ArticlePDF Available

Consequences of agroindustrial sugarcane production to freshwater biodiversity


Abstract and Figures

The environmental benefits of a broad scale adoption of biofuels are critically contingent on what current land uses will be converted for feedstock expansion and how converted land will be managed. We assessed the consequences of land use and land management for the agroindustrial production of sugarcane to the physical, chemical and biological properties of freshwater systems. We surveyed 16 environmental variables and algae, tadpoles, predatory invertebrates and fish in lentic water bodies distributed across a gradient in land use intensity ranging from seasonal Atlantic Forest and cerrado to pastures to sugarcane plantations in SE Brazil, the most important sugarcane-producing region in the world. The gradient in land use intensity was not only an axis of native habitat loss but also of ecosystem productivity, as indicated by increased conductivity, turbidity, and phytoplankton biomass. Land use had a clear signal on community and metacommunity organization, with converted land being impoverished in biodiversity relative to native habitats. However, frequency of occurrence, density, biomass and alpha diversity of tadpoles and their predators were not affected by land use. These results suggest that sugarcane fields function as habitat to a fraction of aquatic biodiversity. Within sugarcane fields, larger wetlands surrounded by buffer strips as required by law appeared comparatively buffered against land management practices and housed a disproportional fraction of animal biomass, likely acting as sources of migrants to other water bodies in the landscape. Conversion of pastures to sugarcane fields, suggested as a strategy to reduce competition for land with food production and biodiversity conservation, does not appear to have strong consequences to lentic freshwater systems, provided that wetlands and surrounding buffer strips are preserved. These observations emphasize the importance of enforcement of legislation regulating land use (i.e. the ′Forest Code′) and certification systems rewarding the voluntary adoption of better land management practices.This article is protected by copyright. All rights reserved.
Content may be subject to copyright.
Consequences of agroindustrial sugarcane production to
freshwater biodiversity
and D
Environmental Management, School of Arts, Sciences and Humanities, University of S~
ao Paulo, Av. Arlindo B
ettio 1000,
03828-000, S~
ao Paulo-SP, Brazil,
The University of Texas at Austin, Department of Integrative Biology, 1 University Station,
C0990, Austin, TX 78712, USA
The environmental benefits of a broad-scale adoption of biofuels are critically contingent on what current land
uses will be converted for feedstock expansion and how converted land will be managed. We assessed the conse-
quences of land use and land management for the agroindustrial production of sugarcane to the physical, chemi-
cal, and biological properties of freshwater systems. We surveyed 16 environmental variables and algae, tadpoles,
predatory invertebrates, and fish in lentic water bodies distributed across a gradient in land-use intensity ranging
from seasonal Atlantic Forest and cerrado to pastures to sugarcane plantations in SE Brazil, the most important
sugarcane-producing region in the world. The gradient in land-use intensity was not only an axis of native habitat
loss but also of ecosystem productivity, as indicated by increased conductivity, turbidity, and phytoplankton bio-
mass. Land use had a clear signal on community and metacommunity organization, with converted land being
impoverished in biodiversity relative to native habitats. However, frequency of occurrence, density, biomass, and
alpha diversity of tadpoles and their predators were not affected by land use. These results suggest that sugarcane
fields function as habitat to a fraction of aquatic biodiversity. Within sugarcane fields, larger wetlands surrounded
by buffer strips as required by law appeared comparatively buffered against land management practices and
housed a disproportional fraction of animal biomass, likely acting as sources of migrants to other water bodies in
the landscape. Conversion of pastures to sugarcane fields, suggested as a strategy to reduce competition for land
with food production and biodiversity conservation, does not appear to have strong consequences to lentic fresh-
water systems, provided that wetlands and surrounding buffer strips are preserved. These observations empha-
size the importance of enforcement of legislation regulating land use (i.e. the ‘Forest Code’) and certification
systems verifying compliance and rewarding the voluntary adoption of better land management practices.
Keywords: agriculture, Atlantic Forest, cerrado, C
odigo Florestal, conservation, ethanol, frog, land-use change, metacommunity,
Received 19 February 2015 and accepted 25 March 2015
Global energy demand is predicted to rise by 37% from
2012 to 2040 and, despite the continuing predominance
of fossil fuels, consumption of biofuels for transporta-
tion is to be increased more than threefold (IEA, 2014).
Biofuels are expected to alleviate human reliance on
nonrenewable fossil fuels while contributing to a reduc-
tion in greenhouse gas emissions and to the promotion
of rural development. However, the environmental ben-
efits of biofuels are critically contingent on which,
where, and how biofuel feedstocks are produced. This
occurs because feedstocks vary broadly in energetic effi-
ciency, because their expansion may promote the con-
version of lands of high conservation value and release
large quantities of stored carbon, and because intensive
agriculture is linked to various forms of environmental
degradation (OECD, 2001, 2006; Clay, 2004; Fargione
et al., 2008; Lapola et al., 2010).
Sugarcane is the most energetically efficient first-gen-
eration source of ethanol, yielding over eight units of
biofuel energy output per-unit energy input when com-
pared with two for beet, wheat, or corn (WWI, 2006).
Echoing a global expansion of biofuel crops, between
2002 and 2012 sugarcane land cover in Brazil almost
doubled to reach a total of 9.7 million hectares, and it
has to expand another 1.44.3 million hectares if the
country is to reach its 2020 ethanol production goals
(UNICA, 2008; Lapola et al., 2010). Brazil currently
responds to 35% of all sugarcane produced in the
world, half of which concentrated in southeastern state
of S~
ao Paulo (UNICA, 2014).
Despite significant recent advances, sugarcane produc-
tion in Brazil has been historically associated with seri-
ous environmental impacts. These include deforestation
Correspondence: Luis Schiesari, tel. +55 11 992346740, fax +55 11
30918120, e-mails:;
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd.
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. 1
GCB Bioenergy (2015), doi: 10.1111/gcbb.12279
of the Atlantic Forest and to a lesser extent the cerrado,
large-scale preharvest burns, extensive soil erosion, high
water consumption in mills, and deliberate release of
vinasse in streams and rivers causing eutrophication,
algal blooms, anoxia, and fish kills (Dean, 1997; Ballester
et al., 1999; Gunkel et al., 2007; Martinelli & Filoso, 2008).
In addition, sugarcane production has a significant
potential for environmental contamination from the
moderately high consumption of inorganic and organic
fertilizers (Cantarella & Rossetto, 2010; UNICA, 2011),
and the moderate consumption of a broad range of pesti-
cides (225 formulations containing 62 active ingredients;
Schiesari & Grillitsch, 2011).
Perhaps surprisingly then, explicit assessments of the
impacts of sugarcane production on biodiversity are
rare. For example, a recent review of the impacts of
global biofuel crop production on biodiversity failed to
find a single article on sugarcane (Immerzeel et al., 2014;
but see, e.g. Corbi et al., 2008; Verdade et al., 2012).
Freshwater systems constitute an appropriate system
to assess the effects of land use and land management
on biodiversity because are intimately connected to sur-
rounding terrestrial systems by the transfer of energy
and matter. Furthermore, water bodies usually sustain
diverse communities comprising species with varying
degrees of dependence on the aquatic environment.
Whereas many species are obligate aquatic, many others
have complex life cycles and as such integrate the
cumulative effects of freshwater and terrestrial degrada-
tion over the course of their life history. Finally, there is
a solid tradition in freshwater theoretical and applied
ecology that can provide a basis for interpreting land-
use effects on traits, individuals, populations, and com-
munities (e.g. Wetzel, 2001).
This study aimed at assessing the consequences of land
use and land management for the production of sugar-
cane on the physical, chemical, and biological properties
of freshwater systems, with an emphasis on the diversity,
composition, and structure of their communities.
This assessment was based on sampling surveys in
replicated lentic (i.e. still) water bodies distributed across
a gradient in land-use intensity ranging from seasonal
Atlantic Forest, wooded cerrado (‘cerrad~
ao’) and cerrado
to pastures to sugarcane plantations in the state of S~
Paulo, and employing amphibian larvae and their preda-
tors as the study system. The sampling design thus cap-
tures three types of land conversion associated with the
historical expansion of sugarcane: the conversion of
native habitats to pasture, the conversion of native habi-
tats to sugarcane, and the conversion of pasture to sugar-
cane (Dean, 1997). The latter, in particular, is central in
the debate regarding the current and future expansion of
biofuel crops in the country. A recently approved
agroecological zoning for the sustainable production of
sugarcane identified as suitable 37 Mha currently
covered by low-profitability pastures and concluded that
sugarcane expansion could occur over pastures with no
impacts on land covered by native vegetation or devoted
to food production (Manzatto et al., 2009; Federal decree
6961/2009). Other studies arrived at a similar recommen-
dation while targeting at a reduction in carbon emissions
(Lapola et al., 2010). Therefore, considering that indus-
trial scale sugarcane production follows an intensive
model of agriculture, this raises the question not only of
the environmental impacts of sugarcane production rela-
tive to native habitats, but also relative to pastures,
which may support moderate levels of biodiversity.
Materials and methods
Study site
Field work was conducted in the Estac
ao Ecol
ogica do Jata
(EEJ) and surrounding farms in the municipality of Luis Anto-
nio, S~
ao Paulo, southeastern Brazil (21°3401100S, 47°4400700W).
The EEJ represents one of the very few preserved (mostly sec-
ond growth) remnants of cerrado, wooded cerrado (‘cerrad~
and seasonal Atlantic Forest vegetation in the state of S~
ao Paulo,
and is embedded in a matrix largely dominated by intensive
sugarcane cultivation, with a smaller occurrence of pastures
and silviculture. Around the EEJ, we were granted access to two
farms cultivating sugarcane in industrial scale and one small
ranch raising cattle and sheep. According to local agronomists,
land management for sugarcane production in these farms are
typical for the broader region of Ribeir~
ao Preto, which is the
most traditional sugarcane-producing region in the country.
Selection of water bodies
We selected 15 lentic water bodies in native habitats, 15 in pas-
tures, and 17 in sugarcane fields based on the following crite-
ria: (i) that the water body was amenable to pipe sampling,
that is, that a considerable fraction of the water body was less
than 60 cm deep; (ii) that the water body in a given land-use
type was at least 100 m away from another land-use type; (iii)
that the water body was independent from and at least 50 m
away from other sampled water bodies; (iv) that each land-use
type included temporary, semipermanent, and permanent
water bodies, given the importance of hydroperiod in influenc-
ing freshwater community composition (Wellborn et al., 1996).
Water bodies comprised puddles, ditches, ponds, farm ponds
(‘cacimbas’), reservoirs, swamps, and marshes varying in size
from 1.4 to 106 000 m
(see Fig. 1). Not all of the selected water
bodies were available for each analysis because some were dry
or did not contain tadpoles at the moment of sampling.
Sampling protocol
Selected water bodies were inspected monthly from December
2010 to November 2011 for phenological record and measure-
ment of water depth for estimation of hydroperiod (% of the
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
monthly visits in which water depth >0 cm). In December 2010
and February 2011, respectively, at the beginning and middle of
the rainy season, water bodies were subject to comprehensive
habitat and physicochemical characterization, amphibian calling
surveys, and freshwater community surveys. Timing of sampling
surveys in this seasonal landscape was chosen based on the phe-
nology of occurrence of water bodies (two-thirds of which only
existed in the rainiest months) and their communities.
