Review
Tansley review
Patterns and mechanisms of spatial variation in
tropical forest productivity, woody residence
time, and biomass
Author for correspondence: Helene C. Muller-Landau1 , K. C. Cushman1 , Eva E. Arroyo2 ,
Helene C. Muller-Landau
Email: mullerh@si.edu Isabel Martinez Cano
3 , Kristina J. Anderson-Teixeira1,4 and
Received: Bogumila Backiel
1
1 April 2020
Accepted: 12 October 2020 1Center for Tropical Forest Science-Forest Global Earth Observatory, Smithsonian Tropical Research Institute, PO Box 0843-03092,
Balboa, Anc
on, Panama; 2Department of Ecology, Evolution and Environmental Biology, Columbia University, 1200 Amsterdam
Avenue, New York, NY 10027, USA; 3Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544,
USA; 4Conservation Ecology Center, Smithsonian Conservation Biology Institute and National Zoological Park, 1500 Remount Rd,
Front Royal, VA 22630, USA
Contents
Summary 1 VI. Disturbance 13
I. Introduction 2 VII. Biogeographic realm 14
II. Methods 3 VIII. Discussion 14
III. Climatic water availability 5 Acknowledgements 16
IV. Temperature and elevation 7 References 16
V. Soil fertility 11
Summary
New Phytologist (2020) Tropical forests vary widely in biomass carbon (C) stocks and fluxes even after controlling for
doi: 10.1111/nph.17084 forest age. A mechanistic understanding of this variation is critical to accurately predicting
responses to global change. We review empirical studies of spatial variation in tropical forest
Key words: biomass carbon stocks, plant biomass, productivity and woody residence time, focusing on mature forests. Woody
functional composition, precipitation, soil productivity and biomass decrease from wet to dry forests and with elevation. Within lowland
fertility, temperature, tropical forests, woody forests, productivity and biomass increase with temperature in wet forests, but decrease with
productivity, woody residence time. temperature where water becomes limiting. Woody productivity increases with soil fertility,
whereas residence timedecreases, andbiomass responses are variable, consistentwith anoverall
unimodal relationship. Areas with higher disturbance rates and intensities have lower woody
residence time and biomass. These environmental gradients all involve both direct effects of
changing environments on forest C fluxes and shifts in functional composition – including
changing abundances of lianas – that substantially mitigate or exacerbate direct effects.
Biogeographic realms differ significantly and importantly in productivity and biomass, even after
controlling for climate and biogeochemistry, further demonstrating the importance of plant
species composition. Capturing these patterns in global vegetation models requires better
mechanistic representation of water and nutrient limitation, plant compositional shifts and tree
mortality.

 2020 The Authors New Phytologist (2020) 1
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
2 Review Tansley review Phytologist
date of overall patterns (Lewis et al., 2009; Wright, 2010). This
II. Introduction
uncertainty is reflected in highly divergent predictions for tropical
Extant tropical forests vary widely in biomass density and thus forest responses in different earth system models (Cavaleri et al.,
carbon (C) stocks, even when controlling for forest age (Becknell 2015; Koven et al., 2015; Rowland et al., 2015).
et al., 2012; Lewis et al., 2013; Poorter et al., 2016; Alvarez-Davila Fundamentally, variation inmature forest aboveground biomass
et al., 2017; Sullivan et al., 2020).Much of this biomass variation is (AGB) arises from variation in aboveground woody productivity
associated with climate and biogeochemistry, which influence (AWP) and/or abovegroundwoody residence time (AWRT). AWP
woody productivity, residence time and biomass both directly and depends on NPP (net primary productivity) and allocation to
indirectly, via shifts in plant functional composition.However, our wood, and ultimately on GPP (gross primary productivity) and C-
understanding of these patterns and their underlying mechanisms use efficiency (Malhi, 2012) (Fig. 1). In recent decades, as interest
remains incomplete (Fig. 1). A mechanistic understanding of in forest C budgets has increased, many studies have investigated
current variation in tropical forest C stocks and fluxes with climate, patterns andmechanisms of spatial variation in tropical forest AWP
soils and other factors is a critical precursor to accurately predicting and AGB with abiotic and biotic factors (e.g. Levine et al., 2016;
forest responses to anthropogenic change. Malhi et al., 2017; Taylor et al., 2017; Moore et al., 2018; Sullivan
Uncertainty about how tropical forest C pools will respond to et al., 2020) (methods summarized in Box 1). This research builds
global change is one of the largest sources of uncertainty in projecting naturally on an older literature on forest structure and composition
future global C budgets and climate (Cavaleri et al., 2015). Tropical (e.g. Richards, 1952; Gentry, 1988). Some consistent large-scale
forests currently account for two-thirds of terrestrial biomass C patterns have become clear, such as increasing dry season length
stocks (Pan et al., 2013) and nearly a third of global soil C to 3 m (and decreasing precipitation) being associated with lower AWP
depth (Jobb
agy & Jackson, 2000). Increasing temperatures, chang- and AGB (Becknell et al., 2012; Poorter et al., 2017; Taylor et al.,
ing precipitation patterns and disturbance regimes, increasing 2017). However, other patterns are inconsistent among studies,
atmospheric carbon dioxide and increasing nutrient deposition have such as AGB increasing with soil fertility in some studies (Slik et al.,
the potential to greatly alter tropical forest C stocks and fluxes, and 2013; Lloyd et al., 2015) and decreasing in others (Lewis et al.,
thus the global C budget (Lewis et al., 2009; Wright, 2010). 2013; Schietti et al., 2016).
However, the combined impacts of these global change drivers on Mechanisms and patterns involving changes in tree mortality or
tropical forests remain unclear, with contrasting effects expected shifts in plant functional composition remain poorly understood,
under different mechanisms and hypotheses, and mixed evidence to whereas those involving changes in productivity of a given plant
Gross primary Autotrophic
productivity (GPP) respiration
Carbon use
efficiency (CUE)
Abiotic and biotic drivers,
and their interactions Net primary Belowground
Climate: productivity (NPP) NPP
precipitation, temperature,
solar radiation, humidity Allocation
aboveground
Geomorphology:
geology, topography, Aboveground NPP Litter
soils (ANPP) production
Biogeographic realm Allocation
and plant functional to wood
composition
(including lianas) Aboveground woody
productivity (AWP)
Aboveground woody
residence time (AWRT)
Aboveground woody
biomass (AGB)
Fig. 1 Climate, geomorphology, biogeographic realm and plant functional composition interact to influence tropical forest aboveground woody productivity
(AWP, units ofmass area
1 time
1), abovegroundwoody residence time (AWRT, time) and thus abovegroundwoody biomass density (AGB,mass area
1) via
multiple pathways. Here blue boxes represent fluxes (mass area
1 time
1), blue arrows represent the factors by which the one quantity is multiplied to obtain
another (e.g. NPP =GPP9CUE), and purple arrows represent causal influences. Note that GPP (gross primary productivity) is the sum of NPP (net primary
productivity) and autotrophic respiration; NPP is the sum of abovegroundNPP (ANPP) and belowgroundNPP (root production); and ANPP is the sum of AWP
and canopy productivity (leaves, fruits, finewoody branches, all measured as litterfall). Box 1 gives basic information onmeasurementmethods for AGB, AWP
and AWRT; Supporting Information Notes S1 provides additional details on these and related variables.
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 3
temporal variation to different factors (Heavens et al., 2013). These
Box 1 Estimating aboveground biomass, woody productivity and models are mechanistic, and attempt to capture hypothesized
residence time. critical processes as gleaned from empirical studies (Heinze et al.,
2019). However, the most recent set of publicly released models
Aboveground biomass (AGB, mass area
1), our central measure of completely fail to reproduce spatial variation in AGB, AWP and
biomassC stocks, is estimatedabovegroundwoodybiomassper area, AWRT in old-growth tropical forests (Fig. 2). This demonstrates
typically of trees above some threshold diameter, omitting smaller
trees and lianas (woody vines). Individual tree AGB is estimated from that the models fail to adequately represent the mechanisms or
tree census data with allometric equations and summed to obtain capture the patterns of spatial variation in tropical forests today,
plot-level totals. AGB also is estimated from lidar and radar and highlights the need for a more mechanistic understanding of
measurements of canopy structure using phenomenological rela- these patterns.
tionships with plot-based AGB estimates. Tree basal area (BA, basal Here we review empirical studies documenting how different
areaof trunkspergroundarea) andmeancanopyheight aregenerally
well-correlated with AGB across sites, and thus are reasonably good environmental factors relate to tropical forest productivity,
proxies for evaluating among-site variation. residence time, biomass, their proxies and related variables. We

 
 first briefly describe the types of studies included, and theirAboveground woody productivity (AWP, mass area 1 time 1), our strengths and weaknesses. We then review empirical findings on
central measure of productivity, is typically estimated from repeat
tree censuses as the sumof the growth in estimated AGB of surviving tropical forest variation with climatic water availability (precipi-
trees plus the AGB of recruits (trees newly above the size threshold), tation regimes), elevation and temperature, soil fertility, distur-
per area per time. Such calculations ignore branch production that bance and biogeographic realm, and discuss hypothesized
merely compensates for branchfall (see Section II, Methods). Like mechanisms underlying observed relationships. We discuss critical
AGB,AWP is basedonallometric equationsandgenerallyomits lianas knowledge gaps and uncertainties in mechanistic understanding
and smaller trees. Parallel calculationsofbasal areaproductivity (BAP)
are good proxies for among-site variation in AWP. and in datasets, and key directions for future research.
Aboveground woody residence time (AWRT, time) is the average
timeC remains in abovegroundwoodybiomass before it becomes dead III. Methods
wood. AWRT is determined by themortality rates of woody plants and
We searched the literature for studies of among-site variation in our
branches,with large treemortality ratesdisproportionately important. In
mature forests, AWRT is most often estimated as the quotient of focal variables in mature, unlogged tropical forests, or in secondary
biomass and productivity (AWRT =AGB/AWP), because productivity forests when controlling for stand age, that included eight or more
fluxes are more constant in time than mortality fluxes and assumed sites. We specifically searched for studies of variation in above-
equal over the long term. When AWP calculations ignore branchfall, ground biomass, woody productivity and woody residence time
AWRTmisses it aswell. AWRT is inversely related to treemortality rates
(AGB, AWP and AWRT) (Box 1), tree mortality rates and tree
and tree turnover rates across sites.
See Section II (Methods) and Supporting Information Notes S1 for turnover rates with respect to elevation, temperature, climatic
details. measures of water availability (e.g. precipitation, dry season length,
climatic water deficit) and/or soil fertility (e.g. soil phosphorus (P),
cation exchange capacity, base cations). We also opportunistically
tabulated studies reporting results for canopy height, basal area
functional type along environmental gradients are relatively well- (BA) and basal area productivity (BAP), which serve as proxies for
understood. Variation in tree mortality and thus AWRT is a key AGB and AWP (Box 1), as well as for the related productivity
driver of spatial variation in AGB within the tropics (Johnson et al., variables of annual net primary productivity (ANPP), Litterfall
2016), yet our understanding of tropical tree mortality remains NPP and gross primary productivity (GPP) (Fig. 1).Where a study
extremely limited (McDowell et al., 2018). Variation in plant included multiple analyses using different measures of the
functional composition also plays a critical role in explaining large- environmental factor of interest (e.g. precipitation and dry season
scale variation in AWP, AWRT and AGB. Different environments length), we report the result for the independent variable showing a
select for different plant functional composition, which in turn stronger relationship. Where both multivariate and bivariate
influences stand-level AWP, AWRT and AGB in ways that may analyses were reported, we report the multivariate analyses.
enhance or counter direct effects of environmental drivers (Fyllas Additional details on the literature search methods are given in
et al., 2009; Fyllas et al., 2017; Turner et al., 2018). For example, the Supporting InformationNotes S1, the geographical distribution of
abundance of lianas (woody climbing plants) varies strongly with data is shown in Figs S1 and S9, and the resulting database is
environmental conditions (DeWalt et al., 2015) and lianas negatively available at Dataset S1. In the remainder of this section, we discuss
affect tree growth and survival and thus AWP, AWRT and AGB the main sources of error in our focal variables.
(Ingwell et al., 2010; Duran&Gianoli, 2013; van der Heijden et al., Most currently available information on our focal variables are
2015; Lai et al., 2017), with differential effects across tree species based on tree plot census data. Because of high local spatial
(Muller-Landau&Visser, 2019). Indeed, experimental liana removal variability in the number and sizes of large trees, these plot-based
increased AWP by 65% and AGB accumulation by 75% in a estimates exhibit considerable sampling error, even for plots of
secondary moist tropical forest (van der Heijden et al., 2015). 1 ha, and this error increases at smaller plot sizes (Muller-Landau
Earth system models (ESMs) are key tools for predicting the et al., 2014). We thus highlight studies based on plots with a
future of the globalC cycle under global change, and for attributing median size of 1 ha or larger (124 of 201 results reviewed). Plot-

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
4 Review Tansley review Phytologist
Model A Model B Model C
400
300
200
100
0
0 100 200 300 400 0 100 200 300 400 0 100 200 300 400
Observed AGB (Mg C ha–1)
Model A Model B Model C
12.5
10 Fig. 2 Earth SystemModel (ESM) predictions
of aboveground woody biomass (AGB, top
7.5 row), aboveground woody productivity
(AWP, middle row) and aboveground woody
5 residence time (AWRT, bottom row) show
2.5 little relation with observational data(Galbraith et al., 2013) for 177 old-growth
0 tropical forests. Both observed and modeled
0 2.5 5 7.5 10 0 2.5 5 7.5 10 0 2.5 5 7.5 10 residence times are calculated as AGB/AWP
(Box1). ESMs simulate vegetationdynamics in
Observed AWP (Mg C ha–1 yr–1) tropical forests around the globe as part of
their simulation of the entire earth system,
Model A Model B Model C including the atmosphere, ocean and land
surface, and their interactions. Spatial
variation in predicted climates in these models
100 translates to spatial variation in predicted
vegetation because of modeled effects of
climateonphotosynthesis and respiration, and
50 thus on woody productivity and potentially
the dominant plant functional type, with
effects that vary depending on the details of
model structure and parameterization. Model
0 predictions are from the most recent set of
0 50 100 0 50 100 0 50 100 publicly released ESMmodels and simulation
Observed AWRT (yr) results, from the Coupled Model
IntercomparisonProject5 (Tayloret al., 2012).
Further details are given in Supporting
Highland Lowland Lowland, highland
Information Notes S1.
based data also may have systematic errors, reflecting nonrandom allometries differ systematically among sites (e.g. Chave et al.,
plot placed. Some studies explicitly choose plot locations to avoid 2014), reflecting differences in height allometries (Feldpausch
canopy gaps or areas of recent natural disturbance (e.g. Kitayama& et al., 2012) and crown form (Ploton et al., 2016), and potentially
Aiba, 2002; Baez et al., 2015), and plot locations tend to be biased also rates of heartrot (Heineman et al., 2015) and crown breakage
towards taller forests even when methods do not explicitly state (Arellano et al., 2019). Such differences are only partially captured
such criteria (Sheil, 1996;Marvin et al., 2014). Plots also tend to be with generalized allometric equations which at best incorporate
located in more accessible areas, which have a stronger signature of local height measurements and associated differences in diameter-
past human land use (McMichael, CNH et al., 2017) and current height allometries, continuous terms for climate variation and/or
human impacts (McMichael, CH et al., 2017). different equations for different regions or forest types (Chave et al.,
Estimation of AGB and AWP depend on biomass allometry 2005, 2014).
equations (Box 1), which are a major source of error. These Estimates of AWP suffer from additional sources of error. They
equations estimate individual tree aboveground woody biomass depend on diameter growth measurements, and thus are highly
from measured tree diameter, and sometimes also tree height and/ sensitive to diameter measurement errors and to data quality
or wood density (e.g. Chave et al., 2005; Chave et al., 2014). The assurance quality control procedures, including procedures for
key issue for analyses of among-site variation is that studies typically estimating diameter change in buttressed trees (Sheil, 1995;
apply the same equation(s) across many sites. However, biomass Cushman et al., 2014; Muller-Landau et al., 2014). AWP is
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
Predicted AWRT (yr) Predicted AWP Predicted AGB
(Mg C ha–1 yr–1) (Mg C ha–1)
New
Phytologist Tansley review Review 5
temporally variable (e.g. Rutishauser et al., 2020), and thus
1. Productivity
sampling errors for short census intervals are high.At the same time,
typical calculations underestimate AWP in longer census intervals Productivity variables are positively associated with climatic water
because they increasingly miss AWP of trees that die between availability across lowland tropical forests over the range from dry to
censuses (Kohyama et al., 2019). Finally, standard methods for wet forests. Across lowland sites, AWP, litterfall and ANPP are
estimating AWP entirely fail to capture wood production to positively related to climatic water availability in most studies
compensate for branchfall, estimated at 15–45% of total AWP (Fig. 3a), with an initial fast increase slowing to a plateau or even a
(Malhi et al., 2014;Marvin&Asner, 2016;Gora et al., 2019). That mild decrease for precipitation above c. 3000mm yr
1 (Poorter
is, as trees grow, they do not simply accrue biomass, they also shed et al., 2017; Taylor et al., 2017). The positive effects of precipitation
old branches as they produce new ones. weaken and reverse in montane tropical forests (e.g. lowland
Residence time variables have particularly high sampling errors, Hofhansl15b vs montane Hofhansl15c in Fig. 3a; Hofhansl et al.,
which may in part explain the dearth of published analyses. Because 2015). Ameta-analysis of 145 tropical forests found that an increase
tree mortality is a binomial process and mortality rates are low, in mean annual precipitation (MAP) from 1000 to 3000mm was
sampling errors inmortality rates are large, especially in small plots and associated with a 2.3-fold increase in ANPP at 28°C, a 1.5-fold
shorter census intervals. Strong temporal variation in mortality – for increase at 24°C, no change at 20°C and a decrease in ANPP at
example resulting from droughts (Bennett et al., 2015) –makes it yet temperatures below 20°C (Taylor et al., 2017).
