   4   Effects of different local (PCI, PCII, richness of flowering plants, sugar production) and landscape (native vegetation coverage,
landscape diversity, edge density) factors on (A) species richness, (B) taxonomic diversity, and (C) functional diversity of hummingbirds across six
landscapes (LUs) in the south central Andes of Ecuador

95% CI

Factor β SE Lower Upper RIV

(A) Species richness

PCI −0.01 0.03 −0.16 0.05 0.15
PCII 0.05 0.06 −0.02 0.20 0.57
Richness of flowering plants 0.10 0.05 0.01 0.20 0.97
Sugar production 0.03 0.05 −0.02 0.17 0.44
Landscape diversity −0.10 0.13 −0.40 0.10 0.65
Edge density 0.01 0.04 −0.06 0.21 0.14
Native vegetation cover −0.04 0.12 −0.46 0.25 0.34
(B) Taxonomic diversity
PCI −0.01 0.02 −0.04 0.00 0.64
Richness of flowering plants 0.04 0.01 0.01 0.05 1
Sugar production 0.00 0.01 −0.01 0.04 0.34
Landscape diversity −0.06 0.04 −0.01 0.01 1
Edge density 0.04 0.01 0.01 0.06 1
Native vegetation cover −0.03 0.04 −0.13 0.02 0.5
(C) Functional diversity
PCI −0.16 0.04 −0.26 −0.07 1
PCII −0.01 0.03 −0.13 0.05 0.17
Richness of flowering plants 0.06 0.04 0.00 0.14 0.83
Sugar production 0.02 0.03 −0.02 0.12 0.43
Landscape diversity 0.30 0.14 0.02 0.58 1
Edge density 0.14 0.05 0.03 0.25 1
Native vegetation cover 0.48 0.15 0.18 0.78 1

Given are standardized averaged estimates (β), their unconditional standard errors (SE), 95% confidence intervals (95% CI), and the relative variable impor-
tance (RVI) in the set of best models (ΔAICc values <2). Factors with significant effects (estimates for which 95% confidence intervals do not overlap zero)
are highlighted in bold.

which indicates that trait diversity may be more affected by land use 2011; Tscharntke et al., 2012). Edge density was positively associated
change than taxonomic diversity (Flynn et al., 2009). At the local scale, with both taxonomic and functional diversities. Edge density, related
we found a weak response of taxonomic and functional diversities to to the amount of habitat boundaries in the landscape, may promote
land use change; these results suggest that landscape-­scale factors species’ movements between complementary habitats (Hughes, Daily,
could be more important than local-­scale factors in mediating how & Ehrlich, 2002), and often, edges contain plant species that provide
hummingbirds respond to anthropogenic disturbance. Moreover, we abundant nectar resources (Hagen & Kraemer, 2010; Tscharntke
found that large body mass, narrow wings, and long tarsi increased et al., 2008). Landscape diversity, positively associated with functional
hummingbirds’ sensitivity to land use change. Taken together, these diversity, can increase niche space (Fahrig et al., 2011), thus promot-
results indicate that different facets of diversity and environmental ing the accumulation of species with different traits and habitat re-
factors at multiple scales should be considered when evaluating how quirements (Haslem & Bennett, 2008; Perović et al., 2015).
anthropogenic activities influence biodiversity (Cadotte et al., 2011; In addition, functional diversity increased with native vegetation
Frishkoff et al., 2014; Monnet et al., 2014). cover, indicating that functional diversity could be more sensitive to
land use change than taxonomic diversity. Taxonomic and functional
diversities may have different responses to land use gradients because
4.1 | Hummingbird diversity
altered environments can set limits to the ranges of traits of coexisting
Our results support the common finding that landscape heterogene- species, and thus constrain the amount of functional variation possible
ity, quantified here as edge density and landscape diversity, promotes in assemblages (Cadotte et al., 2011). For example, species richness
increased animal diversity in human-­modified landscapes (Fahrig et al., often remains the same after anthropogenic disturbance because,
TINOCO et al. |
      9

F I G U R E   2   Graphical depiction of
the first axis of an RLQ analysis for (a)
functional traits and (b) environmental
variables at the landscape scale. Position
of scores relative to the origin indicates
their contribution to RLQ axis, and relative
position of scores along the axis indicates
associations between functional traits
and environmental variables. Species are
plotted within functional trait space

while some forest species may be lost, generalist species often col- increase in plant species richness corresponds to an increase in plants
onize these disturbed landscapes (Graham & Blake, 2001; Renjifo, with morphologically different corollas, potentially more humming-
2001). In our case, species with distinct functional traits, including bird species can coexist through resource partitioning (Abrahamczyk
E. ensifera and P. cyanopterus, were uncommon in more disturbed land- & Kessler, 2014; Feinsinger & Colwell, 1978). A positive correlation
scapes while species with generalized functional traits such as L. nuna between plant and pollinator species richness has also been reported
and M. tyrianthina were more common. This change in abundance pat- in plant–insect pollination systems (Fründ, Linsenmair, & Blüthgen,
terns likely explains why functional diversity, but not species richness 2010; Weiner, Werner, Linsenmair, & Blüthgen, 2011). Future work
or taxonomic diversity, declined as the amount of native vegetation should explore the relationship between functional diversity of flow-
coverage in the landscape declined. ers and functional diversity of pollinators.