Physical and chemical characterization of water bodies
Prior to the community sampling surveys, we conducted a
comprehensive characterization of water bodies. Habitat
characterization comprised recording spatial positioning, land-
use type, any ongoing activities of land management, water
body dimensions, canopy cover over the pond basin (in %,
with a spherical crown densiometer; Forestry Suppliers, Jack-
son, MS, USA), substrate type (rocks, pebbles, sand, clay, mud,
wood, leaf-litter) and cover (each substrate type categorized as
comprising 024%, 2549%, 5074%, 75100% of the pond
basin), aquatic vegetation type (i.e. terrestrial/paludicolous
plants, floating macrophytes, submerged macrophytes) and
cover (as above), and a description of the surrounding terres-
trial vegetation. We then quantified in situ basic physicochemi-
cal water parameters including temperature, pH, conductivity
(with an Orion 3-star pH meter and an Orion 3-star conductiv-
ity meter; Thermo Scientific, Waltham, MA, USA), and dis-
solved oxygen (henceforth DO; with a YSI Pro20 DO meter;
Fig. 1 Effects of land use on selected environmental variables related to water body structure (left), water physicochemistry (middle)
and productivity (right). See Table S1 for univariate statistical analyses. Data from both sampling surveys are lumped for each
response variable. Because in each land-use water bodies were selected so as to cover the full spectrum of the hydroperiod gradient,
the inclusion of area and hydroperiod in this figure is meant to demonstrate that this important gradient was controlled in our sam-
pling surveys. Corrected a=0.002.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
Yellow Springs, OH, USA), as well as turbidity (with an AQ
4500 Advanced Aquafast IV Turbidimeter; Thermo Scientific,
Waltham, MA, USA) and phytoplankton standing crop (in trip-
licate, with an Aquafluor Handheld Fluorometer with in vivo
chlorophyll-aand phycocyanin channels; Turner Designs, Sun-
nyvale, CA, USA). We also collected and preserved water sam-
ples for laboratory analyses of water hardness, total nitrogen
(TN), and total phosphorus (TP). These parameters were ana-
lyzed with a HACH DR 2800 spectrophotometer and appropri-
ate reagent kits (HACH, Loveland, CO, USA).
Calling activity of amphibians
Prior to the freshwater community sampling surveys, we
recorded in each water body the calling activity of amphibian
adults to provide a more comprehensive assessment of
amphibian community composition, to assess the efficiency in
larval sampling, and to aid in larval identification. Calling sur-
veys were semiquantitative (i.e. in the following categories of
abundance: 0, 12, 35, 610, 1120, 2150, or more than 50 call-
ing individuals; Heyer et al., 1994) and conducted for a total of
24 nights in each water body.
Freshwater communities
We sampled the fauna of tadpoles and their predators, namely
nymphal dragonflies (Odonata: Aeshnidae, Libellulidae, Cor-
duliidae, Gomphidae), adult and larval beetles (Coleoptera: Dy-
tiscidae, Hydrophylidae), adult and nymphal heteropterans
(Heteroptera: Notonectidae, Belostomatidae, Nepidae), and
Samples were taken by pipe sampling, dipnetting, and sein-
ing (e.g. Heyer et al., 1994; Werner et al., 2007). Pipe sampling
provided quantitative per-unit-area information on species
composition, abundances, and biomass. Pipes were 60 cm tall
and 33 cm in diameter and therefore sampled 0.09 m
of water
body habitat. The sample was taken by quickly thrusting the
pipe through the water column and into the sediments to seal
the sample area. We then removed by dipnetting all animals
>5 mm from the sampled water column and the first centime-
ter of the sediments. In practice, we considered the pipe to be
empty when at least 10 consecutive sweeps were made without
capturing any more animals. Sampling effort was scaled to
pond surface area: Five pipes were sampled in water bodies
<10 m
, 15 pipes in water bodies >10 m
and <50 m
, 30 pipes
in water bodies >50 and <200 m
, 60 pipes in water bodies
>200 m
and <500 m
and 90 pipes in water bodies >500 m
Pipes were distributed so as to representatively sample all mi-
crohabitats in a water body. In water bodies with deeper
waters (i.e., deeper than sampling pipes; NAT01, NAT05,
PAS09, SUG08, SUG09), we supplemented pipe sampling with
three hauls of a 10-m seine in the deeper water. Tadpoles and
fishes were immediately euthanized and preserved in 10% buf-
fered formalin, invertebrates in 70% ethanol.
In the laboratory, all biological samples were sorted and
identified to family (invertebrates) or species level (vertebrates),
except for the amphibians Scinax and Rhinella, which were
assigned to genus only because could not be confidently
assigned to species (at our study site, there are two species in
the genus Scinax,S. fuscovarius and S. similis, and two in the
genus Rhinella,R. ornata and R. schneideri), counted and
weighed. Fishes with detritivorous or herbivorous feeding hab-
its were excluded from the analysis. Summarizing, in each
water body we obtained as biological response variables com-
munity composition, species or family richness, densities, and
per-unit area biomasses of tadpoles and predators.
Data analyses
Effects of land use on environmental and biological vari-
ables. We first conducted two-way analyses of variance fol-
lowed by Tukey HSD post hoc tests for testing the effects of
land use, survey (i.e. whether in the beginning or middle of the
rainy season), and a land-use-by-survey interaction term on
each individual variable. The effects of survey could not be
assessed for hydroperiod as this is a year-round summary met-
ric, and for conductivity and TN, as these variables were quan-
tified in the first survey only due to technical issues. In all
other cases, a land-use-by-survey interaction term was found
nonsignificant and was dropped from the analysis. When the
assumptions of normality and homogeneity of variances could
not be met even after transformations, we employed Kruskal
Wallis nonparametric tests followed by pairwise comparisons.
In all cases, we applied Bonferroni corrections to avoid Type I
error inflation.
We then conducted two separate permutational multivariate
analyses of variance (perMANOVA) to test for multivariate land-
use differences in the environmental and community datasets.
PerMANOVAs were run on the Euclidean distances of the log-
transformed environmental data (Anderson et al., 2006) and on
the BrayCurtis distances of the Hellinger-transformed tadpole
species density data (Legendre & Gallagher, 2001). Both trans-
formations were implemented in the function decostand of the
R package vegan (Oksanen et al., 2012). The homogeneity of
variances among groups was checked previously to the perMA-
NOVA using the function betadisper from the R package vegan
(Oksanen et al., 2012). When significant differences among
land-use types were detected, we ran multiple comparisons
with Bonferroni-adjusted significance levels. R version 2.15.2 (R
Core Team, 2012) was used in all statistical analyses.
Alpha, beta, and gamma diversity. We compared alpha diver-
sity among land-use types by an analysis of variance on the
total number of larval amphibian species, or predator families,
observed in each water body. We also compared amphibian
beta diversity among water bodies within each land-use type
using the procedure proposed by Baselga (2010, 2012) and
implemented in the R package betapart (Baselga & Orme, 2012).
This procedure decomposes the overall beta diversity (b
into two components: (i) spatial turnover (b
), which is the
fraction due to replacement of some species by others along the
sites, and (ii) nestedness (b
), which is the fraction due to dif-
ference in species richness, such as sites with lower richness
being a subset of the richest site (Baselga, 2010, 2012). To test
the difference between beta diversity components, we used a
resampling procedure where we took 1000 random samples of
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
five sites per area and calculated the average b-values. With
the estimated distribution of bvalues, we assessed the
between-area significance by calculating the probability of the
area with smaller b-value being larger than the other just by
chance (Baselga, 2010). We decided not to compare predator
beta diversity within each land-use type because of the diffi-
culty in interpreting this metric at such a crude taxonomic (i.e.
family) level. We compared gamma diversity among land-use
types by rarefaction curves extrapolating species richness in
each land-use type to 11 samples, that is, the maximum num-
ber of water bodies sampled in a given land-use type (Colwell
et al., 2012). Rarefaction was performed using the software ESTI-
MATES (Colwell, 2013).
Predictors of amphibian species richness. To assess what fac-
tors influence local amphibian species richness (i.e. alpha diver-
sity), we performed a generalized linear model (GLM) with a
model selection and model averaging approach considering
land use and seven variables recognized in the literature as
influencing tadpole richness. These were water body area, hy-
droperiod, canopy cover, the presence of fish, predator density,
competitor density, and distance to forest edge (Wellborn et al.,
1996; Werner et al., 2007). Quantitative variables ranged over
several orders of magnitude and were log-transformed prior to
the analysis (Anderson et al., 2006) using the function deco-
stand of the R package vegan (Oksanen et al., 2012). We assessed
the occurrence of multicollinearity among quantitative variables
by the variance inflation factor (VIF). All VIFs were below the
suggested threshold (VIF <3; Zuur et al., 2009, 2010). We fitted
a global model to the data using a Poisson distribution (no
overdispersion detected). The global model contained all eight
explanatory variables and from that model we generated a set
with all possible submodels using the function dredge from the
R package MuMIn (Barton, 2012). We decided to run all possi-
ble models as we a priori selected only variables known to influ-
ence amphibian richness. We employed Akaike’s information
criterion corrected for small sample sizes (AICc) to rank the
obtained models and selected models with ΔAICc <2 (Burn-
ham & Anderson, 2002) to compute the model-averaged param-
eters using the function model.avg in MuMIn package (Barton,
2012). We did the model averaging with shrinkage, where in
models where a variable is absent its coefficient is set to zero
(Burnham & Anderson, 2002). We also assessed the likelihood
of the models using Akaike weights, which indicates how likely
the model is the best among the ones considered.
Community structure, land use, and associated environmen-
tal variables. To determine the exclusive and shared effects of
environmental variables and land use on community structure
we conducted redundancy analyses (RDA) using the variation
partitioning method with adjusted R
following Borcard et al.
(2011). To objectively reduce dimensionality while maintaining
interpretability, we first separated the 18 environmental vari-
ables in structural properties (area, depth, hydroperiod, canopy,
vegetation cover, organic sediment cover), basic physicochemi-
cal water properties (temperature, pH, DO, conductivity, tur-
bidity, hardness), properties related to ecosystem productivity
(total nitrogen, total phosphorus, chlorophyll-a concentration,
phycocyanin concentration), and predation pressure (density of
fish, density of predatory invertebrates). We ran one PCA on
each of these groups and retained the site scores of the first and
second axes as explanatory environmental variables.
Tadpole density and environmental variables were, respec-
tively, subject to the Hellinger (Legendre & Gallagher, 2001)
and log-transformation (Anderson et al., 2006), both imple-
mented in the function decostand of the R package vegan (Ok-
sanen et al., 2012). We selected the environmental variables to
be used in the variation partitioning by a forward selection
method (Blanchet et al., 2008; Borcard et al., 2011). We tested by
a permutation test with 999 permutations (Borcard et al., 2011)
the significance of the testable fractions of the explanatory vari-
ables in the variation partitioning. All analyses were performed
in the R package vegan (Oksanen et al., 2012).
Analyses using the first and the second sampling surveys,
and a summary dataset comprising both surveys had similar
results. Therefore, we present here only the analyses with the
summary dataset. The summary environmental explanatory
matrix was constructed using the average values of the two
surveys for variables and water bodies sampled in both sur-
veys; the only value available for variables or water bodies
sampled in a single survey; and the maximum density of inver-
tebrate and fish predators across surveys. The summary com-
munity matrix was constructed using the maximum density of
each tadpole species per-pond across surveys.
