Optimum air temperature for tropical forest photosynthesis:mechanisms involved and implications for climate warming

Item Type Article

Authors Tan, Zheng-Hong; Zeng, Jiye; Zhang, Yong-Jiang; Slot, Martijn;Gamo, Minoru; Hirano, Takashi; Kosugi, Yoshiko; da Rocha,Humberto R; Saleska, Scott R; Goulden, Michael L; Wofsy,Steven C; Miller, Scott D; Manzi, Antonio O; Nobre, Antonio D; deCamargo, Plinio B; Restrepo-Coupe, Natalia

Citation Optimum air temperature for tropical forest photosynthesis:mechanisms involved and implications for climate warming 2017,12 (5):054022 Environmental Research Letters

DOI 10.1088/1748-9326/aa6f97

Publisher IOP PUBLISHING LTD

Journal Environmental Research Letters

Rights © 2017 IOP Publishing Ltd

Download date 14/03/2021 09:08:09

Version Final published version

Link to Item http://hdl.handle.net/10150/624639

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 150.135.119.147

This content was downloaded on 06/07/2017 at 18:27

Please note that terms and conditions apply.

Optimum air temperature for tropical forest photosynthesis: mechanisms involved and

implications for climate warming

View the table of contents for this issue, or go to the journal homepage for more

2017 Environ. Res. Lett. 12 054022

(http://iopscience.iop.org/1748-9326/12/5/054022)

Home Search Collections Journals About Contact us My IOPscience

You may also be interested in:

Modeling Long-term Forest Carbon Spatiotemporal Dynamics With Historical Climate and Recent Remote

Sensing Data

Jing M. Chen

An observational study of the carbon-sink strength of East Asian subtropical evergreen forests

Zheng-Hong Tan, Yi-Ping Zhang, Naishen Liang et al.

Tree-ring 13C tracks flux tower ecosystem productivity estimates in a NE temperate forest

Soumaya Belmecheri, R Stockton Maxwell, Alan H Taylor et al.

Warming-induced shift towards forbs and grasses and its relation to the carbon sequestration in an

alpine meadow

Fei Peng, Xian Xue, Manhou Xu et al.

Sensitivity of terrestrial water and energy budgets to CO2-physiological forcing: aninvestigation

using an offline land model

Ranjith Gopalakrishnan, Govindsamy Bala, Mathangi Jayaraman et al.

Uncertainty in the response of transpiration to CO2 and implications for climate change

N Mengis, D P Keller, M Eby et al.

Impact of mountain pine beetle induced mortality on forest carbon and water fluxes

David E Reed, Brent E Ewers and Elise Pendall

A 12-year record reveals pre-growing season temperature and water table level threshold effects on

the net carbon dioxide uptake in a boreal fen

Matthias Peichl, Mats Öquist, Mikaell Ottosson Löfvenius et al.

LETTER

Optimum air temperature for tropical forest photosynthesis:mechanisms involved and implications for climate warming

Zheng-Hong Tan1,17, Jiye Zeng2, Yong-Jiang Zhang3, Martijn Slot4, Minoru Gamo5, Takashi Hirano6,Yoshiko Kosugi7, Humberto R da Rocha8, Scott R Saleska9, Michael L Goulden10, Steven C Wofsy11,Scott D Miller12, Antonio O Manzi13, Antonio D Nobre14, Plinio B de Camargo15 and Natalia Restrepo-Coupe9,16

1 Ecology Program, Institute of Tropical Agriculture and Forestry, Hainan University, Haikou 570228, People’s Republic of China2 National Institute for Environmental Studies, Tsukuba 305-0053, Japan3 Department of Organismic and Evolutionary Biology, Harvard University, Cambridge 02138, United States of America4 Smithsonian Tropical Research Institute, Apartado 0843–3092, Panama5 National Institute of Advanced Industrial Science and Technology, Tsukuba 305–8569, Japan6 Graduate School of Agriculture, Hokkaido University, Sapporo 060–8589, Japan7 Forest Hydrology, Graduate School Agriculture, Kyoto University, Kyoto 606–8502, Japan8 Department of Atmospheric Science, Universidade de Sao Paulo, Sao Paulo 05508-090, Brazil9 Department of Ecology and Evolutionary Biology, University of Arizona, Tucson AZ, 85721, United States of America10 Department of Earth System Science, University of California, Irvine 92697, United States of America11 Division of Applied Science, Harvard University, Cambridge 02138, United States of America12 Atmospheric Sciences Research Center, State University of New York at Albany, New York 12203, United States of America13 Instituto Nacional de Pesquisas da Amazonia (INPA), Manaus 69011, Brazil14 Laboratório de Ecologia Isotópica, Universidade de São Paulo, São Paulo 05468, Brazil15 Department of Biological Sciences, Macquarie University, Sydney 2109, Australia16 Plant Functional Biology and Climate Change Cluster, University of Technology Sydney, Sydney, NSW, 2007, Australia17 Author to whom any correspondence should be addressed.

E-mail: tanzh@xtbg.ac.cn

Introduction

Tropical forests are characterized by a warm andhumid climate (Corlett 2011); however, there iscurrently little consensus on whether climate changewill affect tropical forests. Paleoecological studies showthat neotropical vegetation largely persisted after a 3 to5 °C warming during the Paleocene–Eocene ThermalMaximum (Jaramillo et al 2010). However, thishistorical warming was short-lived and considerablyslower than current warming and future warmingpredicted for the next century. A survey of thetemperatures of broad-leaved forest land coversuggests that climatic warming could have severeconsequences for tropical floras (Wright et al 2009).Closed-canopy forests are found in areas with a meanannual temperature below 28 °C, whereas areas withmean temperatures above 28 °C support shrubs andgrasses instead of broad-leaved evergreen trees. Giventhat excessively high temperatures are typicallyassociated with a high evaporative demand and dryclimate, the absence of closed-canopy forests in areaswith temperatures above 28 °C could also be aconsequence of water limitation. This past recordand the distribution of tropical forests suggest atemperature limit, and therefore the ecosystemsensitivity to this threshold needs to be further studied.

Photosynthetic performance, the basis for carbonsequestration and ecosystem production, is tempera-ture dependent. In general, the light-saturatedphotosynthetic rate increases with temperature to apeak, which is followed by a decline (Sage and Kubien2007). It has been suggested that current temperaturesin regions supporting tropical forests are very close toor even exceed their photosynthetic optimum tem-perature (Topt) (Doughty and Goulden 2008). This is apotentially ominous warning sign for our warmingEarth. Tropical forests store large amounts of carbonin biomass (Dixon et al 1994). Consequently, a slightperturbation in tropical carbon fluxes could havewide-ranging effects on global atmospheric CO2

concentrations (Anderegg et al 2015). Temperaturesin excess of Topt could result in a sharp decline inphotosynthetic carbon sequestration in tropical forests(Doughty and Goulden 2008, Vårhammar et al 2015).A decline in CO2 uptake by the forests could in turnresult in an increase in atmospheric CO2, which wouldfurther accelerate warming through positive feedback.