Our studied landscapes presented continuous variation in the Variables of the vegetation structure at the local scale did not
amount of native vegetation cover and landscape heterogeneity; thus, appear to influence hummingbird taxonomic richness or diversity;
our results are the product of the combined effects of these two fac- however, functional diversity was higher in plots with more open
tors. A more complete understanding of the mechanism influencing vegetation. This result is consistent with the apparent importance
biodiversity responses to land use change that distinguishes the inde- of the landscape factors edge density and landscape diversity. Open
pendent effects of loss of native vegetation and landscape heteroge- areas are often of high quality because of the presence of plant spe-
neity would require replicates of landscapes with the same levels of cies with flowers that produce abundant nectar (Costa & Magnusson,
habitat loss and heterogeneity (see review by Hadley & Betts, 2016). 2003). In our study area, plant species such as Oreocalllis grandiflora
This study design was not possible in our region because there is a and Barnadesia arborea are pioneer colonizers with flowers that at-
long history of anthropogenic disturbance which limits by the number tract many species of hummingbirds (Tinoco et al., 2017). In addition,
of LUs at the same elevation that contain montane forest (White & the strong association found between all three landscape factors and
Maldonado, 1991). In addition, there can be intraspecific behavioral hummingbird functional diversity suggests that relationship between
differences between male and female hummingbirds in their responses local-­scale factors and hummingbirds might be largely dependent on
to land use change (Feinsinger & Colwell, 1978), which could be ob- the landscape context (Renjifo, 2001; Tscharntke et al., 2012). Some
scured with species-­level analyses. Studies, potentially using marked hummingbird species have daily movements of more than 1 km
individuals so sex can be determined, should aim to disentangle the (Hadley & Betts, 2009) and may exploit resources across different land
influence of land use change on the complete functional diversity of cover types; for example, a recent study by Volpe et al. (2016) found
hummingbirds in the tropical Andes. that the presence of the hummingbird Phaethornis guy in small forest
At the local scale, variation across all measures of hummingbird patches is highly dependent in the connectivity of the native forest at
diversity was associated with richness of flowering plants. Trait match- the landscape scale. Thus, the use of different land cover types in a
ing between bill length and corolla length influences hummingbird re- landscape may undermine negative effects of land use change at small
source use (Weinstein & Graham, 2017; Wolf, 1978); therefore, if an local scales for many hummingbird species.
 |
10       TINOCO et al.

F I G U R E   3   Graphical depiction of
the first axis of an RLQ analysis for (a)
functional traits and (b) environmental
variables at the local scale. Position of
scores relative to the origin indicates their
contribution to RLQ axis, and relative
position of scores along the axis indicates
associations between functional traits
and environmental variables. Species are
plotted within functional trait space

their specialized resources (Gibb et al., 2006; Henle et al., 2004).

4.2 | Trait–environment relationships
Trap-­lining hummingbirds can also be affected by land use change
As predicted (Table 1), body mass, tarsus length, and narrow wing if they avoid crossing open areas in their foraging routes (Hadley
aspect appeared to be negatively influenced by land use change et al., 2017). Nonetheless, we acknowledge that flying behavior in
at the landscape scale. In our study, P. cyanopterus, E. ensifera, and hummingbirds is defined by a complex group of parameters influenc-
E. luciani are among those species with high body mass, narrow ing flying aerodynamics (Altshuler, Stiles, & Dudley, 2004), and more
wings, and long tarsi that negatively responded to land use change. research should be carried out to directly measure the influence of
Hadley et al. (2017) found that hummingbirds having specialized wing shape and other traits in the foraging behavior of humming-
morphologies (i.e., large body size) can be sensitive to habitat loss birds. Finally, the positive association between tarsus length and
and fragmentation. Increased sensitivity to land use change in heav- more pristine LUs could be related to the types of flowers available
ier species has been found elsewhere (Gage, de Brooke, Symonds, & across the land use gradient. In hummingbirds, species with long
Wege, 2004; Newbold et al., 2014) and has been attributed to the tarsi frequently perch on flowers for feeding (Stiles, 2008), a behav-
correlation between body mass and demographic parameters, such ior dependent on the availability of flowers with landing structures
as small population size and low reproduction rates (Henle, Davies, (Miller, 1985).