Metacommunity structure, land use, and associated environ-
mental variables. To analyze the structure of the amphibian
metacommunity, we first summarized the distribution pattern
of 21 amphibian species in 26 water bodies (three additional
water bodies were excluded for not containing amphibians) by
an incidence matrix representing the site-by-species presence/
absence data. For completeness, we included both calling sur-
vey and tadpole survey data in this incidence matrix; analyses
of an incidence matrix comprising only tadpole survey data
yielded qualitatively similar results.
We tested whether incidences differed from a null model of
random placement of species in water bodies by ordering rows
and columns of the incidence matrix by reciprocal averaging
(i.e. correspondence analysis), such that water bodies with the
most similar species lists and species with the most similar dis-
tributions were closest together. We then tested this ordinated
matrix for its coherence, that is, whether embedded absences
within species distributions were significantly less common
than expected by chance, using the null model randomization
method r0 (Presley et al., 2010). Significant coherence implies
that there is a strong axis of variation in metacommunity struc-
ture, and justifies further characterization of elements of meta-
community structure in terms of species turnover or nested
subsets, and if there is a significant degree of boundary clump-
ing among species distributions (Leibold & Mikkelson, 2002).
Altogether, these analyses permit testing whether our incidence
matrix fits one of the several idealized patterns of metacommu-
nity structure (Clemmentsian vs Gleasonian gradients, nested
subsets, evenly spaced, checkerboards), which could in some
cases provide insights into hypothesized underlying structur-
ing mechanisms.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
We performed Spearman rank correlations between site ordi-
nation scores and site richness and site environmental variables
to interpret the axis of environmental variation associated with
the latent gradient (as in Meynard et al., 2013) and tested by
means of an ANOVA on site ordination scores whether land use
significantly influences community arrangement along the ordi-
nation axes.
Analyses were conducted following the methodology pro-
posed by Leibold & Mikkelson (2002), modified by Presley
et al. (2010) using R package metacom (Dallas, 2014).
Land-use patterns of environmental properties of water
Land use had a significant univariate effect on seven of
18 measured environmental variables. These were can-
opy cover, temperature, conductivity, turbidity, water
hardness, and chlorophyll-aand phycocyanin concen-
trations (Fig. 1, Table S1). Among the three forms of
land use, water bodies in native habitats had the lowest
turbidity, water hardness, and phycocyanin concentra-
tions. In turn, water bodies in sugarcane plantations
had the highest temperatures and chlorophyll-aconcen-
trations. The only distinctive characteristic of water
bodies in pastures were the lowest temperatures. Irre-
spective of land use, a significant effect of survey was
detected for chlorophyll-aconcentration (greater in
December), and for TP and phycocyanin concentration
(greater in February).
Land use also had a significant effect on the environ-
mental properties of water bodies considered jointly
=6.68, P<0.01; Fig. 2). Environmen-
tal properties of water bodies in native habitats were
significantly different both from those in sugarcane
plantations (perMANOVA,F
=13.27, adj. P<0.01) and
pastures (perMANOVA,F
=4.11, adj. P=0.02), which
could not be differentiated from each other (perMANOVA,
=1.86, adj. P=0.33).
Land-use patterns of amphibians and their predators
Most water bodies in all three forms of land use con-
tained amphibians and their predators (Fig. 3a). Tad-
poles were found in 70% of the water bodies sampled
in native habitats, but in 100% of the water bodies sam-
pled in pastures and sugarcane fields. In turn, predators
were found in 100% of the water bodies sampled in
native habitats and pastures, but in 70% of the water
bodies sampled in sugarcane fields. Interestingly, land
use had no univariate effect on the density and per-unit
area biomass of tadpoles or of their predators (Fig. 3b,c,
Table S1).
Although the three water bodies with the highest
amphibian species richness were all found in native hab-
itats (12, 10 and 7 species), and although the richest
water bodies in pastures and sugarcane fields had half
as many species (6 species), no significant univariate
effect of land use on amphibian species richness was
detected either (Fig. 3d, Table S1). This occurred because
of the high among-pond variability in amphibian species
richness in all three land-use types (012 species in
native habitats, 26 in pastures, and 16 in sugarcane
fields). Likewise, the overall amphibian beta diversity
) and its two components, spatial turnover (b
and nestedness (b
), were similar among land use
–0.10 –0.05 0.00 0.05
–0.06 –0.02 0.02 0.06
–1 0 1 2
–1.0 –0.5 0.0 0.5 1.0
Fig. 2 nMDS plots of the environmental (left) and tadpole community (right) data from water bodies distributed across a gradient in
land-use intensity comprising native habitats, pastures, and sugarcane plantations. Black triangles =sugarcane plantations, gray
squares =pasture, open circles =native habitats. Environment plot based on the Euclidian distance of 16 log-transformed environ-
mental variables. Stress value =0.153. Tadpoles plot based on the BrayCurtis distance of 21 tadpole species Hellinger-transformed
density data. Stress value =0.082.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
types (mean b
, and b
were 0.71, 0.42, 0.29 for
native habitats, 0.69, 0.49, 0.20 for pastures, and 0.69,
0.51, 0.18 for sugarcane; P>0.05 for all land use compar-
isons). In contrast, gamma diversity differed among
land-use types. Total measured amphibian richness was
17 species in native habitats, 8 in pastures, and 10 in sug-
arcane fields (Fig. 3e). Correspondingly, extrapolation of
rarefaction curves of species richness to 11 samples, that
is, the maximum number of water bodies sampled per
land-use type, indicated that species richness in native
habitats (mean 17.62, 95% CI 13.4521.79) was signifi-
cantly higher than species richness in pastures (8.68,
6.2411.12) and sugarcane fields (10, 8.7111.29), which,
in turn, did not differ from each other.
While the analysis of patterns of predator diversity
should be exercised with caution due to the crude taxo-
nomic resolution, no univariate effects of land use on
the per-pond number of predator families were detected
(Fig. 3d; Table S1). Overall, the number of predator fam-
ilies declined from preserved to converted landscapes
(Fig. 3e), but this occurred exclusively due to fish,
which were more common in native habitats because of
three water bodies seasonally linked to streams.
Predictors of amphibian species richness
From the set of models generated to infer the best com-
bination of environmental variables predicting amphib-
ian species richness, five models were best supported
(five models had DAICc <2; Table 1, Table S2). Hydro-
period and tadpole density were present in all five top
models, and therefore, were assigned a relative impor-
tance of 1.00 after model averaging (Table 1). Tadpole
species richness increased with tadpole density and hy-
droperiod, that is, water bodies that contained more
tadpoles per-unit area and that held water for longer
Fig. 3 (a) Frequency of occurrence (b) Density (c) Per-unit-
area biomass (d) Richness per pond and (e) Richness per land-
use type of tadpoles (left) and predators (right) in water bodies
distributed across a gradient in land-use intensity comprising
native habitats, pastures, and sugarcane plantations. Here fre-
quency of occurrence refers to the % of water bodies that con-
tained tadpoles or predators in either the first (December) or
second (February) community surveys. Density and per-unit-
area biomass boxplots lump data from both surveys, and rich-
ness per pond and richness per land-use type consider the
cumulative species list found in a given pond (or land use)
across surveys. Richness for amphibians is species richness, for
invertebrates (which include dragonflies, water bugs, beetles,
and fish) is family richness. Because few fish species were cap-
tured by pipe sampling, estimates of predator richness include
fish captured by seine nets as well. Land use had a significant
effect in (e) but not in (b), (c), or (d). See Table S1 for univariate
statistical analyses.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
periods had higher tadpole richness (Fig. S1). Because
the 95% confidence intervals of all other variables
included zero (Table 1), there is not enough evidence
that distance to native habitats, fish presence, and can-
opy cover affect tadpole species richness in our study.
Amphibian community structure, land use, and associated
environmental properties
As for environmental properties, land use had a signifi-
cant effect on amphibian community structure (perMANO-
=3.74, P<0.01; Fig. 2). Amphibian communities
in native habitats differed significantly from those in sug-
arcane fields (perMANOVA,F
=4.84, adj. P=0.04),
whereas those in pastures could not be differentiated from
either of them (perMANOVA F
=2.83, adj. P=0.09 for
pasture-native habitat comparison and F
=3.31, adj.
P=0.07 for pasture-sugarcane comparison).
Land use, however, comprises several axes of envi-
ronmental change that could, alone or in combination,
influence amphibian community structure. Among the
many descriptors of freshwater habitats, the forward
selection procedure retained four composite environ-
mental variables in our approach to reduce dimension-
ality. These were the first PCA axis of properties related
to ecosystem productivity (prod1), the first PCA axis of
water physicochemical properties (phys_chem1), and the
first and second PCA axes of predation pressure (pred1,
pred2) (see Table S3 and Fig. S2 for details). Axis prod1
explained 65.4% of the total among-site variation and
was positively related to TN. Chlorophyll-aand phyco-
cyanin concentrations, and TP, were weaker positive
contributors to prod1. Axis phys_chem1 explained 69.5%
of the total among-site variation and was positively
related to turbidity. Conductivity and water hardness
were weaker positive contributors to phys_chem1. Axes
pred1 and pred2 explained all total among-site variation
(57.3% and 42.7%, respectively), with pred1 being nega-
tively related to both the maximum density of predatory
invertebrates and fish, and pred2 being positively related
to the maximum density of fish but negatively related
to the maximum density of predatory invertebrates.
The above-mentioned selected composite environ-
mental variables (axes of PCA) and land use together
explained 48.6% of the total tadpole community
variability (R2
adj =0.49, P<0.01). From the total variabil-
ity, 31% was explained exclusively by the selected envi-
ronmental variables (R2
adj =0.31, P<0.01), and over
fourteen percent was explained by the shared compo-
nent between environmental variables and land use
adj =0.14, p not testable). Land-use alone explained
less than three percent of the tadpoles community vari-
ability and was not significant (R2
adj =0.03, P=0.11).
The first RDA axis was negatively related to phys_chem1
and pred1, and less so to prod1 and pred2 (see Table S4
and Fig. S3 for RDA site and species scores, and biplots
of individual selected composite environmental vari-
ables). Therefore, increasing score values along the first
RDA axis, among other characteristics, described a gra-
dient ranging from water bodies with high turbidity
and TN, and less predatory invertebrates and fishes
(water bodies mainly located in sugarcane fields), to
water bodies with low turbidity and TN, and more
predatory invertebrates and fishes (water bodies mainly
located in native habitats and pasture). The second
RDA axis was positively related to pred1 and negatively
related to pred2 (Fig. S3). Therefore, increasing score val-
ues along the second axis described mainly a gradient
from water bodies with more predatory fish and less
predatory invertebrates to water bodies with less preda-
tory fish and more predatory invertebrates. Most spe-
cies were found in water bodies devoid of fish; the
glaring exception was the bufonid Rhinella, typically
found in water bodies with high densities of fish. Scinax
and Physalaemus cuvieri were usually related to less pro-
ductive water bodies, with lower TN and lower turbid-
ity. By contrast, Eupemphix nattereri and Leptodactylus
fuscus were usually associated to more productive water
bodies with higher TN and higher turbidity.