Model simulations indicate that tropical forests arecurrently not at their high-temperature threshold.With the aid of widely-used process models, Lloyd andFarquhar (2008) showed that the temperature oftropical forests was still well below Topt. They arguethat the apparent decrease in photosynthetic rate with

OPEN ACCESS

RECEIVED

5 December 2016

REVISED

23 April 2017

ACCEPTED FOR PUBLICATION

26 April 2017

PUBLISHED

19 May 2017

Original content fromthis work may be usedunder the terms of theCreative CommonsAttribution 3.0 licence.

Any further distributionof this work mustmaintain attribution tothe author(s) and thetitle of the work, journalcitation and DOI.

Environ. Res. Lett. 12 (2017) 054022 https://doi.org/10.1088/1748-9326/aa6f97

© 2017 IOP Publishing Ltd

increasing temperature is predominantly an indirecteffect of stomatal closure (30%) and not a directeffect of warming on mesophyll processes (2%).Temperature increases could reduce photosynthesis,either directly through inhibiting the activity ofphotosynthetic enzymes and electron transport, orindirectly through decreasing stomatal conductance(Farquhar et al 1980). A linear increase in tempera-ture could lead to an exponential growth of watervapour pressure deficit (D) (Campbell and Norman1998), whereas stomatal aperture (conductance)decreases with increased D (Damour et al 2010).Since CO2 enters the mesophyll through stomata,intercellular CO2 (ci) and photosynthesis decreasewhen stomatal conductance (g) declines (Farquharet al 1980). In contrast to the direct effects oftemperature on the photosynthetic apparatus, areduction in photosynthesis caused by increasedstomatal resistance could be offset, at least partly, byelevated CO2 (Lloyd and Farquhar 2008). ElevatedCO2 increases Topt by reducing photorespiration andstomatal resistance, which has a positive effect on theacclimation potential of photosynthesis. Moreover,stomatal closure reduces transpiration and subse-quently reduces its cooling effect (Doughty 2011).This could in turn lead to excessively high temper-atures at the leaf level, which could cause irreversibledamage to the photosynthetic machinery (Berry andBjörkman 1980, Doughty 2011).

In this study, to examine the potential effects ofclimate change on forest photosynthesis, we firstquantified the Topt of ecosystem photosynthesis (ToptE)for seven tropical forests across different continents.Wethen analyzed the relationship between ToptE and meangrowing season air temperature (Ta) to confirm the

widely held consensus that these parameters increasesimultaneously. Ecosystem physiological parameterswere then inverted using a big-leaf analogized processmodel driven by ecosystem photosynthesis measure-ments. Further, we tested the hypothesis proposed byLloyd and Farquhar (2008), which suggests thatstomatal processes play a prominent role in determiningTopt, and that increasing ambient CO2 concentrationswill increase tropical forestTopt, whichwould imply thatthese forest are not as vulnerable to climate change asmay have been indicated by Doughty and Goulden(2008). Finally, we discuss the implications of differentclimate warming scenarios on ecosystem photosynthe-sis in tropical forests.

Material and methods

Studied sitesTropical rainforests are primarily distributed in theAmazon, Southeast Asia, and Africa. In the presentstudy, we investigated seven tropical forests, four ofwhich are located in Southeast Asia and three are in theAmazon (table 1). All three Amazonian sites arelocated near the equator (latitude ∼3°S): from west toeast K34, K67, and K83. The four Asian rainforests arein two different locations: two sites (PDF and PSO) arenear the equator (∼2°S or N) and the other twoThailand forests (MKL and SKR) are located at∼14°N.The selected Asian rainforests are dominated by treesin the Dipterocarpaceae; the exception being the peatswamp forest of the PDF site. Canopy height typicallyexceeds 30 m, although in the peat forest themaximum height is approximately 26 m. The forestat the K83 site has previously been selectively logged

Table 1. Site information.

Lat. Long. Age Hc LAI Ta ppt System Anemometer IRGA Period Country

AmazonianK34 2° 360 S 60° 120 W Primary 30∼35 4.7 26.7 2286 CP Wind master,

Gill

Li-6262,

Li-Cor

1999∼2006 Brazil

K67 2° 510 S 54° 580 W Primary 35∼40 6.0 24.8 1811 CP CSAT3,

Campbell

Li-6262,

Li-Cor

2002∼2006 Brazil

K83 3° 30 S 54° 560 W Selective

logged

35∼40 4.9 24.8 1811 CP CSAT3,

Campbell

Li-7000/

6262,

Li-Cor

2000∼2004 Brazil

SE AsiaMKL 14° 340 N 98° 500 E ∼30 (2008) 30 2∼3 27.5 1650 CP Wind master,

Gill

Li-6262,

Li-Cor

2003∼2004 Thailand

PDF 2° 200 N 114° 20 E — ∼26 5 26.3 2231 OP CSAT3,

Campbell

Li-7500,

Li-Cor

2002∼2005 Indonesia

PSO 2° 580 N 102° 180 E Primary 35∼45 6.52 25.3 1804 OP SAT550, Kaijo Li-7500,

Li-Cor

2003∼2009 Malaysia

SKR 14° 290 N 101° 550 E Mature 35 3.5∼4.0 26.2 1240 CP Wind master,

Gill

Li-6262,

Li-Cor

2001∼2003 Thailand

‘—’, no data available; Lat., Latitude; Long., Longitude; Age, stand age (yr); Hc, canopy height (m); LAI, leaf area index; Ta, mean

annual temperature (°C); ppt, precipitation (mm); System, the open (OP) or closed path (CP) eddy covariance system; IRGA, the

infrared gas analyzer model.

Environ. Res. Lett. 12 (2017) 054022

2

for experimental purposes (Goulden et al 2004). Allthe studied forests have a year-round growing season,with the exception of MKL, in which a proportion ofthe trees shed their leaves during the late dry season.For additional detailed information on these sites andinstrumentation please refer to the previously pub-lished studies of Hirata et al (2008) and Restrepo-Coupe et al (2013).

Eddy flux observations and data processingThe CO2 movement in the lower atmosphericboundary layer is primarily driven by turbulence thatcan be measured using the eddy covariance technique(EC) (Baldocchi 2003). Photosynthetic rates werequantified by examining the ecosystem–atmosphereCO2 exchange. The daytime CO2 exchange measuredusing EC apparatus is conceptually equal to netecosystem photosynthesis (table 2).

