Plant Ecology & Diversity

ISSN: 1755-0874 (Print) 1755-1668 (Online) Journal homepage: http://www.tandfonline.com/loi/tped20

Climate and fragmentation affect forest structure

at the southern border of Amazonia

Simone Matias Reis, Beatriz Schwantes Marimon, Ben Hur Marimon Junior,
Paulo S. Morandi, Edmar Almeida de Oliveira, Fernando Elias, Eder Carvalho
das Neves, Bianca de Oliveira, Denis da Silva Nogueira, Ricardo Keichi
Umetsu, Ted R. Feldpausch & Oliver L. Phillips

To cite this article: Simone Matias Reis, Beatriz Schwantes Marimon, Ben Hur Marimon Junior,
Paulo S. Morandi, Edmar Almeida de Oliveira, Fernando Elias, Eder Carvalho das Neves,
Bianca de Oliveira, Denis da Silva Nogueira, Ricardo Keichi Umetsu, Ted R. Feldpausch & Oliver
L. Phillips (2018): Climate and fragmentation affect forest structure at the southern border of
Amazonia, Plant Ecology & Diversity, DOI: 10.1080/17550874.2018.1455230

To link to this article: https://doi.org/10.1080/17550874.2018.1455230

Published online: 25 Apr 2018.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tped20
Plant Ecology & Diversity, 2018
https://doi.org/10.1080/17550874.2018.1455230

Climate and fragmentation affect forest structure at the southern border of Amazonia
Simone Matias Reisa,b, Beatriz Schwantes Marimonb,c, Ben Hur Marimon Juniorb,c, Paulo S. Morandia,b,
Edmar Almeida de Oliveirab, Fernando Eliasc, Eder Carvalho das Nevesc, Bianca de Oliveirac, Denis da Silva Nogueirac,
Ricardo Keichi Umetsuc, Ted R. Feldpauschc,d and Oliver L. Phillipsa
a
School of Geography, University of Leeds, Leeds, UK; bPrograma de Pós-Graduação da Rede de Biodiversidade e Biotecnologia da
Amazônia Legal (BIONORTE), Universidade do Estado de Mato Grosso–Universidade Federal do Amazonas, Nova Xavantina, Brazil;
c
Programa de Pós-graduação em Ecologia e Conservação, Universidade do Estado de Mato Grosso, Nova Xavantina, Brazil;
d
Geography, College of Life and Environmental Sciences, University of Exeter, Exeter, UK
(Received 27 December 2016; accepted 18 March 2018)

Background: The remaining forests in the extensive contact zone between southern Amazonia (seasonal rain forest) and the
Cerrado (savanna) biomes are at risk due to intense land-use and climate change.
Aims: To explore the vulnerability of these transitional forests to changes in land use and climate, we evaluated the effects of
fragmentation and climatic variables on forest structure.
Methods: We measured the diameter and height of 14,185 trees with diameter ≥10 cm at 24 forest plots distributed over an
area of 25,000 km2. For each plot, we obtained data on contemporary fragmentation and climatic variables.
Results: Forest structure variables (height, diameter, height:diameter allometry, biomass) varied significantly both within
and among plots. The height, H:D and biomass of trees were positively correlated with annual precipitation and fragment
area.
Conclusions: The association between forest structure and precipitation indicates that these forests plots are likely to be
vulnerable to dry season intensification anticipated for the southern edge of the Amazon. Additionally, the reduction in the
fragment area may contribute to reductions in forest biomass and tree height, and consequently ecosystem carbon stocks.
Given the likely susceptibility of these forests, urgent conservation action is needed to prevent further habitat degradation.
Keywords: allometry; Amazon arc of deforestation; biomass; climate change; habitat fragmentation; precipitation; stem
diameter; tree height; transition zone

Introduction these climatic changes may be exacerbated by changes in

Across the Earth’s biomes, environmental conditions are land use (Aragão 2012; Silvério et al. 2015). Finally,
expected to be more variable close to the edges than in the research elsewhere in Amazonia clearly has indicated
core area of each biome, posing potentially ecological and that the structure of the tropical forest vegetation is
evolutionary challenges to biota towards their biogeogra- affected by both climate change (e.g. Phillips et al. 2010;
phical edges (Safriel et al. 1994; Kark and van Rensburg Feldpausch et al. 2016) and fragmentation of habitats (e.g.
2006; Kark et al. 2008). This may be particularly the case Laurance et al. 1997, 2000; Laurance 2004).
in regions subject to rapid environmental change, of which Yet few studies have evaluated structural variation
perhaps the most extreme example are the forests of the among the forests in the southern border region of the
southern edge of the Amazon rain forest biome, an area Amazon forest biome and its covariation with climate
affected by high deforestation rates and subject to signifi- and landscape factors. Exceptions include one analysis of
cant recent and forecast climate change. Thus, here the the effects of the interaction between droughts and wild-
advance of the agricultural frontier has already resulted in fires on tree mortality at one experimental site (Brando
converting most forest to pasture and cropland, increas- et al. 2014), and a landscape study which showed that
ingly fragmenting the landscape over the last few decades habitat fragmentation, combined with droughts, increased
(Alencar et al. 2004, 2015; Nogueira et al. 2008). The the susceptibility of the forests to fire (Alencar et al. 2015).
remaining forests are subject to recent climate change, We are not aware of a single study that has evaluated the
including lengthening of the dry season and increasing effects of habitat fragmentation and different climate vari-
incidence of strong droughts (Marengo et al. 2011; Gloor ables across the region’s forests using direct, on-the-
et al. 2015; Feldpausch et al. 2016), trends which are ground measurement of vegetation structural variables
expected to intensify further (e.g. Boisier et al. 2015). such as tree diameter, height, and biomass.
The land surface temperature has been rising steadily Habitat fragmentation, by decreasing fragment size
recently, especially in the south and east of the Amazon and increasing forest edges and numbers of fragments,
region (Jiménez-Muñoz et al. 2013), and the effects of may modify the forest structure in the remaining fragments

Corresponding author. Email: [email protected]