Amphibian metacommunity structure, land use, and
associated environmental properties
The amphibian metacommunity exhibited strong
coherence along the first axis of ordination (z=3.21,
P<0.001; Fig. 4). There were 141 embedded absences
in a 21 926 species-by-site incidence matrix (546
cells). The community exhibited a nonsignificantly
Table 1 Model averaged estimates (with shrinkage) from the
five generalized linear models (ΔAICc <2; Table S2) predicting
tadpole species richness on 27 water bodies distributed across
a gradient in land-use intensity comprising native habitats, pas-
tures, and sugarcane plantations. SE =standard error;
CI =95% confidence interval. There is a positive relationship
between hydroperiod and richness and between tadpole den-
sity and richness (intervals do not cross zero)
Variable Estimate
(intercept) 1.260 0.73 2.29 0.17
Hydroperiod 0.121 0.05 0.02 0.22 1.00
0.184 0.05 0.08 0.29 1.00
Distance 0.019 0.02 0.07 0.01 0.57
Fish 0.325 0.34 1.28 0.06 0.54
0.007 0.04 0.15 0.03 0.12
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
positive turnover (z=1.39, P=0.164) and was signif-
icantly clumped (Morisita’s index =2.559, df =18,
P<0.001). The first axis of ordination was positively
correlated with land-use intensity, conductivity, turbid-
ity, TN, TP, chlorophyll-a, and phycocyanin concentra-
tions (Spearman rank correlation, P0.027) and
negatively correlated with area, hydroperiod, vegeta-
tion cover, and organic sediment cover (P0.017). As
hypothesized, land-use intensity had a significant effect
on ordination scores (ANOVA F
=3.51 P=0.047): post
hoc analyses indicated that ordination scores of native
habitat communities were significantly different from
those of sugarcane communities, whereas ordination
scores of pasture communities were not significantly
different from either those of native habitats or of sug-
arcane plantations. Species richness was significantly
correlated with site scores of the first axis (Spearman
rho =0.47, P=0.016).
Land use and land management for the production of
sugarcane had a clear signal on the physical, chemical
and biological properties of freshwater systems. In a
Rhinella spp
Leptodactylus mystaceus
Elachistocleis cesarii
Physalaemus cuvieri
Dendropsophus minutus
Scinax spp
Leptodactylus fuscus
Dermatonotus muelleri
Physalaemus centralis
Eupemphix nattereri
Leptodactylus mystacinus
Fig. 4 Amphibian species-by-site incidence matrix for the Estac
ao Ecol
ogica do Jata
ı and surrounding agricultural areas. The matrix
is ordered by reciprocal averaging such that water bodies with the most similar species lists and species with the most similar distri-
butions are closest together. Position of each water body across three important freshwater environmental gradients is represented in
color coded rows at increasing shades of gray at the top of each matrix (intensity of land use: native habitats, pastures and sugarcane
plantations; hydroperiod: 049%, 5099%, and 100% days with water; canopy cover: 033%, 3466%, and 67100% canopy cover).
Arrows display environmental factors significantly associated with site ordination scores according to a Spearman rank correlation.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
multivariate space defined by sixteen environmental
variables, there was virtually no overlap among water
bodies in preserved land and water bodies in sugar-
cane fields. Water bodies in pastures were signifi-
cantly different from those in native habitats, but not
from those in sugarcane fields. Likewise, land use
had a clear signal on community and metacommunity
organization. This signal was revealed both by guided
(i.e. nMDS and associated perMANOVA) and by
unguided ordination tools (i.e. RDA, reciprocal aver-
aging in the analysis of elements of metacommunity
structure) and emerged even after accounting for
other gradients of known importance in structuring
freshwater metacommunities such as hydroperiod,
canopy cover, and water body surface area (Wellborn
et al., 1996; Skelly et al., 1999; Werner et al., 2007).
Amphibian communities in native habitats were sig-
nificantly different from those in sugarcane fields;
pastures were intermediate in community structure in
that they were not significantly different neither from
native habitats nor from sugarcane fields. Therefore,
both abiotic and biotic data lend general support to
our hypothesis of a gradient in land-use intensity
and therefore environmental degradation ranging
from preserved cerrados and seasonal Atlantic Forests
to pastures to sugarcane fields.
Land-use patterns of environmental properties of water
The transition of native habitats to productive agricul-
tural land was associated with a decrease in canopy
cover and to an increase in turbidity, water hardness,
and abundance of cyanobacteria in lentic water
bodies. In turn, the transition of pastures to sugar-
cane plantations was associated with a further reduc-
tion in canopy cover, to an increase in phytoplankton
standing crop, and to a sharp increase in water con-
ductivity. Overall, the physical, chemical, and biologi-
cal characterization of water bodies indicated that, in
addition to an axis of habitat loss and fragmentation,
the gradient in land-use intensity is clearly a gradient
in ecosystem productivity. In fact, every multivariate
analysis conducted in this study strongly pointed to a
gradient in land-use intensity on indicators of produc-
tivity, especially phytoplankton standing crop. This
enhanced productivity is likely a product of increased
incident solar radiation due to canopy opening (Schie-
sari, 2006), and of the increase in nutrients from cattle
grazing and defecation in pastures, and from fertiliza-
tion in sugarcane fields. Indeed, sugarcane is an
intense consumer of fertilizers, and overuse is the
rule (Martinelli & Filoso, 2008). The likely fate of a
considerable fraction of applied nutrients is water
bodies, both by leaching (as is the case of nitrate,
which is highly soluble) and by erosion (as is the
case of phosphorus), which high rates are well docu-
mented for sugarcane and reinforced by our high
measures of turbidity and water hardness. Although
it may seem somewhat surprising that in this study
TN or TP did not show a univariate increase with
land-use intensity, 9 of the 10 highest values of TP
and of TN were found in converted land, and land-
use intensity was accompanied by a sharp increase in
water conductivity.
Land-use patterns of biodiversity
Sugarcane fields and pastures were impoverished in
amphibian biodiversity with 10 and 9 species rela-
tive to preserved land with 18 species. Species sam-
pled exclusively in native habitats included Hypsiboas
faber, H. albopunctatus, H. lundii, Trachycephalus venulo-
sus, Leptodactylus labyrinthicus, L. mystaceus, Odontophry-
nus cultripes and Chiasmocleis albopunctata; in turn,
Rhinella sp. was sampled exclusively in pastures and Le-
ptodactylus mystacinus and Phyllomedusa hypochondrialis
in sugarcane fields. At least part of these land use-spe-
cific species occurrences, however, are likely sampling
effects: all species sampled exclusively in converted
land are known to occur in the Estac
ao Ecol
ogica do
ı (Prado et al., 2009, R. Sawaya pers. obs., L. Schie-
sari pers. obs.).
The amphibian metacommunity exhibited strong
coherence along the first axis of ordination, implying
that, overall, species responded to the same environ-
mental gradient. This environmental gradient ranged
from larger water bodies with longer hydroperiods,
richer in aquatic and semiaquatic vegetation and
organic sediment, and containing water with lower
conductivity, turbidity, nutrient concentrations, and
phytoplankton standing crop, to water bodies with
the opposite characteristics. This environmental gra-
dient also ranged from water bodies embedded in
preserved landscapes to those embedded in land-
scapes with high intensity of land use that, as
hypothesized, was accompanied by decreasing spe-
cies richness.
One might expect that intensification in land use
would be associated with a cumulative pattern of
species losses such that communities in intensively
managed landscapes would be a subset of communi-
ties in moderately intensively managed landscapes,
which, in turn, would be a subset of communities in
preserved, native habitats. In operational terms, this
would be the case if the metacommunity exhibited
positive boundary clumping and negative turnover
(Presley et al., 2010). This pattern was not strictly
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
observed, as we found positive boundary clumping
but positive turnover albeit nonsignificantly so.
Such structure corresponds to a ‘quasi-Clementsian’
metacommunity structure. In a strictly Clementsian
metacommunity structure (i.e. with significant positive
turnover), multiple species show coincident range
boundaries because of shared evolutionary history
and interdependent ecological relationships. Therefore,
groups of adjacent sites across the latent environmen-
tal gradient have shared species compositions termed
‘compartments’ (Lewinsohn et al., 2006). We identified
two compartments. The first comprised 1315 sites in
all land use types and across the full range of hyd-
roperiods that presented comparatively low conduc-
tivity, turbidity, and primary productivity, and up to
17 amphibian species. The second comprised eight
water bodies in converted land, characterized by
being nonpermanent, mostly devoid of vegetation,
turbid, and productive. This compartment contained
ten species including Physalaemus cuvieri, P. centralis,
Eupemphix nattereri, Dendropsophus minutus, Scinax spp,
Leptodactylus fuscus, L. mystacinus, and Dermatonotus
muelleri. Five to seven species occurred in both com-
As is frequently observed in Clementsian metacom-
munity structures, many species ranges were very
broad and had boundaries clumped at the ends of the
gradient. This could be a result either from the sam-
pled environmental gradient being too short, or from
species presenting broad niches. Considering the
remarkable breadth of lentic habitats sampled, we con-
clude that the niches of several species are indeed very
broad, including their tolerances to land-use and land
management practices. The range of Leptodactylus fus-
cus, Scinax spp (mostly S. fuscovarius), P. cuvieri and
D. minutus comprised no less than 92% (counting
embedded absences), 85%, 77%, and 73% of the latent
gradient. This is in strong contrast with species such
as Odontophrynus cultripes and Hypsiboas lundii, each
found in a single site and therefore only 4% of the gra-
In contrast to patterns in prey distributions, land use
had no effect on predatory invertebrate faunas at
least at a crude taxonomic resolution (i.e. family level).
All three sampled bug families, all two beetle families
and two of four dragonfly families were found across
the gradient; the two remaining dragonfly families
(Corduliidae, Gomphidae) were found in at least one
converted land use type. The number of fish families
in the cerrado (8) was much larger than that in pas-
tures (4) and sugarcane fields (1), but this effect is
likely due to some water bodies in native habitats
being seasonally linked to lotic water bodies, and not
to land use.
Potential drivers of biodiversity change
Considering that most changes in community composi-
tion occurred in the transition from cerrados and sea-
sonal Atlantic Forests to pastures, and not in the
transition from pastures to sugarcane fields, it is rea-
sonable to hypothesize that habitat loss, habitat
fragmentation and habitat split that is, the spatial
segregation between adult and larval habitat are
important contributors to species losses in productive
land (Becker et al., 2007). Indeed, organisms with com-
plex life cycles appear to be particularly vulnerable to
land conversion not only because integrate environ-
mental degradation in aquatic habitats as larvae, and
in terrestrial habitats as juvenile and adults (each of
which alone could contribute to population declines),
but also because the spatial segregation of adult and
larval habitat obliges adults and metamorphs to risky
Differences in land management between pastures
and sugarcane fields also are likely contributors to
changes in local species compositions and/or abun-
dances. Sugarcane fields are subject to intense physical
disturbance every year (the harvesting) and every 5
6 years (the field reform), plus the occasional pulses of
chemical disturbance following the application of pesti-
cides and fertilizers. Small ponds, puddles, and ditches
in sugarcane fields were typically devoid of dragonfly
larvae; we observed the same pattern in soybean fields,
where dragonfly mortality was tightly synchronized
with pesticide application (D. Negri and L. Schiesari,
pers. obs.). Some species of tadpoles were frequent and
apparently endured pesticide applications in both sug-
arcane and soybean fields, but we witnessed tadpole
dieoffs in ponds adjacent to sugarcane fields which
could plausibly be a result of pesticide application.