more difficult to capture long-term mean mortality rates. Tree Lower forest productivity at lower precipitation reflects limita-
turnover rates, calculated as the average of mortality and recruitment tion by water availability and/or drought stress when potential
rates, suffer these same problems. Syntheses of among-site patterns in evapotranspiration exceeds precipitation, combined with alloca-
mortality and turnover are further hindered by variability in methods tional changes and compositional shifts towards drought-tolerant
for calculating mortality rates, inadequate reporting of calculation species (Flack-Prain et al., 2019). Limited water availability
methods, and systematic biases in many estimators (Kohyama et al., translates into reduced GPP through both reduced leaf area
2018) (seeNotes S1). Calculating AWRT as the quotient AGB/AWP maintained (including drought deciduous leaf phenology) and
(Box 1) only partially avoids this issue, as AWP estimates also depend reduced photosynthesis per available leaf area as plants close their
on mortality (because trees that die do not contribute to AWP). Such stomata and/or invest in more drought-tolerant organs with lower
estimates ofAWRTalsomaybebiasedby the equilibriumassumption light-use efficiency (LUE) (Tan et al., 2013; Guan et al., 2015;Wu
that underlies them (see Notes S1). et al., 2016; Pfeifer et al., 2018). Higher precipitation also is
Finally, most estimates of AGB, AWP and AWRT omit smaller associated with higher allocation of aboveground NPP to AWP
trees, lianas, epiphytes, herbaceous plants and nonwoody tissues, (Hofhansl et al., 2015) and taller trees for a given diameter (Banin
and (by definition) belowground biomass; these are generally et al., 2012), further contributing to higher AWP. Compositional
assumed to be relatively small and/or to vary proportionately. shifts also contribute: species found in drier forests have lower
These assumptions, and other aspects of measurement methods growth rates than those restricted to wetter forests (Baltzer &
and associated errors are discussed in more detail in Notes S1. Davies, 2012; Brenes-Arguedas et al., 2013; Kupers et al., 2019),
because drought-tolerance traits, such as narrower xylemvessels, are
costly (Gorel et al., 2019), whereas the ‘drought-avoiding’
IV. Climatic water availability
deciduous strategy involves foregoing photosynthesis in part of
Precipitation patterns vary among tropical forests from those that the year (Brenes-Arguedas et al., 2013).
receive abundant precipitation year-round (wet tropical forests) to Although the direct effects of water availability on productivity
those that experience limitations in water availability during one or are positive, higher rainfall also is associated with increased
two dry seasons (moist and dry tropical forests), variation we cloudiness and decreased soil fertility, both of which depress
encompass under the term climatic water availability. This productivity, and may explain declining productivity at very high
variability is evident in the large range ofmean annual precipitation rainfall and lower temperatures (Taylor et al., 2017). Wetter sites
among tropical forests (Fig. S2). In general, the length and intensity on average have higher cloudiness and thus reduced light
of dry seasons aremore important than total annual precipitation in availability (Wagner et al., 2016). High precipitation also is
determining forest C stocks and fluxes. Further, water limitation associated with soil-mediated reductions in productivity as a
depends not only on precipitation, but also on potential evapo- consequence of leaching of nutrients and reduced soil redox
transpiration (itself dependent on temperature, solar radiation), as potential; these influences are relatively more important at cooler
well as soil depth, soil water-holding capacity and topographic temperatures. Decreases in productivity with precipitation at the
position. Many analyses thus evaluate relationships with more very highest levels of precipitation, especially in cooler sites (Taylor
integrativemeasures of climatic water availability such as dry season et al., 2017) likely reflect these correlated increases in limitation by
length or maximum climatological water deficit, which are light and nutrients.
generally better predictors of forest structure and dynamics (e.g.
Alvarez-Davila et al., 2017). Here, we discuss how our focal
2. Residence time
variables vary with climatic water availability, and evaluate patterns
in relation to the range of annual precipitation and temperature Few studies have evaluated how among-site variation in AWRT,
within studies (Figs 3, S3). mortality or turnover relate to climatic water availability, and those

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
6 Review Tansley review Phytologist
(a) Productivity vs moisture (c) Biomass vs moisture
AWP* AGB*Poorter17(201) Levine16b(na)
Quesada12(59) AWP* AGB*Poorter16(43)
deSouza19(90) AWP* AGB*
Vilanova18(50) AWP*
Poorter17(201)
AGB*
Rozendaa17(61) AWP* Lewis13(260) AGB*
Taylor17b(244) AWP* Moore18a(14)
Moore18b(8) AWP
AGB*
Slik13(120)
Malhi15a(10) AWP AGB*Reis18(16)
Sullivan20(590) AWP AGB*Turner18(32)
AGB*
Duran13(145)
Banin14(28) BAP* AGB*
Dattaraj18(19) BAP Becknell12b(178)
AGB*
Baez15(45) BAP Becknell12a(51)
AGB*
Vilanova18(50)
Hofhansl15b(62) ANPP AGB*
Taylor17a(145) ANPP*
Rozendaa17(61)
AGB*
ANPP Lloyd15b(9)Hofhansl15c(43) AGB
AlvarezD17(200)
AGB
Taylor17c(428) Litter* Slik10(83)
Chave10(51) Litter
AGB
Silver00a(143)
AGB
NPP Alamgir16b(34)Moore18a(14) AGB*Sullivan20(590)
Malhi15a(10) NPP AGB
Quesada12(59)
AGB
Malhi15a(10) GPP* deSouza19(90)
AGB
500 1000 2000 5000 10 000 Schietti16(55) AGB
Mean annual precipitation (mm) Hofhansl20(20)
AGB
Alamgir16a(24)
(b) Residence time vs moisture CanHt*
Reis18(16)
AWRT CanHt*
Moore18b(8) Lloyd15b(9)
AWRT*
Sullivan20(590)
AWRT BA*
Galbrait13a(105) Toledo11(220) BA
Dattaraj18(19)
BA
Malhi06(227)
Turn* BA
Quesada12(59) Slik10(83)
Turn* BA
Vilanova18(50) Alamgir16b(34)
Turn BA
Baez15(45) Alamgir16a(24)
500 1000 2000 5000 10 000 500 1000 2000 5000 10 000
Mean annual precipitation (mm) Mean annual precipitation (mm)
Fig. 3 Literature results on spatial variation in productivity (a), residence time (b) and aboveground biomass (c) with precipitation, dry season length and other
measuresof climaticwateravailability,graphed in relation to the rangeofprecipitation in thestudysites (ona log-scale).Blue indicates thatproductivity, residence time
or biomass tend to be higher in wetter sites; orange indicates that they tend to be higher in drier sites; dashed blue and orange variable pattern that depends on the
range of the independent variable or on temperature; and black indicates no relationship. Asterisks indicate statistically significant effects. Bold highlights studies in
whichmedian plot area is ≥ 1 ha,whereas results for studieswith smaller plot sizes are shown in italics. Note that the patterns always are reported here in terms of the
responseofproductivity, residence timeorbiomass, even if the responsemetric is inversely related to these (e.g. ablue turnover result indicates that inwetter sites tree
turnover is lower implying residence time is higher). These results are graphed in relation to temperature range in Supporting Information Fig. S3. AGB, aboveground
biomass; ANPP, aboveground net primary productivity; AWP, abovegroundwoody productivity; AWRT, abovegroundwoody residence; BA, basal area; BAP, basal
areaproductivity;CanHt, canopyheight;GPP,grossprimaryproductivity; Litter, litterfall; na, notavailable;NPP,netprimaryproductivity;Turn, tree turnover rate. See
Box1,Fig. 1andNotesS1fordefinitions,measurementmethodsand inter-relationshipsof these responsevariables.Literature resultsarecodedbythefirsteight letters
of the first author’s name, the last twodigits of the year, a letter indicatingwhich set of siteswithin the publication (if there ismore thanone set of sites for the study in
the database), and the number of sites included within parentheses (Dataset S1).
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 7
that do have found at best weak relationships (e.g. Quesada et al., Additional data and analyses are needed to establish whether/how
2012; Vilanova et al., 2018). More studies have found trends for mortality rates vary spatially with climatic water availability, and to
AWRT to be higher (and turnover lower) in wetter sites than the investigate the role of compositional shifts in contributing to
opposite, but overall patterns are inconsistent (Fig. 3b). This may variation in C fluxes and stocks. The role of lianas deserves more
reflect contrasting trends in different mortality threats with attention, as lianas are more abundant in drier sites (DeWalt et al.,
precipitation regimes. Drier sites are more likely to experience fire 2010), and could contribute to their lower tree productivity and
(Cochrane, 2011) and drought stress elevates mortality through possibly lower residence time.
hydraulic damage (Choat et al., 2018), whereas higher rainfall is
associated with greater risks of mortality from treefalls, lightning
V. Temperature and elevation
and landslides (Espirito-Santo et al., 2010; Yanoviak et al., 2020).
By contrast with the paucity of studies of spatial variation, there Most temperature variation across tropical forests is explained by
have been multiple studies of temporal variation. Many studies have elevation (Pearson r =
0.96 across 14,643 1-km pixels; Fig. 5a),
documented elevated mortality in drought years (reviewed in Phillips and thus our understanding of temperature influences is based
et al., 2010; Bennett et al., 2015), whereas a few have found higher largely on elevational variation. However, it is important to keep in
mortality in wetter years (Aubry-Kientz et al., 2015) or wetter seasons mind that elevational temperature variation is confounded with
(Brokaw, 1982; Fontes et al., 2018). Patterns of temporal variation in other factors. Atmospheric pressure decreases systematically with
mortality with water availability do not necessarily predict among-site elevation, which affects photosynthesis both directly and indirectly
variation because compositional shifts at least partially compensate for by altering selection on photosynthetic traits (Wang et al., 2017).
shifts in mortality threats. For example, tree species common in drier Cloud cover (and thus solar radiation) and precipitation also
sites have higher survival under drought than those common in wetter change with elevation (Fig. 5b,c), as do other climate variables and
sites (Engelbrecht et al., 2007; Baltzer & Davies, 2012; Brenes- geomorphology (Porder et al., 2007). Indeed, across tropical forests
Arguedas et al., 2013; Esquivel-Muelbert et al., 2017). globally,mean cloud cover increases from57%at 29°C to c. 89%at
8°C (Fig. S4). Here we synthesize results for the many observa-
tional studies of variation with elevation and the few with
3. AGB
temperature, and graph results in relation to the ranges of
Aboveground biomass is positively related to climatic water temperature, elevation and precipitation represented in each study
availability in tropical forests in 16 of 16 studies finding a statistically (Figs 6, S5).
significant relationship (Fig. 3c). The relationship of AGB with
precipitation exhibits an initially steep increase below2000mmyr
1
1. Productivity
gradually saturating at higher precipitation (Becknell et al., 2012;
Poorter et al., 2016; Alvarez-Davila et al., 2017). Increases are All productivity variables decline with elevation (Fig. 6a), suggest-
roughly parallel in old-growth and secondary forests: over 1000– ing a positive effect of temperature, but analyses with temperature
3000mm MAP, AGB increases two-fold in 20-yr-old secondary find both positive and negative effects (Fig. 6a,d). Overall patterns
forests (Poorter et al., 2016) and c. 2.3-fold in mature forests seem consistent with a positive effect of temperature inwet sites and
(Alvarez-Davila et al., 2017). Qualitatively the same patterns are a negative effect in dry sites. This is particularly apparent in studies
found for tree basal area and canopy height, for both plot-based and that evaluate interactions of climatic water availability and
remote sensing studies, and inbothold-growth and secondary forests temperature (Taylor et al., 2017; Sullivan et al., 2020). A meta-
of a given age (Fig. 3c).Measures of drought stress such as dry season analysis found that ANPP (litterfall) decreased with temperature
length or dry season water deficit are generally better predictors of for precipitation below c. 1400 mm yr
1 (1600 mm yr
1), and
AGB than precipitation alone, and exhibit more linear relationships increased with temperature for precipitation above that level, with
with AGB (Poorter et al., 2016; Alvarez-Davila et al., 2017). At ever-faster increases for higher precipitation (Taylor et al., 2017).
extremely high precipitation levels above c. 4000mm yr
1, AGB At 2500 mm MAP, ANPP doubles between 10 and 22°C and
maydecreasewith further increases inprecipitation, but there are few triples by 28°C (Taylor et al., 2017).
data for such sites, and spatial variation in precipitation may be Spatial variation in AWP with temperature can be explained in
confounded with solar radiation, soil fertility and other factors large part by the temperature responses of plant metabolic rates –
(Alvarez-Davila et al., 2017). Overall the patterns in AGB parallel photosynthesis and respiration. Across sites, the optimum temper-
those in AWP, consistent with what would be expected given little ature for photosynthesis is strongly positively correlated with mean
variation in AWRT with precipitation (Fig. 4a). growing season temperature (Tan et al., 2017), and the photosyn-
thetic rate at the temperature optimum increases with temperature,
meaning warmer sites are expected to have higher photosynthetic
4. Synthesis
rates, if water is not limiting (Farquhar et al., 1980). Maintenance
Overall, patterns of variation in tropical forest productivity and respiration rates also increase with temperature within sites – but
biomass with climatic water availability are relatively well-docu- acclimation means that respiration rates at growth temperatures
mented and well-understood, and the underlying mechanisms are increase very little or not at all (Atkin et al., 2015; Malhi et al.,
increasingly well-represented in forest and vegetation models 2017). Biomass accumulation rates increase with temperature in
(Christoffersen et al., 2016; Levine et al., 2016; Xu et al., 2016). well-watered conditions (Cheesman & Winter, 2013), likely

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
8 Review Tansley review Phytologist
(a) Climatic water availability
AWP × AWRT = AGB AWP × AWRT = AGB
(b) Elevation
AWP × AWRT = AGB AWP × AWRT = AGB
(c) Soil fertility
AWP × AWRT = AGB
AWP × AWRT = AGB AWP × AWRT = AGB
Infertile Fertile
(d) Disturbance
Fig. 4 Schematic of patterns of variation in
tropical forest aboveground woody
productivity (AWP), residence time (AWRT)
and biomass (AGB) with climatic water
AWP × AWRT = AGB availability (a), elevation in moist or wet sites
AWP × AWRT = AGB (b), soil fertility (c) and disturbance (d). Text
size reflects variation in a given variable along
the environmental gradient (e.g. AWP and
AGB increase with climatic water availability)
(watercolors by K. T. Anderson-Teixeira).
reflecting an increase in biosynthesis rates. By contrast, where water LUE and, thus, stand-level productivity (Reich, 2014). Cooler,
is limiting, photosynthesis decreases with temperature as a result of higher elevation sites also tend to have higher allocation below-
increased stomatal closure and higher respiratory costs (Schippers ground, a pattern consistent with increased nutrient limitation
et al., 2015). Overall, for any given plant and site, net photosyn- (Hofhansl et al., 2015). This allocational shift could reconcile
thesis is expected to be a unimodal function of temperature, stronger elevational decreases in ANPP with weaker patterns in
reflecting biochemically determined unimodal responses of max- total NPP. Among water-limited sites, increasing temperature
imum photosynthetic rates in combination with stomatal conduc- increases drought stress, potentially leading to the same types of
tance and respiration (Slot & Winter, 2017). allocational and compositional shifts expected under reduced
Allocational and compositional shifts also contribute to spatial climatic water availability.