© 2018 Botanical Society of Scotland and Taylor & Francis

2 S. M. Reis et al.

(Fahrig 2003; Haddad et al. 2015). For example, fragment of Amazonia with the neighbouring Cerrado (savanna)
edges are subject to a greater incidence of insolation and biome, to test hypotheses related to the variation in the forest
increased velocity of winds, resulting in higher tempera- structure and the factors that determine this variation. We
tures and a drier microclimate than the forest interior addressed two questions. First, does habitat fragmentation
(D’Angelo et al. 2004; Laurance 2004; Haddad et al. affect the forest structure? We expected that forest cover loss
2015), which increases tree mortality rates, principally and forest plots present in smaller fragments and/or nearer
for larger trees (Laurance et al. 2000; Laurance 2004). the edge would have trees with lower height and smaller
The death of bigger trees reduces total biomass, height, diameter stems, or with smaller height:diameter (H:D) allo-
mean diameter and basal area, especially in the smaller metric relationships and reduced biomass, since work else-
fragments and the areas closest to the forest edge, although where has shown mortality rates are greater in smaller, more
with some mortality effects also propagating a few hun- edge-affected fragments, especially for bigger trees (e.g.
dred metres into the forest (Laurance 2004; Haddad et al. Laurance et al. 1997, 1998, 2000; Laurance 2004;
2015; Rocha-Santos et al. 2016). Recently, it has even Chaplin-Kramer et al. 2015). Second, how does the forest
been suggested, based on interpretation of pantropical structure vary in relation to the climate? We expected that
satellite imagery, that in tropical forests the negative the height and the diameter of stems, the H:D ratio, and
effects on standing biomass and forest structure penetrate biomass were all greater in forest plots that have greater
as much as 1.5 km into forests (Chaplin-Kramer et al. precipitation, and consequently less deficit water, since the
2015). greater water availability favours the height growth of the
In addition to landscape-scale factors, regional climate trees, accumulating more biomass (e.g. Feldpausch et al.
is related to variation in the forest structure (e.g. Banin 2011; Pan et al. 2013; Chave et al. 2014).
et al. 2015). For example, where precipitation and tem-
perature are higher, forests generally have taller trees that
accumulate more biomass (Koch et al. 2004; Way and Materials and methods
Oren 2010; Feldpausch et al. 2011; Pan et al. 2013; Study area
Chave et al. 2014). However, in the very warmest forests
We studied 24 forest plots distributed in the so-called
the forest structural responses are unclear. There is some
“arc of deforestation” (Nogueira et al. 2008) over an
evidence that here plants may photosynthesise less and
area of ca. 25,000 km2 (Figure 1 and Table 1). The
expend more energy on respiration, so potentially accu-
regional climate is of the Aw (tropical with dry winters)
mulating less biomass (Lloyd and Farquhar 2008; Lewis
and Am (tropical monsoon) types in the Köppen classi-
et al. 2013). However, the temperature sensitivity of key
fication system (Alvares et al. 2013), and originally
respiration processes appears to decline in warmer envir-
supported evergreen or semi-evergreen forest vegetation
onments (Atkin et al. 2015; Heskel et al. 2016), rather than
in all cases. Mean annual precipitation and temperature
increasing exponentially as simple Q10 formulations in
range from 1511 to 2353 mm and from 24.1 to 27.3 °C,
earlier global vegetation models suggested (Cox et al.
respectively (Table 1).
2000), suggesting that the overall sensitivity of biomass
stocks to high temperatures might be weaker than many
models indicated.
Forest fragments
Extreme drought events may alter the forest struc-
ture. Drought causes mortality, principally in the big- The largest and best preserved regional fragments of mature
ger trees, which are more susceptible to damage in forests were selected for the study, using Google Earth
their vascular system (Phillips et al. 2010; Bennett imagery in order to capture regional variation in floristics
et al. 2015; Rowland et al. 2015; Feldpausch et al. and physiognomy, and with at least three plots for each
2016). During drought events, tropical trees may also forest type. All forest fragments are surrounded by exten-
grow less (e.g. Worbes 1999; Doughty et al. 2015), sive cattle-ranching or soybean fields. The fragments sur-
and if droughts are prolonged or repeated forests even- veyed varied in size from 5 to 45,459 ha (Table 1).
tually accumulate less biomass (Rowland et al. 2015;
Feldpausch et al. 2016).
In the context of regional land-use and climatic changes Forest structure
in southern Amazonia, and the projected high regional cli- In each fragment we established an inventory plot of 1 ha,
mate sensitivity to global warming (IPCC 2015), it is there- which was subdivided into 25 contiguous subplots of 20 m
fore extremely important to understand how the forest × 20 m. The forest plots were established between 2008
structure is affect by abiotic factors. It may for example and 2016 within the private properties and in conservation
help to improve the conservation measures to protect the units; locations varied between 1 and 5440 m from the
remaining forest fragments. In this study, we evaluated nearest edge of the fragment. Six plots were seasonally
whether, and to what extent, climatic factors and fragmenta- flooded (Table 1) and occasionally affected by fire; the
tion determine variation in the forest structure of the south- others have no recent record of fire and were either on
ern Amazon border. We assembled data from permanent anthropogenic black earth (terra preta de índio), open rain
plots established across the region close to the natural border forests, seasonal evergreen forests, or seasonal semi-
 Climate and fragmentation affect forest structure 3

Figure 1. Location of the forests sampled in the southern Amazon border, between eastern and northern Mato Grosso and southern Pará,
Brazil, showing the approximate biome boundaries based in IBGE (2004). The classification of forest and no forest was based on the
PRODES (Amazon Deforestation Monitoring Project) (INPE 2017). All plots sampled lie within mature, evergreen or semi-evergreen
forest fragments.

deciduous forests (Table 1). For this study, we used the Habitat fragmentation
latest available censuses between 2013 and 2016. To evaluate the effect of habitat fragmentation on forest
We identified and tagged all the woody individuals structure, we measured distance from each plot to the forest
with a diameter at breast height (1.3 m) of ≥10 cm, for a edge, the size of each fragment and the forest cover in
total of 14,185 (range = 338–1599; standard devia- surrounding landscapes. Whenever possible we measured
tion = 31) trees and at least 410 (range = 9–135; the distance to the nearest edge in the field. When this was
standard deviation = 256) taxa identified to species not possible, we estimated this value using Google Earth,
level. We identified species in the field or by compar- which provided a spatial resolution of approximately 20–
ison of collections with herbarium (NX, UFMT, UB and 30 m depending on available imagery, and based on our own
IAN) material of known identity, and with the help of detailed knowledge, having explored the local context of
specialists. After identification, the material was incor- each plot on foot. In our definition of forest habitat edge,
porated into Herbarium NX Nova Xavantina, Mato we included all other vegetation and land-use such as planta-
Grosso (Coleção Zoobotânica James Alexander Ratter). tions, pastures, and roads at least 25 m wide, as well as
We determined the classification of families based on natural grasslands in the six floodplain forests.
APG III (Angiosperm Phylogeny Group 2009) and We calculated the area of the fragment where each plot
reviewed and updated the nomenclature of the taxa was located using Google Earth and ZONUM software
using the Lista de Espécies da Flora do Brasil (2016) (http://zonums.com/online/kmlArea/). These edge and
(http://floradobrasil.jbrj.gov.br/2015). fragment data were collected at the closest possible date
We measured the diameter of each tree following stan- to the field sampling and in no case were they collected
dard protocols of the RAINFOR network (http://www.rain more than 2 years after the last forest census.
for.org/). We measured the total height using a Leica We calculated the percentage of forest cover surrounding
DISTO laser measurement device. Data were deposited each plot using buffers of radius size of 1000 m (314 ha),
in the ForestPlots.net forest monitoring database (Lopez- following recommendations of Rocha-Santos et al. (2016).
Gonzalez et al. 2011). For this we used the land-based metrics in the Fragstats
4 S. M. Reis et al.