Indeed, Moutinho (2013) found that glyphosate, amet-
ryn, and acetochlor, three of the most common herbi-
cides applied in sugarcane fields, caused mortality of
frog larvae when manipulated at application doses rec-
ommended by the manufacturer. Likewise, ammonium
concentrations measured in ditches embedded in sugar-
cane plantations reached concentrations sufficient to
cause larval amphibian mortality (Ilha & Schiesari,
2014), especially at earlier developmental stages (Bellezi
et al., 2015).
Implications for the conservation of aquatic and
semiaquatic biodiversity in managed land
Increasing environmental concerns of further conversion
of native habitats to agricultural land, and of competi-
tion between biofuel and food crops for land, led inde-
pendent researchers (e.g., Lapola et al., 2010), NGOs
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
(WWF, 2009) and governmental agencies (Manzatto
et al., 2009) to argue for an expansion of sugarcane over
pastures. While we agree that this is the best conserva-
tion option face to an expansion of the agroindustrial
production of biofuel feedstocks, it is nonetheless
important to acknowledge that large-scale conversion of
pastures into sugarcane fields following prevailing land
management practices is expected to lead to further ter-
restrial habitat homogenization, and increased siltation
and eutrophication of freshwater systems. Expansion of
sugarcane fields over pastures is also expected to lead
to increased environmental contamination with pesti-
cides (Schiesari & Grillitsch, 2011). Assuming conserva-
tion of wetlands and surrounding buffer strips, our data
suggest no substantial effects of further conversion of
pastures into sugarcane fields on amphibian frequency
of occurrence, density, per-unit-area biomass, and alpha
diversity. Likewise, conversion of pastures into sugar-
cane fields appears to have no effect on predator abun-
dance or per-unit-area biomass. However, a more
careful evaluation of the effects of land use on predator
alpha diversity awaits for a species-level analysis, which
is underway.
The observation that 100% and 73% of the water
bodies in sugarcane plantations contained tadpoles and
predatory invertebrates, respectively, and that tadpole
and invertebrate predator densities and per-unit-area
biomass were not affected by land use imply that (i) len-
tic freshwater habitat is limiting for organisms with com-
plex life cycles in this markedly seasonal landscape but
also that (ii) sugarcane fields are permeable to at least
part of the freshwater community. It is important to real-
ize, however, that distribution of biodiversity in sugar-
cane fields was highly uneven: three large (0.310.7 ha)
marshes contained 11 of the 14 species of amphibians
found in sugarcane fields, six of which found exclusively
there, and 82% of all individuals sampled. Correspond-
ingly, they contained all eight families of invertebrate
predators found in sugarcane fields, two of which found
exclusively there, and 94% of all individuals sampled.
Contrary to small ponds, farm ponds, puddles and
ditches, preservation of buffer zones around streams and
wetlands greater than 1 ha are mandated by the Forest
Code, federal legislation regulating land use in every pri-
vate property in the country. Therefore, these water
bodies were at least 30 m from sugarcane fields and, not
surprisingly, then, had less extreme abiotic characteris-
tics when compared with other water bodies embedded
in sugarcane fields, including lower turbidity, conductiv-
ity, and chlorophyll-a and phycocyanin concentrations
(Fig. S4). These larger wetlands very likely act as regional
sources of biodiversity. As an example, multiplying
organismal densities by surface area, in December 2010 a
10.7-ha marsh contained no less than ~34 million tad-
poles [31 million Scinax spp, 2.3 million Dendropsophus
minutus, and 0.5 million Physalaemus cuvieri] and ~3 mil-
lion larval insect predators [1.8 million dragonflies,
0.6 million bugs, 0.5 million beetles]. Part of these indi-
viduals eventually reached metamorphosis and likely
dispersed across the terrestrial environment colonizing
other water bodies in the landscape.
Another evident argument for the importance of lar-
ger wetlands as sources of biodiversity in sugarcane
fields is the observation that large-scale burnings were
until recently the standard preharvesting practice in
sugarcane fields throughout Brazil. Indeed, 43% of the
area in Luis Antonio, including part of the land sur-
rounding sampled water bodies, was burned in the year
preceding our surveys (data provided by CANASAT; Rud-
orff et al., 2010; Aguiar et al., 2011; INPE, 2014). Under
such conditions, buffered streams, streamside marshes,
and permanent and semipermanent wetlands are more
than likely important sources of biodiversity in sugar-
cane landscapes, especially to amphibians, which are at
the same time sensitive to desiccation and dispersal lim-
ited in comparison with dragonflies and other predatory
Whereas our observations reinforce the importance of
the Forest Code for biodiversity conservation, farmers
compliance with these the so-called Permanent Preser-
vation Areas is well below 30% in the most important
sugarcane-producing regions in the country (Sparovek
et al., 2012). Therefore, in a country with relatively
strong environmental legislation but subject to deficient
enforcement, Bonsucro (Better Sugar Cane Initiative)
and any other certification scheme that proposes to fos-
ter the sustainability of the sugarcane sector are helpful.
Bonsucro Standards demand compliance with all rele-
vant applicable laws, including the Forest Code and
those regulating the storage, delivery, and application
rates of vinasse and pesticides. Bonsucro Standards also
prevent greenfield expansion over High Conservation
Value Areas and require development and implementa-
tion of an Environmental Management Plan taking into
account endangered species, habitats, and ecosystems in
the farm. Finally, Bonsucro Standards include applica-
tion of fertilizers according to soil or leaf analysis and
set a moderately restrictive maximum dosage of inor-
ganic fertilizers, pesticides, water consumption and of
tilled land area.
Considering that most native habitats in Central-
South Brazil even if fragmented and in early phases of
regeneration are contained within private properties,
enforcement of current legislation and recognition of
better land management practices by means of certifica-
tion systems are clearly important for the conservation
of biodiversity and, more broadly, the environmental
sustainability of biofuels.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
We thank the Instituto Florestal for granting permission to con-
duct research in Estac
ao Ecol
ogica do Jata
ı and farmers around
Luis Antonio for granting permission to conduct research in
their lands. We thank ICMBio and Instituto Florestal for collec-
tion permits (ICMBio 17559) and FAPESP/BIOEN Bioenergy
Research Program (Young Researcher Award 2008/57939-9),
52321-7, CNPq 563075/2010-4) and CAPES (Ci^
encia sem
Fronteiras 1176/13-7) for funding. We thank Ti~
ao, Antonio Car-
los da Silva, Daniel Din Betin Negri, Mariana Fekete Moutinho,
Bianca Gonc
ßalves dos Santos, Paulo Ilha, Tiago Gabriel Correia,
and Marquinhos for field assistance and Ana Cristina Mach-
ado, Daniel Din Betin Negri, and Ilberto Calado for lab assis-
tance. We thank Edson Montilha de Oliveira for logistical
support, Britta Grillitsch for comments in earlier phases of pro-
ject design, Mathew Leibold and his laboratory for discussion
on patterns in metacommunity structure, Denise Rossa-Feres
for sharing her expertise in tadpole identification, Victor Gio-
vannetti and Kleber Leite for fish identification, and Fernando
Rodrigues da Silva and Thiago Gonc
ßalves-Souza for statistical
Aguiar DA, Rudorff BFT, Silva WF, Adami M, Mello MP (2011) Remote sensing
images in support of environmental protocol: monitoring the sugarcane harvest
in S~
ao Paulo State, Brazil. Remote Sensing,3, 26822703.
Anderson MJ, Ellingsen KE, McArdle BH (2006) Multivariate dispersion as a mea-
sure of beta diversity. Ecology Letters,9, 683693.
Ballester MVR, Martinelli LA, Krusche AV, Victoria RL, Bernardes M, Camargo PB
(1999) Effects of increasing organic matter loading on the dissolved O
, free CO
and respiration rates in the Piracicaba river basin, Brazil. Water Resources Research,
33, 21192129.
Barton K (2012) MuMIn: Multi-model inference. R package version 1.7.11. Available
at: (accessed 30 September 2012)
Baselga A (2010) Partitioning the turnover and nestedness components of beta diver-
sity. Global Ecology and Biogeography,19, 134143.
Baselga A (2012) The relationship between species replacement, dissimilarity
derived from nestedness, and nestedness. Global Ecology and Biogeography,21,
Baselga A, Orme CDL (2012) betapart : an R package for the study of beta diversity.
Methods in Ecology and Evolution,3, 808812.
Becker CG, Fonseca CR, Haddad CFB, Batista RF, Prado PI (2007) Habitat split and
the global decline of amphibians. Science,318, 17751777.
Bellezi L, Ilha P, Schiesari L (2015) Ontogenetic variation in the sensitivity of the
gladiator frog, Hypsiboas faber, to inorganic nitrogen. Copeia,103,1421.
Blanchet FG, Legendre P, Borcard D (2008) Forward selection of explanatory vari-
ables. Ecology,89, 26232632.
Borcard D, Gillet F, Legendre P (2011) Numerical Ecology with R. Springer, New York,
Burnham KP, Anderson DR (2002) Model Selection and Multimodel Inference: A Pratical
Information-Theoretic Approach, 2nd edn. Springer, New York, NY.
Cantarella H, Rossetto R (2010) Fertilizantes para a cana-de-ac
ucar. In: Bioetanol de
ucar. P&D para produtividade e sustentabilidade (eds Cortez LAB), pp.
405422. Blucher, S~
ao Paulo, Brazil.
Clay J (2004) World Agriculture and the Environment. A Commodity-By-Commodity
Guide to Impacts and Practices. World Wildlife Fund, Island Press, Washington DC.
Colwell RK (2013) EstimateS: Statistical estimation of species richness and shared
species from samples. Version 9. User’s Guide and application published at: (accessed 27 March 2013)
Colwell RK, Chao A, Gotelli NJ, Lin SY, Mao CX, Chazdon RL, Longino JT (2012)
Models and estimators linking individual-based and sample-based rarefaction,
extrapolation and comparison of assemblages. Journal of Plant Ecology,5,321.
Corbi JJ, Trivinho-Strixino S, dos Santos A (2008) Environmental evaluation of met-
als in sediments and dragonflies due to sugar cane cultivation in Neotropical
streams. Water, Air, and Soil Pollution,195, 325333.
Dallas T (2014) metacom: an R package for the analysis of metacommunity structure.
Ecography,37, 402405.
Dean W (1997) With Broadax and Firebrand: The Destruction of the Brazilian Atlantic For-
est. University of California Press, Oakland.
Fargione J, Hill J, Tilman D, Polasky S, Hawthorne P (2008) Land clearing and the
biofuel carbon debt. Science,319, 12351236.
Gunkel G, Kosmol J, Sobral M, Rohn H, Montenegro S, Aureliano J (2007) Sug-
arcane industry as a source of water pollution case study on the situation
in Ipojuca River, Pernambuco, Brazil. Water, Air, and Soil Pollution,180, 261
Heyer WR, Donnelly MA, McDiarmid RW, Hayek LAC, Foster MS (1994) Measuring
and Monitoring Biological Diversity Standard Methods for Amphibians. Smithsonian
Institution Press, Washington, DC.
Ilha P, Schiesari L (2014) Lethal and sublethal effects of inorganic nitrogen on gladia-
tor frog tadpoles (Hypsiboas faber, Hylidae). Copeia,2014, 221230.