variation in AWPwith temperature. Cooler sites tend to have plant Finally, correlated variation in other environmental factors also
species with higher nutrient use efficiencies, longer-lived leaves, influences patterns with temperature among tropical sites. Cooler
higher LMA (Asner & Martin, 2016) and other slow life-history tropical forests are found overwhelmingly at higher elevations,
traits (Dalling et al., 2016; Bahar et al., 2017). These traits increase where cloud cover is higher and fog is more frequent, thereby
competitiveness in lower resource environments, while reducing decreasing solar radiation and increasing light limitation
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 9
(a) 30
20
10
0
(b) 100
Fig. 5 Variation in the distributions of mean
annual temperature (a), mean cloud cover (b) 75
andmeanannual precipitation (c) in relation to
elevation in tropical forests. Panels show violin
plots of the distribution across 1-km pixels, 50
with the red dots indicating medians. Tropical
forest area was defined based on SYNMAP
(Jung et al., 2006) as land between 23.44°S 25
and 23.44°N latitude, in land cover types
classified as ‘trees’ (see Supporting
Information Fig. S6; see also Figs S7, S8 for (c)
versions including additional land-cover 6000
types). Mean elevation data from SRTM
(https://cgiarcsi.community/data/srtm-90m-
digital-elevation-database-v4-1/); mean 4000
annual temperature and precipitation from
CHELSA (http://chelsa-climate.org/); and
cloud cover fromWilson & Jetz (2016) 2000
(https://journals.plos.org/plosbiology/artic
le?id=10.1371/journal.pbio.1002415). The
violin plots for annual precipitation are 0
truncated at 6000mm for graphing (at most
00 00 00 00 00 00 00 00 00 00 00
0.7% of data were above 6000mm in any < 3 –60 0–
9 12 15 18 21 24 7 0 0– – – – – –2 –3  3
elevation class); the form of the plots and the 30 60 00 00 00 09 2 5 80 10
0 00 00 >4 7
location of the medians are based on the 1 1 1 2 2 2
complete untruncated datasets. Elevation range (m)
(Bruijnzeel et al., 2011). Cooler temperatures also slow decompo- is consistent with the global pattern of a positive correlation
sition (Taylor et al., 2017) and reduce biological nitrogen (N) between tree productivity andmortality (Stephenson&Mantgem,
fixation (Houlton et al., 2008), which tends to reduce nutrient 2005), given that higher elevations tend to be associated with lower
availability, especially N availability (Wilcke et al., 2008; Notting- productivity and slower life histories (e.g. lower LMA; Asner &
ham et al., 2015).However, higher elevation and thus cooler forests Martin, 2016).
tend to be found on geochemically young substrates with eroding
slopes, which are associated with relatively higher availability of
3. AGB
rock-derived nutrients (Porder et al., 2007). Thus, for any given
area, elevational variation in cloud cover, rainfall and soils can Aboveground biomass decreases with elevation inmost studies, and
magnify or counter the patterns expected based on temperature canopy height decreases with elevation in almost all studies, but
alone, and interact with compositional shifts (Peng et al., 2020). patterns of basal area variation are decidedly mixed, as are patterns
of AGBwith temperature (Fig. 6c,f). It is notable that some studies
find very high or even the highest AGB at intermediate or high-
2. Residence time
elevation sites (e.g. Girardin et al., 2010); the mechanisms
Few studies have evaluated howAWRT,mortality or turnover rates underlying these exceptions are an important area for future
vary with temperature or elevation, and relationships were not research. In terms of the quantitative strength of these effects,
statistically significant in most studies (Fig. 6b,e). Of the four regressions of AGB on elevation in Bolivia, Peru and Ecuador
studies finding significant relationships with elevation, three show found that AGB decreases 32, 34 and 50Mgha
1 per 1000 m
higher AWRT (lower turnover) at higher elevation (Fig. 6b). This elevation, respectively (Girardin et al., 2014). Overall, the patterns

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
Precipitation (mm) Cloud cover (% d) Temperature (°C)
New
10 Review Tansley review Phytologist
(a) Productivity vs elevation (c) Biomass vs elevation (d) Productivity vs temperature
AGB*
Malhi17a(16) AWP* Girardin14c(43) Taylor17b(244)
AWP*
AGB* AWP*
Vilanova18(50) AWP* Girardin14d(8)
AWP* Girardin14e(20) AGB*
Sullivan20(590)
Unger12b(32) AGB* deSouza19(90)
AWP*
Kitayama02(8) AWP* Gonmadje17(15)
AWP* Phillips19(20) AGB*Sherman12(75)
Sherman12(75) AGB* ANPP*
Baez15(45) BAP* Alamgir16b(34)
AGB Hofhansl15c(43)
BAP* Grubb77a(12) AGB ANPP*Wolf11(54) Raich06b(11)
Homeier10b(11) BAP* Carey94(17)
AGB
Hofhansl15b(62) ANPP
Wilcke08(8) BAP* Unger12a(80)
AGB
Taylor17a(145) ANPP*
Unger12b(32) BAP Kitayama02(8)
AGB
BAP Aiba99(8) AGBClark15(9) Litter*
Slik10(83) AGB Raich06d(30) Litter*
ANPP* Venter17b(172) AGB Taylor17c(428)Girardin10a(9)
ANPP* Alamgir16a(24) AGBKitayama02(8) 10 15 20 25 30
Alves10(13) AGB*
AGB* Mean annual temperature (°C)
Grubb77c(9) Litter Vilanova18(50)
Kitayama02(8) Litter
Wolf11(54) Litter* Girardin14c(43)
CanHt*
Asner14(20) CanHt*
CanHt*
Malhi17a(16) NPP* Girardin14d(8) CanHt* (e) Residence time vs temperature
Girardin10a(9) NPP Girardin14e(20) AWRT
Wolf16c(2264) CanHt* Galbrait13a(105)
GPP* Unger12c(29) CanHt*Malhi17b(8)
Clark15(9) CanHt
AWRT
Sullivan20(590)
0 1000 2000 3000 4000 Wolf16b(1631) CanHt
Elevation (m) Grubb77d(na) CanHt
10 15 20 25 30
Mean annual temperature (°C)
Aiba99(8) CanHt
(b) Residence time vs elevation Gonmadje17(15)
BA*
Homeier10a(17) BA* (f) Biomass vs temperature
AWRT BA*
Malhi17a(16) Wilcke08(8) AGB*
AWRT Alamgir16b(34)
BA Raich06a(20)
Galbrait13c(17) Naveenku17(15) BA
Clark15(9) AWRT*
AGB
Aiba99(8) BA deSouza19(90)
Girardin14c(43) BA
Girardin14d(8) BA AGB*Sullivan20(590)
Carey94(17) Mort Girardin14e(20)
BA
Wolf11(54) BA AGB
BA Lewis13(260)Slik10(83)
Turn*
Bellingh09(16) Lieberma96(11)
BA
BA AGB*Clark15(9) Slik13(120)
Vilanova18(50) Turn* Alamgir16a(24) BA
Turn* AGB*
Baez15(45) Unger12a(80)
BA* Duran13(145)
0 1000 2000 3000 4000 0 1000 2000 3000 4000 10 15 20 25 30
Elevation (m) Elevation (m) Mean annual temperature (°C)
Fig. 6 Literature results on spatial variation in productivity (a, d), residence time (b, e) and abovegroundbiomass (c, f)with elevation (a–c) or temperature (d–f),
graphed in relation to the range in elevation or temperature, respectively, in the study sites. Red indicates that productivity, residence timeor biomass tend to be
higher in lower elevation or warmer sites; purple indicates that they tend to be higher in higher elevation or cooler sites; black indicates no relationship; and
dashed red andpurple that they exhibit a variable relationshipdependingeither on the rangeof the independent variableor onaprecipitation variable.Asterisks
indicate statistically significant effects. Bold highlights studies inwhichmedian plot area is ≥ 1 ha,whereas results for studieswith smaller plot sizes are shown in
italics. These results are graphed in relation to precipitation range in Supporting Information Fig. S8. AGB, aboveground biomass; ANPP, aboveground net
primary productivity; AWP, aboveground woody productivity; AWRT, aboveground woody residence; BA, basal area; BAP, basal area productivity; CanHt,
canopy height; GPP, gross primary productivity; Litter, litterfall; na, not available; NPP, net primary productivity; Turn, tree turnover rate. Literature results are
codedby thefirst eight letters of thefirst author’s name, the last twodigits of theyear, a letter indicatingwhich set of siteswithin thepublication, and thenumber
of sites included within parentheses (Dataset S1). Response variable and study abbreviations as in Fig. 3 (Dataset S1).
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 11
in AGB with elevation and temperature largely mirror those in nutrients in plant function. Higher soil nutrients enable higher
AWP. plant nutrient content (Fyllas et al., 2009; Cleveland et al., 2011;
Asner & Martin, 2016), which in turn enables greater plant LUE
(Elser et al., 2010). Higher soil nutrient availability also means that
4. Synthesis
plants need to spend fewer resources on nutrient acquisition,
The biochemical and physiological mechanisms by which temper- whether in constructing roots or supporting microbial symbionts,
ature interacts with water availability to affect plant productivity are which enables higher fertility forests to turn a higher proportion of
relatively well understood. These are central to responses to short- their GPP into AGB production (Vicca et al., 2012;Doughty et al.,
term temporal variation in temperature within sites, which is 2018). However compositional shifts partly compensate, as low-
reasonably well captured in mechanistic models (Schippers et al., fertility sites have species with better nutrient acquisition abilities
2015). By contrast, responses to spatial variation in temperature and higher nutrient-use efficiencies, reducing productivity differ-
regimes depend in large part on acclimation, allocational shifts and ences with soil fertility (Gleason et al., 2009; Dalling et al., 2016;
compositional variation, and remain poorly understood. Composi- Turner et al., 2018). In addition, herbivory and liana abundance
tional patterns, such as the decline in lianas and palms with elevation increase with soil fertility; it may be that these consumers and
(e.g. Lieberman et al., 1996), are likely to be major contributors to structural parasites capture a disproportionate share of the benefits
among-site variation in tropical forest C cycling with elevation and of elevated nutrient availability (Schnitzer & Bongers, 2002;
temperature; they deserve more attention. Finally, among-site Campo & Dirzo, 2003). The consequence of these compositional
patterns may vary not only with mean temperatures, but also with shifts and biotic interactions is that the increase in stand-level AWP
extremes (e.g. relationships with maximum temperature were more with fertility is lower than would be expected based on single-
often negative than those with mean temperature) (Dataset S1). species responses in isolation, and may even be absent (e.g. Turner
et al., 2018).
VI. Soil fertility
2. Residence time
Tropical forests exhibit great heterogeneity in their biogeochem-
istry, reflecting wide variation in soil age, chemistry, and suscep- Soil fertility is positively associated with tree mortality rates and
tibility to erosion or uplift, as well as high plant diversity; diversity thus negatively associated with AWRT across tropical forests
matters because plants can affect soil properties under their crowns (Fig. 7b). This pattern has been found at local (de Toledo et al.,
(Townsend et al., 2008; Waring et al., 2015). Soil fertility is 2011; Sawada et al., 2015), regional (Quesada et al., 2012) and
multidimensional, involvingmany different nutrients important in global (Galbraith et al., 2013) scales. The variation is substan-
different ways (Kaspari & Powers, 2016), and available in different tial, eclipsing both variation in productivity with soil fertility
concentrations and forms at different soil depths, that covary across and variation in AWRT with climate. For example, across 59
sites (e.g.Quesada et al., 2010).Many studies thus evaluate patterns sites in the Amazon, turnover increased three-fold from low to
with respect to principal components axes or soil classes that reflect high soil P (Quesada et al., 2012). Pantropical analyses also
covariation in multiple nutrients (‘Multi’ in Fig. 7). In cases where found strong relationships, with median AWRT increasing
individual studies investigated relationships with multiple soil c. 50% from young to old soils in Neotropical forests, and from
fertility variables, we report results relative to the variable showing intermediate to old soils in Paleotropical forests (Galbraith
the strongest relationship with the dependent variable. et al., 2013).
Three classes of mechanisms likely contribute to higher
mortality at higher soil fertility. First, higher growth at higher soil
1. Productivity
fertility speeds the rate of self-thinning, thereby increasing
Values for AWP, BAP, ANPP and litterfall are positively related to associated mortality rates (Stephenson & Mantgem, 2005).
soil fertility in tropical forests. Of 22 analyses of among-site Second, more productive environments select for tree species with
variation, 21 showed a positive trend and 16 were significantly ‘fast’ life-history strategies such as lowwood density (Quesada et al.,
positive (Fig. 7a). Fertilization experiments further demonstrate 2012), and given underlying tradeoffs, these species also have
that tropical forest productivity is limited by P and by N, and higher mortality rates (Stephenson&Mantgem, 2005; Kraft et al.,
suggest that potassium (K) and calcium (Ca) alsomight be limiting 2010;Wright et al., 2010; Reich, 2014). Third, higher soil fertility
– only one tropical forest fertilization experimentmanipulatedKor is associated with higher liana abundance (Putz & Chai, 1987;
Ca (Wright, 2019). However, the range of AWP variation Laurance et al., 2001; Schnitzer & Bongers, 2002; DeWalt et al.,
explained by fertility seems to be relatively smaller than that 2006), and higher liana abundance is associated with higher tree
explained by climate; for example, AWP on high-P soils averages mortality in observational and experimental studies (Ingwell et al.,
c. 20% higher than AWP on low-P soils in the Amazon and Sierra 2010; van der Heijden et al., 2015; Wright et al., 2015).
Leone (Quesada et al., 2012; Jucker et al., 2016). This may in part
reflect shifts in allocation with fertility, with increased allocation to
3. AGB
reproduction in more fertile sites (Wright et al., 2011).
The increase in woody productivity with soil fertility is The combination of increasing AWP and decreasing AWRT with
consistent with our mechanistic understanding of the role of fertility would lead to the expectation of a unimodal relationship of

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
12 Review Tansley review Phytologist
(a)  Productivity vs fertility (c) Biomass vs fertility
Banin14(28) AWP* * * Slik13(120) AGB*
Quesada12(59) AWP* * ~ ~ Hofhansl20(20) AGB*
Toledo17(72) AWP* Toledo17(72) AGB*
Toledo17(72) AWP* Laurance99(63) * ~ ~ * AGB*deSouza19(90) * AWP* Grau17(9) AGB*
Kitayama02(8) AWP*
Paoli07(30) AWP* ~ Paoli08(30) AGB** * *
Jucker16(142) AWP* Poorter17(201)
AGB
Sullivan20(590) AWP Aiba99(8) AGB
Poorter17(201) AWP Soong20a(10) AGB
Grau17(9) AWP ~ ~ deSouza19(90) AGB ~
Poorter16(43) AGB
Banin14(28) BAP* AGB** Lewis13(260)
Homeier10c(15) BAP* Turner18(32) AGB
Sullivan20(590) AGB
Hofhansl15b(62) ANPP* Schietti16(55) AGB*
Hofhansl15c(43) ANPP* Quesada12(59) ~ AGB*
Kitayama02(8) ANPP * *
Paoli05(30) ANPP Aiba99(8) CanHt
Clevelan11b(32) ANPP
vanSchai85(17) Litter* Homeier10d(8) BA* *
Chave10(51) Litter* Sellan19(16) BA* ~ ~
Paoli07(30) Litter* Aiba99(8) BA** *
Toledo11(220) BA
Aragao09(10) NPP* Baraloto11(74) BA*
Multi P CEC Bases Other Multi P CEC Bases Other
(b) Residence time vs fertility
Sullivan20(590) AWRT
Galbrait13b(96) AWRT*
Grau17(9) Mort ~ ~
Toledo17(72) Mort
Sawada15(9) Mort
Soong20b(9) Mort*
Sawada15b(9) Mort*
Quesada12(59) ** * Turn *
Bellingh09(16) Turn*
Multi P CEC Bases Other
Fig. 7 Literature results on spatial variation in productivity (a), residence time (b) and aboveground biomass (c) with soil fertility, graphed in relation to the soil
fertilitymeasure used (Multi, a soil fertility axis or classification that encompassedmultiple nutrients; P, phosphorus; CEC, cation exchange capacity; Bases, total
soil bases; Other includes studies using nitrogen, potassium, magnesium and calcium. Green indicates that productivity, residence time or biomass tend to be
higher in more fertile sites; brown indicates that they tend to be higher in less fertile sites, and black indicates no relationship or an inconsistent relationship.
Asterisks indicate statistically significant effects. Bold highlights studies inwhichmedian plot area is ≥ 1 ha,whereas results for studieswith smaller plot sizes are
shown in italics. For studies that investigatemultiple soil fertilitymeasure, the text denoting the response variable is graphed in the columncorresponding to the
variable that exhibited the strongest relationship; additional results for other types of soil variables are indicatedwith an asterisk for significant results and a tilde
for others. In somecases results for secondaryvariables reflectweaker tests of effects (e.g. correlations) than themain results (e.g.multiple regression), and thus
the secondary results can be significant while the primary results are not (e.g. turnover results for Quesada et al., 2012). AGB, aboveground biomass; ANPP,
aboveground net primary productivity; AWP, aboveground woody productivity; AWRT, aboveground woody residence; BA, basal area; BAP, basal area
productivity; CanHt, canopy height; GPP, gross primary productivity; Litter, litterfall; NPP, net primary productivity; Turn, tree turnover rate. Literature results
are coded by the first eight letters of the first author’s name, the last two digits of the year, a letter indicating which set of sites within the publication, and the
number of sites included within parentheses (Supporting Information Dataset S1).