Table 1. Characteristics of plots sampled in different tropical forest ecosystems at the southern Amazon border.

Plot Code Forest plot code Geographical coordinate Local AF (ha) DE (m) Prec (mm) Temp (°C) TB (Mg)

FEP-01 FLO-01 −12.8S and −51.9W PP 870 1,030 1613 25.5 111.1
FEP-02 FLO-02 −12.8S and −51.9W PP 2,035 1,000 1621 25.6 144.7
FEP-03 TAN-02 −13.1S and −52.4W PP 8,432 990 1625 24.9 143.5
FEP-04 TAN-03 −12.8S and −52.3W PP 16,901 520 1679 25.1 127.4
FEP-05 TAN-04 −12.9S and −52.4W PP 16,901 329 1662 25 138.3
FEP-06 FRP-01 −11.5S and −51.5W PP 45,459 3,600 1634 26.9 135.1
FEP-07 POA-01 −11.0S and −52.2W PP 9,789 1,180 1772 26.1 140.1
FES-01 VCR-02 −14.8S and −52.2W PP 4,968 1,350 1511 25.2 196.8
FES-02 GAU-02 −13.4S and −53.3W PP 3,499 160 1701 24.1 91.7
FES-03 SAT-01 −9.8S and −50.5W PP 17,624 90 1821 26.7 121.8
FES-04 SAA-01 −9.8S and −50.4W PP 13,039 860 1815 26.8 187.7
FES-05 SAA-02 −9.6S and −50.4W PP 15,680 2,980 1778 26.6 166.3
FOA-01 SIP-01 −11.4S and −55.3W PP 12,066 900 1848 25.1 79.2
FOA-02 ALF-01 −9.6S and −55.9W CU 17,628 5,440 2350 25.5 98.8
FOA-03 ALF-02 −9.6S and −55.9W CU 17,628 5,410 2353 25.6 160.5
FSI-01 PEA-01 −12.1S and −50.8W CU 21 1 1631 27.3 133.7
FSI-02 PEA-02 −12.3S and −50.7W CU 378 1 1637 27.2 154.7
FSI-03 PEA-03 −12.4S and −50.9W CU 164 1 1621 27.1 131.4
FSI-04 PEA-04 −12.4S and −50.7W CU 605 1 1637 27.1 210.4
FSI-05 PEA-07 −12.5S and −50.9W CU 5 1 1621 27.1 226.8
FSI-06 PEA-08 −12.5S and −50.7W CU 8 1 1632 27 222.5
FTP-01 GAU-04 −13.1S and −53.3W PP 234 150 1795 24.7 145.8
FTP-02 GAU-05 −13.0S and −52.9W PP 29,560 2,720 1757 24.9 250.2
FTP-03 GAU-06 −13.3S and −53.4W PP 85 80 1729 24.7 176.9

FA: fragment area; DE: distance to the forest edge; Prec: total mean annual precipitation; Temp: mean annual temperature; TB: total above-ground biomass
per hectare; PP: private properties; and CU: conservation unit. In this study, we used codes (“Plot code”) to represent the different types of vegetation. FEP:
floresta estacional perenifólia (seasonal evergreen forest); FTP: floresta estacional perenifólia em terra preta de índio (seasonal evergreen forest on
anthropogenic black earth); FES: floresta estacional semidecidual (seasonal semi-deciduous forest); FOA: floresta ombrófila aberta (open rainforest); and
FSI: floresta sazonalmente inundável (seasonally flooded forest). Equivalent forest plot codes are given to indicate equivalency to those codes used in the
ForestPlots.net database (Lopez-Gonzalez et al. 2011) where the data have been deposited.

software, that computes descriptors of forest patch and land- weighted Lorey’s height values, based on basal area per
scape attributes (McGarigal and Cushman 2002). subplot, using the equation
X X
ABi
Hi = ABi
Climate variables
To evaluate the climate effect on the forest structure, we where ABi is the basal area of an individual and Hi is its
obtained data on 19 bioclimatic variables (Table S1) from height (e.g. Saatchi et al. 2011). To evaluate the H:D
the WorldClim 1.4 database, with a horizontal resolution of relationship, independently of disturbance, such as the
ca. 1 km (Hijmans et al. 2005). We also used data from the damage caused by recently opened clearings, we excluded
Tropical Rainfall Monitoring Mission (TRMM) (NASA from the analyses all trees with broken stems or those with
2012) to derive the mean of the annual maximum clima- more than 50% of the crown broken off.
tological water deficit (MCWD) (Aragão et al. 2007) We also calculated the mean, median, and total biomass
between January 1999 and December 2011, including the of trees per plot. We estimated the biomass (B) based on the
droughts of 2005, 2007 and 2010 (Figure S2). To estimate Pantropical model revised by Chave et al. (2014), which is
this, we first calculated MCWD for each year, and then derived from the equation in Chave et al. (2005), that is,
took the mean of all years. MCWD was defined as the
most negative value of climatological water deficit (pre-  0:976
B ¼ 0:0673 x ρD2 H
cipitation lower than evapotranspiration) among all the
months in each year.
where D is the diameter in cm, H is the total height of
the tree in m, and ρ is the density of the wood. We
obtained wood density values from the ForestPlots data-
Data analysis base (https://www.ForestPlots.net/). We chose this equa-
In each plot, we calculated the minimum, maximum, med- tion to calculate the biomass because it is the most robust
ian, and 95 percentile of tree diameter (D), height (H) and approach, given that it takes into consideration the dia-
their allometric (H:D) relationship. We also calculated the meter and height of each tree.
 Climate and fragmentation affect forest structure 5