Kleyer, Margules, & Settele, 2004). Species with these demographic At the local scale, the RLQ indicated that hummingbirds with nar-
characteristics are often prone to extinction via environmental and row wings, large body mass, and large bills were more common in plots
demographic stochasticity (Lande, 1993). Moreover, heavier hum- with high resource abundance. Species that have those functional
mingbirds also had longer bills (note that our measure of bill length traits included P. cyanopterus, A. cupripennis, and E. ensifera. Moreover,
accounted for body mass which may be why this trait is not strongly we found that plots with large trees and low abundance of sugar
associated with the RLQ plot), which could increase hummingbirds′ were associated with small hummingbirds with short bills. These plots
sensitivity to land use change because longer billed species have likely corresponded to areas with an important presence of the ex-
more specialized diets (Tinoco et al., 2017) and may decline because otic Eucalyptus trees, where species like L. nuna and M. tyrianthina can
their resources are often less common in disturbed environments be abundant. Other studies have reported that Eucalyptus trees can
(Newbold et al., 2014). sustain hummingbird populations (Montaldo, 1984; Renjifo, 2001),
Narrow-­winged species were also associated with less disturbed but as found here, this observation might apply only to species with
landscapes. This trait confers reduced power requirements during particular traits, such as small body size and short bill. More generally,
flight, a characteristic that would benefit trap-­lining behavior in the result that the association between some hummingbird traits and
hummingbirds (Feinsinger & Colwell, 1978). Species that forage for the environment were as predicted highlights the importance of trait-­
resources that are patchy in the environment are thought to be more based approaches for understanding the factors that influence species
sensitive to land use change because of unpredictable variation in responses to land use change.
TINOCO et al. |
      11

O RC I D
4.3 | Conservation implications
Boris A. Tinoco  http://orcid.org/0000-0003-1054-8945
Heterogeneous landscapes with significant coverage of native vegeta-
tion correspond to intermediately disturbed landscapes. While hum-
mingbird functional diversity might benefit from some level of land
REFERENCES
use change, as found here, there is likely a threshold beyond which
Abrahamczyk, S., & Kessler, M. (2014). Morphological and behavioural
heterogeneity can negatively affect diversity. This is because an in-
adaptations to feed on nectar: How feeding ecology determines the
crease in heterogeneity leads to a loss of native vegetation (Betts
diversity and composition of hummingbird assemblages. Journal of
et al., 2007; Cerezo, Conde, & Poggio, 2011), producing a negative Ornithology, 156, 333–347.
impact to species with unique functional roles (Carrara et al., 2015). Altshuler, D. L., Stiles, F. G., & Dudley, R. (2004). Of hummingbirds
Functional diversity of pollinators can increase fruit productiv- and helicopters: Hovering costs, competitive ability, and forag-
ing strategies. The American Naturalist, 163, 16–25. https://doi.
ity in plants (Hoehn, Tscharntke, Tylianakis, & Steffan-­Dewenter,
org/10.1086/380511
2008); thus, maintenance of this diversity might be key for sustain-
Barton, K. (2011). Available at: http://CRAN.R-project.org/package=MuMIn.
ing high-­quality pollination services (Rader, Bartomeus, Tylianakis, & Betts, M. G., Forbes, G. J., & Diamond, A. W. (2007). Thresholds in songbird
Laliberté, 2014). In the future, it will be important to directly test occurrence in relation to landscape structure. Conservation Biology, 21,
how changes in the functional diversity of hummingbirds can alter 1046–1058. https://doi.org/10.1111/j.1523-1739.2007.00723.x
Bivand, R. (2013). spdep: Spatial dependence: Weighting schemes, statistics
pollination services. Nonetheless, this study points out the impor-
and models. R package version 0.5-56.