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 13
AGBwith fertility,withAWP limiting at the low end andAWRTat and intensity of large-scale tropical cyclones (known regionally as
the high end (Fig. 4c). Empirical studies have variously found hurricanes, typhoons or cyclones) is near zero in tropical forests
positive, negative and no relationships of tropical forest AGB to soil with latitudes < 10°, and varies strongly among other areas (Ibanez
fertility (Fig. 7c). For example, AGB decreased 1.4-fold from low et al., 2019). Convective thunderstorms and lightning occur across
to high soil P across 59 plots in the Amazon (Quesada et al., 2012), the tropics, and both show strong geographical variation in
and decreased c. two-fold from the lowest to highest total base frequency (Pereira-Filho et al., 2015; Gora et al., 2020). Within
cations across 260 plots in Africa (Lewis et al., 2013), whereas it sites, storm impacts vary topographically, reflecting variation in
increased 1.4-fold with soil nitrogen across 63 plots in the central wind exposure (highest on ridges; Boose et al., 1994), soil saturation
Amazon (Laurance et al., 1999). These different patterns are (highest in floodplains and concave topographies; Margrove et al.,
consistent with what we might expect if studies span different parts 2015), and landslide risk (highest on steep slopes; Larsen&Torres-
of an overall unimodal relationship. Because the decrease inAWRT Sanchez, 1998). Wildfire risk increases with dry season length and
is greater than the increase in AWPwith fertility, we expect the peak intensity, as well as with proximity to anthropogenic disturbance
to be located closer to the lower fertility end of the gradient. The (Cochrane, 2011).
location of the peak in AGB with respect to soil fertility is likely to Disturbance directly increases tree mortality and decreases
vary across regions, reflecting compositional differences among AWRT, thereby reducing AGB (Fig. 4d). Both large-scale
regions and strong interspecific variation in mortality rates and cyclones and local convective storms increase tree mortality
responses to soil fertility (Condit et al., 2006, 2013). from treefalls (including landslides) (Larsen & Torres-Sanchez,
1998; Ostertag et al., 2005; Negr
on-Ju
arez et al., 2017; Hall
et al., 2020) and convective thunderstorms also kill trees via
4. Synthesis
lightning (Yanoviak et al., 2020). Across tropical forests, higher
It has long been clear that soil fertility plays a critical role in tropical lightning frequency is associated with higher biomass turnover
forest structure and function (Vitousek & Sanford, 1986), and the rates and lower old-growth forest biomass (Gora et al., 2020).
broad outlines of its importance are evident in studies to date Higher tropical cyclone frequency is associated with lower
(Fig. 7). A central challenge is that tropical tree species display a canopy height and higher stem density, reflecting an increasing
wide diversity of strategies for nutrient acquisition and use, number of smaller stems (Ibanez et al., 2019). In humid tropical
strategies that are critical to compositional shifts and stand-level forests, median canopy height was 1.3-fold higher where cyclone
responses to soil fertility, and their regional variation (Laliberte frequency averaged less than one per century than where it
et al., 2017). Yet our understanding of these strategies – which averaged greater than one per decade (Ibanez et al., 2019).
include not only root morphology and foraging behavior, but also Topographic variation in storm impacts is evident in mortality
chemical root exudates and interactions with microbial symbionts patterns; e.g. cyclone mortality rates are higher in areas with
– remains very limited, reflecting the general paucity of data on greater wind exposure (Negron-Juarez et al., 2014). Fires kill
roots and belowground interactions. trees directly and also increase mortality rates in subsequent
New data, analyses and modeling are needed to advance our years, especially in wetter forests (Barlow et al., 2003), and areas
understanding of soil fertility’s role in structuring variation in that have experienced fires have lower biomass stocks than
tropical forests. More, better and more consistent data on tropical unburned areas for decades afterwards (Gerwing, 2002; Sato
soils are a critical component, especially in enabling better analyses et al., 2016).
of large-scale patterns (Hengl et al., 2017). The ability to estimate Disturbance also influences functional composition, as tropical
foliar nutrients from airborne hyperspectral imaging has enabled tree species differ strongly in how they are affected by disturbances
large-scale data collection of these quantities and their relation to (Zimmerman et al., 1994; Curran et al., 2008; Slik et al., 2010b;
soils (e.g. Chadwick & Asner, 2018); and satellite hyperspectral Paz et al., 2018; Staver et al., 2019). In general, species with ‘faster’
missions promise further advances (Schimel et al., 2013). Earth life histories are able to rebound more quickly following distur-
systemmodels are starting to incorporate nutrientsmechanistically, bances, and thus are more common in areas with recent
and can provide useful tools to explore associated mechanisms and disturbances (Paz et al., 2018). Associated tradeoffs mean that
link them to patterns at different levels (Medvigy et al., 2019; disturbances generally increase the relative abundance of tree
Sulman et al., 2019). species with fast life histories, which tend to have low wood
densities and achieve low biomass (Carreno-Rocabado et al., 2012;
Paz et al., 2018). Lianas also proliferate after disturbances, and thus
VII. Disturbance
high disturbance frequency increases liana abundance (Schnitzer&
Tropical forests vary strongly in the frequency and intensity of Bongers, 2011). Different disturbances also can favor particular
natural disturbances, with important consequences for forest traits; for example, species with higher wood density are less likely
structure, dynamics and composition. Here, we focus specifically to suffer stem breaks during a hurricane (Zimmerman et al., 1994).
on short-term natural disturbances such as storms, landslides and Whereas shifts towards more disturbance-resistant species would
wildfires, excluding disturbance by chronic stressors such as tend to mitigate the direct effects of disturbance on mortality and
drought (addressed under water availability above) and flooding biomass, increases in the abundance of lianas and of tree species
(addressed byDaskin et al., 2019). Variation in natural disturbance with fast life-history strategies would tend to further increase
rates across the tropics is substantial and systematic. The frequency mortality and reduce biomass. Thus, compositional responses to

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
14 Review Tansley review Phytologist
disturbances also need to be considered to determine the total the abundance of small stems and favor the growth of fewer larger
impacts of disturbance regimes on tropical forest structure and trees of higher wood density, resulting in elevated forest C stocks
dynamics. (Berzaghi et al., 2019).
VIII. Biogeographic realm IX. Discussion
Tropical forests on different continents have significantly different Our review of spatial variation in tropical forest C stocks and fluxes
productivity, residence time and biomass. AWP is 25% higher in documented considerable qualitative consistency across studies,
Asian than in Latin American forests (Taylor et al., 2019). Mean while also illuminating areas of divergent results and limited data.
AWRT in old-growth tropical forests also is higher in Asia and AWP and other measures of productivity examined here decrease
Africa than in Latin America, by 22% and 33%, respectively strongly with seasonal water limitation and elevation, and increase
(Galbraith et al., 2013). Consistent with higher AWP and AWRT, weakly with soil fertility. This is consistent with our understanding
AGB is higher in Paleotropical than in Neotropical forests, in both of how water availability, temperature and nutrients affect
plot-based and satellite-based datasets (Lewis et al., 2013; Slik et al., photosynthesis, allocation and functional composition. Favorable
2013; Avitabile et al., 2016; Sullivan et al., 2017; Taylor et al., conditions for photosynthesis (i.e. moist, warm and fertile) lead to
2019). For example, plot-based studies find thatmean AGB is 29% greater allocation to AWP as well as functional shifts towards
higher in Asian than Latin American forests (Taylor et al., 2019), species with greater LUE, such that these indirect effects reinforce
and 26% higher in central Africa than in central Amazonia (Lewis the direct ones. This variation in AWP in turn contributes to AGB
et al., 2013). The dearth of studies of African forests is particularly variation with the same factors, but AGB patterns with climate are
concerning in light of these important biogeographical differences much noisier than AWP patterns, and AGB variation with fertility
(Figs S1, S9). does not necessarily align with AWP (Fig. 4). This reflects the
Tropical forests in different biogeographic regions differ signif- importance of AWRT as a dominant driver of empirical variation
icantly in plant allocation, tree allometry and forest structure. in AGB (Johnson et al., 2016), the limited variation in AWRT that
African forests have a larger proportion of their biomass in the is explained by climate and the strong decrease in AWRT with soil
largest trees than do Neotropical forests (Bastin et al., 2018). fertility. In general, our knowledge of AWRT drivers remains
Allocation of NPP to AWP is substantially higher in Asian than in limited, although we know disturbance decreases AWRT. Overall,
Neotropical forests (Paoli & Curran, 2007; Malhi et al., 2011; high tropical biodiversity challenges our ability to explain patterns
Taylor et al., 2019), which could contribute to the differences in in tropical forest C stocks and fluxes, most obviously in the
AWP. Tropical trees in Asia are taller for the same diameter than substantial differences among biogeographic regions.
those in other tropical regions (Feldpausch et al., 2012), with Africa
intermediate and American trees shortest (Banin et al., 2012).
1. Residence time
These differences in tree height persist even after controlling for
differences in climate and soils, and even when comparing related Abovegroundwoody residence time is determined by treemortality
taxa among regions; for example, Asian trees in the family Fabaceae and branch turnover rates, both of which remain poorly under-
are taller than confamilials in Africa and the Americas (Banin et al., stood, especially in comparison with productivity. Failure to better
2012). understand tree mortality is reflected in models that currently have
Differences in continental averages in part reflect differences in very limited and mostly phenomenological representations of tree
the frequencies of different climate regimes (Parmentier et al., mortality, and thus completely fail to reproduce empirical variation
2007), but substantial differences remain even after controlling for in mortality and AGB (Fig. 2) (Galbraith et al., 2013; Friend et al.,
climate (Corlett & Primack, 2011). These can be explained by 2014; Koven et al., 2015). Our limited understanding of tropical
differences in the composition of plant and animal communities tree mortality ultimately reflects the dearth of high-quality data on
related to historical contingency and evolutionary legacy (Caven- mortality patterns and mechanisms (McDowell et al., 2018). The
der-Bares et al., 2016). Taxonomic composition of tropical forests binomial nature of mortality, the low mortality rates in tropical
varies strongly across biogeographic realms, which align to a large forests, and the relatively high temporal variation inmortalitymean
degree with continents (Slik et al., 2018). Asian tropical forests are that sampling errors in mortality and woody residence time are
dominated by trees in the Dipterocarpaceae, a family that is almost large, such that very large sample sizes (in area and time) are needed
absent in the Americas and Africa. Dipterocarp trees are distinctive to quantify geographical variation with useful precision
in their combination of ectomycorrhizal associations, tall archi- (McMahon et al., 2019). Calculation of woody residence time as
tecture, seed dispersal by wind and mast fruiting (Ghazoul, 2016). the quotient of AGB and AWP provides an alternative approach
Essentially, Asian tropical forests have a plant functional type that is that circumvents some of these problems, but is of course
substantially different from those in other tropical forests, and this dependent on high-quality estimates of AGB and AWP, and has
leads to differences in stand-level AWP and AGB (Cavender-Bares its own pitfalls (Ge et al., 2019). There is an urgent need for much
et al., 2016), as well as selective pressures on co-occurring trees to be more data on tropical tree mortality and woody residence time.
tall also (Banin et al., 2012). Differences among biogeographic Satellite-based methods have the potential to enable these to be
regionsmay also in part reflect differences in the animal community estimated over much larger areas at much finer temporal resolution
(Corlett & Primack, 2011). For example, African elephants reduce (Clark et al., 2004), but this potential has yet to be realized.
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 15
Branch turnover rates also contribute to woody residence time patterns, yet the mechanisms underlying variation in liana
and are even less well understood than mortality. Branch turnover abundance remain little understood (Schnitzer, 2018; Muller-
encompasses both ‘planned’ branchfall as trees drop old branches Landau & Pacala, 2020). Trees with heavy liana infestations had
and build newones, and ‘unplanned’ branchfall (e.g. resulting from approximately half the growth and twice the mortality rates of
damage when a neighboring tree falls). Relatively few studies have liana-free trees in observational studies (Ingwell et al., 2010;Wright
measured branchfall rates directly (but see Palace et al., 2008;Malhi et al., 2015; Visser et al., 2018), and experimental liana removal
et al., 2017; Moore et al., 2018), and spatiotemporal variability in increased tree growth by 25–372% (Estrada-Villegas & Schnitzer,
branchfall is so high that sampling errors in such data are invariably 2018). Thus, lianas decrease AWP, AWRT and, thereby, AGB.
large (Gora et al., 2019).Most AWP estimates from plot recensuses Mean AGB decreases more than two-fold with increasing liana
include only net increases in standing woody biomass without abundance across sites (Duran&Gianoli, 2013), and experimental
considering branch turnover, and thus are systematic underesti- liana removal increased AGB accumulation in secondary forests by
mates. Branchfall also is ignored by most AWRT calculations, 75% (van der Heijden et al., 2015). Further, lianas differentially
which are thus systematic overestimates. These AWP and AWRT affect trees of different species (Muller-Landau & Visser, 2019),
estimates are mutually consistent, but a poor basis for modeling, and thus likely influence tree community functional composition,
because they underestimate the cost of tree growth. Incorporating which maymagnify or mitigate the direct effects of lianas. Tropical
the cost of branch turnover to dynamic vegetation models reduces lianas are themselves very diverse, with local species richness
tree biomass accumulation rates, improving estimates of forest size typically on the order of a third to half of that of trees, and thus liana
structure (Mart
ınez Cano et al., 2020). More measurements of functional composition also may play a role. Liana species vary in
branch turnover are needed to provide information on this critical their traits and effects on trees (Ichihashi & Tateno, 2011), and
parameter, including its variation among tree species and with shifts in liana composition among sites may thus contribute to
environmental conditions. variation in forest C dynamics (Muller-Landau & Visser, 2019).
The incorporation of lianas in models involves unique challenges
because of the complexities of their interactions with host trees, but
2. Community ecology
may be critical to reproducingmajor changes in forest structure and
In order to understand spatial variation in tropical forest C stocks functioning associated with variation in liana abundance along
and fluxes it is critical to understand the drivers of variation in plant successional, climate and disturbance gradients (Brugnera et al.,
functional composition– in the relative abundance of plants varying 2019).
in life-history strategy and functional traits. As detailed in this Most research on variation in plant functional composition has
review, every major environmental gradient in tropical forests is focused on direct environmental influences on plant performance.
characterized by shifts in tree functional composition that influence However, environmental conditions also may influence plants via
patterns of productivity, mortality and biomass along these changes in antagonistic andmutualistic interactions withmicrobes,
gradients (e.g. Gleason et al., 2009; Dalling et al., 2016). invertebrates, and vertebrates. For example, there is some evidence
Understanding functional composition is a complex problem of higher herbivory in sites with higher soil fertility, where plant
involving historical biogeographical influences on species pools, tissue nutrient concentrations are higher (Campo&Dirzo, 2003).
species sorting by environmental filters, competition among species Differences in vertebrate abundance and community composition
and phenotypic variation within species (McGill & Brown, 2007). contribute to savanna–forest boundaries and possibly differences in
Empirical research provides considerable information on spatial forest structure among biogeographic regions (Corlett, 2016). In
variation in tropical tree species and functional composition, how addition it has long been hypothesized that pest pressures are higher
species traits relate to performance under different environmental at wetter sites, and may drive compositional shifts and higher plant
conditions, and on associated tradeoffs (e.g. Poorter & diversity (Janzen & Schoener, 1968; Givnish, 1999), although
Markesteijn, 2008; Gleason et al., 2009; Brenes-Arguedas et al., evidence to date remains limited (but see Spear et al., 2015). The
2013; Asner & Martin, 2016; Staver et al., 2019). Better influences of biotic interactions have been assumed to be secondary
representation of the diversity of tropical plant physiology and to more direct environmental influences, and have been ignored in
life-history strategies in models is critical to capturing turnover in vegetation models; however, they may be critical to predicting
functional composition and associated shifts in forest functioning future forest C dynamics under global change, including defauna-
along environmental gradients (Levine et al., 2016) and among tion (Dirzo et al., 2014).
floristic realms (Slik et al., 2018; Taylor et al., 2019), as well as the
diversity of locally coexisting functional types that determines
3. Conclusions and future directions
functioning and responses to temporal climatic variation (Verhei-
jen et al., 2015; Sakschewski et al., 2016; Powell et al., 2018). An overview of decades of empirical research in tropical forests
Liana abundance varies greatly among tropical forests, and suggests general patterns in productivity, residence time, and
strongly influences forest C stocks and fluxes. Liana abundance estimated AGB variation, but studies to date have important
increases with soil fertility and disturbance, and decreases with limitations. First, essentially all studies have sizable sampling errors
rainfall and elevation (Schnitzer & Bongers, 2002); it also varies (see Section II, Methods), and these are especially large for studies
greatly within individual tropical forest sites (e.g. Schnitzer et al., with smaller plot sizes, smaller numbers of sites and shorter
2012). Multiple hypotheses have been proposed to explain these measurement periods (Clark et al., 2017). Second, studies to date

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
16 Review Tansley review Phytologist
all rest on the application of one or a few allometric equations across vital if thesemissions are to fulfill their promise (Chave et al., 2019).