We developed a correlation matrix of the Kendall’s tau (FES-01–02-05) (Figure 2 and Table S6) and like the
values of the environmental and forest structure variables other 11 forest plots (FEP-01–07; FES-03–04 and FTP-
mentioned above (Table S3). Multiple variables share 02–03). The H:D ratio varied in a similar fashion to tree
similar source data, leading to high correlation amongst height, with the lowest ratios (i.e. the lowest heights for a
them, so we excluded those with greatest correlations given diameter) being recorded in two of the seasonally
(r > 0.7) to avoid repetition of largely redundant forest flooded forest plots (FSI-05 and FSI-06). Tree diameter
structure and climate variables (Tables S3 and S4). For all and biomass did not vary systematically among the plots,
variables, the maximum values and the 95 percentiles were except for FSI-03, which had lower diameter and biomass
highly correlated; we included only the 95 percentile in than the most of others plots (Figure 2).
order to avoid the influence of outliers. Finally, we
excluded predictor variables that correlated poorly
(r < 0.1) with the vegetation descriptors (Tables S3 Relationship between forest structure, fragmentation and
and S4). climate variables
To verify possible differences among all forest plots in The structural variables were associated with the precipita-
the structural variables (95 percentiles of the D, H and H: tion and with fragment area and distance from the edge
D, and mean B), we applied the Kruskal-Wallis analysis of (Figure 3 and Table 2). Tree height, allometry (H:D) and
variance with the Dunnett post hoc test and a Bonferroni biomass all correlated positively with precipitation and
correction (Zar 2010). fragment area (Figure 3). Tree height also correlated with
We evaluated the influence of habitat fragmentation the MCWD (Figure 3). Tree diameter did not correlate
and climatic variables on forest structure using simple with any of the variables. Additionally, the precipitation
correlation and Generalised Linear Models (GLM). We and MCWD correlated positively with the fragment area
also included in the models the forest type for each forest (P < 0.05; Kendal’s τ = 0.44 and 0.60, respectively).
plot. Simple correlation showed that, six seasonally Based on the best GLM models for each forest struc-
flooded plots and two plots on anthropogenic black earth ture variable, forest type and precipitation were most
were unduly influential, with extreme structure and cov- related to tree height (Table 2). Forest type was also a
arying extreme climatic and fragmentation conditions. To strongly related to tree diameter and biomass. Annual
avoid these outliers driving the regional results we mean precipitation and distance from the edge were impor-
excluded them from the GLM and correlation analyses tant factors for mean plot H:D (Table 2). The percentage
described above. of forest cover around each plot was not selected in the
To build the GLM, we first standardised the data and best models and was not correlated with any forest struc-
removed the collinearities on the basis of Variance ture variables. All plots presented more than 50% forest
Inflation Factors (VIFs) of less than 10 (Quinn and cover in surrounding landscapes.
Keough 2002). We conducted model selection using the Precipitation and MCWD were not selected in the
Akaike’s Information Criterion (AIC), with a model con- same model, indicating that each had similar (but inverse)
sidered to be the best if it had the lowest AIC value effects on forest structure. Thus, all structural parameters
(Barton 2016). To access the spatial autocorrelation in affected positively by precipitation (Table 2) are affected
the residuals for each model we used Moran’s I. Here, negatively by moisture stress (MCWD) (Table S7).
no spatial dependence was detected among plots, indicat-
ing that the data were not spatially structured (Figure S5).
Thus, we considered the plots as independent samples in Discussion
our subsequent analyses. Our results show that the forests of the southern border zone
We conducted the analyses using SAM 4.0 pro- of Amazonia vary remarkably in their structure, principally
gramme (Rangel et al. 2010) and R 2.15.1 (R Core in terms of their tree height and tree height:diameter ratio.
Team 2012). The applied R packages were vegan Most of the structural variation in these forests was statis-
(Oksanen et al. 2016), spdep (Bivand et al. 2013), tically related to fragment area and precipitation, supporting
spacemakeR (Dray 2013), MuMIn (Barton 2016) and our overall expectations and largely consistent with our
packfor (Dray et al. 2016). We adopted a 5% significance hypotheses. Here, we briefly first discuss this overall varia-
level for all analyses and used 999 randomisations for the bility and its potential ultimate drivers, before proceeding to
permutation methods. discuss the results in more detail.

Structural variability of the forests of the southern Amazon

Results border zone
Forest structure Our general expectation was that climatic variation in the
In general, the three open rainforest plots (FOA-01–03), a region would be a fundamental determinant of the varia-
forest plot on anthropogenic black earth (FTP-01), were bility in forest structure here, principally because drought
significantly taller than the six seasonally flooded forest events and seasonality may be more intense at the southern
plots (FSI-01–06), three seasonal semi-deciduous forest border in relation to the core area of the Amazonas basin
6 S. M. Reis et al.

Figure 2. Variation in the vertical structure of forests at the southern Amazon border. Box-plots show subplot-level values in each
location, statistical comparisons are made for among-forest analyses based on the non-parametric Kruskal–Wallis test (H). The
complementary pair-wise analysis of forest structure is provided in Table S7. = FTP (seasonal evergreen forest on anthropogenic
black earth), = FOA (open rainforest), = FEP (seasonal evergreen forest), = FES (seasonal semi-deciduous forest), = FSI
(seasonally flooded forest).

with evergreen non-seasonal rain forests (Lewis et al. increase tree mortality, as observed in a recent study in
2011). In particular, water deficit may kill large trees these forest plots (Maracahipes et al. 2014). It therefore
(McIntyre et al. 2015), taller trees tend to be most affected appears likely that the combined effects of reduced frag-
by these conditions (Rowland et al. 2015). As these trees ment area and precipitation and higher temperatures,
die and break-up or fall, large clearings are opened, together with fire and its potential interactions with
favouring the establishment of species of different ecolo- droughts (Brando et al. 2014), contribute to forest struc-
gical groups (Lawton and Putz 1988). The frequent for- ture here.
mation of clearings in these hyperdynamic transitional
forests, as documented by Marimon et al. (2014), may
thus also contribute to the structural variability found Response of the forest structure to the fragmentation and
here. Finally, the forests of the southern border of the climate variables
Amazon are located within a mosaic of vegetation types Temperature appears to be an important factor determin-
with many species typical of the adjacent biomes (Ratter ing the height of the trees worldwide, including poten-
et al. 1973), which may have a direct influence on the tially in tropical forests (Koch et al. 2004; Way and
structural diversity of these forests. Oren 2010; Feldpausch et al. 2011; Lines et al. 2012;
Pan et al. 2013), but here the absence of a clear statis-
tical relationship between structure and temperature
Seasonally flooded forest plots (P > 0.05, Kendal’s τ = 0.31) suggests it is not critical
The lowest height and H:D allometric ratio in the season- at the southern Amazon transition zone. Rather, in our
ally flooded forest plots may be explained by their smaller study the greater forest heights, H:D ratio and biomass
fragment size and proximity to edges. These factors as that were observed with increasing precipitation suggest
well as higher temperatures and lower precipitation water supply is the dominant climate control on forest
(Table 1) may intensify the fire effects. Fires in the wider structure, and is consistent with some work elsewhere in
grassland matrix can penetrate into forest fragments and the tropics (e.g. Álvarez-Dávila et al. 2017), given
 Climate and fragmentation affect forest structure 7

Figure 3. Significant (P ≤ 0.05) relationships between forest structure and climatic and fragmentation variables of the southern Amazon
border forest plots. H95 = height 95 percentile, A95 = allometric ratio (H:D) 95 percentile, MB = mean biomass (Mg), FA = fragment
area (ha), PrecWM = precipitation of wettest month (mm), MCWD = maximum climatological water deficit (mm).

especially that tropical plants tend to grow faster and The negative correlation between the cumulative water
taller as water is more available (Givnish et al. 2014; deficit and tree height may be related to the mortality of the
Vlam et al. 2014). In addition to apparent effects of largest individuals during extreme drought events (Phillips
annual rainfall, we also found that climatological water et al. 2010). Such droughts have been directly observed in
deficit was associated with reduced investment by the the study region in 2005, 2007, and 2010 (e.g. Brando et al.
trees in height growth, consistent with the hypothesis 2014), and these have indeed tended to kill larger trees
that tree height is constrained by the availability of (Phillips et al. 2010; Feldpausch et al. 2016), as is often the
water (Ryan et al. 2004; Givnish et al. 2014). A sig- case with droughts in other tropical forests (Bennett et al.
nificant positive correlation was also found between 2015). In Amazonia, recent strong droughts appear also to
precipitation and tree height along a precipitation gra- be a major cause of the recent basin-wide increase in tree
dient in Australia, which Givnish et al. (2014) related to mortality rates (Phillips et al. 2009; Brienen et al. 2015). In
the increase in leaf area and rates of photosynthesis with the near future, more frequent extreme droughts, especially if
increasing precipitation. combined with warming of the Amazon region and thermal
8 S. M. Reis et al.