tance of the maintenance of native vegetation to promote high levels Blake, J. G., & Loiselle, B. A. (2001). Bird assemblages in second-­growth and
of functional diversity of hummingbirds in heterogeneous agricul- old-­growth forests, Costa Rica: Perspectives from mist nets and point
tural landscapes. Cattle ranching is one of the main causes of the counts. The Auk, 118, 304–326. https://doi.org/10.1642/0004-8038(2
001)118[0304:BAISGA]2.0.CO;2
reduction in native vegetation coverage across the tropical Andes
Botta-Dukát, Z. (2005). Rao’s quadratic entropy as a measure of functional
(Viña & Cavelier, 1999), but several management actions could be diversity based on multiple traits. Journal of Vegetation Science, 16,
implemented to both retain the commercial value of cattle ranching 533–540. https://doi.org/10.1111/j.1654-1103.2005.tb02393.x
activities and promote hummingbird functional diversity. The pro- Brawn, J. D., Robinson, S. K., &Thompson, III, F. R. (2001). The role of dis-
turbance in the ecology and conservation of birds. Annual Review of
tection of native vegetation remnants, the incorporation of native
Ecology and Systematics, 32, 251–276. https://doi.org/10.1146/an-
vegetation hedgerows, the maintenance of native vegetation along nurev.ecolsys.32.081501.114031
rivers, and the use of native flowering plants on fences are all im- Brown, J. H., & Kodric-Brown, A. (1979). Convergence, competition, and
portant practices that could contribute to the conservation of hum- mimicry in a temperate community of hummingbird-­pollinated flowers.
Ecology, 60, 1022–1035. https://doi.org/10.2307/1936870
mingbird taxonomic and functional biodiversity and the pollination
Burnham, K. P., & Anderson, D. R. (2002). Model selection and multi-model
services they provide. inference: A practical information-theoretic approach. New York, NY:
Springer.
Cadotte, M. W., Carscadden, K., & Mirotchnick, N. (2011). Beyond spe-
ACKNOWLE DG ME NTS cies: Functional diversity and the maintenance of ecological processes
and services. Journal of Applied Ecology, 48, 1079–1087. https://doi.
We thank J. M. Aguilar, A. Nieto, and C. Silva for assistance with
org/10.1111/j.1365-2664.2011.02048.x
fieldwork, E. Zarate from Universidad del Azuay and F. Sanchez Calder, W. A., & Calder, L. L. (1995). Size and abundance: Breeding popu-
from ETAPA for logistic arrangements, G. Stiles for sharing mor- lation density of the Calliope Hummingbird. The Auk, 112, 517–521.
phological data of hummingbirds, and S. Padron for aesthetics https://doi.org/10.2307/4088746
Cardinale, B. J., Duffy, J. E., Gonzalez, A., Hooper, D. U., Perrings, C., Venail,
of graphs. Funding was provided by NASA grant NNX11AO28G
P., … Naeem, S. (2012). Biodiversity loss and its impact on humanity.
and Parque Nacional Cajas. B.A.T. was supported by a graduate Nature, 486, 59–67. https://doi.org/10.1038/nature11148
fellowship from SENESCYT and Decanato de Investigaciones— Carrara, E., Arroyo-Rodríguez, V., Vega-Rivera, J. H., Schondube, J. E., de
Universidad del Azuay. Idea Wild supported this research with field Freitas, S. M., & Fahrig, L. (2015). Impact of landscape composition
and configuration on forest specialist and generalist bird species in the
equipment. We also thank D. Padilla and S. Latta for comments on
fragmented Lacandona rainforest, Mexico. Biological Conservation, 184,
this manuscript. 117–126. https://doi.org/10.1016/j.biocon.2015.01.014
Celleri, R., Willems, P., Buytaert, W., & Feyen, J. (2007). Space–time rainfall
variability in the Paute basin, Ecuadorian Andes. Hydrological Processes,
CO NFLI CT OF I NTE RE S T 21, 3316–3327. https://doi.org/10.1002/(ISSN)1099-1085
Cerezo, A., Conde, M. C., & Poggio, S. L. (2011). Pasture area and landscape
None declared. heterogeneity are key determinants of bird diversity in intensively man-
aged farmland. Biodiversity and Conservation, 20, 2649–2667. https://
doi.org/10.1007/s10531-011-0096-y
AUT HOR CONTRI B UTI O N S Chao, A. (1984). Nonparametric estimation of the number of classes in a
population. Scandinavian Journal of Statistics, 11, 265–270.
B.A.T. and C.H.G. designed the study. B.A.T. collected field data. B.A.T.