We collected EC flux data for the seven forests fromflux networks. The fluxes have a temporal resolution of30minor 1h, and span at least twoyears. Themajorfluxdata used in this study include the following: net CO2

exchange (NEE, after storageflux correction), latent heatflux (LE), sensible heat flux (Hs), net radiation flux (Rn),and soil heat flux (G). In addition to flux data, we alsoused meteorological measurements, including airtemperature (Ta, °C), relative humidity (hs, %), watervapour pressure deficit (DE, kPa), and soil water content(SWC, m3 m�3). For reproducibility, the data areavailable at the following sites:

AsiaFLUX dataset: https://db.cger.nies.go.jp/asiafluxdb/

BrasilFLUX dataset: www.climatemodeling.org/lba-mip/

Determining the optimum temperature (Topt) ofphotosynthesisDetermining of Topt could be done based on grossphotosynthesis (GPP) dataset. The advantage of thisway is reducing uncertainties related to respiratoryprocesses which are most significant in eddy flux cases(Yi et al 2000, Yi et al 2004). However, a reliablemethod to obtain GPP by portioning NEE is currentlyunavailable either because light inhibition of leafrespiration or inconsistence of temperature depen-dency of autotrophic respiration (cf. Yi et al 2004 andreference therein).

Topt could also be determined by fitting a peakfunction to the temperature response of light-saturatedphotosynthesis (Lange et al 1974). In leaf-level studies,temperature is specified to leaf temperature (TL), andlight in the leaf chamber is set to a saturating levelduring CO2 exchange measurements. There are,however, some modifications required when theseequations are applied at the ecosystem level. Instead ofleaf temperature, we used air temperature near thecanopy level (Niu et al 2012). Therefore, in the presentstudy, the ecosystem photosynthesis Topt (ToptE) wasdetermined in terms of optimum air temperature. Todetermine values for light-saturated photosynthesis, weomitted all data points below site-specific saturatinglight levels. The site-specific light saturation point wascalculated by applying a non-rectangular hyperbola tothe stand-level photosynthesis–light response curve(Lasslop et al 2010).

There are several peak functions that could be usedto fit the temperature response curve to determineTopt. We adopted a modified function of the modelproposed by June et al (2004):

NEEsat ¼ NEE25expðbðTK � 298Þ=ð298RTkÞÞ=½1þ expðcðTk � ToptEÞÞ�2; ð1Þ

where NEEsat is the measured net ecosystem photo-synthesis rate under light saturation (mmol m�2 s�1)(note that a positive NEE indicates photosynthesisuptake, in order to make it comparable to that of leaf-level conventions), Tk is the ambient temperature indegrees Kelvin, R is the gas constant, andNEE25 (mmolm�2 s�1), b, c, and ToptE (Kelvin) are fitted parameters.

The Farquhar–von Caemmerer–Berry (FvCB) modelA process-based photosynthesis model was used inthis study. The model is a combination of theFarquhar–von Caemmerer–Berry photosynthesis(FvCB) model (Farquhar et al 1980) and theBall–Berry stomatal conductance model (Ball et al1987), with some additional parameterization infor-mation provided by von Caemmerer et al (2009). Thedetailed model equations are listed in table A1 in theappendix. We used an iteration method to solveintercellular CO2 (ci). The model was coded in theCþþ environment and is available upon request.

Table 2. Terms (and their abbreviations) used at the leaf-level andthe corresponding abbreviations used at ecosystem-level. Pg: leafgross photosynthetic rate; GPP: gross primary production ofecosystem; RL: leaf respiration; RE: ecosystem respiration, which isthe sum of autotrophic and heterotrophic respiration; TL: leaftemperature; Ta: air temperature near the top of the forest canopy;ciL: leaf intercellular CO2 concentration; ciE: bulk canopyintercellular CO2 concentration; DL: leaf-to-air water vapourdeficit; DE: atmospheric water deficit; gs: stomatal conductance; Gc:canopy conductance; Pn: net leaf photosynthesis rate, where Pn ¼Pg � RL; NEE: net ecosystem exchange, where NEE ¼ GPP � RE.

General description Leaf Ecosystem

Gross photosynthesis Pg GPP

Dark respiration: Rd RL RE

Temperature T TaIntercellular CO2 concentration: ci ciL ciEWater vapour deficit: D DL DE

Stomatal or canopy conductance: g gs Gc

Maximum carboxylation rate: Vcmax VcmaxL VcmaxE

Maximum electronic transport rate: Jmax JmaxL JmaxE

Conductance sensitivity: g1 g1L g1ETemperature curve factor: S SL SETemperature curve factor: H HL HE

Net photosynthesis rate Pn NEE

Optimal temperature: Topt ToptL ToptE

Environ. Res. Lett. 12 (2017) 054022

3

The big-leaf analogyThe factorsused indeterminingTopt canbe summarizedas photosynthetic biochemical (biochemical hereafter),respiratory, and stomatal processes (Hikosaka et al2006,Lin et al 2012). In order to separate the relativecontributions of each process, we implemented theFvCB model with a big-leaf analogy.

Firstly, the ecosystem as a whole was abstractedinto a ‘big leaf ’. This is consistent with the philosophyof the EC method and enabled us to directly use theleaf-level FvCB model at an ecosystem-level. The ECsystem measures the gas exchange at ecosystem levelanalog to leaf chamber measurements do a small scale(table 2). Thus, daytime net ecosystem exchange (NEE)was regarded as equivalent to net photosynthesis rate(Pn) at the leaf level. At the ecosystem level, the airtemperature near the canopy (Ta), air water vapourdeficit (DE), and canopy bulk intercellular CO2

concentration (ciE) corresponded to leaf temperature(TL), leaf-to-air water vapour deficit (DL) andintercellular CO2 concentration (ciL) at the leaf level,respectively. Critically, ecosystem respiration (RE) atthe ecosystem level was considered analogous to leafrespiration (RL) at the leaf level. After analogizing,ecosystem terms were derived that corresponded tothe leaf terms, and the FvCB model was applied at theecosystem level and driven by ecosystem measure-ments.

This type of big-leaf analogy differs from that ofthe big-leaf model (de Pury and Farquhar 1997), inthat here the ecosystem as a whole was treated as a bigleaf. Therefore, the parameters derived here for theecosystem are not directly comparable to those used inleaf studies. The overall motivation for us inconducting this analogy was to make the parameterinversion as simple as possible but with necessaryphysiological considerations. Parameter inversion isvery sensitive to initial parameter values because ofmany non-linear processes, and consequently it ismore practical and helpful to construct simple modelswith certain assumptions than to execute a complexmulti-layer model (Wang et al 2007).

Inverting photosynthesis parameters bycombining the FvCB model and ecosystemfluxes

Parameters of theFvCBmodel at the ecosystem levelwereinverted from eddy flux observations after abstracting.For inversion, we used the Levenberg–Marquardtoptimization algorithm. We also examined whetherthe inversion method would have a strong impact onthe inverted parameters. A Bayesian statisticalmethod, the ‘adaptive population Monte Carloapproximate Bayesian computation’ method (Lenor-mand et al 2013), was included for comparison. Thealgorithm pseudo-code of the Bayesian method waspresented in Zeng et al (2017).