Table 2. The relationship between environmental variables and can reduce the density and diversity of species (Mantyka-
forest structure, using generalised linear models, of the southern Pringle et al. 2012). Such effects can also increase the sus-
Amazonia forests, Brazil.
ceptibility of fragmented forest structure and their biota to
Factors Estimate Standard error t P fire (Laurance and Williamson 2001; Laurance 2004). In the
southern Amazon region, these different effects are all likely
Height 95 percentile to be relevant, but clearly further analysis is needed, includ-
Intercept −0.276 0.109 −2.531 0.003
ing long-term monitoring evaluation of the climatic and
FES −0.008 0.161 −0.050 0.961
FOA 1.392 0.328 4.249 0.001 dynamic processes in these forests.
PrecWM 0.431 0.140 3.082 0.010
Diameter 95 percentile
Intercept −0.356 0.290 −1.228 0.243 Conclusions
FES 0.039 0.445 0.089 0.931
FOA 1.715 0.530 3.237 0.007 Our analysis across different locations, spanning a large
H:D 95 percentile part of the southern Amazon zone, suggests climate
Intercept <0.001 0.174 <0.001 1.000 sensitivity in forest structure here. Climate change, and
DE −0.785 0.302 −2.597 0.023 especially any reduction in annual or seasonal precipita-
PrecWM 1.260 0.302 4.167 0.001 tion, is thus likely to have a significant effect on the
Mean biomass
Intercept −0.540 0.166 −3.249 0.007 forest structure in the southern border of the Amazon.
FES 0.244 0.257 0.949 0.361 Secondly, our results also suggest that the effects of
FOA 2.291 0.303 7.555 <0.001 reduction in the annual precipitation may be further
exacerbated in smaller fragments. This suggests that
DE: distance to the edge; PrecWM: precipitation of wettest month; H:D:
allometric H:D ratio; FES: seasonal semi-deciduous forest-plots; FOA: habitat fragmentation may intensify the negative effects
open rainforest-plots. Significant effects (P ≤ 0.05) are shown in bold of climate change and burning in forests in the southern
type. Amazon border, resulting in a substantial risk of
increases in tree mortality. Given the likely susceptibil-
ity of the remaining southern Amazon border forests to
peaks in El Niño events such as in 2015–16 (Jiménez-Muñoz environmental change, strong conservation strategies are
et al. 2016), may therefore have profound implications for the urgently needed to guarantee the persistence of these
forest structure of the southern Amazon border, located as they habitats, especially for the smaller fragments and those
are in a region that is already naturally close to their distribu- close to agricultural frontiers.
tional and hydraulic limits. In this scenario, large trees are
more susceptible to damage to the xylem, which can ulti-
mately result in the death of the plant (e.g. McIntyre et al. Acknowledgements
2015) and eventually lead to forests of lower stature We are grateful to the team of the Laboratório de Ecologia
(McDowell et al. 2008; Rowland et al. 2015). Trees being Vegetal – Plant Ecology Laboratory at the UNEMAT campus
smaller in drier areas with greater water deficiency is directly in Nova Xavantina, especially to Henrique A. Mews, Nayane
C. Prestes, Ana Paula G. da Silva, Laís F. S. Neves, Mônica
be related to conservative modifications in the hydraulic struc-
Forsthofer and Leonardo Maracahipes, for help in collecting
ture of the plants under hydrological stress to avoid embolism field data. We also thank the National Council for Scientific
(e.g. Lines et al. 2012; Claeys and Inzé 2013). Thus, as have and Technological Development (CNPq) for financial support
recently argued in both tropical and temperate zone contexts of the projects PELD “Cerrado-Amazon Forest Transition:
(e.g. Stegen et al. 2011; Banin et al. 2012; McIntyre et al. ecological and socio-environmental bases for Conservation”
(stage II) – process nr. 403725/2012-7), PVE “special visiting
2015) it is likely that trees in forests subject either to more
researcher” (401279/2014-6), PPBIO “Phytogeography of the
extreme climatic events, or to more disturbance (including Amazon-Cerrado Transition Zone” (457602/2012-0) and
seasonally flooded habitats), or both, will in general tend to FAPEMAT – Fundação de Amparo à Pesquisa do Estado de
be shorter at a given diameter in order to avoid risks of Mato Grosso (164131/2013). We also thank CNPq for
hydraulic and/or mechanical failure, whereas trees in forests research productivity grants to B.S. Marimon and B.H.
Marimon Junior, a post-doctoral scholarship to D. Nogueira,
with high rainfall, such as our FOA-01 and FOA-02, will have
and international doctoral grants to S.M.A. Reis and P.S.
greater heights and hence greater biomass. Morandi. We are also grateful to CAPES and FAPEMAT
Besides the correlation with the climatic variables, both for scholarships to S.M.A. Reis, P.S. Morandi, F. Elias, E.
height and the biomass of trees were positively correlated A. Oliveira and E.C. Neves.
with fragment area. This result may be related to the inci-
dence of wind in smaller fragments which have a higher
proportion of forest edge (D’Angelo et al. 2004; Laurance Disclosure statement
2004; Haddad et al. 2015). These disturbances are known to No potential conflict of interest was reported by the authors.
be able to generate high mortality, especially of the tallest
trees (Laurance et al. 2000; Laurance 2004), and conse-
quently in our dataset such edge-generated disturbances Funding
may have affected the height and biomass of trees. We also thank the National Council for Scientific and
Elsewhere, local climatic changes as a result of fragmentation Technological Development (CNPq) for financial support of the
 Climate and fragmentation affect forest structure 9