Costa, F. R. C., & Magnusson, W. E. (2003). Effects of selective logging on
and V.E.S. performed data analysis. B.A.T. and C.H.G. wrote the manu- the diversity and abundance of flowering and fruiting understory plants
script. All authors critically read and commented on the manuscript. in a central Amazonian forest. Biotropica, 35, 103–114.
 |
12       TINOCO et al.

Cushman, S. A., & McGarigal, K. (2002). Hierarchical, multi-­scale decom- in tropical hummingbird communities. Biotropica. https://doi.
position of species-­environment relationships. Landscape Ecology, 17, org/10.1111/btp.12487
637–646. https://doi.org/10.1023/A:1021571603605 Hagen, M., & Kraemer, M. (2010). Agricultural surroundings support flower–
Dearborn, D. D. (1998). Interspecific territoriality by a Rufous-­ visitor networks in an Afrotropical rain forest. Biological Conservation,
Tailed Hummingbird (Amazilia tzacatl): Effects of intruder 143, 1654–1663. https://doi.org/10.1016/j.biocon.2010.03.036
size and resource value. Biotropica, 30, 306–313. https://doi. Haslem, A., & Bennett, A. F. (2008). Birds in agricultural mosaics: The in-
org/10.1111/j.1744-7429.1998.tb00064.x fluence of landscape pattern and countryside heterogeneity. Ecological
Doledec, S., Chessel, D., terBraak, C. J. F., & Champely, S. (1996). Matching Applications, 18, 185–196. https://doi.org/10.1890/07-0692.1
species traits to environmental variables: A new three-­table ordination Henle, K., Davies, K. F., Kleyer, M., Margules, C., & Settele, J. (2004). Predictors
method. Environmental and Ecological Statistics, 3, 143–166. https://doi. of species sensitivity to fragmentation. Biodiversity & Conservation, 13,
org/10.1007/BF02427859 207–251. https://doi.org/10.1023/B:BIOC.0000004319.91643.9e
Dray, S., & Dufour, A.-B. (2007). The ade4 package: Implementing the dual- Hoehn, P., Tscharntke, T., Tylianakis, J. M., & Steffan-Dewenter, I. (2008).
ity diagram for ecologists. Journal of Statistical Software, 22, 1–20. Functional group diversity of bee pollinators increases crop yield.
Esquivel, A. M., & Peris, S. (2008). Influence of time of day, duration and Proceedings of the Royal Society of London B: Biological Sciences, 275,
number of counts in point count sampling of birds in an Atlantic forest 2283–2291. https://doi.org/10.1098/rspb.2008.0405
of Paraguay. Ornitologia Neotropical, 19, 229–242. Hughes, J. B., Daily, G. C., & Ehrlich, P. R. (2002). Conservation of tropical
ESRI. (2011). ARCGIS: 9.0. Redlands, California: Environmental Systems forest birds in countryside habitats. Ecology Letters, 5, 121–129. https://
Research Institute. doi.org/10.1046/j.1461-0248.2002.00294.x
Fahrig, L., Baudry, J., Brotons, L., Burel, F. G., Crist, T. O., Fuller, R. J., … James, F. C., & Shugart, H. H. Jr. (1970). A quantitative method of habitat
Martin, J.-L. (2011). Functional landscape heterogeneity and animal description. Audubon Field Notes, 24, 727–736.
biodiversity in agricultural landscapes. Ecology Letters, 14, 101–112. Kearns, C. A., & Inouye, D. W. (1993). Techniques for pollination biologists.
https://doi.org/10.1111/j.1461-0248.2010.01559.x Niwot, Colorado: University Press Colorado.
Feinsinger, P., Busby, W. H., Murray, K. G., Beach, J. H., Pounds, W. Z., & Laliberte, E., & Legendre, P. (2010). A distance-­based framework for mea-
Linhart, Y. B. (1988). Mixed support for spatial heterogeneity in spe- suring functional diversity from multiple traits. Ecology, 91, 299–305.
cies interactions: Hummingbirds in a tropical disturbance mosaic. The https://doi.org/10.1890/08-2244.1
American Naturalist, 131, 33–57. https://doi.org/10.1086/284772 Lande, R. (1993). Risks of population extinction from demographic and
Feinsinger, P., & Colwell, R. (1978). Community organization among neo- environmental stochasticity and random catastrophes. The American
tropical nectar-­feeding birds. American Zoologist, 18, 779–795. https:// Naturalist, 142, 911–927. https://doi.org/10.1086/285580
doi.org/10.1093/icb/18.4.779 Loiselle, B., & Blake, J. (1994). Annual variation in birds and plants of a
Flynn, D. F. B., Gogol-Prokurat, M., Nogeire, T., Molinari, N., Richers, B. tropical 2nd-­growth woodland. Condor, 96, 368–380. https://doi.