multiple sites, and almost none involve site-specific measurements Every type of evidence on its own has key limitations; triangulation
of branch turnover. Systematic differences in biomass allometries across multiple lines of evidence is needed to reach robust
and/or branch turnover along environmental gradients could lead conclusions (Munafo & Smith, 2018). We must integrate
patterns in true AGB, AWP and AWRT to diverge substantially empirical studies and mechanistic modeling to make progress on
from those estimated by currentmethods. Third, study sites are not the big questions of the mechanisms of extant variation in tropical
well-distributed across tropical forests, owing to local and global forests today and the implications for their future trajectories
bias in plot placement and research effort (Figs S1, S9). There is a (Hofhansl et al., 2016; Fisher et al., 2018).
critical need and opportunity for future empirical research that
overcomes these limitations by taking advantage of new technolo-
Acknowledgements
gies such as laser scanning to more directly measure biomass
allometries, branch turnover and their variation among sites We thank Deborah Clark, Joe Wright, Ben Turner, Martijn Slot,
(Stovall et al., 2018), and of new and forthcoming satellite remote Ed Tanner, Evan Gora, Jeff Hall, Bert Leigh and two anonymous
sensing products that will provide much larger and better reviewers for thoughtful comments. KCC was supported as part of
distributed datasets on forest C cycling (Schimel et al., 2019). the Next Generation Ecosystem Experiments-Tropics, funded by
We also critically need a mechanistic understanding of the the U.S. Department of Energy, Office of Science, Office of
emergence of observed empirical patterns, so that we can reproduce Biological and Environmental Research.
them in models for the right reasons and have some hope of
correctly predicting responses to future novel climate conditions
Author contributions
(Wright et al., 2009). Research to date provides considerable
support for various hypotheses regarding contributing mecha- HCM planned and designed the research; HCM, KCC and EEA
nisms. However, every environmental pattern involves multiple conducted the literature review; HCM, KCC, IMC and BB
mechanisms, and we lack an understanding of the relative analyzed data; HCM, KCC, IMC, KAT and BB prepared figures;
importance of different mechanisms and their interactions. A and HCM drafted the manuscript. All authors contributed to
combination of mechanistic empirical studies and mechanistic revisions.
modeling is key to resolving this uncertainty, yet many of the
hypothesized underlying processes are not yet represented in
ORCID
models, which currently fail to reproduce key patterns (Fig. 2). This
is not surprising considering the models’ very limited representa- Kristina J. Anderson-Teixeira https://orcid.org/0000-0001-
tion of tree mortality (Galbraith et al., 2013; Johnson et al., 2016), 8461-9713
tropical tree functional diversity (Sakschewski et al., 2016) and Eva E. Arroyo https://orcid.org/0000-0002-8918-9721
many other processes. Bogumila Backiel https://orcid.org/0000-0002-9429-2600
Fortunately, a new generation of models has been developed in K. C. Cushman https://orcid.org/0000-0002-3464-1151
the last decade that better captures some spatial variation in tropical IsabelMartinez Cano https://orcid.org/0000-0003-4205-8596
forest biomass.Whereas oldermodels represented forest vegetation Helene C. Muller-Landau https://orcid.org/0000-0002-3526-
as a ‘big leaf’, new vegetation demographic approaches explicitly 9021
model the growth, survival and reproduction of trees or cohorts of
trees (Fisher et al., 2018). When run with prescribed meteorolog-
ical conditions, these models have succeeded in reproducing a References
multitude of patterns within individual tropical sites, as well as Aiba S, Kitayama K. 1999. Structure, composition and species diversity in an
general patterns of among-site variation along some environmental altitude-substrate matrix of rain forest tree communities on Mount Kinabalu,
et al et al et al Borneo. Plant Ecology 140: 139–157.gradients (Seiler ., 2014; Levine ., 2016; Xu ., 2016;
Alamgir M, Turton SM, Macgregor CJ, Pert PL. 2016. Assessing regulating and
Longo et al., 2019; Medvigy et al., 2019; Koven et al., 2020; provisioning ecosystem services in a contrasting tropical forest landscape.
Mart
ınez Cano et al., 2020). However, most still contain large Ecological Indicators 64: 319–334.
systematic errors, such as predicting too many large trees (Koven Alvarez-Davila E, Cayuela L, Gonzalez-Caro S, Aldana AM, Stevenson PR,
et al., 2020), and/or excessively high tree mortality rates (Longo Phillips O, Cogollo A, Penuela MC, von Hildebrand P, Jimenez E et al. 2017.
et al Forest biomass density across large climate gradients in northern SouthAmerica is., 2019). Furthermore, they mostly lack the mechanisms
related to water availability but not with temperature. PLoS ONE 12: e0171072.
needed to capture temporal responses to drought or spatial Alves LF, Vieira SA, Scaranello MA, Camargo PB, Santos FAM, Joly CA,
variation with soil fertility, disturbance and biogeographic region. Martinelli LA. 2010. Forest structure and live aboveground biomass variation
Tree mortality, branch turnover, tree functional composition, along an elevational gradient of tropical Atlantic moist forest (Brazil). Forest
and biotic interactions of trees with lianas and other organisms are Ecology and Management 260: 679–691.
Aragao L, Malhi Y, Metcalfe DB, Silva-Espejo JE, Jimenez E, Navarrete D,
key areas for further research, both for empirical data collection as
Almeida S, Costa ACL, Salinas N, Phillips OL et al. 2009. Above- and below-
well as modeling. Advances in remote sensing promise to yield ground net primary productivity across ten Amazonian forests on contrasting
much more and more widely distributed data on tropical forest soils. Biogeosciences 6: 2759–2778.
structure and function (Schimel et al., 2019), but adequate Arellano G, Medina NG, Tan S, Mohamad M, Davies SJ. 2019. Crown damage
investment in concurrent ground data collection in the tropics is and the mortality of tropical trees. New Phytologist 221: 169–179.
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 17
Asner GP, Anderson CB,Martin RE, Knapp DE, Tupayachi R, Sinca F, Malhi Y. Campo J, Dirzo R. 2003. Leaf quality and herbivory responses to soil nutrient
2014. Landscape-scale changes in forest structure and functional traits along an addition in secondary tropical dry forests of Yucat
an, Mexico. Journal of Tropical
Andes-to-Amazon elevation gradient. Biogeosciences 11: 843–856. Ecology 19: 525–530.
Asner GP,Martin RE. 2016.Convergent elevation trends in canopy chemical traits Carey EV, Brown S, Gillespie AJR, Lugo AE. 1994. Tree mortality in mature
of tropical forests. Global Change Biology 22: 2216–2227. lowland tropical moist and tropical lower montane moist forests of Venezuela.
Atkin OK, Bloomfield KJ, Reich PB, Tjoelker MG, Asner GP, Bonal D, Bonisch Biotropica 26: 255–265.
G, Bradford MG, Cernusak LA, Cosio EG et al. 2015.Global variability in leaf Carreno-Rocabado G, Pena-Claros M, Bongers F, Alarcon A, Licona JC, Poorter
respiration in relation to climate, plant functional types and leaf traits. New L. 2012. Effects of disturbance intensity on species and functional diversity in a
Phytologist 206: 614–636. tropical forest. Journal of Ecology 100: 1453–1463.
Aubry-KientzM, Rossi V,Wagner F, Herault B. 2015. Identifying climatic drivers Cavaleri MA, Reed SC, Smith WK, Wood TE. 2015. Urgent need for warming
of tropical forest dynamics. Biogeosciences 12: 5583–5596. experiments in tropical forests. Global Change Biology 21: 2111–2121.
Avitabile V, Herold M, Heuvelink GBM, Lewis SL, Phillips OL, Asner GP, Cavender-Bares J, Ackerly DD, Hobbie SE, Townsend PA. 2016. Evolutionary
Armston J, Ashton PS, Banin L, Bayol N et al. 2016. An integrated pan-tropical legacy effects on ecosystems: biogeographic origins, plant traits, and implications
biomass map using multiple reference datasets.Global Change Biology 22: 1406– for management in the era of global change. Annual Review of Ecology, Evolution,
1420. and Systematics 47: 433–462.
Baez S,Malizia A,Carilla J, BlundoC,AguilarM,AguirreN,AquirreZ, Alvarez E, Chadwick KD, Asner GP. 2018. Landscape evolution and nutrient rejuvenation
Cuesta F, Duque A et al. 2015. Large-scale patterns of turnover and basal area reflected in Amazon forest canopy chemistry. Ecology Letters 21: 978–988.
change in Andean forests. PLoS ONE 10: e0126594. Chave J, Andalo C, Brown S, Cairns MA, Chambers JQ, Eamus D, Folster H,
Bahar NH, Ishida FY, Weerasinghe LK, Guerrieri R, O’Sullivan OS, Bloomfield Fromard F, Higuchi N, Kira T et al. 2005. Tree allometry and improved
KJ, AsnerGP,MartinRE, Lloyd J,Malhi Y et al. 2017.Leaf-level photosynthetic estimation of carbon stocks and balance in tropical forests.Oecologia 145: 87–99.
capacity in lowland Amazonian and high-elevation Andean tropical moist forests Chave J, Davies SJ, Phillips OL, Lewis SL, Sist P, Schepaschenko D, Armston J,
of Peru. New Phytologist 214: 1002–1018. Baker TR, Coomes D, Disney M et al. 2019. Ground data are essential for
Baltzer JL,Davies SJ. 2012.Rainfall seasonality and pest pressure as determinants of biomass remote sensing missions. Surveys in Geophysics 40: 863–880.
tropical tree species’ distributions. Ecology and Evolution 2: 2682–2694. Chave J, Navarrete D, Almeida S, Alvarez E, Aragao L, Bonal D, Chatelet P, Silva-
Banin L, Feldpausch TR, Phillips OL, Baker TR, Lloyd J, Affum-Baffoe K, Arets Espejo JE, Goret JY, von Hildebrand P et al. 2010. Regional and seasonal
EJMM,BerryNJ, BradfordM, BrienenRJW et al. 2012.What controls tropical patterns of litterfall in tropical South America. Biogeosciences 7: 43–55.
forest architecture? Testing environmental, structural and floristic drivers.Global Chave J, Rejou-MechainM, Burquez A, Chidumayo E, ColganMS,DelittiWBC,
Ecology and Biogeography 21: 1179–1190. Duque A, Eid T, Fearnside PM, Goodman RC et al. 2014. Improved allometric
BaninL,Lewis SL,Lopez-GonzalezG,BakerTR,QuesadaCA,ChaoK-J, Burslem models to estimate the aboveground biomass of tropical trees. Global Change
DFRP, Nilus R, Abu Salim K, Keeling HC et al. 2014. Tropical forest wood Biology 20: 3177–3190.
production: a cross-continental comparison. Journal of Ecology 102: 1025–1037. CheesmanAW,WinterK. 2013.Elevated night-time temperatures increase growth
Baraloto C, Rabaud S, Molto Q, Blanc L, Fortunel C, Herault B, Davila N, in seedlings of two tropical pioneer tree species.NewPhytologist 197: 1185–1192.
Mesones I, Rios M, Valderrama E et al. 2011. Disentangling stand and Choat B, Brodribb TJ, Brodersen CR, Duursma RA, Lopez R,Medlyn BE. 2018.
environmental correlates of aboveground biomass in Amazonian forests. Global Triggers of tree mortality under drought. Nature 558: 531–539.
Change Biology 17: 2677–2688. Christoffersen BO, Gloor M, Fauset S, Fyllas NM, Galbraith DR, Baker TR,
Barlow J, Peres CA, Lagan BO, Haugaasen T. 2003. Large tree mortality and the Kruijt B, Rowland L, Fisher RA, BinksOJ et al. 2016.Linking hydraulic traits to
decline of forest biomass following Amazonian wildfires. Ecology Letters 6: 6–8. tropical forest function in a size-structured and trait-driven model (TFS vol 1-
Bastin JF, Rutishauser E, Kellner JR, Saatchi S, Pelissier R, Herault B, Slik F, Hydro). Geoscientific Model Development 9: 4227–4255.
Bogaert J, De Canniere C, Marshall AR et al. 2018. Pan-tropical prediction of ClarkDA,AsaoS, FisherR,ReedS,ReichPB,RyanMG,WoodTE,YangX. 2017.
forest structure from the largest trees.Global Ecology and Biogeography 27: 1366– Reviews and syntheses: Field data to benchmark the carbon cycle models for
1383. tropical forests. Biogeosciences 14: 4663–4690.
Becknell JM, Kucek LK, Powers JS. 2012. Aboveground biomass in mature and Clark DB, Hurtado J, Saatchi SS. 2015. Tropical rain forest structure, tree growth
secondary seasonally dry tropical forests: A literature review and global synthesis. and dynamics along a 2700-m elevational transect in Costa Rica. PLoS ONE 10:
Forest Ecology and Management 276: 88–95. e0122905.
Bellingham PJ, Sparrow AD. 2009.Multi-stemmed trees in montane rain forests: ClarkDB,Read JM,ClarkML,CruzAM,DottiMF,ClarkDA. 2004.Application
their frequency and demography in relation to elevation, soil nutrients and of 1-M and 4-M resolution satellite data to ecological studies of tropical rain
disturbance. Journal of Ecology 97: 472–483. forests. Ecological Applications 14: 61–74.
Bennett AC,McDowell NG, Allen CD, Anderson-Teixeira KJ. 2015. Larger trees Cleveland CC, Townsend AR, Taylor P, Alvarez-Clare S, Bustamante MM,
suffer most during drought in forests worldwide. Nature Plants 1: 15139. Chuyong G, Dobrowski SZ, Grierson P, Harms KE, Houlton BZ et al. 2011.
Berzaghi F, Longo M, Ciais P, Blake S, Bretagnolle F, Vieira S, Scaranello M, Relationships among net primary productivity, nutrients and climate in tropical
Scarascia-Mugnozza G, Doughty CE. 2019. Carbon stocks in central African rain forest: a pan-tropical analysis. Ecology Letters 14: 939–947.
forests enhanced by elephant disturbance. Nature Geoscience 12: 725–729. Cochrane MA. 2011. The past, present, and future importance of fire in tropical
Boose ER, Foster DR, FluetM. 1994.Hurricane impacts to tropical and temperate rainforests. In: Bush MB, Flenley JR, Gosling WD, eds. Tropical rainforest
forest landscapes. Ecological Monographs 64: 369–400. responses to climate change. New York, NY, USA: Springer, 213–240.
Brenes-Arguedas T, Roddy AB, Kursar TA, Tjoelker M. 2013. Plant traits in ConditR,AshtonP,BunyavejchewinS,DattarajaHS,Davies S, Esufali S, Ewango
relation to the performance and distribution of woody species in wet and dry C, Foster R, Gunatilleke I, Gunatilleke CVS et al. 2006. The importance of
tropical forest types in Panama. Functional Ecology 27: 392–402. demographic niches to tree diversity. Science 313: 98–101.
BrokawNVL. 1982.Treefalls: frequency, timing, and consequences. In: Leigh EG, Condit R, Engelbrecht BMJ, Pino D, Perez R, Turner BL. 2013. Species
Rand AS, Windsor DM eds. The ecology of a tropical forest: seasonal rhythms and distributions in response to individual soil nutrients and seasonal drought across a
long-term changes. Washington, DC, USA: Smithsonian Institution, 101–108. community of tropical trees. Proceedings of the National Academy of Sciences, USA
Brugnera MDE, Meunier F, Longo M, Moorthy SMK, De Deurwaerder H, 110: 5064–5068.
Schnitzer SA, BonalD, FaybishenkoB, VerbeeckH. 2019.Modeling the impact Corlett RT. 2016. Ecological roles of animals in tropical forests. In: Pancel L, K€ohl
of liana infestation on the demography and carbon cycle of tropical forests.Global M eds. Tropical forestry handbook. Berlin, Germany: Springer, 503–510.
Change Biology 25: 3767–3780. Corlett RT, Primack RB. 2011. Tropical rain forests: an ecological and
Bruijnzeel LA, Mulligan M, Scatena FN. 2011.Hydrometeorology of tropical biogeographical comparison. Oxford, UK: Wiley-Blackwell.
montane cloud forests: emerging patterns. Hydrological Processes 25: Curran TJ, Gersbach LN, Edwards W, Krockenberger AK. 2008.Wood density
465–498. predicts plant damage and vegetative recovery rates caused by cyclone disturbance

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
18 Review Tansley review Phytologist
in tropical rainforest tree species of north Queensland, Australia. Austral Ecology Fyllas NM, Bentley LP, Shenkin A, Asner GP, Atkin OK, Diaz S, Enquist BJ,
33: 442–450. Farfan-Rios W, Gloor E, Guerrieri R et al. 2017. Solar radiation and functional
Cushman KC, Muller-Landau HC, Condit RS, Hubbell SP. 2014. Improving traits explain the decline of forest primary productivity along a tropical elevation
estimates of biomass change in buttressed trees using tree tapermodels.Methods in gradient. Ecology Letters 20: 730–740.