Immerzeel DJ, Verweij PA, Fvd Hilst, Faaij APC (2014) Biodiversity impacts of bio-
energy crop production: a state-of-the-art review. Global Change Biology Bioenergy,
6, 183209.
INPE (2014) CANASAT: Monitoramento da cana-de-ac
ucar via imagens
de sat
elite. Available at: (accessed 1
November 2014).
International Energy Agency/Organization for Economic Co-operation and Devel-
opment (2014) World Energy Outlook.
Lapola DM, Schaldach R, Alcamo J, Bondeau A, Koch J, Koelking C, Priess JA (2010)
Indirect land-use changes can overcome carbon savings from biofuels in Brazil.
Proceedings of the National Academy of Sciences,107, 33883393.
Legendre P, Gallagher E (2001) Ecologically meaningful transformations for ordina-
tion of species data. Oecologia,129, 271280.
Leibold MA, Mikkelson GM (2002) Coherence, species turnover, and boundary
clumping: elements of meta-community structure. Oikos,97, 237250.
Lewinsohn TM, Prado PI, Jordano P, Bascompte J, Olensen JM (2006) Structure in
plant-animal interaction assemblages. Oikos,113, 174184.
Manzatto CV, Assad ED, Bacca JFM, Zaroni MJ, Pereira SEM (2009) Zoneamento ag-
ogico da cana-de-ac
ucar. Expandir a Producß
ao, Preservar a Vida, Garantir o Fu-
turo. Embrapa Solos, Rio de Janeiro.
Martinelli LA, Filoso S (2008) Expansion of sugarcane ethanol production in Brazil:
environmental and social challenges. Ecological Applications,18, 885898.
Meynard CN, Lavergne S, Boulangeat I, Garraud L, van Es J, Mouquet N, Thuiller
W (2013) Disentangling the drivers of metacommunity structure across spatial
scales. Journal of Biogeography,40, 15601571.
Moutinho M (2013) Ecotoxicidade comparativa dos herbicidas da cana-de-ac
para larvas de anf
ıbios. MSc Thesis. University of S~
ao Paulo.
Oksanen J, Blanchet FG, Kindt R et al. (2012) vegan: Community Ecology Package.
Organization for Economic Co-operation and Development (OECD) (2001) Environ-
mental Indicators for Agriculture. Volume 3. Methods and Results.
Organization for Economic Co-operation and Development (OECD) (2006) Water
and Agriculture: Sustainability, Markets and Policies Conclusions and Recom-
Prado VHM, Silva FR, Dias NYN, Pires JSR, Rossa-Feres DC (2009) Anura, Estac
ogica de Jata
ı, S~
ao Paulo state, southeastern Brazil. Checklist,5, 495502.
Presley SJ, Higgins CL, Willig MR (2010) A comprehensive framework for the evalu-
ation of metacommunity structure. Oikos,119, 908917.
R Core Team (2012) R: A Language and Environment for Statistical Computing. R Foun-
dation for Statistical Computing, Vienna, Austria.
Rudorff BFT, Aguiar DA, Silva WF, Sugawara LM, Adami M, Moreira MA (2010)
Studies on the rapid expansion of sugarcane for ethanol production in S~
ao Paulo
State (Brazil) using Landsat data. Remote Sensing,2, 10571076.
Schiesari L (2006) Pond canopy cover: a resource gradient for anuran larvae. Fresh-
water Biology,51, 412423.
Schiesari L, Grillitsch B (2011) Pesticides meet megadiversity in the expansion of bio-
fuel crops. Frontiers in Ecology and the Environment,4, 215221.
Skelly DK, Werner EE, Cortwright SA (1999) Long-term distributional dynamics of a
Michigan amphibian assemblage. Ecology,80, 23262337.
Sparovek G, Berndes G, Barretto AGOP, Klug ILF (2012) The revision of the Brazil-
ian Forest Act: increased deforestation or a historic step towards balancing agri-
cultural development and nature conservation? Environmental Science and Policy,
UNICA (2008) A industria da cana-de-ac
ucar: etanol, ac
ucar, bioeletricidade. 34 pp.
UNICA (2011) Usina virtual. Available at:
tual.php (accessed 12 October 2014).
UNICA (2014). Available at: (accessed 1 November 2014).
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
Verdade LM, Gheler-Costa C, Penteado M, Dotta G (2012) The impacts of sugarcane
expansion on wildlife in the state of Sa
˜o Paulo, Brazil. Journal of Sustainable Bioe-
nergy Systems,2, 138144.
Wellborn GA, Skelly DK, Werner EE (1996) Mechanisms creating community struc-
ture across a freshwater habitat gradient. Annual Review of Ecology and Systematics,
27, 337363.
Werner EE, Skelly DK, Relyea RA, Yurewicz KL (2007) Amphibian species richness
across environmental gradients. Oikos,116, 16971712.
Wetzel RG (2001) Limnology. Academic Press, San Diego.
World Watch Institute/German Agency for Technical Cooperation (2006) Biofuels
for transportation: global potential and implications for sustainable agriculture
and energy in the 21st Century. 8 pp.
WWF (2009)O impacto do mercado mundial de biocombust
ıveis na expans~
ao da agricultura bra-
sileira e suas consequ^
encias para as mudancßas clim
aticas. WWF, Bras
ılia, DF, Brasil. 72pp.
Zuur A, Ieno EN, Walker N, Saveliev AA, Smith GM (2009) Mixed Effects Models and
Extensions in Ecology with R. Springer, New York, NY.
Zuur AF, Ieno EN, Elphick CS (2010) A protocol for data exploration to avoid com-
mon statistical problems. Methods in Ecology and Evolution,1,314.
Supporting Information
Additional Supporting Information may be found in the online version of this article:
Table S1. Univariate analyses for the effects of land use and survey on 16 environmental variables and 8 community variables
measured in sampled water bodies.
Table S2. Top generalized linear models predicting tadpole species richness on 27 water bodies distributed across a gradient in
land use intensity comprising native habitats, pastures, and sugarcane plantations.
Table S3. Scores from the Principal Component Analyses used to reduce dimensionality in the 18 environmental variables mea-
sured in each water body.
Table S4. Scores from the Redundancy Analysis (RDA) of 25 larval amphibian communities in water bodies distributed across a
gradient of environmental degradation comprising native habitats, pastures and sugarcane plantations.
Figure S1. Positive relationship between tadpole richness and hydroperiod, and tadpole richness and tadpole density on 27 water
bodies distributed across a gradient in land use intensity (native habitats, pastures, and sugarcane plantations).
Figure S2. Biplots of the Principal Component Analyses used to reduce dimensionality in the 18 environmental variables mea-
sured in each water body.
Figure S3. RDA triplot of 25 larval amphibian communities in water bodies distributed across a gradient of environmental degra-
dation comprising native habitats, pastures and sugarcane plantations.
Figure S4. A comparison of representative physico-chemical properties of small and large water bodies in sugarcane fields.
©2015 The Authors. Global Change Biology Bioenergy published by John Wiley & Sons Ltd., GCB Bioenergy, doi: 10.1111/gcbb.12279
... food crops and native habitats. In Brazil, suggestions to solve this dilemma, according to governmental agencies, NGOs and academia, converged to promote the intensification in cattle production to free up space in extensive pastures to expand both food and biofuel production without need to convert additional native habitats (Schiesari & Corrêa, 2016). In each land use, we installed six 4000 L aquatic mesocosms simulating ponds typical of agricultural fields. ...
... Mesocosms received 60 L of inorganic soil, 2 kg of Brachiaria leaf-litter, water from an artesian well, and a plankton inoculum. As in natural waterbodies (Schiesari & Corrêa, 2016), mesocosms were spontaneously colonized by algae, aquatic and semi-aquatic insects, and amphibian larvae. ...
... Unfortunately, the ecosystem stability we documented in our experimental plots despite severe environment change may not manifest in most, conventionally designed crop fields. Sampling surveys in sugarcane-producing landscapes showed that the median number of predatory insect genera in temporary and semi-permanent ponds declined from 9 in natural savannas to 6.5 in pastures and 1 in sugarcane plantations (Schiesari & Corrêa, 2016). Judging by Figure 4, ...
Full-text available
Sustainable intensification may depend on associating precision farming with the harnessing of ecological principles in crop fields, and integrating farms and non‐farmed land in productive landscapes. Small wetlands could play an important role in both pursuits for having high per‐unit‐area rates of element cycling and species richness while deeply penetrating crop fields. However, their potential for ecosystem service provisioning is unlikely to be met if land management intensification promotes ecosystem destabilization in biomass production. We tested the consequences of land use intensification on various dimensions of freshwater ecosystem stability by means of a large‐scale field experiment converting extensive pastures to intensive pastures and sugarcane plantations in Southeastern Brazil. Nested within experimental plots were 4000 L mesocosms simulating ponds and puddles commonly found in productive landscapes. Mesocosms were monitored for basic physico‐chemical parameters, nutrients, pesticides, phytoplankton standing crop, and the spontaneously colonizing biodiversity. Despite severe environmental change, the stability of sugarcane communities was no different from that of extensive and intensive pastures. This occurred because the local extinction of a sensitive top dragonfly predator following the application of vinasse and insecticide was compensated by colonization of a suite of more tolerant invertebrate mesopredators such as beetles and bugs. Community stability tended to increase with biomass asynchrony and species richness, evidencing a portfolio effect of biodiversity. Unfortunately, the species richness necessary to stabilize biomass production is unlikely to be available in many sugarcane fields and several other row crops. Synthesis and applications . Ponds and puddles could be effective centers of irradiation of ecosystem service provisioning in agricultural fields in terms of pest control; nutrient accumulation, cycling and export back to fields; and habitat and stepping stones for freshwater biodiversity. However, the impoverished biodiversity that results from a combination of harsh local conditions and spatial isolation renders pond communities inherently unstable. Given the unlikely, immediate reduction in agrochemical use in much of the intensively managed crop area, a combination of large, protected source wetlands at the margin of fields and small constructed or naturally forming ponds and puddles in plantations could contribute to sustainable intensification.
... However, Neotropical metacommunities in disturbed landscapes generally do not exhibit nested subsets. This is true for bats in Costa Rica (Cisneros et al. 2015) or the Amazon (Martins 2016), as well as for small mammals (de la Sancha et al. 2014;Delciellos et al. 2018) or amphibians (Schiesari and Corrêa 2016) in Atlantic Forest. In contrast, nested structure did manifest for bats in Atlantic Forest (Teixeira 2019; Gomes 2020). ...
... The crucial aspects of scale in a metacommunity context are that the definitions of a site (i.e., community) and the extent of the environmental domain are consistent with the theoretical questions addressed in the analysis, as well as with the explanatory factors and mechanisms invoked to account for empirical patterns. For example, to evaluate effect of landscape structure on metacommunity structure of forest fragments (Cisneros et al. 2015;Schiesari and Corrêa 2016;Delciellos et al. 2018), communities are restricted to forest fragments, explanatory variables are characteristics of fragments and the surrounding landscape, and the spatial extent of the metacommunity is confined to one habitat type to ensure that only variation in landscape structure, and not variation in habitat type, affects the composition of communities. In contrast, to evaluate biogeographical attributes on metacommunity structure in an island system , communities are defined by islands, explanatory variables are island characteristics, and the spatial extent is a group of islands that represent variation in important biogeographical factors (e.g., area, habitat heterogeneity, distance to source populations). ...