T., Lin, B. B., … DeClerck, F. (2009). Loss of functional diversity under org/10.2307/1369321
land use intensification across multiple taxa. Ecology Letters, 12, 22–33. Luck, G. W., Carter, A., & Smallbone, L. (2013). Changes in bird functional
https://doi.org/10.1111/j.1461-0248.2008.01255.x diversity across multiple land uses: Interpretations of functional re-
Frishkoff, L. O., Karp, D. S., M’Gonigle, L. K., Mendenhall, C. D., Zook, J., dundancy depend on functional group identity. PLoS ONE, 8, e63671.
Kremen, C., … Daily, G. C. (2014). Loss of avian phylogenetic diversity https://doi.org/10.1371/journal.pone.0063671
in neotropical agricultural systems. Science, 345, 1343–1346. https:// Maglianesi, M. A., Blüthgen, N., Böhning-Gaese, K., & Schleuning, M.
doi.org/10.1126/science.1254610 (2014). Morphological traits determine specialization and resource
Fründ, J., Linsenmair, K. E., & Blüthgen, N. (2010). Pollinator diversity and use in plant–hummingbird networks in the neotropics. Ecology, 95,
specialization in relation to flower diversity. Oikos, 119, 1581–1590. 3325–3334.
https://doi.org/10.1111/j.1600-0706.2010.18450.x Maglianesi, M. A., Blüthgen, N., Böhning-Gaese, K., & Schleuning, M. (2015).
Gage, G. S., de Brooke, M. L., Symonds, M. R. E., & Wege, D. (2004). Functional structure and specialization in three tropical plant–hum-
Ecological correlates of the threat of extinction in neotropical bird mingbird interaction networks across an elevational gradient in Costa
species. Animal Conservation, 7, 161–168. https://doi.org/10.1017/ Rica. Ecography, 38, 1119–1128. https://doi.org/10.1111/ecog.01538
S1367943004001246 Martensen, A. C., Ribeiro, M. C., Banks-Leite, C., Prado, P. I., & Metzger,
Gibb, H., Hjältén, J., Ball, J. P., Pettersson, R. B., Landin, J., Alvini, O., & J. P. (2012). Associations of forest cover, fragment area, and con-
Danell, K. (2006). Wing loading and habitat selection in forest beetles: nectivity with neotropical understory bird species richness and
Are red-­listed species poorer dispersers or more habitat-­specific than abundance. Conservation Biology, 26, 1100–1111. https://doi.
common congenerics? Biological Conservation, 132, 250–260. https:// org/10.1111/j.1523-1739.2012.01940.x
doi.org/10.1016/j.biocon.2006.04.017 McConkey, K. R., & O’Farrill, G. (2015). Cryptic function loss in animal
Graham, C. H., & Blake, J. G. (2001). Influence of patch and land- populations. Trends in Ecology & Evolution, 30, 182–189. https://doi.
scape level factors on bird assemblages in a fragmented tropi- org/10.1016/j.tree.2015.01.006
cal landscape. Ecological Applications, 11, 1709–1721. https://doi. Miller, R. S. (1985). Why hummingbirds hover. The Auk, 102, 722–726.
org/10.1890/1051-0761(2001)011[1709:IOPALL]2.0.CO;2 Monnet, A.-C., Jiguet, F., Meynard, C. N., Mouillot, D., Mouquet, N.,
Graham, C. H., Parra, J. L., Tinoco, B. A., Stiles, F. G., & McGuire, J. A. (2012). Thuiller, W., & Devictor, V. (2014). Asynchrony of taxonomic, functional
Untangling the influence of ecological and evolutionary factors on trait and phylogenetic diversity in birds. Global Ecology and Biogeography, 23,
variation across hummingbird assemblages. Ecology, 93, S99–S111. 780–788. https://doi.org/10.1111/geb.12179
https://doi.org/10.1890/11-0493.1 Montaldo, N. (1984). Asociación de dos especies de picaflores con árboles
Hadley, A., & Betts, M. (2009). Tropical deforestation alters humming- del genero Eucalyptus (Myrtaceae) en la provincia de Buenos Aires. El
bird movement patterns. Biology Letters, 5, 207–210. https://doi. Hornero, 12, 159–162.
org/10.1098/rsbl.2008.0691 Mouillot, D., Graham, N. A. J., Villéger, S., Mason, N. W. H., & Bellwood,
Hadley, A., & Betts, M. (2016). Refocusing habitat fragmentation research D. R. (2013). A functional approach reveals community responses to
using lessons from the last decade. Current Landscape Ecology Reports, disturbances. Trends in Ecology & Evolution, 28, 167–177. https://doi.