Ecology and Evolution 5: 573–582. Fyllas NM, Patino S, Baker TR, Nardoto GB, Martinelli LA, Quesada CA, Paiva
Dalling JW, Heineman K, Gonzalez G, Ostertag R. 2016. Geographic, R, SchwarzM,HornaV,Mercado LM et al. 2009.Basin-wide variations in foliar
environmental and biotic sources of variation in the nutrient relations of tropical properties of Amazonian forest: phylogeny, soils and climate. Biogeosciences 6:
montane forests. Journal of Tropical Ecology 32: 368–383. 2677–2708.
Daskin JH, Aires F, Staver AC. 2019. Determinants of tree cover in tropical Galbraith D, Malhi Y, Affum-Baffoe K, Castanho ADA, Doughty CE, Fisher
floodplains. Proceedings of the Royal Society of London. Series B: Biological Sciences RA, Lewis SL, Peh KSH, Phillips OL, Quesada CA et al. 2013. Residence
286: 20191755. times of woody biomass in tropical forests. Plant Ecology & Diversity 6:
Dattaraja HS, Pulla S, SureshHS, NagarajaMS,Murthy CAS, Sukumar R. 2018. 139–157.
Woody plant diversity in relation to environmental factors in a seasonally dry Ge R, HeHL, Ren XL, Zhang L, Yu GR, Smallman TL, Zhou T, Yu SY, Luo YQ,
tropical forest landscape. Journal of Vegetation Science 29: 704–714. Xie ZQ et al. 2019. Underestimated ecosystem carbon turnover time and
DeWalt SJ, Ickes K, Nilus R, Harms KE, Burslem DFRP. 2006. Liana habitat sequestration under the steady state assumption: A perspective from long-term
associations and community structure in a Bornean lowland tropical forest. Plant data assimilation. Global Change Biology 25: 938–953.
Ecology 186: 203–216. Gentry AH. 1988.Changes in plant community diversity and floristic composition
DeWalt SJ, Schnitzer SA, Alves LF, Bongers F, Burnham RJ, Cai Z, Carson WP, on environmental and geographical gradients. Annals of the Missouri Botanical
Chave J, ChuyongGB, Costa FRC et al. 2015. Biogeographical patterns of liana Garden 75: 1–34.
abundance and diversity. In: Schnitzer SA, Bongers F, BurnhamRJ, Putz FE, eds. Gerwing JJ. 2002. Degradation of forests through logging and fire in the eastern
Ecology of lianas. Hoboken, NJ, USA: John Wiley & Sons, 131–146. Brazilian Amazon. Forest Ecology and Management 157: 131–141.
DeWalt SJ, Schnitzer SA, Chave J, Bongers F, Burnham RJ, Cai ZQ, Chuyong G, Ghazoul J. 2016.Dipterocarp biology, ecology, and conservation. Oxford,UK:Oxford
Clark DB, Ewango CEN, Gerwing JJ et al. 2010. Annual rainfall and seasonality University Press.
predict pan-tropical patterns of liana density and basal area. Biotropica 42: 309–317. GirardinCAJ, Farfan-RiosW,Garcia K, FeeleyKJ, Jorgensen PM,Murakami AA,
DirzoR, YoungHS,GalettiM,CeballosG, IsaacNJ,CollenB. 2014.Defaunation Perez LC, Seidel R, Paniagua N, Claros AFF et al. 2014. Spatial patterns of
in the Anthropocene. Science 345: 401–406. above-ground structure, biomass and composition in a network of six Andean
Doughty CE, Goldsmith GR, Raab N, Girardin CAJ, Farfan-Amezquita F, elevation transects. Plant Ecology & Diversity 7: 161–171.
Huaraca-Huasco W, Silva-Espejo JE, Araujo-Murakami A, da Costa ACL, GirardinCAJ,Malhi Y, Aragao L,MamaniM,HuascoWH,Durand L, Feeley KJ,
Rocha W et al. 2018.What controls variation in carbon use efficiency among Rapp J, Silva-Espejo JE, Silman M et al. 2010. Net primary productivity
Amazonian tropical forests? Biotropica 50: 16–25. allocation and cycling of carbon along a tropical forest elevational transect in the
Duran SM, Gianoli E. 2013. Carbon stocks in tropical forests decrease with liana Peruvian Andes. Global Change Biology 16: 3176–3192.
density. Biology Letters 9: 20130301. Givnish TJ. 1999.On the causes of gradients in tropical tree diversity. Journal of
Elser JJ, Fagan WF, Kerkhoff AJ, Swenson NG, Enquist BJ. 2010. Biological Ecology 87: 193–210.
stoichiometry of plant production:metabolism, scaling and ecological response to Gleason SM, Read J, Ares A,Metcalfe DJ. 2009. Phosphorus economics of tropical
global change. New Phytologist 186: 593–608. rainforest species and stands across soil contrasts in Queensland, Australia:
Engelbrecht BMJ, Comita LS, Condit R, Kursar TA, Tyree MT, Turner BL, understanding the effects of soil specialization and trait plasticity. Functional
Hubbell SP. 2007. Drought sensitivity shapes species distribution patterns in Ecology 23: 1157–1166.
tropical forests. Nature 447: 80–U82. Gonmadje C, Picard N, Gourlet-Fleury S, Rejou-Mechain M, Freycon V,
Espirito-Santo FDB, Keller M, Braswell B, Nelson BW, Frolking S, Vicente G. Sunderland T, McKey D, Doumenge C. 2017. Altitudinal filtering of large-tree
2010. Storm intensity and old-growth forest disturbances in the Amazon region. species explains above-ground biomass variation in an Atlantic Central African
Geophysical Research Letters 37: L11403. rain forest. Journal of Tropical Ecology 33: 143–154.
Esquivel-Muelbert A, Galbraith D, Dexter KG, Baker TR, Lewis SL, Meir P, Gora EM, Burchfield JC, Muller-Landau HC, Bitzer PM, Yanoviak SP. 2020.
Rowland L, da Costa ACL, Nepstad D, Phillips OL. 2017. Biogeographic Pantropical geography of lightning-caused disturbance and its implications for
distributions of neotropical trees reflect their directly measured drought tropical forests. Global Change Biology 26: 5017–5026.
tolerances. Scientific Reports 7: 8334. Gora EM, Kneale RC, Larjavaara M, Muller-Landau HC. 2019. Dead wood
Estrada-Villegas S, Schnitzer SA. 2018.A comprehensive synthesis of liana removal necromass in amoist tropical forest: stocks, fluxes, and spatiotemporal variability.
experiments in tropical forests. Biotropica 50: 729–739. Ecosystems 22: 1189–1205.
Farquhar GD, von Caemmerer S, Berry JA. 1980. A biochemical model of Gorel AP, Steppe K, Beeckman H, De Baerdemaeker NJF, Doucet JL, Ligot G,
photosynthetic CO2 assimilation in leaves of C3 species. Planta 149: 78–90. Dainou K, Fayolle A. 2019. Testing the divergent adaptation of two congeneric
Feldpausch TR, Lloyd J, Lewis SL, Brienen RJW,GloorM,Mendoza AM, Lopez- tree species on a rainfall gradient using eco-physio-morphological traits.
Gonzalez G, Banin L, Abu Salim K, Affum-Baffoe K et al. 2012. Tree height Biotropica 51: 364–377.
integrated into pantropical forest biomass estimates.Biogeosciences 9: 3381–3403. GrauO, Penuelas J, Ferry B, Freycon V, Blanc L, DesprezM, Baraloto C, Chave J,
Fisher RA, Koven CD, Anderegg WRL, Christoffersen BO, Dietze MC, Farrior Descroix L,Dourdain A et al. 2017.Nutrient-cyclingmechanisms other than the
CE, Holm JA, Hurtt GC, Knox RG, Lawrence PJ et al. 2018. Vegetation direct absorption from soil may control forest structure and dynamics in poor
demographics in Earth system models: a review of progress and priorities. Global Amazonian soils. Scientific Reports 7: 45017.
Change Biology 24: 35–54. Grubb PJ. 1977. Control of forest growth and distribution on wet tropical
Flack-Prain S,Meir P,Malhi Y, SmallmanTL,WilliamsM. 2019.The importance mountains: with special reference to mineral nutrition. Annual Review of Ecology
of physiological, structural and trait responses to drought stress in driving spatial and Systematics 8: 83–107.
and temporal variation in GPP across Amazon forests. Biogeosciences 16: 4463– GuanKY,PanM,LiHB,WolfA,Wu J,MedvigyD,CaylorKK, Sheffield J,Wood
4484. EF, Malhi Y et al. 2015. Photosynthetic seasonality of global tropical forests
Fontes CG, Chambers JQ, Higuchi N. 2018. Revealing the causes and temporal constrained by hydroclimate. Nature Geoscience 8: 284–289.
distribution of tree mortality in Central Amazonia. Forest Ecology and Hall J, Muscarella R, Quebbeman A, Arellano G, Thompson J, Zimmerman JK,
Management 424: 177–183. Uriarte M. 2020.Hurricane-induced rainfall is a stronger predictor of tropical
Friend AD, Lucht W, Rademacher TT, Keribin R, Betts R, Cadule P, Ciais P, forest damage in Puerto Rico than maximum wind speeds. Scientific Reports 10:
ClarkDB,Dankers R, Falloon PD et al. 2014.Carbon residence time dominates 4318.
uncertainty in terrestrial vegetation responses to future climate and atmospheric Heavens NG, Ward DS, Natalie MM. 2013. Studying and projecting climate
CO2. Proceedings of the National Academy of Sciences, USA 111: 3280–3285. change with Earth System Models. Nature Education Knowledge 4: 4.
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 19
van der Heijden GM, Powers JS, Schnitzer SA. 2015. Lianas reduce carbon Koven CD, Chambers JQ, Georgiou K, Knox R, Negron-Juarez R, Riley WJ,
accumulation and storage in tropical forests.Proceedings of theNationalAcademy of Arora VK, Brovkin V, Friedlingstein P, Jones CD. 2015.Controls on terrestrial
Sciences, USA 112: 13267–13271. carbon feedbacks by productivity versus turnover in the CMIP5 Earth System
HeinemanKD, Russo SE, Baillie IC,Mamit JD, Chai PPK, Chai L, Hindley EW, Models. Biogeosciences 12: 5211–5228.
Lau BT, Tan S, Ashton PS. 2015. Evaluation of stem rot in 339 Bornean tree Koven CD, Knox RG, Fisher RA, Chambers JQ, Christoffersen BO, Davies SJ,
species: implications of size, taxonomy, and soil-related variation for aboveground Detto M, Dietze MC, Faybishenko B, Holm J et al. 2020. Benchmarking and
biomass estimates. Biogeosciences 12: 5735–5751. parameter sensitivity of physiological and vegetation dynamics using the
Heinze C, Eyring V, Friedlingstein P, Jones C, Balkanski Y, Collins W, Fichefet T, Functionally Assembled Terrestrial Ecosystem Simulator (FATES) at Barro
Gao S, Hall A, IvanovaD et al. 2019. ESDReviews: Climate feedbacks in the Earth Colorado Island, Panama. Biogeosciences 17: 3017–3044.
system and prospects for their evaluation. Earth System Dynamics 10: 379–452. Kraft NJ, Metz MR, Condit RS, Chave J. 2010. The relationship between wood
HenglT,Mendes de Jesus J,HeuvelinkGBM,RuiperezGonzalezM,KilibardaM, density and mortality in a global tropical forest data set. New Phytologist 188:
Blagoti
c A, ShangguanW,Wright MN, Geng X, Bauer-Marschallinger B et al. 1124–1136.
2017. SoilGrids250m: Global gridded soil information based on machine Kupers SJ, Engelbrecht BMJ, Hernandez A, Wright SJ, Wirth C, Ruger N. 2019.
learning. PLoS ONE 12: e0169748. Growth responses to soil water potential indirectly shape local species
Hofhansl F, Andersen KM, Fleischer K, Fuchslueger L, Rammig A, Schaap KJ, distributions of tropical forest seedlings. Journal of Ecology 107: 860–874.
Valverde-Barrantes OJ, Lapola DM. 2016. Amazon Forest ecosystem responses Lai HR, Hall JS, Turner BL, van Breugel M. 2017. Liana effects on biomass
to elevated atmospheric CO2 and alterations in nutrient availability: filling the dynamics strengthen during secondary forest succession. Ecology 98: 1062–1070.
gaps with model-experiment integration. Frontiers in Earth Science 4: 19. Laliberte E, Kardol P, Didham RK, Teste FP, Turner BL, Wardle DA. 2017. Soil
Hofhansl F, Chac
on-Madrigal E, Fuchslueger L, Jenking D, Morera-Beita A, fertility shapes belowground food webs across a regional climate gradient. Ecology
Plutzar C, Silla F, Andersen KM, Buchs DM, Dullinger S et al. 2020. Climatic Letters 20: 1273–1284.
and edaphic controls over tropical forest diversity and vegetation carbon storage. Larsen MC, Torres-Sanchez AJ. 1998. The frequency and distribution of recent
Scientific Reports 10: 5066. landslides in three montane tropical regions of Puerto Rico. Geomorphology 24:
Hofhansl F, Schnecker J, SingerG,WanekW.2015.New insights intomechanisms 309–331.
driving carbon allocation in tropical forests. New Phytologist 205: 137–146. Laurance WF, Fearnside PM, Laurance SG, Delamonica P, Lovejoy TE, Rankin-
Homeier J, Breckle SW, Gunter S, Rollenbeck RT, Leuschner C. 2010. Tree de Merona JM, Chambers JQ, Gascon C. 1999. Relationship between soils and
diversity, forest structure and productivity along altitudinal and topographical Amazon forest biomass: a landscape-scale study. Forest Ecology and Management
gradients in a species-rich Ecuadorian montane rain forest. Biotropica 42: 140– 118: 127–138.
148. Laurance WF, Perez-Salicrup D, Delamonica P, Fearnside PM, D’Angelo S,
Houlton BZ, Wang YP, Vitousek PM, Field CB. 2008. A unifying framework for Jerozolinski A, Pohl L, Lovejoy TE. 2001. Rain forest fragmentation and the
dinitrogen fixation in the terrestrial biosphere. Nature 454: 327–330. structure of Amazonian liana communities. Ecology 82: 105–116.
Ibanez T, Keppel G, Menkes C, Gillespie TW, Lengaigne M, Mangeas M, Rivas- LevineNM, ZhangK, LongoM, Baccini A, PhillipsOL, Lewis SL, Alvarez-Davila
Torres G, BirnbaumP. 2019.Globally consistent impact of tropical cyclones on E, de Andrade ACS, Brienen RJW, Erwin TL et al. 2016. Ecosystem
the structure of tropical and subtropical forests. Journal of Ecology 107: 279–292. heterogeneity determines the ecological resilience of the Amazon to climate
IchihashiR,TatenoM.2011.Strategies to balance between light acquisition and the change. Proceedings of the National Academy of Sciences, USA 113: 793–797.
risk of falls of four temperate liana species: to overtop host canopies or not? Journal Lewis SL, Lloyd J, Sitch S,Mitchard ETA, LauranceWF. 2009.Changing ecology
of Ecology 99: 1071–1080. of tropical forests: evidence and drivers. Annual Review of Ecology, Evolution and
Ingwell LL,Wright SJ, BecklundKK,Hubbell SP, Schnitzer SA. 2010.The impact Systematics 40: 529–549.
of lianas on 10 years of tree growth and mortality on Barro Colorado Island, Lewis SL, Sonke B, Sunderland T, Begne SK, Lopez-Gonzalez G, van der Heijden
Panama. Journal of Ecology 98: 879–887. GMF, Phillips OL, Affum-Baffoe K, Baker TR, Banin L et al. 2013. Above-
Janzen DH, Schoener TW. 1968. Differences in insect abundance and diversity ground biomass and structure of 260 African tropical forests. Philosophical
between wetter and drier sites during a tropical dry season. Ecology 49: 96–110. Transactions of the Royal Society of London. Series B: Biological Sciences 368:
Jobb
agy EG, JacksonRB. 2000.The vertical distribution of soil organic carbon and 20120295.
its relation to climate and vegetation. Ecological Applications 10: 423–436. Lieberman D, Lieberman M, Peralta R, Hartshorn GS. 1996. Tropical forest
Johnson MO, Galbraith D, Gloor M, De Deurwaerder H, Guimberteau M, structure and composition on a large-scale altitudinal gradient in Costa Rica.
Rammig A, Thonicke K, Verbeeck H, von Randow C, Monteagudo A et al. Journal of Ecology 84: 137–152.