projects PELD “Cerrado-Amazon Forest Transition: ecological Alencar AA, Brando PM, Asner GP, Putz FE. 2015. Landscape
and socio-environmental bases for Conservation” (stage II) – fragmentation, severe drought, and the new Amazon forest
process nr. 403725/2012-7), PVE “special visiting researcher” fire regime. Ecological Applications 25:1493–1505.
(401279/2014-6), PPBIO “Phytogeography of the Amazon- Alvares CA, Stape JL, Sentelhas PC, de Moraes G, Leonardo
Cerrado Transition Zone” (457602/2012-0) and FAPEMAT – J, Sparovek G. 2013. Köppen’s climate classification map
Fundação de Amparo à Pesquisa do Estado de Mato Grosso for Brazil. Meteorologische Zeitschrift 22:711–728.
[164131/2013]. Álvarez-Dávila E, Cayuela L, González-Caro S, Aldana AM,
Stevenson PR, Phillips O, Cogollo A, Peñuela MC, von
Hildebrand P, Jiménez E, et al. 2017. Forest biomass density
Supplemental data across large climate gradients in northern South America is
related to water availability but not with temperature. Plos
Supplemental data for this article can be accessed here. One 12:e0171072.
Angiosperm Phylogeny Group. 2009. An update of the angios-
perm Phylogeny group classification for the orders and
Notes on contributors families of flowering plants: APG III. Botanical Journal of
Simone Matias Reis is a Ph.D. student focussing on the effects of the Linnean Society 161:105–121.
biotic and abiotic factors on forest structure in the southern Aragão LE. 2012. Environmental science: the rainforest’s water
border of Amazonia. pump. Nature 489:217–218.
Aragao LEO, Malhi Y, Roman-Cuesta RM, Saatchi S, Anderson
Beatriz Schwantes Marimon is a professor and an ecologist with LO, Shimabukuro YE. 2007. Spatial patterns and fire
experience in working with structure and dynamics of different response of recent Amazonian droughts. Geophysical
vegetation types of the southern Amazon boundaries in Central Research Letters34: L07701.
Brazil. Her current work focuses on ecology and management of Atkin OK, Bloomfield KJ, Reich PB, Tjoelker MG, Asner GP,
forests in the transition zone between Cerrado and Amazon forest Bonal D, Bönisch G, Bradford MG, Cernusak LA, Cosio
biomes, and currently has great interest in understanding changes EG, et al. 2015. Global variability in leaf respiration in
in tropical plants communities as a result of climate change. relation to climate, plant functional types and leaf traits.
Ben Hur Marimon-Junior is a professor. He has experience in New Phytologist 206:614–636.
forest ecology, studying biogeochemical cycles, carbon stocks, Banin L, Feldpausch TR, Phillips OL, Baker TR, Lloyd J,
pyrogenic carbon in mineral nutrition of plants, biodiversity and Affum-Baffoe K, Arets EJMM, Berry NJ, Bradford M,
ecosystem functions. His current work focuses on understanding Brienen RJW, et al. 2012. What controls tropical forest
changes in tropical plants communities as a result of climate architecture? Testing environmental, structural and floristic
change. drivers. Global Ecology and Biogeography 21:1179–1190.
Banin LF, Phillips OL, Lewis SL. 2015. Tropical Forests.
Paulo S. Morandi is a Ph.D. student focussing on structural Routledge Handbook of Forest Ecology 5:56–75.
patterns, diversity and distribution of species under the effect of Barton K 2016. MuMIn: multi-model inference. R package;
environmental variables and vegetation succession in the [accssed 2017 Feb 20]. https://cran.r-project.org/web/
Amazon-Cerrado transition. packages/MuMIn/MuMIn.pdf.
Edmar Almeida de Oliveira is a Ph.D. student focussing on Bennett AC, McDowell NG, Allen CD, Anderson-Teixeira KJ.
historical effect of soil disturbance and pyrogenic carbon on the 2015. Larger trees suffer most during drought in forests
composition and structure of forests in southern Amazonia. worldwide. Nature Plants 1:15139.
Bivand RS, Hauke J, Kossowski T. 2013. Computing the
Fernando Elias is a Ph.D. student focussing on the role of Jacobian in Gaussian spatial autoregressive models: an illu-
climate, land management and landscape on the resilience of strated comparison of available methods. Geographical
secondary forests in the Amazon. Analysis 45:150–179.
Eder Carvalho das Neves is M.Sc. student focussing on the Boisier JP, Ciais P, Ducharne A, Guimberteau M. 2015. Projected
physiology of forests in the Amazon-Cerrado transition. strengthening of Amazonian dry season by constrained cli-
mate model simulations. Nature Climate Change 5:656–660.
Bianca de Oliveira is a biologist and conducted her master’s Brando PM, Balch JK, Nepstad DC, Morton DC, Putz FE, Coe
research on nutrient cycling in the Amazonia-Cerrado transition. MT, Silvério D, Macedo MN, Davidson EA, Nóbrega CC,
Denis Nogueira is a post-doctoral researcher focusing on forest et al. 2014. Abrupt increases in Amazonian tree mortality
dynamics of the Amazon-Cerrado transition in Mato Grosso. due to drought-fire interactions. Proceedings of the National
Academy of Sciences 111:6347–6352.
Ricardo Keichi Umetsu is a lecturer. He has experience in envir- Brienen RJW, Phillips OL, Feldpausch TR, Gloor E, Baker TR,
onmental planning with emphasis on water resources, hydro- Lloyd J, Lopez-Gonzalez G, Monteagudo-Mendoza A, Malhi
graphic basins, hydrogeochemistry of streams, plant ecology Y, Lewis SL, et al. 2015. Long-term decline of the Amazon
and landscape. carbon sink. Nature 519:344–348.
Ted R. Feldpausch is a lecturer in tropical ecology. His research Chaplin-Kramer R, Ramler I, Sharp R, Haddad NM, Gerber JS,
focuses on the ecology and effects of global change on tropical West PC, Mandle L, Engstrom P, Baccini A, Sim S, et al.
forests. 2015. Degradation in carbon stocks near tropical forest
edges. Nature Communications 6: 10158.
Oliver L. Phillips is professor of tropical ecology. He leads the Chave J, Andalo C, Brown S, Cairns MA, Chambers JQ, Eamus
RAINFOR network of scientists in Amazonian forests. D, Folster H, Fromard F, Higuchi N, Kira T, et al. 2005. Tree
allometry and improved estimation of carbon stocks and
balance in tropical forests. Oecologia 145:87–99.
References Chave J, Réjou-Méchain M, Búrquez A, Chidumayo E, Colgan
Alencar A, Nepstad DC, McGrath D, Moutinho P, Pacheco P, MS, Delitti WB, Duque A, Eid T, Fearnside PM, Goodman
Diaz MDCV, Soares Filho B. 2004. Desmatamento na RC, et al. 2014. Improved allometric models to estimate the
Amazônia: indo além da “emergência crônica. Belém: aboveground biomass of tropical trees. Global Change
Intituto de Pesquisa Ambiental da Amazônia. Biology 20:3177–3190.
10 S. M. Reis et al.