1, 55–66. https://doi.org/10.1007/s40823-016-0007-8 org/10.1016/j.tree.2012.10.004
Hadley, A., Frey, S. J., Robinson, W. D., & Betts, M. (2017). Forest frag- Nakagawa, S., & Schielzeth, H. (2013). A general and simple
mentation and loss reduce richness, availability, and specialization method for obtaining R2 from generalized linear mixed-­effects
TINOCO et al. |
      13

models. Methods in Ecology and Evolution, 4, 133–142. https://doi. Stouffer, P. C., & Bierregaard, R. O. (1995). Effects of forest fragmentation on
org/10.1111/j.2041-210x.2012.00261.x understory hummingbirds in Amazonian Brazil. Conservation Biology, 9,
Newbold, T., Scharlemann, J. P. W., Butchart, S. H. M., Şekercioğlu, Ç. H., 1085–1094. https://doi.org/10.1046/j.1523-1739.1995.9051072.x-i1
Joppa, L., Alkemade, R., & Purves, D. W. (2014). Functional traits, land-­ Tinoco, B. A., Astudillo, P. X., Latta, S. C., Strubbe, D., & Graham, C. H.
use change and the structure of present and future bird communities (2013). Influence of patch factors and connectivity on the avifauna of
in tropical forests. Global Ecology and Biogeography, 23, 1073–1084. fragmented Polylepis forest in the Ecuadorian Andes. Biotropica, 45,
https://doi.org/10.1111/geb.12186 602–611. https://doi.org/10.1111/btp.12047
Nichols, J. D., Hines, J. E., Sauer, J. R., Fallon, F. W., Fallon, J. E., & Heglund, P. Tinoco, B. A., Graham, C. H., Aguilar, J. M., & Schleuning, M. (2017). Effects
J. (2000). A double-­observer approach for estimating detection proba- of hummingbird morphology on specialization in pollination net-
bility and abundance from point counts. The Auk, 117, 393–408. https:// works vary with resource availability. Oikos, 126, 52–60. https://doi.
doi.org/10.1642/0004-8038(2000)117[0393:ADOAFE]2.0.CO;2 org/10.1111/oik.02998
O’Dea, N., Watson, J. E. M., & Whittaker, R. J. (2004). Rapid assess- Tscharntke, T., Sekercioglu, C. H., Dietsch, T. V., Sodhi, N. S., Hoehn, P., &
ment in conservation research: A critique of avifaunal assess- Tylianakis, J. M. (2008). Landscape constraints on functional diversity
ment techniques illustrated by Ecuadorian and Madagascan of birds and insects in tropical agroecosystems. Ecology, 89, 944–951.
case study data. Diversity & Distributions, 10, 55–63. https://doi. https://doi.org/10.1890/07-0455.1
org/10.1111/j.1472-4642.2004.00050.x Tscharntke, T., Tylianakis, J. M., Rand, T. A., Didham, R. K., Fahrig, L., Batáry,
Oksanen, J. F., Blanchet, G., Kindt, R., Legendre, P., Minchin, P. R., O’Hara, R. P., … Westphal, C. (2012). Landscape moderation of biodiversity pat-
B., … Wagner, H. (2015). vegan: Community ecology package. R pack- terns and processes—eight hypotheses. Biological Reviews, 87, 661–
age version 2.3-0. 685. https://doi.org/10.1111/j.1469-185X.2011.00216.x
Paton, D. C., & Carpenter, F. L. (1984). Peripheral foraging by territorial Villéger, S., Miranda, J. R., Hernández, D. F., & Mouillot, D. (2010).
Rufous Hummingbirds: Defense by exploitation. Ecology, 65, 1808– Contrasting changes in taxonomic vs. functional diversity of tropical
1819. https://doi.org/10.2307/1937777 fish communities after habitat degradation. Ecological Applications, 20,
Perović, D., Gámez-Virués, S., Börschig, C., Klein, A.-M., Krauss, J., 1512–1522. https://doi.org/10.1890/09-1310.1
Steckel, J., … Westphal, C. (2015). Configurational landscape het- Viña, A., & Cavelier, J. (1999). Deforestation rates (1938–1988) of trop-
erogeneity shapes functional community composition of grassland ical lowland forests on the Andean foothills of Colombia. Biotropica,
butterflies. Journal of Applied Ecology, 52, 505–513. https://doi. 31, 31–36.
org/10.1111/1365-2664.12394 Volpe, N. L., Robinson, W. D., Frey, S. J. K., Hadley, A. S., & Betts, M. G.