2016. Variation in stem mortality rates determines patterns of above-ground Lloyd J, Domingues TF, Schrodt F, Ishida FY, Feldpausch TR, Saiz G, Quesada
biomass in Amazonian forests: implications for dynamic global vegetation CA, SchwarzM, Torello-RaventosM, GilpinM et al. 2015. Edaphic, structural
models. Global Change Biology 22: 3996–4013. andphysiological contrasts acrossAmazonBasin forest-savanna ecotones suggest a
Jucker T, Sanchez AC, Lindsell JA, Allen HD, Amable GS, Coomes DA. 2016. role for potassium as a key modulator of tropical woody vegetation structure and
Drivers of aboveground wood production in a lowland tropical forest of West function. Biogeosciences 12: 6529–6571.
Africa: teasing apart the roles of tree density, tree diversity, soil phosphorus, and LongoM, Knox RG, Levine NM, Swann ALS,Medvigy DM, DietzeMC, Kim Y,
historical logging. Ecology and Evolution 6: 4004–4017. Zhang K, Bonal D, Burban B et al. 2019. The biophysics, ecology, and
Jung M, Henkel K, Herold M, Churkina G. 2006. Exploiting synergies of global biogeochemistry of functionally diverse, vertically and horizontally heterogeneous
land cover products for carbon cycle modeling. Remote Sensing of Environment ecosystems: the Ecosystem Demography model, version 2.2 – Part 2: Model
101: 534–553. evaluation for tropical SouthAmerica.GeoscientificModelDevelopment12: 4347–
KaspariM,Powers JS. 2016.Biogeochemistry and geographical ecology: embracing 4374.
all twenty-five elements required to build organisms. American Naturalist 188: Malhi Y. 2012. The productivity, metabolism and carbon cycle of tropical forest
S62–S73. vegetation. Journal of Ecology 100: 65–75.
Kitayama K, Aiba SI. 2002. Ecosystem structure and productivity of tropical rain Malhi Y, Amezquita FF, Doughty CE, Silva-Espejo JE, Girardin CAJ, Metcalfe DB,
forests along altitudinal gradients with contrasting soil phosphorus pools on AragaoL,Huaraca-Quispe LP,Alzamora-Taype I, Eguiluz-MoraL et al. 2014.The
Mount Kinabalu, Borneo. Journal of Ecology 90: 37–51. productivity, metabolism and carbon cycle of two lowland tropical forest plots in
Kohyama TS, Kohyama TI, Sheil D. 2018.Definition and estimation of vital rates south-western Amazonia, Peru. Plant Ecology & Diversity 7: 85–105.
from repeated censuses: Choices, comparisons and bias corrections focusing on Malhi Y, Doughty C, Galbraith D. 2011. The allocation of ecosystem net primary
trees.Methods in Ecology and Evolution 9: 809–821. productivity in tropical forests. Philosophical Transactions of the Royal Society of
KohyamaTS,KohyamaTI, Sheil D. 2019.Estimating net biomass production and London. Series B: Biological Sciences 366: 3225–3245.
loss from repeated measurements of trees in forests and woodlands: Formulae, Malhi Y, Doughty CE, Goldsmith GR, Metcalfe DB, Girardin CAJ, Marthews
biases and recommendations. Forest Ecology and Management 433: 729–740. TR, del Aguila-Pasquel J, Aragao LEOC, Araujo-Murakami A, Brando P et al.

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
20 Review Tansley review Phytologist
2015.The linkages between photosynthesis, productivity, growth and biomass in NottinghamAT,Turner BL,Whitaker J,OstleNJ,McNamaraNP, Bardgett RD,
lowland Amazonian forests. Global Change Biology 21: 2283–2295. SalinasN,MeirP. 2015.Soilmicrobial nutrient constraints along a tropical forest
Malhi Y, Girardin CAJ, Goldsmith GR, Doughty CE, Salinas N, Metcalfe DB, elevation gradient: a belowground test of a biogeochemical paradigm.
HuaracaHuascoW, Silva-Espejo JE, del Aguilla-Pasquell J, Farf
anAm
ezquita F Biogeosciences 12: 6071–6083.
et al. 2017. The variation of productivity and its allocation along a tropical OstertagR, SilverWL, LugoAE. 2005.Factors affectingmortality and resistance to
elevation gradient: a whole carbon budget perspective. New Phytologist 214: damage following hurricanes in a rehabilitated subtropicalmoist forest.Biotropica
1019–1032. 37: 16–24.
Malhi Y,WoodD,BakerTR,Wright J, PhillipsOL,CochraneT,Meir P,Chave J, PalaceM, Keller M, Silva H. 2008.Necromass production: Studies in undisturbed
Almeida S, Arroyo L et al. 2006. The regional variation of aboveground live and logged Amazon forests. Ecological Applications 18: 873–884.
biomass in old-growthAmazonian forests.Global Change Biology12: 1107–1138. Pan YD, Birdsey RA, Phillips OL, Jackson RB. 2013. The structure, distribution,
Margrove JA, BurslemDFRP, Ghazoul J, Khoo E, Kettle CJ, Maycock CR. 2015. and biomass of the world’s forests. Annual Review of Ecology, Evolution, and
Impacts of an extreme precipitation event on dipterocarp mortality and habitat Systematics 44: 593–622.
filtering in a Bornean tropical rain forest. Biotropica 47: 66–76. Paoli GD, Curran LM. 2007. Soil nutrients limit fine litter production and tree
Mart
ınez Cano I, Shevliakova E, Malyshev S, Wright SJ, Detto M, Pacala SW, growth in mature lowland forest of Southwestern Borneo. Ecosystems 10: 503–
Muller-Landau HC. 2020. Allometric constraints and competition enable the 518.
simulation of size structure and carbon fluxes in a dynamic vegetation model of Paoli GD, Curran LM, Slik JWF. 2008. Soil nutrients affect spatial patterns of
tropical forests (LM3PPA-TV). Global Change Biology 26: 4478–4494. aboveground biomass and emergent tree density in southwestern Borneo.
Marvin DC, Asner GP. 2016. Branchfall dominates annual carbon flux across Oecologia 155: 287–299.
lowland Amazonian forests. Environmental Research Letters 11: 094027. PaoliGD,CurranLM,ZakDR. 2005.Phosphorus efficiency of Bornean rain forest
MarvinDC, Asner GP, KnappDE, AndersonCB,Martin RE, Sinca F, Tupayachi productivity: Evidence against the unimodal efficiency hypothesis. Ecology 86:
R. 2014.Amazonian landscapes and the bias in field studies of forest structure and 1548–1561.
biomass.Proceedings of theNational Academy of Sciences,USA111: E5224–E5232. Parmentier I, Malhi Y, Senterre B, Whittaker RJ, Alonso A, Balinga MPB,
McDowell N, AllenCD, Anderson-Teixeira K, Brando P, BrienenR, Chambers J, Bakayoko A, Bongers F, Chatelain C, Comiskey JA et al. 2007. The odd man
Christoffersen B, Davies S, Doughty C, Duque A et al. 2018. Drivers and out? Might climate explain the lower tree alpha-diversity of African rain forests
mechanisms of tree mortality in moist tropical forests.New Phytologist 219: 851– relative to Amazonian rain forests? Journal of Ecology 95: 1058–1071.
869. Paz H, Vega-Ramos F, Arreola-Villa F. 2018. Understanding hurricane resistance
McGill BJ, Brown JS. 2007. Evolutionary game theory and adaptive dynamics of and resilience in tropical dry forest trees: A functional traits approach. Forest
continuous traits. Annual Review of Ecology, Evolution, and Systematics 38: 403– Ecology and Management 426: 115–122.
435. Peng YK, Bloomfield KJ, Prentice IC. 2020. A theory of plant function helps to
McMahon SM, Arellano G, Davies SJ. 2019. The importance and challenges of explain leaf-trait and productivity responses to elevation. New Phytologist 226:
detecting changes in forest mortality rates. Ecosphere 10: e02615. 1274–1284.
McMichael CH, FeeleyKJ,DickCW,PipernoDR,BushMB. 2017.Comment on Pereira-Filho AJ, Carbone RE, Tuttle JD, KaramHA. 2015.Convective rainfall in
“Persistent effects of pre-Columbian plant domestication on Amazonian forest Amazonia and adjacent tropics. Atmospheric and Climate Sciences 5: 137–161.
composition”. Science 358: eaan8347. Pfeifer M, Gonsamo A, Woodgate W, Cayuela L, Marshall AR, Ledo A, Paine
McMichael CNH, Matthews-Bird F, Farfan-Rios W, Feeley KJ. 2017. Ancient TCE, Marchant R, Burt A, Calders K et al. 2018. Tropical forest canopies and
human disturbances may be skewing our understanding of Amazonian forests. their relationships with climate and disturbance: results from a global dataset of
Proceedings of the National Academy of Sciences, USA 114: 522–527. consistent field-based measurements. Forest Ecosystems 5: 7.
MedvigyD,WangG, ZhuQ,RileyWJ, Trierweiler AM,WaringBonnieG, XuX, Phillips J, Ramirez S, Wayson C, Duque A. 2019. Differences in carbon stocks
Powers JS. 2019.Observed variation in soil properties can drive large variation in along an elevational gradient in tropicalmountain forests of Colombia.Biotropica
modelled forest functioning and composition during tropical forest secondary 51: 490–499.
succession. New Phytologist 223: 1820–1833. Phillips OL, van der Heijden G, Lewis SL, Lopez-Gonzalez G, Aragao L, Lloyd J,
Moore S, Adu-Bredu S, Duah-Gyamfi A, Addo-Danso SD, Ibrahim F,Mbou AT, Malhi Y, Monteagudo A, Almeida S, Davila EA et al. 2010.Drought-mortality
de Grandcourt A, Valentini R, Nicolini G, Djagbletey G et al. 2018. Forest relationships for tropical forests. New Phytologist 187: 631–646.
biomass, productivity and carbon cycling along a rainfall gradient inWest Africa. Ploton P, Barbier N, Momo ST, Rejou-Mechain M, Bosela FB, Chuyong G,
Global Change Biology 24: E496–E510. Dauby G, Droissart V, Fayolle A, Goodman RC et al. 2016. Closing a gap in
Muller-Landau HC, Detto M, Chisholm RA, Hubbell SP, Condit R. 2014. tropical forest biomass estimation: taking crown mass variation into account in
Detecting and projecting changes in forest biomass from plot data. In: Coomes pantropical allometries. Biogeosciences 13: 1571–1585.
DA, Burslem DFRP, eds. Forests and global change. Cambridge, UK: Cambridge Poorter L, Markesteijn L. 2008. Seedling traits determine drought tolerance of
University Press, 381–415. tropical tree species. Biotropica 40: 321–331.
Muller-Landau HC, Pacala SW. 2020.What determines the abundance of lianas Poorter L, Ongers FB, Aide TM, Zambrano AMA, Balvanera P, Becknell JM,
and vines? In: Dobson A, Tilman D, Holt R, eds. Unsolved problems in ecology. Boukili V, Brancalion PHS, Broadbent EN, Chazdon RL et al. 2016. Biomass
Princeton, NJ, USA: Princeton University Press, 239–264. resilience of Neotropical secondary forests. Nature 530: 211–214.
Muller-Landau HC, Visser MD. 2019.How do lianas and vines influence Poorter L, van der SandeMT, Arets E, AscarrunzN, Enquist B, Finegan B, Licona
competitive differences and niche differences among tree species? Concepts and a JC, Martinez-Ramos M, Mazzei L, Meave JA et al. 2017. Biodiversity and
case study in a tropical forest. Journal of Ecology 107: 1469–1481. climate determine the functioning of Neotropical forests. Global Ecology and
MunafoMR, Smith GD. 2018.Repeating experiments is not enough.Nature 553: Biogeography 26: 1423–1434.
399–401. Porder S, VitousekPM,ChadwickOA,ChamberlainCP,HilleyGE. 2007.Uplift,
Naveenkumar J, Arunkumar KS, Sundarapandian SM. 2017. Biomass and carbon erosion, and phosphorus limitation in terrestrial ecosystems. Ecosystems 10: 158–
stocks of a tropical dry forest of the Javadi Hills, Eastern Ghats, India. Carbon 170.
Management 8: 351–361. Powell TL, Koven CD, Johnson DJ, Faybishenko B, Fisher RA, Knox RG,
Negron-JuarezRI, Chambers JQ,HurttGC, AnnaneB,Cocke S, PowellM, StottM, McDowell NG, Condit R, Hubbell SP, Wright SJ et al. 2018. Variation in
Goosem S, Metcalfe DJ, Saatchi SS. 2014. Remote sensing assessment of forest hydroclimate sustains tropical forest biomass and promotes functional diversity.
disturbance across complex mountainous terrain: the pattern and severity of impacts New Phytologist 219: 932–946.
of tropical cyclone Yasi on Australian rainforests. Remote Sensing 6: 5633–5649. Putz FE, Chai P. 1987. Ecological studies of lianas in Lambir National-Park,
Negr
on-Ju
arez R, Jenkins H, Raupp C, RileyW, Kueppers L,MagnaboscoMarra Sarawak, Malaysia. Journal of Ecology 75: 523–531.
D, Ribeiro G, Monteiro M, Candido L, Chambers J et al. 2017.Windthrow Quesada CA, Lloyd J, Schwarz M, Patin~o S, Baker TR, Czimczik C, Fyllas NM,
variability in Central Amazonia. Atmosphere 8: 28. Martinelli L, Nardoto GB, Schmerler J et al. 2010. Variations in chemical and
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 21
physical properties of Amazon forest soils in relation to their genesis.Biogeosciences Sheil D. 1995. A critique of permanent plot methods and analysis with examples
7: 1515–1541. from Budongo Forest, Uganda. Forest Ecology and Management 77: 11–34.
Quesada CA, Phillips OL, Schwarz M, Czimczik CI, Baker TR, Patino S, Fyllas SheilD. 1996. Species richness, tropical forest dynamics and sampling:Questioning
NM, Hodnett MG, Herrera R, Almeida S et al. 2012. Basin-wide variations in cause and effect. Oikos 76: 587–590.
Amazon forest structure and function are mediated by both soils and climate. Sherman RE, Fahey TJ, Martin PH, Battles JJ. 2012. Patterns of growth,
Biogeosciences 9: 2203–2246. recruitment, mortality and biomass across an altitudinal gradient in a
Raich JW, Russell AE, Kitayama K, PartonWJ, Vitousek PM. 2006.Temperature neotropical montane forest, Dominican Republic. Journal of Tropical Ecology
influences carbon accumulation in moist tropical forests. Ecology 87: 76–87. 28: 483–495.
Reich PB. 2014. The world-wide ’fast-slow’ plant economics spectrum: a traits Silver WL, Ostertag R, Lugo AE. 2000. The potential for carbon sequestration
manifesto. Journal of Ecology 102: 275–301. through reforestation of abandoned tropical agricultural and pasture lands.
Reis SM, Marimon BS, Marimon B, Morandi PS, de Oliveira EA, Elias F, das Restoration Ecology 8: 394–407.
Neves EC, de Oliveira B, Nogueira DD, Umetsu RK et al. 2018. Climate and Slik JWF, Aiba SI, Brearley FQ, Cannon CH, ForshedO, KitayamaK, Nagamasu
fragmentation affect forest structure at the southern border of Amazonia. Plant H, Nilus R, Payne J, Paoli G et al. 2010a. Environmental correlates of tree
Ecology & Diversity 11: 13–25. biomass, basal area, wood specific gravity and stem density gradients in Borneo’s
Richards PW. 1952. The tropical rain forest: an ecological study. London, UK: tropical forests. Global Ecology and Biogeography 19: 50–60.
Cambridge University Press. Slik JWF, Breman FC, Bernard C, van Beek M, Cannon CH, Eichhorn KAO,
Rowland L, Harper A, Christoffersen BO, Galbraith DR, Imbuzeiro HMA, Sidiyasa K. 2010b. Fire as a selective force in a Bornean tropical everwet forest.
Powell TL,Doughty C, LevineNM,Malhi Y, Saleska SR et al. 2015.Modelling Oecologia 164: 841–849.
climate change responses in tropical forests: similar productivity estimates across Slik JWF, Franklin J, Arroyo-Rodriguez V, Field R, Aguilar S, Aguirre N,
five models, but different mechanisms and responses. Geoscientific Model Ahumada J, Aiba SI, Alves LF, AnithaK et al. 2018.Phylogenetic classification of
Development 8: 1097–1110. the world’s tropical forests. Proceedings of the National Academy of Sciences, USA
Rozendaal DMA, Chazdon RL, Arreola-Villa F, Balvanera P, Bentos TV, Dupuy 115: 1837–1842.