Claeys H, Inzé D. 2013. The agony of choice: how plants Amazon rainforest during the course of El Niño 2015-2016.
balance growth and survival under water-limiting conditions. Scientific Reports 6:33130.
Plant Physiology 162:1768–1779. Jiménez-Muñoz JC, Sobrino JA, Mattar C, Malhi Y. 2013.
Cox PM, Betts RA, Jones CD, Spall SA, Totterdell IJ. 2000. Spatial and temporal patterns of the recent warming of the
Acceleration of global warming due to carbon-cycle feed- Amazon forest. Journal of Geophysical Research:
backs in a coupled climate model. Nature 408:184–187. Atmospheres 118:5204–5215.
D’Angelo SA, Andrade ACS, Laurance SG, Laurance WF, Kark S, Hadany L, Safriel UN, Noy-Meir I, Eldredge N,
Mesquita RCG. 2004. Inferred causes of tree mortality in Tabarroni C, Randi E. 2008. How does genetic diversity
fragmented and intact Amazonian forests. Journal of Tropical change towards the range periphery? Evolutionary Ecology
Ecology 20:243–246. Research 10:391–414.
Doughty CE, Metcalfe DB, Girardin CAJ, Amézquita FF, Kark S, van Rensburg BJ. 2006. Ecotones: marginal or central
Cabrera DG, Huasco WH, Silva-Espejo JE, Araujo- areas of transition? Israel Journal of Ecology & Evolution
Murakami A, da Costa MC, Rocha W, et al. 2015. Drought 52:29–53.
impact on forest carbon dynamics and fluxes in Amazonia. Koch GW, Sillett SC, Jennings GM, Davis SD. 2004. The limits
Nature 519:78–82. to tree height. Nature 428:851–854.
Dray S 2013. spacemakeR: spatial modelling. R package version Laurance WF. 2004. Forest-climate interactions in fragmented
0.0-5/r113. https://R-Forge.R-project.org/projects/sedar/ tropical landscapes. Philosophical Transactions of the
Dray S, Legendre P, Blanchet G 2016. packfor: forward Selection Royal Society of London B: Biological Sciences
with permutation (Canoco p. 46). R package version 0.0-8/ 359:345–352.
r136. https://R-Forge.R-project.org/projects/sedar/ Laurance WF, Delamônica P, Laurance SG, Vasconcelos HL,
Fahrig L. 2003. Effects of habitat fragmentation on biodiversity. Lovejoy TE. 2000. Conservation: rainforest fragmentation
Annual Review of Ecology, Evolution, and Systematics kills big trees. Nature 404:836.
34:487–515. Laurance WF, Ferreira LV, Rankin-de Merona JM, Laurance SG.
Feldpausch TR, Banin L, Phillips OL, Baker TR, Lewis SL, 1998. Rain forest fragmentation and the dynamics of
Quesada CA, Affum-Baffoe K, Arets EJMM, Berry NJ, Amazonian tree communities. Ecology 79:2032–2040.
Bird M, et al. 2011. Height-diameter allometry of tropical Laurance WF, Laurance SG, Ferreira LV, Merona JMR, Gascon
forest trees. Biogeosciences 8:1081–1106. C. 1997. Biomass collapse in Amazonian Forest Fragments.
Feldpausch TR, Phillips OL, Brienen RJW, Gloor E, Lloyd J, Science 278:1117–1118.
Lopez-Gonzalez G, Monteagudo-Mendoza A, Malhi Y, Laurance WF, Williamson GB. 2001. Positive feedbacks among
Alarcón A, Álvarez Dávila E, et al. 2016. Amazon forest forest fragmentation, drought, and climate change in the
response to repeated droughts. Global Biogeochemical Amazon. Conservation Biology 15:1529–1535.
Cycles 30:1–19. Lawton RO, Putz FE. 1988. Natural disturbance and gap-phase
Givnish TJ, Wong SC, Stuart-Williams H, Holloway-Phillips M, regeneration in a wind-exposed tropical cloud forest.
Farquhar GD. 2014. Determinants of maximum tree height in Ecology 69:764–777.
Eucalyptus species along a rainfall gradient in Victoria, Lewis SL, Brando PM, Phillips OL, van der Heijden GM,
Australia. Ecology 95:2991–3007. Nepstad D. 2011. The 2010 amazon drought. Science
Gloor M, Barichivich J, Ziv G, Brienen R, Schöngart J, Peylin P, 331:554.
Cintra BBL, Feldpausch T, Phillips O, Baker J. 2015. Recent Lewis SL, Sonké B, Sunderland T, Begne SK, Lopez-Gonzalez
Amazon climate as background for possible ongoing and G, van der Heijden GM, Phillips OL, Affum-Baffoe K,
future changes of Amazon humid forests. Global Baker TR, Banin L, et al. 2013. Above-ground biomass
Biogeochemical Cycles 29:1384–1399. and structure of 260 African tropical forests. Philosophical
Haddad NM, Brudvig LA, Clobert J, Davies KF, Gonzalez A, Transactions of the Royal Society B 368:20120295.
Holt RD, Lovejoy TE, Sexton JO, Austin MP, Collins CD, Lines ER, Zavala MA, Purves DW, Coomes DA. 2012.
et al. 2015. Habitat fragmentation and its lasting impact on Predictable changes in aboveground allometry of trees
Earth’s ecosystems. Science Advances 1:e1500052. along gradients of temperature, aridity and competition.
Heskel MA, O’Sullivan OS, Reich PB, Tjoelker MG, Global Ecology and Biogeography 21:1017–1028.
Weerasinghe LK, Penillard A, Egerton JJ, Creek D, Lista de Espécies da Flora do Brasil. 2016. Jardim Botânico do
Bloomfield KJ, Xiang J, et al. 2016. Convergence in the Rio de Janeiro; [accessed 2016 Jul 17]. http://floradobrasil.
temperature response of leaf respiration across biomes and jbrj.gov.br/2016
plant functional types. Proceedings of the National Academy Lloyd J, Farquhar GD. 2008. Effects of rising temperatures and
of Sciences 113:3832–3837. [CO2] on the physiology of tropical forest trees.
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Philosophical Transactions of the Royal Society B:
Very high resolution interpolated climate surfaces for global Biological Sciences 363:1811–1817.
land areas. International Journal of Climatology 25:1965– Lopez-Gonzalez G, Lewis SL, Burkitt M, Phillips OL. 2011.
1978. ForestPlots.net: a web application and research tool to man-
IBGE - Instituto Brasileiro de Geografia e Estatística. 2004. Área age and analyse tropical forest plot data. Journal of
territorial brasileira. [accessed 2017 Jun 25]. http://www. Vegetation Science 22:610–613.
ibge.gov.br/ Mantyka-Pringle CS, Martin TG, Rhodes JR. 2012. Interactions
INPE – Instituto Nacional de Pesquisas Espaciais. 2017. between climate and habitat loss effects on biodiversity: a
Coordenação geral de observação da terra. Projeto systematic review and meta-analysis. Global Change Biology
PRODES: monitoramento da Floresta Amazônica Brasileira 18:1239–1252.
por Satélite. [accessed 2017 Jun 29]. http://www.obt.inpe.br/ Maracahipes L, Marimon BS, Lenza E, Marimon-Junior BH, de
prodes/index.php Oliveira EA, Mews HA, Gomes L, Feldpausch TR. 2014.
IPCC - Intergovernmental Panel on Climate Change. 2015. Post-fire dynamics of woody vegetation in seasonally
Climate change 2014: mitigation of climate change. flooded forests (impucas) in the Cerrado-Amazonian Forest
Cambridge: Cambridge University Press. transition zone. Flora-Morphology, Distribution, Functional
Jiménez-Muñoz JC, Mattar C, Barichivich J, Santamaría-Artigas Ecology of Plants 209:260–270.
A, Takahashi K, Malhi Y, Sobrino JA, van der Schrier G. Marengo JA, Tomasella J, Alves LM, Soares WR, Rodrigues
2016. Record-breaking warming and extreme drought in the DA. 2011. The drought 800 of 2010 in the context of
 Climate and fragmentation affect forest structure 11