Results

The ToptE of tropical forestsThe temperature dependence of light-saturated eco-system photosynthesis (NEEsat) is shown in figure 1general, there was a clear unimodal pattern for mostof the sites. The ToptE determined by fitting equation1 to the observations varied from 23.7 to 28.1 °Cacross sites.

A close relationship was found between meanannual air temperature (Ta) and ToptE (figure 2(a)).Since most tropical forests maintain a year-roundgrowing season, the mean annual Ta could roughly betreated as the growth temperature. Therefore, tropicalforests growing under higher growth temperature tendto have a higher ToptE. The slope of the linearrelationship is close to one (1.12). ToptE was alsorelated to mean air temperature under light saturatedcondition (figure 2(b)). When the sites with seasonallyclimate were omitted, a very close relationship wasfound between ToptE and mean air temperature underlight saturated condition.

Contribution of physiological parameters tothe change in ToptE across sites

The goodness-of-fit was shown in figure 3 whenimplemented the FvCB model to these datasets. Ingeneral, themodelfitted results have a good relationshipwith that of observations. It suggests high reliability ofthese inverted parameters (table 3). Principal compo-nent analysis of these parameters identified threecomponents that could explain over 86% of thevariance. However, none of the three components weresignificantly correlated with ToptE (data not shown). Afurther correlation analysis showed that the activationenergy of ecosystem respiration (RE) was the onlyparameter significantly correlated with ToptE (figure 4(a)). We found that two sites (PDF and MKL indicatedby open circles in figure 4(b)) differ from the remainingsites with respect to the relationship between ToptE andstomatal sensitivity (g1E): with the exception of PDF andMKL, the sites showanegative correlationbetweenToptEand g1E. These two excluded sites have special waterconditions which discussed latter in the discussionsection.

The contribution of stomatal processes indetermining ToptE

The PSO site, which has eight years’ continuous fluxdata, was taken as an example of per-humid site toillustrate the contribution of stomatal processes indetermine ToptE. The overall ToptE for the PSO site was26.5 °C for the entireDE range, as shown by the dashedline in figure 5(a) When the whole dataset was dividedintodifferentDE levels,we found that the light-saturated

Environ. Res. Lett. 12 (2017) 054022

4

photosynthesis rate (NEEsat) increased with Ta even atvalues greater than 30 °C (see data points and regressionline in different colours, figure 5(a). Only at the highestDE level was NEEsat found to decrease with Ta. In thisregard, ToptE (when NEEsat starts to decrease with anincrease in Ta) should at least be 30 °C when DE iscontrolled. This contrasts with the value of 26.5 °Cobtained using the full DE range and implies strongstomatal control ofToptE.We also inferredparameters ofFvCB model for all DE sub-ranges (table 4). Most ofthese parameters showed a unimodal pattern alongwithincrease of DE. A similar case was found in a site withstrongly seasonal climate (figure 5(b)).

Discussion

The ecosystem ToptE we quantified differs from leafToptL in several aspects. Leaf ToptL is specified to leaftemperature, not air temperature. The leaf surface is a

direct light interceptor, which leads to strongertemperature variations in leaves than in ambient air,i.e. the transitional leaf temperature can easily reach40 °Cunder full light (Doughty and Goulden 2008). Inaddition, the dark respiration term (Rd) at theecosystem level (RE) is the sum of respirations fromdifferent organisms, litter, woody debris, and soilorganic matter, whereas at the leaf level, the respirationterm (RL) is specified to leaf respiration. Despite thesedifferences, however, the ecosystem ToptE obtained inour study is very close to that of leaf ToptL. Forexample, two Costa Rican tropical forest species grownunder a daily temperature of 27 °C showed a leaf ToptLof 27 °C (Vargas and Cordero 2013), which is close tothe ecosystem ToptE value we determined for tropicalforests. Similarly, Slot and Winter (2017) reportedmean ToptE values of 30.4 °C and 29.2 °C for theupper-canopy leaves of 42 species in two lowlandforests in Panama, which were close to the meanafternoon air temperature. The higher ToptL values

(a) K34 (b) K67

(c) K83 (d) MKL

(e) PDF

(g) SKR

(f) PSO

25

20

15

10

5

25

20

15

10

5

25

20

15

10

5

25

20

15

10

5

0

20

15

10

5

20

15

10

5

0

35

30

25

20

15

10

5

ToptE : 28.1 °C

ToptE : 25.2 °C

ToptE : 26.8 °C

ToptE : 26.3 °C

ToptE : 26.4 °C

ToptE : 27.1 °C

ToptE : 23.7 °C

NE

Esa

t (µm

ol m

-2 s

-1)

NE

Esa

t (µm

ol m

-2 s

-1)

Air temperature (K)

Air temperature (K)

285 295 305 310300290

295 305 310300290

Figure 1. Temperature response of light saturated ecosystem photosynthesis (NEEsat). A modified June et al (2004) function was fittedto the data point and optimum temperature (ToptE) was determined.

Environ. Res. Lett. 12 (2017) 054022

5

(a) K34 (b) K67

(c) K83 (d) MKL

(e) PDF

(g) SKR

(f) PSO

Observed daytime NEE (µmol m-2 s-1)

Day

time N

EE c

alcu

late

d ba

se o

n fit

ting

FvC

B m

odel

(µm

ol m

-2 s

-1)

Observed daytime NEE (µmol m-2 s-1)

30

20

10

0

-10

30

20

10

0

-10

-20

40

30

20

10

0

-10

-20

30

20

10

0

-10

-20

30

20

10

0

-10

20

10

0

-10

20

10

0

-10

-10 0 10 20 30

-10 0 10 20 30

-10 0 10 20

-20 -10 0 10 20

-10 0 10 20

30 -20 -10 0 10 20 30

-20 -10 0 10 20 30

40

Figure 3. A comparison on observed and FvCB model fitted daytime net ecosystem exchange (NEE) to illustrate the goodness-of-fit.The fitted parameters were listed in table 3. The solid line is 1:1 line.

(a) (b)29

28

27

26

25

24

23

32

30

28

26

24

2223 24 25 26 27 28 29 22 24 26 28 30 32

Mean annual air temperature (°C) Light saturated mean air temperature (°C)

slope: 1.11; r: 0.80; p: 0.03 slope: 3.0; r: 0.99; p: 0.05

Opt

imum

pho

tosy

nthe

sis

tem

pera

ture

, Top

t (°C

)

K34

K34MKL

MKL

PDF

PDFSKR

SKR

PSOPSO

K83K83

K67K67

1:1 1:1

Figure 2. A comparison on photosynthesis optimum temperature (ToptE) with mean annual air temperature (Ta) (a) and mean lightsaturated air temperature (b). Black line is linear regression line, and grey line is 1:1 line. The open circles in subpanel (b) were notincluded in regression. The statistic information was shown in the top of figures.