Potts, S. G., Vulliamy, B., Dafni, A., Ne’eman, G., & Willmer, P. (2003). (2016). Tropical forest fragmentation limits movements, but not oc-
Linking bees and flowers: How do floral communities structure pollina- currences of a generalist pollinator species. PLoS ONE, 11, e0167513.
tor communities? Ecology, 84, 2628–2642. https://doi.org/10.1890/ https://doi.org/10.1371/journal.pone.0167513
02-0136 Weiner, C. N., Werner, M., Linsenmair, K. E., & Blüthgen, N. (2011). Land
R Development Core Team. (2013). A language for environmental and sta- use intensity in grasslands: Changes in biodiversity, species compo-
tistical computing. sition and specialisation in flower visitor networks. Basic and Applied
Rader, R., Bartomeus, I., Tylianakis, J. M., & Laliberté, E. (2014). The Ecology, 12, 292–299. https://doi.org/10.1016/j.baae.2010.08.006
winners and losers of land use intensification: Pollinator commu- Weinstein, B. G., & Graham, C. H. (2017). Persistent bill and corolla matching
nity disassembly is non-­random and alters functional diversity. despite shifting temporal resources in tropical hummingbird-­plant inter-
Diversity and Distributions, 20, 908–917. https://doi.org/10.1111/ actions. Ecology Letters, 20, 326–335. https://doi.org/10.1111/ele.12730
ddi.12221 White, S., & Maldonado, F. (1991). The use and conservation of natural
Rahbek, C., & Graves, G. R. (2000). Detection of macro-­ecological pat- resources in the Andes of Southern Ecuador. Mountain Research and
terns in South American hummingbirds is affected by spatial scale. Development, 11, 37–55. https://doi.org/10.2307/3673526
Proceedings of the Royal Society of London. Series B: Biological Sciences, Whittaker, J. (1984). Model interpretation from the additive elements of
267, 2259–2265. https://doi.org/10.1098/rspb.2000.1277 the likelihood function. Journal of the Royal Statistical Society. Series C
Reist, J. D. (1985). An empirical evaluation of several univariate methods (Applied Statistics), 33, 52–64.
that adjust for size variation in morphometric data. Canadian Journal of Wolf, L. L. (1978). Aggressive social organization in nectarivorous birds.
Zoology, 63, 1429–1439. https://doi.org/10.1139/z85-213 American Zoologist, 18, 765–778. https://doi.org/10.1093/icb/18.4.765
Renjifo, L. (2001). Effect of natural and anthropogenic landscape matrices Zuur, A., Ieno, E. N., Walker, N., Saveliev, A. A., & Smith, G. M. (2009). Mixed
on the abundance of subandean bird species. Ecological Applications, effects models and extensions in ecology with R. Berlin: Springer. https://
11, 14–31. https://doi.org/10.1890/1051-0761(2001)011[001 doi.org/10.1007/978-0-387-87458-6
4:EONAAL]2.0.CO;2
Sala, O. E., Chapin III, F. S., Armesto, J. J., Berlow, E., Bloomfield, J.,
Dirzo, R., … Wall, D. H. (2000). Global biodiversity scenarios for the S U P P O RT I NG I NFO R M AT I O N
year 2100. Science, 287, 1770–1774. https://doi.org/10.1126/
science.287.5459.1770 Additional Supporting Information may be found online in the sup-
SIGTIERRAS - MAGAP. (2010) Available at: www.sigtierras.gob.ec. porting information tab for this article. 
Smith, W. P., Twedt, D. J., Hamel, P. B., Ford, R. P., Wiedenfeld, D. A., &
Cooper, R. J. (1998). Increasing point-­count duration increases stan-
dard error. Journal of Field Ornithology, 69, 450–456.
How to cite this article: Tinoco BA, Santillán VE, Graham CH.
Stiles, F. G. (1981). Geographical aspects of bird-­flower coevolution, with
particular reference to Central America. Annals of the Missouri Botanical Land use change has stronger effects on functional diversity
Garden, 68, 323–351. https://doi.org/10.2307/2398801 than taxonomic diversity in tropical Andean hummingbirds.
Stiles, F. G. (2008). Ecomorphology and phylogeny of hummingbirds: Ecol Evol. 2018;00:1–13. https://doi.org/10.1002/ece3.3813
Divergence and convergence in adaptations to high elevations.
Ornitología Neotropical, 19, 511–519.