historical droughts in the Amazon region. Geophysical Ratter JA, Richards PW, Argent G, Gifford DR. 1973. Observations
Research Letters 2011:38: L12703. on the vegetation of the northeastern Mato Grosso. I. The
Marimon BS, Marimon-Junior BH, Feldpausch TR, Oliveira- woody vegetation types of the Xavantina-Cachimbo
Santos C, Mews HA, Lopez-Gonzalez G, Lloyd J, Expedition area. Philosophical Transactions of the Royal
Franczak D, de Oliveira EA, Maracahipes L, et al. 2014. Society of London B 266:449–492.
Disequilibrium and hyperdynamic tree turnover at the forest– Rocha-Santos L, Pessoa MS, Cassano CR, Talora DC, Orihuela
cerrado transition zone in southern Amazonia. Plant Ecology RL, Mariano-Neto E, Morante-Filho JC, Faria D, Cazetta E.
& Diversity 7:281–292. 2016. The shrinkage of a forest: landscape-scale deforesta-
McDowell N, Pockman T, Allen CD, Breshears DD, Cobb N, tion leading to overall changes in local forest structure.
Kolb T, Plaut J, Sperry J, West A, Williams DG. 2008. Biological Conservation 196:1–9.
Mechanisms of plant survival and mortality during drought: Rowland L, da Costa ACL, Galbraith DR, Oliveira RS, Binks
why do some plants survive while others succumb to OJ, Oliveira AAR, Pullen AM, Doughty CE, Metcalfe DB,
drought? New Phytologist 178:719–739. Vasconcelos SS, et al. 2015. Death from drought in tropical
McGarigal K, Cushman S. 2002. Comparative evaluation of forests is triggered by hydraulics not carbon starvation.
experimental approaches to the study of habitat fragmenta- Nature 528:119–122.
tion effects. Ecological Applications 12:335–345. Ryan MG, Binkley D, Fownes JH, Giardina CP, Senock RS.
McIntyre PJ, Thorne JH, Dolanc CR, Flint AL, Flint LE, Kelly 2004. An experimental test of the causes of forest growth
M, Ackerly DD. 2015. Twentieth-century shifts in forest decline with stand age. Ecological Monographs 74:393–
structure in California: denser forests, smaller trees, and 414.
increased dominance of oaks. Proceedings of the National R Core Team. 2012. The R Project for Statistical Computing.
Academy of Sciences 112:1458–1463. Vienna (AT): R Foundation for Statistical Computing;
NASA. 2012. TRMM 3B43(7): monthly 0.25 x 0.25 degree [accessed 2017 Oct 10]. http://www.R-project.org.
merged TRMM and other sources estimates. http://disc.sci. Saatchi SS, Harris NL, Brown S, Lefsky M, Mitchard ET, Salas
gsfc.nasa.gov/precipitation/documentation/TRMM_ W, Zuttaa BR, Buermannb W, Lewis SL, Hagenf S, etal.
README/TRMM_3B43_readme.shtml edited, NASA 2011. Benchmark map of forest carbon stocks in tropical
Distrib. Active Arch. Cent., Goddard Space Flight Cent. regions across three continents. Proceedings of the National
Earth Sci., Greenbelt, Md. Academy of Sciences 108:9899–9904.
Nogueira EM, Fearnside PM, Nelson BW, Barbosa RI, Keizer Safriel UN, Volis S, Kark S. 1994. Core and peripheral popula-
EWH. 2008. Estimates of forest biomass in the Brazilian tions and global climate change. Israel Journal of Plant
Amazon: new allometric equations and adjustments to bio- Sciences 42:331–345.
mass from wood-volume inventories. Forest Ecology and Silvério DV, Brando PM, Macedo MN, Beck PS, Bustamante
Management 256:1853–1857. M, Coe MT. 2015. Agricultural expansion dominates cli-
Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, mate changes in southeastern Amazonia: the overlooked
McGlinn D, Minchin PR, O’Hara RB, Simpson GL, non-GHG forcing. Environmental Research Letters
Solymos P, et al. 2016. Vegan: Community Ecology 10:104015.
Package. R package version 2.4-0. https://CRAN.R-project. Stegen JC, Swenson NG, Enquist BJ, White EP, Phillips OL,
org/package=vegan Jørgensen PM, Weiser MD, Monteagudo Mendoza A, Núñez
Pan Y, Birdsey RA, Phillips OL, Jackson RB. 2013. The struc- Vargas P. 2011. Variation in above-ground forest biomass
ture, distribution, and biomass of the world’s forests. Annual across broad climatic gradients. Global Ecology and
Review of Ecology, Evolution, and Systematics 44:593–622. Biogeography 20:744–754.
Phillips OL, Aragão LEOC, Lewis SL, Fisher JB, Lloyd J, Vlam M, Baker PJ, Bunyavejchewin S, Zuidema PA. 2014.
López-González G, Malhi Y, Monteagudo A, Peacock J, Temperature and rainfall strongly drive temporal growth
Quesada CA, et al. 2009. Drought sensitivity of the variation in Asian tropical forest trees. Oecologia
Amazon Rainforest. Science 323:1344–1347. 174:1449–1461.
Phillips OL, Van Der Heijden G, Lewis SL, López-González G, Way DA, Oren R. 2010. Differential responses to changes in
Aragão LE, Lloyd J, Malhi Y, Monteagudo A, Almeida S, growth temperature between trees from different functional
Dávila EA, et al. 2010. Drought–mortality relationships for groups and biomes: a review and synthesis of data. Tree
tropical forests. New Phytologist 187:631–646. Physiology 30:669–688.
Quinn GP, Keough PR. 2002. Experimental design and data analysis Worbes M. 1999. Annual growth rings, rainfall-dependent
for biologists. Cambridge: Cambridge University Press. growth and long-term growth patterns of tropical trees from
Rangel TF, Diniz-Filho JAF, Bini LM. 2010. SAM: a compre- the Caparo Forest Reserve in Venezuela. Journal of Ecology
hensive application for spatial analysis in macroecology. 87:391–403.
Ecography 33:46–50. Zar JH. 2010. Biostatistical analysis. NJ: Prentice Hall.