Environ. Res. Lett. 12 (2017) 054022

6

reported by Slot andWinter (2017), compared with thevalues reported in the present study and those reportedbyVargas andCordero (2013), couldbe explainedby thefact that upper canopy leaves experience higher lightintensity and leaf temperatures compared to the wholecanopy mean values. In the following sections, wediscuss the possible mechanism of ToptE changes acrosssites, thecontributionof stomatal processes toToptE, andthe implications of our findings.

The mechanisms of ToptE changes acrosssites

Our cross-site analysis shows that tropical forestsgrowing in a warmer climate tend to exhibit higherToptE (figure 2(a)). We separated the contributions ofbiochemical, respiratory, and stomatal processes toToptE by means of parameter inversions. Respiratoryprocess play a role in ToptE, as suggested by the

significant relationship between ToptE and the activa-tion energy (Ea) of respiration (RE) (figure 4(a)). Thisfinding differs from those of leaf level studies, whichindicate that leaf respiration plays a negligible role inToptL (Lin et al 2012). At the ecosystem level, RE is thesum of the autotrophic respiration of all organisms(above- and below-ground) and heterotrophic respi-ration of soil organic matter and litter. At the leaf-level,however, RL reflects only leaf respiration, whichtypically represents a small fraction of net photosyn-thesis. These differences emphasize the importance ofrespiratory processes in studying ToptE at the ecosys-tem level, even though it is negligible at the leaf level.

It is known that activation energy (Ea) canrepresent the temperature sensitivity (Q10) of RE,the relationship of which can be expressed as follows:

Q10 ¼ exp10Ea

RTaðT a þ 10Þ� �

: ð2Þ

(a) (b)

PDFMKL

0 10 20 30 40 50 60 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0

29

28

27

26

25

24

23

Opt

imal

pho

tosy

nthe

sis

tem

pera

ture

, Top

tE (°

C)

Activation energy of ecosystem respiration (kJ) Stomatal sensitivity factor, g1E

r=-0.85p=0.01

r=-0.99p<0.01

Figure 4. Optimum photosynthesis temperature (ToptE) was related to activation energy (Ea) of dark respiration (RE) and stomatalsensitivity factor (g1E). Two sites indicated by open circles were not included in regression for subfigure (b).

Table 3. Parameters of a photosynthesis model inverted using a nonlinear regression method.

Site Rate at 25 °C (mmol m�2 s�1) Activation energy (J) SE HE g1E

VcmaxE JmaxE RdE VcmaxE JmaxE RdE

K34 294 190 9.03 65 340 76 340 10 000 706 218 782 2.31

K67 297 212 10.00 60 503 35 255 66 673 702 217 451 3.27

K83 290 215 9.64 63 009 33 047 39 721 408 126 451 2.97

MKL 295 235 9.15 63 724 56 013 25 547 630 195 119 3.83

PDF 230 212 10.00 61 913 38 906 44 375 697 215 911 3.40

PSO 185 209 9.44 59 110 70 404 15 779 798 247 355 2.60

SKR 252 239 9.59 62 342 56 916 35 102 710 220 120 2.69

Vcmax, maximum Rubisco activity; Jmax, maximum electron transport rate; Rd, dark respiration rate; S, a term similar to an entropy

factor, H, the rate of decrease in the function above the optimum, g1, stomatal sensitivity factor.

Environ. Res. Lett. 12 (2017) 054022

7

The relationship between decreasing ToptE andincreasing Q10 suggests that the RE of tropical forestswith higher ToptE and mean Ta is less sensitive to Tavariations than forests with lower ToptE. This isconsistent with previous reports showing that Q10

decreases with rising temperature, as a consequence ofthermal acclimation (Tjoelker et al 2009, Slot andKitajima 2015).

When we excluded PDF and MKL (two secondaryforests with unique hydrological conditions) from theanalysis, we identified a strong correlation between g1E

and ToptE(figure 4(b)). The forest in the PDF site isdrained peat swamp forest (Hirano et al 2007), whichis generally waterlogged. By contrast, the MKL siteexperiences seasonal water deficits (Gamo et al 2005).Since stomatal conductance or g1E is highly sensitive towater availability, it seems appropriate to treat thesetwo secondary forests as outliers when investigatingToptE–g1E relationships.

Theoretically, g1E, which is a stomatal sensitivityfactor, would be expected to increase with increasinggrowth temperature, as does ToptE (Leuning 1990,

Table 4. The inverted parameters under different water vapor pressure deficit (DE) levels. This was carried out in the per-humidPasoh site. This table could correspond to figure 4(a).

DE levels (kPa) <0.7 0.7∼0.9 0.9∼1.1 1.1∼1.2 1.2∼1.3 1.3∼1.4 1.4∼1.5 1.5∼1.6 1.6∼1.8 1.8∼2.1 >2.1

Rate at 25 °C(mmol m�2 s�1)

VcmaxE 91 100 114 254 288 289 273 268 165 279 101

JmaxE 125 123 107 110 140 118 129 129 107 91 102

RdE 0.10 0.51 0.10 1.91 8.10 5.67 8.50 8.98 6.59 5.00 2.90

Activation energy (J)

VcmaxE 62 714 59 957 60 577 64 505 68 994 70 017 66 453 66 916 63 578 66 943 59 348

JmaxE 44 315 41 412 47 631 51 075 66 799 65 574 79 995 79 573 79 994 79 998 41 830

RdE 63 481 63 637 63 254 63 565 55 985 60 646 64 301 61 926 62 723 58 468 64 409

SE 742 729 756 798 1052 1126 903 896 859 874 742

HE 22 9765 22 5983 23 4348 24 7302 32 5870 34 8879 27 9685 27 7651 26 6097 27 0846 22 9800

g1E 11.00 10.51 13.00 20.43 36.50 30.72 13.85 14.31 10.97 3.44 2.08

Vcmax, maximum Rubisco activity; Jmax, maximum electron transport rate; Rd, dark respiration rate; S, a term similar to an entropy

factor, H, the rate of decrease in the function above the optimum, g1, stomatal sensitivity factor.