A metacommunity perspective provides ecological insight into spatiotemporal dynamics because it explicitly considers the structure and organization of communities along environmental gradients, and seeks to understand the local (e.g., biotic interactions, environmental tolerances, habitat preferences) and regional (e.g., dispersal, habitat fragmentation, landscape structure) processes that generate these patterns. More specifically, a metacommunity is a network of communities or sites that are potentially connected to each other via dispersal of individuals among constituent communities or sites. Local emigration and immigration, when coupled with other spatially explicit ecological processes such as species sorting, habitat filtering, priority effects, or interspecific competition, imbue the network with an emergent structure that corresponds to underlying environmental gradients. High biodiversity, a complex biogeographical history, and a heterogeneous topography associated with orogenic events combine to make the Neotropics an ideal test bed for theories related to ecological gradients and the assembly of communities.Metacommunity structure is an emergent property defined by the relationships among species distributions along particular environmental gradients. These patterns are based on three characteristics of species ranges called elements of metacommunity structure: coherence, range turnover, and range boundary clumping. Coherence (species distributions that are molded by a common gradient) characterizes many structures that have been described in ecology, with the extent and locations of species distributions along this gradient determining metacommunity structure. Nested structures are coherent structures that exhibit less range turnover than expected by chance. In contrast, multiple distinct coherent structures are characterized by more turnover than expected by chance, and these structures are distinguished by range boundaries that are clumped (Clementsian structure), randomly distributed (Gleasonian structure), or hyperdispersed (evenly spaced structure). These idealized structures form the framework of a continuum of possible structures, from those with high species turnover (e.g., Clementsian, Gleasonian) to those with low species turnover (e.g., nested subsets), and from those with coincident range boundaries (i.e., Clementsian) to those with hyper-dispersed range boundaries (i.e., evenly spaced). Although elements of metacommunity structure can identify particular structures based on the distributions of species, supporting (e.g., canonical correspondence analysis, hierarchical partitioning of biodiversity, general linear models), and complementary (e.g., variation partitioning) analyses are required to determine the nature of environmental gradients along which the metacommunity is structured, the number and locations of compartments, or the relative influence of potential structuring mechanisms.Metacommunity approaches have revealed structures associated with many types of environmental gradients in the Neotropics, have done so from small to large spatial scales, and in a hierarchical fashion. Neotropical metacommunities are structured by gradients associated with abiotic variation, biotic interactions, edaphic features, habitat type, disturbance, contamination, land use, or legacies of historical factors. By simultaneously evaluating mechanisms that affect patterns at different time scales (e.g., biogeographical history versus modern land use change), metacommunity approaches have generated and tested hypotheses associated with a wide range of ecological, biogeographical, evolutionary, and conservation questions. Large-scale metacommunity structures are typically associated with historical factors (evolutionary histories, biogeography), whereas smaller scale patterns arise in response to variation in local factors (e.g., habitat type, disturbance history) or anthropogenic activities (e.g., habitat loss and fragmentation, pollution and contamination). Finally, ecological differences between taxa have resulted in taxon-specific structure along the same gradients, confirming that the study of multiple metacommunities in the same system may be required to understand the relative influence of historical and contemporary mechanisms.Two aspects of biodiversity monitoring in the Neotropics can improve the understanding of metacommunity dynamics. First, long-term studies that repeatedly sample at regular intervals to quantify changes in metacommunities along gradients subject to combinations of press (e.g., eutrophication, contamination, pollution, climate change) and pulse (e.g., extreme weather events, logging) disturbances are necessary to understand spatiotemporal dynamics. Second, comparative studies of multiple taxa (e.g., bats, rodents, birds, amphibians, arthropods, trees, fungi, microbes) along the same spatial or environmental gradients are needed to determine which groups are particularly vulnerable to changing environmental conditions, as well as to identify emerging conservation concerns. This is particularly urgent as the tempo and mode of human-induced disturbances associated with the Anthropocene will likely have great impact on the structure and functioning of vital ecological systems in the New World tropics, and beyond.
... Among them, amphibians are highly susceptible to pesticides, especially during their larval stage, when tadpoles have high skin permeability (Yan et al. 2008). Several tadpole species inhabit temporary water bodies formed in flooded soils after prolonged periods of rain (Schiesari and Corrêa 2016). Many of these habitats are inserted or located close to sugarcane crops, which favors contamination by agrochemicals through leaching (Bridges et al. 2004;Castro Berman et al. 2018;Etchegoyen et al. 2017;Mann et al. 2009). ...
... In each plot, an artificial pond (mesocosm) with 8 m long, 1 m wide, and 1 m deep was built (Fig. 1). Mesocosms were constructed to simulate lentic aquatic systems usually formed in Brazilian agricultural lands and grasslands, where amphibians commonly occur during their larval stages (Schiesari and Corrêa 2016). Ponds were lined with a polyethylene sheath to prevent water infiltration and received a ~ 10 cm deep layer of inorganic soil and ~ 9000 L of well water. ...
Full-text available
This study evaluated the effects of environmental contamination caused by pasture intensification and pasture-sugarcane conversion on oxidative stress, biotransformation, esterase enzymes, and development of Scinax fuscovarious and Physalaemus nattereri. Tadpoles were exposed in mesocosms allocated in three treatments: (1) untreated extensive pasture (EP); (2) intensive-pasture conversion (IP) (2,4-D herbicide + fertilizers); and (3) pasture-sugarcane conversion (SC) (fipronil + 2,4-D + fertilizers). After 7 days of exposure, IP reduced catalase (CAT) and increased malondialdehyde (MDA) levels in P. nattereri, while this treatment decreased glucose-6-phosphate dehydrogenase (G6PDH) and CAT activities in S. fuscovarious. SC decreased CAT, G6PDH, and glutathione S-transferase (GST) activities in P. nattereri. In S. fuscovarius, SC reduced G6PDH, acetylcholinesterase (AChE), and carboxylesterase (CbE) activities. MDA was raised in both tadpole species exposed to SC, evidencing oxidative stress. Integrated biomarker responses showed higher scores in both species exposed to SC. Our results warn that management practices currently applied to sugarcane cultivation in Brazil can negatively impact the functional responses of amphibians at natural systems.
... One of these changes is a notable increase in the number of temporary ponds and puddles in interfluves (here, meaning the entire land surface comprised between two streams) (Schiesari et al., 2020;Swartz & Miller, 2021). These ponds work both as the main habitat and as stepping stones for the dispersal of organisms with complex life cycles such as amphibians and aquatic insects with aerial adults, forming complex metacommunity structures (Prado & de Rossa-Feres, 2014;Schiesari & Corrêa, 2016) that are subject to contamination by these very same agrochemicals through direct application, drift and runoff (Carvalho, 2017). ...
Full-text available
Since the Green Revolution, the management of agrochemicals has become a cornerstone of agriculture, allowing it to expand over terrestrial ecosystems previously deemed unsuitable for cultivation. Such expansion altered the connectedness of aquatic habitats to organisms with complex life cycles by increasing the number of temporary ponds in interfluves. Here, we investigate how spatial isolation and intensification in the use of agrochemicals can interact to increase or decrease the permeability of agricultural fields to these organisms by affecting community structure in temporary ponds. We carried out a field experiment that simulates realistic scenarios of agrochemical contamination at different levels of spatial isolation. We constructed 45 1200‐L artificial ponds in a Savanna landscape in Brazil at three distances from a source wetland (30, 120, and 480 m). The ponds were spontaneously colonized by aquatic insects and amphibians and treated with no agrochemicals (control), fertilizers (as in managed pastures) or fertilizers and a single pulse of the insecticide fipronil and the herbicide 2,4‐D (as in sugarcane fields) following realistic dosages and application schedules. Communities treated only with fertilizers generally exhibited a higher abundance of aquatic insects at the peak of the rainy season, suggesting that fertilization caused bottom‐up effects. However, because low spatial isolation and fertilization favoured different predatory taxa, their insect prey eventually became less abundant in less isolated fertilized ponds. Communities treated with fertilizers and pesticides diverged strongly from other treatments after the application of the insecticide, when a decrease in insect abundance indirectly benefitted amphibian populations. In this case, spatial isolation delayed the recovery of the abundance of several herbivore and detritivore insects in isolated ponds. Synthesis and applications . These results have direct consequences for the emergence of insects and amphibians, possibly affecting important ecosystem services. Such consequences can be controlled by avoiding agrochemical contamination at the peak of most taxa reproductive season or through changes in spatial isolation by deliberately constructing wetlands in interfluves. In all cases, the effects of agrochemicals and isolation must be addressed jointly for a successful management of biodiversity in agricultural landscapes.
... However, regarding functional diversity we could detect significant differences only for the functional fil-rot trait, which showed higher densities in citrus crop land use. Aquatic environments within citrus crops have a higher concentration of chlorophyll-a and nutrients than in pastures (Schiesari and Corrêa, 2016;Santos et al., 2021). This may occur once agricultural activities cause nutrients to run off into aquatic environments, which can alter water pH and TP and TN concentrations (Nhiwatiwa et al., 2017;Tromboni et al., 2019). ...
Changes in land use are one of the main impacts on aquatic ecosystems. The conversion of natural areas into agropastoral, such as pasture or monocultures, can affect the limnological characteristics of the water that in turn influence the structure of aquatic communities. But the impact is still unclear, particularly on zooplankton communities. The objective of this study was to evaluate the effect of water parameters from eight reservoirs inserted in an agropastoral landscape on the functional structure of zooplankton. The functional characterization of the zooplankton community was based on four traits: body size, feeding type, habitat type, and trophic group. Functional diversity indices (FRic FEve and FDiv) were estimated and modeled water parameters, using generalized additive mixed models (GAAM). The water parameters investigated were total nitrogen (TN) and total phosphorus (TP), dissolved oxygen (OD), temperature and pH. Moreover, we used RDA to analyze the influence of these environmental variables on the sharing of traits among the sample sites. FRic was high in the reservoirs with low TN concentration and low pH values. FEve was also high in low pH and high TP concentration. FDiv was high with unsharp increases of pH and high concentrations of TN and DO. Our analyses showed pH as a key variable for functional diversity, since it was related to the variation in all indices. Data also pointed out changes in diversity functional with small pH changes. The big and medium size, raptorial-cop and filtration-clad functional traits were positively associated with high concentrations of TN and alkaline pH. The small size and filtration-rot were negatively associated with high concentrations of TN and alkaline pH. Density of filtration-rot was lower in pasture landscapes. In conclusion, our study indicated that pH and TN are key variables for the functional structure of zooplanktonic communities in an agropastoral landscape.
... We measured local and landscape variables known to affect frog species richness and composition in tropical open areas (da Silva et al., 2011b(da Silva et al., , 2012aSchiesari & Corrêa, 2016). ...