JM, Luis Hernandez-Stefanoni J, Jakovac CC, Lebrija-Trejos EE, Lohbeck M Slik JWF,PaoliG,McGuireK,Amaral I, Barroso J, BastianM,BlancL,Bongers F,
et al. 2017. Demographic drivers of aboveground biomass dynamics during Boundja P, Clark C et al. 2013. Large trees drive forest aboveground biomass
secondary succession in neotropical dry and wet forests. Ecosystems 20: 340–353. variation in moist lowland forests across the tropics. Global Ecology and
Rutishauser E, Wright SJ, Condit R, Hubbell SP, Davies SJ, Muller-Landau HC. Biogeography 22: 1261–1271.
2020. Testing for changes in biomass dynamics in large-scale forest datasets. Slot M, Winter K. 2017. In situ temperature relationships of biochemical and
Global Change Biology 26: 1485–1498. stomatal controls of photosynthesis in four lowland tropical tree species. Plant,
Sakschewski B, von BlohW, Boit A, Poorter L, Pen~a-Claros M, Heinke J, Joshi J, Cell & Environment 40: 3055–3068.
Thonicke K. 2016. Resilience of Amazon forests emerges from plant trait Soong JL, Janssens IA,GrauO,MargalefO, Stahl C, Van Langenhove L,Urbina I,
diversity. Nature Climate Change 6: 1032–1036. Chave J, Dourdain A, Ferry B et al. 2020. Soil properties explain tree growth and
Sato LY, Gomes VCF, Shimabukuro YE, Keller M, Arai E, Nara dos-Santos M, mortality, but not biomass, across phosphorus-depleted tropical forests. Scientific
Brown IF, De Aragao L. 2016. Post-fire changes in forest biomass retrieved by Reports 10: 2302.
airborne LiDAR in Amazonia. Remote Sensing 8: 839. de Souza FC, Dexter KG, Phillips OL, Pennington RT, Neves D, Sullivan MJP,
Sawada Y, Aiba S-i, Takyu M, Repin R, Nais J, Kitayama K. 2015. Community Alvarez-Davila E, Alves A, Amaral I, Andrade A et al. 2019. Evolutionary
dynamics over 14 years along gradients of geological substrate and topography in diversity is associated with wood productivity in Amazonian forests. Nature
tropical montane forests onMount Kinabalu, Borneo. Journal of Tropical Ecology Ecology & Evolution 3: 1754–1761.
31: 117–128. Spear ER, Coley PD, Kursar TA. 2015. Do pathogens limit the distributions of
van SchaikCP,Mirmanto E. 1985. Spatial variation in the structure and litterfall of tropical trees across a rainfall gradient? Journal of Ecology 103: 165–174.
a Sumatran rain forest. Biotropica 17: 196–205. Staver AC, Brando PM, Barlow J, Morton DC, Paine CET, Malhi Y, Araujo
Schietti J, Martins D, Emilio T, Souza PF, Levis C, Baccaro FB, Pinto JLPdV, Murakami A, del Aguila Pasquel J. 2019. Thinner bark increases sensitivity of
Moulatlet GM, Stark SC, Sarmento K et al. 2016. Forest structure along a 600 wetter Amazonian tropical forests to fire. Ecology Letters 23: 99–106.
km transect of natural disturbances and seasonality gradients in central-southern Stephenson NL, Mantgem PJ. 2005. Forest turnover rates follow global and
Amazonia. Journal of Ecology 104: 1335–1346. regional patterns of productivity. Ecology Letters 8: 524–531.
SchimelD, Schneider FD, Carbon JPL, EcosystemP. 2019. Flux towers in the sky: Stovall AEL, Anderson-Teixeira KJ, Shugart HH. 2018. Assessing terrestrial laser
global ecology from space. New Phytologist 224: 570–584. scanning for developing non-destructive biomass allometry. Forest Ecology and
Schimel DS, Asner GP, Moorcroft P. 2013.Observing changing ecological Management 427: 217–229.
diversity in the Anthropocene. Frontiers in Ecology and the Environment 11: 129– Sullivan MJP, Lewis SL, Affum-Baffoe K, Castilho C, Costa F, Sanchez AC,
137. Ewango CEN, Hubau W, Marimon B, Monteagudo-Mendoza A et al. 2020.
Schippers P, Sterck F, Vlam M, Zuidema PA. 2015. Tree growth variation in the Long-term thermal sensitivity of Earth’s tropical forests. Science 368: 869.
tropical forest: understanding effects of temperature, rainfall and CO2. Global Sullivan MJ, Talbot J, Lewis SL, Phillips OL, Qie L, Begne SK, Chave J, Cuni-
Change Biology 21: 2749–2761. Sanchez A, Hubau W, Lopez-Gonzalez G et al. 2017. Diversity and carbon
Schnitzer SA.2018.Testing ecological theorywith lianas.NewPhytologist220: 366– storage across the tropical forest biome. Scientific Reports 7: 39102.
380. Sulman BN, Shevliakova E, Brzostek ER, Kivlin SN, Malyshev S, Menge DNL,
Schnitzer SA, Bongers F. 2002.The ecology of lianas and their role in forests.Trends ZhangX. 2019.Diversemycorrhizal associations enhance terrestrialC storage in a
in Ecology & Evolution 17: 223–230. global model. Global Biogeochemical Cycles 33: 501–523.
Schnitzer SA, Bongers F. 2011. Increasing liana abundance and biomass in tropical Tan ZH, Cao M, Yu GR, Tang JW, Deng XB, Song QH, Tang Y, Zheng Z, Liu
forests: emerging patterns and putativemechanisms. Ecology Letters 14: 397–406. WJ, Feng ZL et al. 2013.High sensitivity of a tropical rainforest to water
Schnitzer SA,Mangan SA,Dalling JW, Baldeck CA,Hubbell SP, Ledo A,Muller- variability: Evidence from 10 years of inventory and eddy flux data. Journal of
LandauH,TobinMF,Aguilar S, BrassfieldD. 2012.Liana abundance, diversity, Geophysical Research-Atmospheres 118: 9393–9400.
and distribution on Barro Colorado Island, Panama. PLoS ONE 7: e52114. TanZH,Zeng JY,ZhangYJ, SlotM,GamoM,HiranoT,Kosugi Y, daRochaHR,
Seiler C, Hutjes RWA, Kruijt B, Quispe J, Anez S, Arora VK, Melton JR, Hickler Saleska SR,GouldenML et al. 2017.Optimumair temperature for tropical forest
T, Kabat P. 2014.Modeling forest dynamics along climate gradients in Bolivia. photosynthesis: mechanisms involved and implications for climate warming.
Journal of Geophysical Research-Biogeosciences 119: 758–775. Environmental Research Letters 12: 054022.
SellanG,Thompson J,MajalapN,BrearleyFQ.2019.Soil characteristics influence Taylor KE, Stouffer RJ, Meehl GA. 2012. An overview of CMIP5 and the
species composition and forest structure differentially among tree size classes in a experiment design. Bulletin of the American Meteorological Society 93:
Bornean heath forest. Plant and Soil 438: 173–185. 485–498.

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com
New
22 Review Tansley review Phytologist
Taylor PG, Cleveland CC, Soper F, Wieder WR, Dobrowski SZ, Doughty CE, Wright SJ. 2010. The future of tropical forests. Annals of the New York Academy of
Townsend AR. 2019.Greater stem growth, woody allocation, and aboveground Sciences 1195: 1–27.
biomass in Paleotropical forests than inNeotropical forests. Ecology 100: e02589. Wright SJ. 2019. Plant responses to nutrient addition experiments conducted in
Taylor PG, Cleveland CC, Wieder WR, Sullivan BW, Doughty CE, Dobrowski tropical forests. Ecological Monographs 89: e01382.
SZ, Townsend AR. 2017. Temperature and rainfall interact to control carbon Wright SJ, Kitajima K, Kraft NJB, Reich PB, Wright IJ, Bunker DE, Condit R,
cycling in tropical forests. Ecology Letters 20: 779–788. Dalling JW, Davies SJ, Diaz S et al. 2010. Functional traits and the growth-
Toledo JJ, Castilho CV, Magnusson WE, Nascimento HEM. 2017. Soil controls mortality trade-off in tropical trees. Ecology 91: 3664–3674.
biomass and dynamics of an Amazonian forest through the shifting of species and WrightSJ,Muller-LandauHC,Schipper J. 2009.The future of tropical species on a
traits. Brazilian Journal of Botany 40: 451–461. warmer planet. Conservation Biology 23: 1418–1426.
de Toledo JJ,MagnussonWE,CastilhoCV,NascimentoHEM. 2011.Howmuch Wright SJ, Sun I-F, Pickering M, Fletcher CD, Chen Y-Y. 2015. Long-term
variation in tree mortality is predicted by soil and topography in Central changes in liana loads and tree dynamics in aMalaysian forest. Ecology 96: 2748–
Amazonia? Forest Ecology and Management 262: 331–338. 2757.
Toledo M, Poorter L, Pena-Claros M, Alarcon A, Balcazar J, Leano C, Carlos Wright SJ, Yavitt JB, Wurzburger N, Turner BL, Tanner EVJ, Sayer EJ, Santiago
Licona J, Bongers F. 2011. Climate and soil drive forest structure in Bolivian LS, Kaspari M, Hedin LO, Harms KE et al. 2011. Potassium, phosphorus, or
lowland forests. Journal of Tropical Ecology 27: 333–345. nitrogen limit root allocation, tree growth, or litter production in a lowland
TownsendAR, AsnerGP, ClevelandCC. 2008.The biogeochemical heterogeneity tropical forest. Ecology 92: 1616–1625.
of tropical forests. Trends in Ecology & Evolution 23: 424–431. Wu J, Albert LP, Lopes AP, Restrepo-CoupeN,HayekM,WiedemannKT,Guan
Turner BL, Brenes-Arguedas T, Condit R. 2018. Pervasive phosphorus limitation K, Stark SC, Christoffersen B, Prohaska N et al. 2016. Leaf development and
of tree species but not communities in tropical forests. Nature 555: 367–370. demography explain photosynthetic seasonality in Amazon evergreen forests.
UngerM,Homeier J, LeuschnerC. 2012.Effects of soil chemistry on tropical forest Science 351: 972–976.
biomass and productivity at different elevations in the equatorial Andes.Oecologia Xu X, Medvigy D, Powers JS, Becknell JM, Guan K. 2016. Diversity in plant
170: 263–274. hydraulic traits explains seasonal and inter-annual variations of vegetation
Venter M, Dwyer J, Dieleman W, Ramachandra A, Gillieson D, Laurance S, dynamics in seasonally dry tropical forests. New Phytologist 212: 80–95.
Cernusak LA, Beehler B, Jensen R, Bird MI. 2017.Optimal climate for large Yanoviak SP, Gora EM, Bitzer PM, Burchfield JC,Muller-LandauHC, DettoM,
trees at high elevations drives patterns of biomass in remote forests of Papua New Paton S, Hubbell SP. 2020. Lightning is a major cause of large tree mortality in a
Guinea. Global Change Biology 23: 4873–4883. lowland neotropical forest. New Phytologist 225: 1936–1944.
Verheijen LM, Aerts R, Brovkin V, Cavender-Bares J, Cornelissen JHC, Kattge J, Zimmerman JK, Everham EM III, Waide RB, Lodge DJ, Taylor CM, Brokaw
Van Bodegom PM. 2015. Inclusion of ecologically based trait variation in plant NVL. 1994.Responses of tree species to hurricane winds in subtropical wet forest
functional types reduces the projected land carbon sink in an earth systemmodel. in Puerto Rico: implications for tropical tree life histories. Journal of Ecology 82:
Global Change Biology 21: 3074–3086. 911–922.
Vicca S, Luyssaert S, Penuelas J, Campioli M, Chapin FS, Ciais P, Heinemeyer A,
Hogberg P, Kutsch WL, Law BE et al. 2012. Fertile forests produce biomass
more efficiently. Ecology Letters 15: 520–526. Supporting Information
VilanovaE, Ramirez-AnguloH,Torres-LezamaA, AymardG,GamezL,DuranC,
Hernandez L, Herrera R, van der Heijden G, Phillips OL et al. 2018. Additional Supporting Information may be found online in the
Environmental drivers of forest structure and stem turnover across Venezuelan Supporting Information section at the end of the article.
tropical forests. PLoS One 13: e0198489.
Visser MD, Schnitzer SA, Muller-Landau HC, Jongejans E, de Kroon H, Comita Dataset S1 Database of the literature results on environmental
LS, Hubbell SP, Wright SJ. 2018. Tree species vary widely in their tolerance for
variation in tropical forest productivity, woody residence time and
liana infestation: A case study of differential host response to generalist parasites.
Journal of Ecology 106: 781–794. biomass that appear in Figs 3, 6 and 7.
Vitousek PM, Sanford RL. 1986.Nutrient cycling in moist tropical forest. Annual
Review of Ecology and Systematics 17: 137–167. Fig. S1 Global distribution of data underlying the studies of
Wagner FH, Herault B, Bonal D, Stahl C, Anderson LO, Baker TR, Becker GS, tropical forest productivity, woody residence time and biomass
Beeckman H, Souza DB, Botosso PC et al. 2016.Climate seasonality limits leaf
carbon assimilation and wood productivity in tropical forests. Biogeosciences 13: reviewed here.
2537–2562.
Wang H, Prentice IC, Davis TW, Keenan TF, Wright IJ, Peng CH. 2017. Fig. S2 Distribution of tropical land area and forest area with
Photosynthetic responses to altitude: an explanation based on optimality respect to mean annual precipitation and mean annual tempera-
principles. New Phytologist 213: 976–982. ture.
Waring BG, Alvarez-Cansino L, Barry KE, Becklund KK, Dale S, GeiMG, Keller
AB, LopezOR,Markesteijn L,Mangan S et al. 2015.Pervasive and strong effects
of plants on soil chemistry: a meta-analysis of individual plant ’Zinke’ effects. Fig. S3 Literature results on spatial variation in productivity,
Proceedings of the Royal Society of London. Series B: Biological Sciences 282: 91–98. residence time, abovegroundbiomass, and associated variables with
Wilcke W, Oelmann Y, Schmitt A, Valarezo C, Zech W, Homeier J. 2008. Soil precipitation, dry season length and other measures of climatic
properties and tree growth along an altitudinal transect in Ecuadorian tropical water availability, graphed in relation to the range of temperature in
montane forest. Journal of Plant Nutrition and Soil Science 171: 220–230.
Wilson AM, JetzW. 2016.Remotely sensed high-resolution global cloud dynamics the study sites.
for predicting ecosystem and biodiversity distributions. PLoS Biology 14:
e1002415. Fig. S4 Mean annual cloud cover in relation to temperature in
Wolf J, Brocard G, Willenbring J, Porder S, Uriarte M. 2016. Abrupt change in tropical forests.
forest height along a tropical elevation gradient detected using airborne Lidar.
Remote Sensing 8.
Wolf K, VeldkampE,Homeier J,MartinsonGO. 2011.Nitrogen availability links Fig. S5 Literature results on spatial variation in productivity,
forest productivity, soil nitrous oxide and nitric oxide fluxes of a tropical montane residence time, aboveground biomass and associated variables with
forest in southern Ecuador. Global Biogeochemical Cycles 25: GB4009.
New Phytologist (2020) 
 2020 The Authors
www.newphytologist.com New Phytologist
 2020 New Phytologist Foundation
New
Phytologist Tansley review Review 23
elevation or temperature, graphed in relation to the range in include land cover types ‘trees and shrubs’ and ‘trees and grasses’ in
precipitation in the study sites. addition to ‘trees’.
Fig. S6Map of relevant SYNMAP land cover classes in the tropics. Fig. S9 Interactive version of Fig. S1, showing the global
distribution of data underlying the studies of tropical forest
Fig. S7Variation in the distributions of mean annual temperature, productivity, woody residence time and biomass reviewed here.
mean cloud cover and mean annual precipitation in relation to
elevation in tropical forests, when tropical forests are defined to Notes S1 Additional information on methods.
include land cover type ‘trees and shrubs’ in addition to ‘trees’.
Please note: Wiley Blackwell are not responsible for the content or
Fig. S8Variation in the distributions of mean annual temperature, functionality of any Supporting Information supplied by the
mean cloud cover and mean annual precipitation in relation to authors. Any queries (other than missing material) should be
elevation in tropical forests, when tropical forests are defined to directed to the New Phytologist Central Office.
New Phytologist is an electronic (online-only) journal owned by the New Phytologist Foundation, a not-for-profit organization
dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews and
Tansley insights.
Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. 
We are committed to rapid processing, from online submission through to publication ‘as ready’ via Early View – our average time
to decision is <26 days. There are no page or colour charges and a PDF version will be provided for each article. 
The journal is available online at Wiley Online Library. Visit www.newphytologist.com to search the articles and register for table
of contents email alerts.
If you have any questions, do get in touch with Central Office (np-centraloffice@lancaster.ac.uk) or, if it is more convenient,
our USA Office (np-usaoffice@lancaster.ac.uk)
For submission instructions, subscription and all the latest information visit www.newphytologist.com

 2020 The Authors New Phytologist (2020)
New Phytologist
 2020 New Phytologist Foundation www.newphytologist.com