(a) PSO

(b) MKL

NE

Esa

t (µm

ol m

-2 s

-1)

NE

Esa

t (µm

ol m

-2 s

-1)

22

20

18

16

14

12

10

8

6

40

35

30

25

20

15

10

5

0

22 24 26 28 30 32 34

222018 24 26 28 30 32 34 36

Air temperature, Ta (°C)

DE level (kPa)

<0.70.7~0.90.9~1.11.1~1.21.2~1.31.3~1.41.4~1.51.5~1.61.6~1.81.8~2.1>2.1

All

<0.700.70~0.850.85~0.950.95~1.05

1.05~1.151.15~1.301.30~1.451.45~1.601.60~1.851.85~2.10>2.10

1.53 1.13 1.120.57 0.55

0.620.34

0.26

0.47

0.57

-0.35

-1.13

-0.12-0.47

-0.582.91

2.802.91

2.802.963.32 3.53

Figure 5. The temperature response of light saturated photosynthesis rate (NEEsat) under different water vapor deficit (DE) levels. (a)represents the per-humid environment collected from the PSO site; (b) represents seasonal climate collected from the MKL site. Thesymbols and regression line with different colours represent different DE levels. Values of different color represent the correspondingslope of the regression line. The dashed line in subpanel (a) is fully in the DE range which is the same as figure 1(f).

Environ. Res. Lett. 12 (2017) 054022

8

Medlyn et al 2011). Our numeric simulation alsosupports this theoretic expectation for a specific site bychecking ToptE with varied g1E (data not shown).Nevertheless, we found that at the ecosystem level fortropical forests, ToptE tends to decrease with increasingg1E (figure 4(b)). Since g1E is strongly correlated withthe Ea of RE (Pearson’s r = 0.96), it is would be difficultto state that the strong correlation shown in figure 4(b)is solely attributable to g1E or whether it is merely anindirect reflection of the Ea–ToptE relationship shownin figure 4(a). In situ warming experiments at both leafand ecosystem levels might be helpful in reconcilingthese contrasts (Cavaleri et al 2015).

Interestingly, we found that biochemical processesdid not play a significant role in ToptE changes acrosssites. Traditionally, Topt acclimation studies haveprimarily focused on biochemical processes (Hikosakaet al 2006). However, in the present study these keyprocesses were found to make a negligible contribu-tion to ToptE changes across sites. Subsequent tofurther confirmation that thermal acclimation ofrespiration rather than biochemical processes is amore important determinant of ToptE, these findingsshould be implemented in global change models(Lombardozzi et al 2015).

The role of stomatal processes indetermining ToptE

Previous studies have shown that stomatal processesare potentially important (Lin et al 2012, Duursmaet al 2014, Slot and Winter 2017), or even the mostimportant factors determining ToptL (Lloyd andFarquhar 2008, Rowland et al 2015). Nevertheless,how stomatal processes control Topt is still not wellunderstood. This uncertainty is partly caused by theconfounding effects of temperature and water factorson Topt.

Relative humidity (hs) and vapour pressure deficit(D) are strongly dependent on temperature (Campbelland Norman 1998). When temperature rises, thesaturated water vapour pressure will increase expo-nentially. This, in turn, will alter both hs and D, andhence stomatal conductance. Therefore, there is anindirect effect of temperature on Topt through its effecton stomatal conductance. This effect is illustrated infigure 5. The dashed line in figure 5(a) shows thetemperature response curve represented in figure 1(f)under the fullDE range. However, when we divided thewhole dataset into different DE levels (as shown by thedifferent colours in figure 5(a)), it was apparent thatNEEsat increases with temperature within these subsetsuntil the temperature exceed 30 °C . This indicates thatthe ToptE should be at least 30 °C if there is no DE

limitation on the stomatal response, which isconsiderably higher than the value estimated fromthe entire DE range (26.4 °C). This analysis lendssupport to the idea that stomatal processes play a

significant role in determining ToptE, as temperaturemay indirectly influence photosynthesis throughchanging D.

The implications under future climatechange

In the future, the Earth’s surface air is predicted tobecome richer in CO2 and higher inmean temperature(Corlett 2011). At present, however, there seems littleconsensus on how tropical forest ecosystem willrespond under such a climatic scenario (see the reviewby Lloyd and Farquhar (2008)). Our findings,however, provide certain insights into how tropicalforest ecosystems might respond and could serve ascomplement to previous studies in our pursuit of amore complete understanding future changes in forestphotosynthesis.

Firstly, we revealed the role of stomatal limitationin determining ToptE at the ecosystem level, which islargely consistent with leaf-level findings. The role ofstomatal effects in shaping ToptL have been welldemonstrated in leaf-level measurements (Koch et al1994, Ishida et al 1996, Carswell et al 2000, Slot andWinter 2017) and have been verified by a leaf-levelmodel (Lloyd and Farquhar 2008). Our ecosystem fluxanalysis showed that without DE limitation onstomatal conductance, tropical forests could have ahigher photosynthetic performance (NEEsat) underhigh Ta, as indicated by their increased ToptE (figure 5).The direct implication of this finding is that factorsaffecting stomatal conductance will contribute sub-stantially to the modification of ToptE. Among thesefactors, the most prominent is ambient air CO2

concentration. Given unchanged moisture conditions(e.g. soil water or D), ToptE is expected to increase withCO2 and will decrease stomatal limitation onphotosynthesis (Lloyd and Farquhar 2008). Accord-ingly, this can be considered as a positive signal fortropical forests given the prospect of increasing CO2

levels.Secondly, tropical forests in environments with

higher Ta tend to have higher ToptE (figure 2), which isconsistent with growth chamber cultivation experi-ments (Kositsup et al 2009) and cross-seasonobservations (Lange et al 1974). This pattern suggestspotential acclimation of tropical forests to Ta, which isa further adaptive strategy that will increase theresilience of tropical forest given the predicted climatewarming scenarios.

Thirdly, the significant relationship between theactivation energy of respiration and ToptE implies thepossible thermal acclimation of RE (figure 4(a)). Thetemperature acclimation of ecosystem respiration, i.e.the decrease in the sensitivity of respiration totemperature changes as growth temperature increases,would have a positive effect on net photosynthesis andlead to increases in ToptE.

Environ. Res. Lett. 12 (2017) 054022

9

Collectively, our findings indicate an optimisticfuture for tropical forests under the predictedconditions of global climate change. Nevertheless,some uncertainties remain. Firstly, increasing CO2 willreduce stomatal conductance and water losses andhence the cooling effect of transpiration. This couldpotentially result in excessively high leaf temperaturesand consequently heat damage and declines inphotosynthesis and carbon sequestration. Further-more, because ecosystem respiration in the tropics andsubtropics is generally more sensitive to warming thanthat of photosynthesis (Yi et al 2010, Zhang et al 2016),it remains unclear to what extent the warming-induced increase in night-time ecosystem respirationwould offset the positive effect of thermal acclimationin photosynthesis on net carbon sequestration.

Conclusions and implications

In conclusion, we quantified ecosystem ToptE fortropical forests, which ranges from 23.7 to 28.1 °C.Moreover, we found that tropical forests with highergrowth temperatures tend to have higher ToptE,suggesting the acclimation potential for many tropicalforests. In contrast to previous studies, however, ourresults show that biochemical processes make only aminor contribution to the ToptE changes across sites.Instead, respiratory processes, which are generallynegligible at the leaf level, play an important role inexplaining ToptE variation across sites. Consistent with

leaf level studies, stomatal processes are also critical indetermining ToptE at the ecosystem level. StrongD andstomatal limitation on ToptE suggests that increasingCO2 concentrations may increase the ToptE of tropicalforests.