Global demand for crops will continue increasing over the next few decades to cover both food and biofuel needs. This demand will put further pressure to expand arable land and replace natural habitats. However, we are only beginning to understand the combined effects of agrochemicals and land use change on tropical freshwater biodiversity. Here, we analyzed how pond‐dwelling anuran larvae respond to pond characteristics, landscape composition, and agrochemical contamination in a sugarcane‐dominated agroecosystem in Brazil. Then, we used an information theoretical approach with generalized linear models to relate species richness and abundance to predictor variables. The variation in tadpole abundance was associated with both agrochemical concentration (e.g., Ametryn, Diuron, and Malathion) and landscape variables (e.g., percentage of forest, percentage of agriculture, and distance to the closest forest). The relationship between species abundance and agrochemicals was species‐specific. For example, the abundance of Scinax fuscovarius and Physalaemus nattereri were negatively associated with Ametryn, and Dendropsophus nanus was negatively associated with Tebuthiuron, while that of Leptodactylus fuscus was positively associated with Malathion. Conversely, species richness was associated with distance to forest fragments and aquatic vegetation heterogeneity, but not agrochemicals. Although we were unable to assign a specific mechanism to the variation in tadpole abundance based on field observations, the lower abundance of three species in ponds with high concentrations of agrochemicals suggest they negatively impact some frog species inhabiting agroecosystems. We recommend conserving ponds near forest fragments, with abundant stratified vegetation, and far from agrochemical runoffs to safeguard more sensitive pond‐breeding species.
... Here, tadpoles from ponds surrounded by smaller crop areas bore superficial resemblance to their reference values, while tadpoles present in ponds with larger crop areas showed larger fins, demonstrating that these organisms, according to changes in land use around the ponds, have the ability to adjust their morphology. While agricultural runoff has strong teratogenic and genotoxic effects on tadpoles (Mann et al. 2009;Silva et al. 2021), the water availability associated with land preparation and crop cycle can be particularly challenging for amphibians in agricultural landscapes (Moreira and Maltchik 2014;Schiesari and Corrêa 2016). These stressors can influence behavioral and morphological changes in the individual traits of organisms. ...
Full-text available
Geographical patterns of phenotypic variation are the outcome of a complex array of evolutionary and environmental factors. Studies on the correlates of the geographic variation in morphological characters can be useful to understand the drivers of phenotypic differences and because intraspecific variation in morphology can impact either local adaptation of any given species or higher-level ecosystem processes. The morphology of larval amphibians responds to multiple factors, including wetland level and surrounding land use. However, understanding the relative influence of such environmental factors on the morphology of tadpoles remains challenging. We used geometric morphometrics to describe the intraspecific morphological variation in tadpoles of Scinax squalirostris (Anura: Hylidae) in freshwater wetlands ranging along the transition between the Pampa and Atlantic Forest biomes in southern Brazil. Specifically, we assessed the relative influence of spatial and environmental (aquatic predators, water chemistry and hydroperiod, crop area) factors to tadpole body size and shape. The size of S. squalirostris tadpoles was affected by spatial factors and pH. Smaller-sized tadpoles were observed in the Pampa–Atlantic Forest transition. Allometry-free changes in the shape of tadpoles were associated both to wetland and upland factors. Tadpoles in larger ponds showed globular bodies and higher ventral fins, while tadpoles in ponds with higher electrical conductivity showed smaller tails. Tadpoles in ponds surrounded by larger crop areas showed longer fins. Our results are useful to understand the importance of wetland and upland effects on the morphology of widely distributed aquatic species, especially given the ongoing scenarios of land cover modification and climate change.
... Our experiment ran from 18 January 2017 to 24 April 2017 and was intended to simulate the dynamics of temporary or semi-permanent ponds during the rainy season, which experience drought every dry season. Such ponds are filled with water during heavy rains and dry out slowly for the next few months (da Silva et al. 2012, Schiesari andCorrêa 2016) being largely recolonized by macroinvertebrates every rainy season. Therefore, our experiment should be interpreted as applying to understand the dynamics of a single such colonization cycle. ...
Full-text available
In the absence of environmental heterogeneity, spatial variation among local communities can be mostly attributed to random dispersal, demographic stochasticity and species interactions causing historical contingency in colonization (e.g. priority effects). The consequences of demographic stochasticity and random dispersal are highly dependent on community size, regional richness and local richness, which, along with historical contingency can be strongly affected by spatial isolation and the presence of predators. Here, we tested whether and how the presence of a generalist predatory fish and spatial isolation can change spatial community variability in artificial ponds by manipulating the presence/absence of fish (redbreast tilapia) at three different distances from a source wetland. We then used null models to compute how much the observed community variability within treatments deviates from the expected from randomly reassembled communities (i.e. beta‐deviation). We found that spatial isolation can have both negative and positive effects on community variability, and their importance may depend on the presence of predatory fish. Negative effects likely occurred because predatory insects cannot always successfully colonize highly isolated ponds, causing the abundance of herbivores and detritivores to increase, consequently increasing community size, which is known to decrease the importance of demographic stochasticity. However, when fish are absent, the more isolated communities exhibit more community variability than is expected from random community assembly. We believe that such an increase in variability is most likely due to an increase in the importance of historical contingency in colonization, generating more distinct communities in isolated ponds. Such effects were absent in the presence of generalist predatory fish, likely because fish prevented any early colonizing taxa from becoming too abundant. Our findings show that the presence of generalist predators can change how community variability responds to spatial isolation.
... In tropical areas, agricultural activities and the consequent application of pesticides are commonly increased during the rainy season, mainly because the molecules are better absorbed in moist soils (Moreira et al., 2012). It is also during the rainy summer that temporary ponds and puddles used by several amphibian species for reproduction and larval development are formed (Schiesari and Corrêa, 2016). In these habitats, pesticide concentrations can be severalfold higher than those found in other aquatic environments, such as lake and rivers, due to the low volume of water (Bridges et al., 2004;Mann et al., 2009). ...
Expansion of sugarcane crops may have contributed to the increased contamination of native habitats in Brazil. Several species of amphibians inhabit ponds formed in flooded farmlands, where pesticide concentrations are usually high. This study evaluated the ecotoxicological effects of the sugarcane pesticides fipronil and 2,4-D, as well as the fertilizer vinasse (isolated and mixed), on physiological responses of Leptodactylus fuscus and Lithobates catesbeianus tadpoles. In situ assays were conducted in mesocosms with concentrations based on the doses recommended by the manufacturer. Vinasse (1.3% dilution) caused 100% tadpoles’ mortality immediately after its application. Fipronil and/or 2,4-D altered antioxidant and biotransformation responses, induced neurotoxicity and changed lipid contents in tadpoles. A multivariate approach indicated that the mixture of pesticides induced most of the sublethal effects in both tadpole species, in addition to the isolated fipronil in L. fuscus. Fipronil alone increased glucose-6-phosphate dehydrogenase (G6PDH) activity, decreased acetylcholinesterase (AChE) and total lipid contents, and altered some individual lipid classes (e.g., free fatty acids and acetone-mobile polar lipids) in L. fuscus. The interaction between fipronil and 2,4-D in this species were more evident for lipid contents, although enzymatic alterations in G6PDH, AChE and glutathione-S-transferase (GST) were also observed. In L. catesbeianus, the mixture of pesticides reduced triglycerides and total lipids, as well as increased GST and decreased AChE activities. The detoxifying enzyme carboxylesterase was reduced by 2,4-D (alone or in mixture) in both species. Isolated pesticides also modulated specific lipid classes, suggesting their disruptive action on energy metabolism of tadpoles. Our study showed that fipronil, 2,4-D, and vinasse, individually or mixed, can be harmful to amphibians during their larval phase, causing mortality or impairing their functional responses.
Full-text available
Here, we summarize examples of significant advances in amphibian research supported by the São Paulo Research Foundation (FAPESP), focusing on recent discoveries in the fields of community ecology, habitat change, infection diseases, and multipurpose DNA sequencing. We demonstrated that FAPESP has been fundamental not only by directly funding research projects and scholarships, but also through its science training policy, fostering international collaborations with world-class research institutions, improving and consolidating new lines of research that often depended on a synergetic combination of different knowledge and complex tools. We emphasized that future studies will continue to focus on basic questions, such as description of new species, as well as taxonomic and systematic corrections. Furthermore, we also expect that there will be a strong integration among different disciplines using novel bioinformatics tools and modeling approaches, such as machine learning. These new approaches will be critical to further develop our understanding of foundational questions of amphibian life-history trait variation, disease transmission, community assembly, biogeography, and population forecasts under different global change scenarios such as agricultural expansion, agrochemical use, habitat loss, and climate change.
Full-text available
Full-text available
In a biosphere transformed by human activities, understanding the relationship between age, developmental stage, or body mass and toxicological sensitivity is a relevant endeavor because the environmental release of some quantitatively important contaminants follows predictable seasonal patterns, as do wildlife demographic events. We tested the hypothesis that the sensitivity of the gladiator frog, Hypsiboas faber, to components and/or dissolution products of fertilizers varies predictably through ontogeny by sequentially exposing eggs, embryos, and larvae to ammonium nitrate, sodium nitrate, sodium nitrite, and ammonium sulfate. Nitrate was the least toxic nitrogen ion regardless of stage; the most deadly compound was ammonium or nitrite, depending on stage. Overall, embryos were more sensitive to inorganic nitrogen than more advanced developmental stages, indicating that the jelly capsule envelopes that surround the egg do not provide effective protection against contaminants. However, for most compounds tested, the relationship between lethal concentrations (LCx), no-observed effect concentrations (NOEC), and lowest observed effect concentrations (LOEC) and developmental stage or mass was not monotonic. Whereas for sodium nitrate sensitivity declined asymptotically with developmental stage, for ammonium nitrate, sodium nitrite, and ammonium sulfate, sensitivity declined through early development and then increased again at Gosner stage 25. Together with previously published literature, this finding indicates no single and predictable trend in sensitivity through development, even if we restrict our analysis to a relatively narrow range of organisms and contaminants. Such a relationship creates complex scenarios of risk as a function of the phenology of application of fertilizers and of the temporal variation in environmental factors that modulate nitrogen speciation and toxicity in and near agricultural fields.
Tools for performing model selection and model averaging. Automated model selection through subsetting the maximum model, with optional constraints for model inclusion. Model parameter and prediction averaging based on model weights derived from information criteria (AICc and alike) or custom model weighting schemes. [Please do not request the full text - it is an R package. The up-to-date manual is available from CRAN].
Limitations of linear regression applied on ecological data. - Things are not always linear additive modelling. - Dealing with hetergeneity. - Mixed modelling for nested data. - Violation of independence - temporal data. - Violation of independence spatial data. - Generalised linear modelling and generalised additive modelling. - Generalised estimation equations. - GLMM and GAMM. - Estimating trends for Antarctic birds in relation to climate change. - Large-scale impacts of land-use change in a Scottish farming catchment. - Negative binomial GAM and GAMM to analyse amphibian road killings. - Additive mixed modelling applied on deep-sea plagic bioluminescent organisms. - Additive mixed modelling applied on phyoplankton time series data. - Mixed modelling applied on American Fouldbrood affecting honey bees larvae. - Three-way nested data for age determination techniques applied to small cetaceans. - GLMM applied on the spatial distribution of koalas in a fragmented landscape. - GEE and GLMM applied on binomial Badger activity data.
Metacommunity theory is an extension of metapopulation theory with the goal of understanding how ecological communities vary through space and time. One off-shoot of metacommunity theory deals with understanding how community structure varies along biotic or environmental gradients. The Elements of Metacommunity Structure framework is a three-tiered analysis of metacommunity structure that enables the user to identify metacommunity properties that arise in site-by-species incidence matrices. These properties can then be related to underlying variables that influence species distributions. The EMS framework is now implemented in metacom, an open source R package that allows for the analysis and plotting of metacommunities.