Appendix

Acknowledgments

The study was supported by the C-project of Talent,Hainan University and National Natural ScienceFoundation of China (31660142). Brazil flux datawas produced with funds from the NASA LBA-DMIPproject (# NNX09AL52G) (NASA), NASA LBAinvestigation CD-32, and the National ScienceFoundation’s Partnerships for International Researchand Education (PIRE). We thank two reviewers and aneditor board member for their insightful andconstructive comments on this work. The AsiaFLUXwas acknowledged for permission on accessing andusing their dataset.

References

Anderegg W R L et al 2015 Tropical nighttime warming as adominant driver of variability in the terrestrial carbon sinkProc. Natl Acad. Sci. 112 15591–6

Baldocchi D D 2003 Assessing the eddy covariance technique forevaluating carbon dioxide exchange rates of ecosystems:past, present and future Glob. Change Biol. 9 479–92

Table A1. Equations used for the leaf photosynthesis biochemical model (FvCB).

Number Equation

1 Pn ¼ minfPc;P j;Psg2 Pc ¼ ðc i�G� ÞV cmax

c iþK cð1þOi=KoÞ � Rd

3 Pj ¼ ðc i�G� ÞJ4c iþ8G� � Rd

4 Ps ¼ 0:5V cmax � Rd

5 0:7J2 � JðIabs þ JmaxÞ þ IabsJmaxgb ¼ 0

6 Iabs ¼ I�f abs ð1� f Þ=27 ci ¼ ca � Pn

1gbþ 1

g s

� �

8 gb ¼ 0:147�ffiffiffiffiffiffiffiffiffiffiffiffiffi

W s

0:752�Lw

q9 g s ¼ g1hs

csPn þ g0

100:98P2

v � ðPc þ PjÞPv þ PcP j ¼ 0

0:98P2n � ðPv þ PsÞPn þ PvPs ¼ 0

11 fK c;Ko;Rd;V cmax;G� g ¼ fK c;K o;Rd;Vmax;G� g25eEa

TK�298298�R�TK

� �

12 fJmaxg ¼ fJmaxg25eEa

TK�298298�R�TK

� �� 1þ eð298S�HÞ=ð298RÞ

1þ eðTKS�HÞ=ðTKRÞ

Pn: net photosynthetic rate, Pc: Rubisco-limited photosynthesis, Pj: electron transport-limited photosynthesis, Ps: export-limited

photosynthesis, Rd: dark respiration rate, ci: CO2 partial pressure at the carboxylating site, Oi: O2 partial pressure at the carboxylating

site, Kc: Michaelis–Menten constant of Rubisco for CO2, Ko: Michaelis–Menten constant of Rubisco for O2, Vcmax: maximum Rubisco

activity, G�: CO2 compensation point in the absence of Rd, J: electron transport rate, Jmax: maximum electron transport rate, Iabs:

absorbed light, fabs: leaf absorbance (∼0.85), I: light intensity, f: correction factor for the spectral quality of light (∼0.15), gs: stomatal

conductance, ca: ambient CO2 concentration, cs: leaf surface CO2 concentration, gb: laminar boundary layer conductance, Ws: wind

speed, Lw: leaf width, hs: relative humidity, g0 and g1 are two model parameters, Pv is used for smoothing the transition between Pc,

Pj, and Ps, Ea: activation energy, TK: temperature in degrees Kelvin, R: gas constant, S: term similar to an entropy factor, H: describes

the rate of decrease in the function above the optimum.

Environ. Res. Lett. 12 (2017) 054022

10

Ball J T, Woodrow I E and Berry J A 1987 A model predictingstomatal conductance and its contribution to the controlof photosynthesis under different environmental conditionsProgress in Photosynthesis Research ed I Biggins(Dordrecht: Martinus-Nijhoff)

Berry J and Björkman O 1980 Photosynthetic response andadaptation to temperature in higher plants Annu. Rev.Plant Physiol. 31 491–543

Campbell G S and Norman J M 1998 An Introduction toEnvironmental Biophysics (New York: Springer)

Carswell F E, Meir P, Wandelli E V, Bonates L C M, Kruijt B,Barbosa E M, Nobre A D, Grace J and Jarvis P G 2000Photosynthetic capacity in a central Amazonian rain forestTree Physiol. 20 179–86

Cavaleri M A, Reed S C, Smith W K and Wood T E 2015Urgent need for warming experiments in tropical forestsGlob. Change Biol. 21 2111–21

Corlett R T 2011 Impacts of warming on tropical lowlandrainforests Trends Ecol. Evol. 26 606–13

Damour G, Simonneau T, Cochard H and Urban L 2010 Anoverview of stomatal conductance at the leaf level PlantCell Environ. 33 1419–38

de Pury D G and Farquhar G D 1997 Simple scaling ofphotosynthesis from leaves to canopies without the errorsof big-leaf models Plant Cell Environ. 20 537–57

Dixon R K, Brown S, Houghton R A, Solomon A M, Trexler MC and Wisniewski J 1994 Carbon pools and flux of globalforest ecosystems Science 263 185–90

Doughty C E 2011 An in situ leaf and branch warmingexperiment in the Amazon Biotropica 43 658–65

Doughty C E and Goulden M L 2008 Are tropical forests near ahigh temperature threshold? J. Geophys. Res. Biogeosci.113 G00B07

Duursma R A, Barton C V M, Lin Y S, Medlyn B E, Eamus D,Tissue D T, Ellsworth D S and McMurtrie R E 2014 Thepeaked response of transpiration rate to vapour pressuredeficit in field conditions can be explained by thetemperature optimum of photosynthesis Agric. Meteorol.189–190 2–10

Farquhar G D, von Caemmerer S and Berry J 1980 Abiochemical model of photosynthetic CO2 assimilation inleaves of C3 species Planta 149 78–90

Gamo M et al 2005 Carbon flux observation in the tropicalseasonal forests and tropical rain forest. Proc. Int.Workshop on Advanced Flux Network and Flux Evaluation(AsiaFlux Workshop 2005) (Fujiyoshida, Japan)

Goulden M L, Miller S D, da Rocha H R, Menton M C, deFreitas H C, Figureira A M E S and de Sousa C A D 2004Diel and seasonal patterns of tropical forest CO2 exchangeEcol. Appl. 14 s42–54

Hikosaka K, Ishikawa K, Borjigidai A, Muller O and Onoda Y2006 Temperature acclimation of photosynthesis:mechanisms involved in the changes in temperaturedependence of photosynthetic rate J. Exp. Bot. 57 291–302

Hirano T, Seagh H, Harada T, Limin S, June T, Hirata R andOsaki M 2007 Carbon dioxide balance of a tropical peatswamp forest in Kalimantan, Indonesia Glob. Change Biol.13 412–25

Hirata R, Saigusa N, Yamamoto S, Ohtani Y, Ide R, Asanuma J,Gamo M, Hirano T, Kondo H and Kosugi Y 2008 Spatialdistribution of carbon balance in forest ecosystems acrossEast Asia Agric. Meteorol. 148 761–75

Ishida A, Toma T, Matsumoto Y, Yap S K and Maruyama Y 1996Diurnal changes in leaf gas exchange characteristics in theuppermost canopy of a rain forest tree, Dryobalanopsaromatic Gaertn. f Tree Physiol. 16 779–85

Jaramillo C et al 2010 Effects of rapid global warming at thePaleocene-Eocene boundary on neotropical vegetationScience 330 957–61

June T, Evans J R and Farquhar G D 2004 A simple newequation for the reversible temperature dependence ofphotosynthetic electron transport: a study on soybean leafFunct. Plant Biol. 31 275–83

Koch G W, Amthor J S and Goulden M L 1994 Diurnal patternsof leaf photosynthesis, conductance and water potential atthe top of a lowland rain forest canopy in Cameroon:measurements from the Radeau des Cimes Tree Physiol. 14347–60

Kositsup B, Montpied P, Kasemsap P, Thaler P, Ameglio T andDreyer E 2009 Photosynthetic capacity and temperatureresponse of photosynthesis of rubber trees (Heveabrasiliensis Müll. Arg.) acclimate to changes in ambienttemperatures Trees 23 357–65

Lange O L, Schulze E-D, Evenari M, Kappen L and Buschbom U1974 The temperature-related photosynthetic capacity ofplants under desert conditions. I. Seasonal changes of thephotosynthetic response to temperature Oecologia 1797–110

Lasslop G, Reichstein M, Papale D, Arneth A D, Barr A, Stoy Aand Wohlfahrt P 2010 Separation of net ecosystemexchange into assimilation and respiration using a lightresponse curve approach: critical issues and globalevaluation Glob. Change Biol. 16 187–208

Lenorman M, Jabot F and Deffuant G 2013 Adaptiveapproximate Bayesian computation for complex modelsComput. Stat. 28 2777–96

Leuning R 1990 Modelling stomatal behavior and photosynthesisof Eucalyptus grandis Aust. J. Plant Physiol. 17 339–55

Lloyd J and Farquhar G D 2008 Effect of rising temperatures and[CO2] on the physiology of tropical forest trees Phil.Trans. R. Soc. B 363 1811–17

Lin Y S, Medlyn B E and Ellsworth D S 2012 Temperatureresponse of leaf net photosynthesis: the role of componentprocesses Tree Physiol. 32 219–31

Lombardozzi D L, Bonan G B, Smith N G, Dukes J S and FisherR A 2015 Temperature acclimation of photosynthesis andrespiration: a key uncertainty in the carbon cycle-climatefeedback Geophys. Res. Lett. 42 8624–31

Medlyn B E, Duursma R A, Eamus D, Ellsworth D S, Prentice CV M, Barton K Y, Crous P, Angelis M, Freeman M andWingate L 2011 Reconciling the optimal and empiricalapproaches to modeling stomatal conductance Glob.Change Biol. 17 2134–44

Niu S et al 2012 Thermal optimality of net ecosystem exchangeof carbon dioxide and underlying mechanisms New Phytol.194 775–83

Restrepo-Coupe N et al 2013 What drives the seasonality ofphotosynthesis across the Amazon basin? A cross-siteanalysis of eddy flux tower measurements from the Brasilflux network Agric. Meteorol. 182 128–44

Rowland L et al 2015 Modelling climate change responses intropical forests: similar productivity estimates across fivemodels, but different mechanisms and responses Geosci.Model Dev. 8 1097–110

Sage R F and Kubien D S 2007 The temperature response of C3

and C4 photosynthesis Plant Cell Environ. 30 1086–106Slot M and Kitajima K 2015 General patterns of acclimation of

leaf respiration to elevated temperatures across biomes andplant types Oecologia 177 885–900

Slot M and Winter K 2017 In situ temperature response ofphotosynthesis of 42 tree and liana species in the canopyof two Panamanian lowland tropical forests withcontrasting rainfall regimes New Phytol. 214 1103–17

Tjoelker M G, Oleksyn J, Lorenc-Plucinska G and Reich P B2009 Acclimation of respiratory temperature response innorthern and southern populations of Pinus banksianaNew Phytol. 181 218–29

Vargas G G and Cordero S R A 2013 Photosynthetic responsesto temperature of two tropical rainforest tree species fromCosta Rica Trees 27 1261–70

Vårhammar A, Wallin G, McLean C M, Dusenge M E, Medlyn BE, Hasper T B, Dsabimana D and Uddling J 2015Photosynthetic temperature response of tree species inRwanda: evidence of pronounced negative effects of hightemperature in montane rainforest climax species NewPhytol. 206 1000–12

Environ. Res. Lett. 12 (2017) 054022

11

von Caemmerer S, Farquhar G and Berry J 2009 Biochemical modelof C3 photosynthesis Photosynthesis in Silico: UnderstandingComplexity from Molecules to Ecosystems ed A Laisk, L Nedbaland Govindjee (New York: Springer) pp 209–30

Wang Y P, Baldocchi D, Leuning R, Falge E and Vesala T 2007Estimating parameters in land-surface model by applyingnonlinear inversion to eddy covariance flux measurementsfrom eight Fluxnet sites Glob. Change Biol. 13 652–70

Wright S J, Muller-Landau H C and Schipper A J 2009 The future oftropical species on a warmer planet Conserv. Biol. 23 1418–26

Yi C, Davis K J, Bakwin P S, Berger B W and Marr L C 2000Influence of advection on measurements of the netecosystem-atmosphere exchange of CO2 from a very talltower J. Geophys. Res. Atmos. 105 9991–9

Yi C et al 2004 A nonparametric method for separatingphotosynthesis and respiration components in CO2 fluxmeasurements Geophys. Res. Lett. 31 L17107

Yi C et al 2010 Climate control of terrestrial carbon exchangeacross biomes and continents Environ. Res. Lett. 5034007

Zeng J, Tan Z H and Saigusa N 2017 Using approximateBayesian computation to infer photosynthesis modelparameters Chinese J. Plant Ecol. 41 378–85

Zhang Y J, Cristiano P M, Zhang Y F, Campanello P I, Tan ZH, Zhang Y P, Cao K and Goldstein G 2016 Tropical treephysiology Carbon Economy of Subtropical Forests ed GGoldstein and L S Santiago (Cham: Springer)pp 337–55

Environ. Res. Lett. 12 (2017) 054022

12