Climate change imposes phenological trade-offson forest net primary productivityMatthew J. Duveneck1 and Jonathan R. Thompson1

1Harvard Forest, Harvard University, Petersham, Massachusetts, USA

Abstract Climate warming is expected to lengthen growing seasons of temperate forest ecosystems andincrease gross primary productivity. Simultaneously, warming is expected to increase summer ecosystemrespiration, which could offset gains accrued from longer growing seasons. These responses have beenobserved during anomalously warm years, but the role of future climate change on phenologicaltrade-offs and how they affect net primary productivity (NPP) at regional scales in temperate forestsremain unexplored. We simulated scenarios of climate change on monthly forest NPP throughout 18million hectares of temperate forests in New England, USA, through year 2100. Using an ecophysiologicalmodel coupled to a forest landscape model, we simulated scenarios of climate change on monthly NPP. Ahigh emission scenario (RCP 8.5), resulted in longer growing seasons that offset midsummer ecosystemrespiration costs and produced greater annual NPP throughout the study landscape compared tosimulations using the current climate. In spring and autumn months, temperature was positivelyassociated with greater NPP; in summer months, the relationship was negative. Spatially, the greatestincrease in NPP occurred in the warmer southern region under a warm climate scenario with increasedprecipitation. Under a warm scenario with drier conditions, the greatest increase in NPP occurred in thecooler northern region. Phenological trade-offs will affect NPP of future forests and their potential to serveas a negative feedback to climate change. Barring other limitations, longer growing seasons will offsetgreater respiratory demands and contribute to increases in NPP throughout the temperate forests of NewEngland in the future.

Plain Language Summary Climate change is expected to make growing seasons longer butwarmer. This may increase forest growth in the spring and autumn but decrease growth in the summer.We simulated future forest growth in New England over the next 90 years. Climate change produced earlierand longer growing seasons that offset declines in growth due to warmer summers. In spring and autumn,temperature was positively associated with greater growth, but in the summer, the relationship was negative.Seasonal trade-offs will affect growth of future forests in New England and their potential to serve as anegative feedback to climate change.

1. Introduction

Climate warming is expected to increase both gross primary productivity (GPP) and ecosystem respiration intemperate forests [Hollinger et al., 2004; Richardson et al., 2010; Keenan et al., 2014]. Increases in GPP will resultfrom an earlier start to the growing season in the spring [Hollinger et al., 2004; Schwartz et al., 2006; Richardsonet al., 2010] and a later end to the growing season in autumn [Dragoni et al., 2011; Jeong et al., 2011; Xie et al.,2015]. Warmer summers, however, will impose greater midsummer respiratory demands that could offsetproductivity gains [Cook and Johnson 1989; Sage et al., 2008; Buermann et al., 2013]. So while a gross increasein productivity is expected in many parts of the globe, a decline in net primary productivity (NPP) will resultwhenever ecosystem respiration exceeds production under climate change [Cramer et al., 2001; Dymondet al., 2016]. As temperature increases, ecosystem respiration is expected to increase before acclimating tohigher temperatures [Atkin et al., 2005; Wythers et al., 2013; Reich et al., 2016]. Greater productivity (i.e., lesscarbon released and more carbon entering the forest) under climate change will also depend on sufficientwater, nutrients, and photosynthetic active radiation (PAR) [Gauthier et al., 2015]. Lower productivity (i.e.,more carbon released and less carbon entering the forest) will result where and when these resources areotherwise limited.

The net effect of climate change on carbon flux has been inferred from long-term eddy covariance instru-ments, which allow precise measurements of carbon exchange between forest and atmosphere. Within

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 1

PUBLICATIONSJournal of Geophysical Research: Biogeosciences

RESEARCH ARTICLE10.1002/2017JG004025

Key Points:• Longer growing seasons offset greaterrespiratory demands to productivityof future forests of the northeasternU.S. under climate change

• Simulated increases in temperaturewere associated with greaterproductivity in spring and autumn butlower productivity in summer

• The strength of the balance betweenproductivity and respiration variesspatially

Correspondence to:M. J. Duveneck,mduveneck@gmail.com

Citation:Duveneck, M. J., and J. R. Thompson(2017), Climate change imposesphenological trade-offs on forest netprimary productivity, J. Geophys. Res.Biogeosci., 122, doi:10.1002/2017JG004025.

Received 6 JUL 2017Accepted 17 JUL 2017Accepted article online 16 AUG 2017

©2017. American Geophysical Union.All Rights Reserved.

observed anomalous years (i.e., years with greater than average growing season length) at 21 NorthernHemisphere FLUXNET eddy covariance sites in boreal and temperate forests, greater GPP during springand smaller increases in ecosystem respiration during summer resulted in a net increase in forestproductivity [Richardson et al., 2010]. Likewise, warmer than average conditions were positively corre-lated with growth observed in black spruce tree rings in Quebec, Canada [D’Orangeville et al., 2016].Wolf et al. [2016] used 22 FLUXNET eddy covariance sites coupled with satellite remote sensing to deter-mine that an earlier spring offset and exceeded the carbon losses occurring in the 2012 continentalscale drought. These results imply that increased forest productivity could partially mitigate increasingcarbon dioxide in the atmosphere. However, findings based on historically observed climate variationare not necessarily representative of future temperature increases, which are expected to far exceedclimatic precedent [Intergovernmental Panel on Climate Change (IPCC), 2013]. Furthermore, rising CO2

concentrations in the atmosphere may have a fertilizing effect. As water use efficiency increases,stomatal conductance may be reduced, resulting in greater forest carbon gain that would otherwisebe lost to respiration [Keenan et al., 2011, 2013; De Kauwe et al., 2013; Franks et al., 2013]. However,greater carbon dioxide, longer growing seasons, and enhanced forest growth may reduce nutrientavailability (e.g., nitrogen) which could result in a reduction in net productivity [Elmore et al., 2016]. Insummary, how future climate will affect carbon uptake driven by the phenological trade-offs of thecombination of longer growing seasons and increased summer ecosystem respiration within temperateforests remains unknown.

The trade-off between longer growing seasons and increased summer ecosystem respiration will varyspatially and temporally and determine the direction and magnitude of carbon flux at regional scales.Indeed, several approaches to modeling climate change impacts on carbon flux have shown widevariation within the last century [Gauthier et al., 2015], across forest types [Duveneck et al., 2016;Dymond et al., 2016], latitude [Cramer et al., 2001], and aspect [Walker et al., 2015]. For example, inthe northeastern United States, NPP has been modeled and shown to increase in northern hardwoodforests over the next 30 years of moderate climate change. However, NPP in spruce-fir forests declinedafter 80 years of more intense climate change [Ollinger et al., 2008]. And Xie et al. [2015] found thatautumn senescence, directly effecting growing season length, was related to nonlinear interactionsbetween cold, frost, moisture, and heat stress. Under future climate projections, they predicted delayedsenescence of autumn foliage under moderate temperature as found in northern New England andearlier senescence of autumn foliage under higher heat stress as found in southern New England [Xieet al., 2015].

As climate change is expected to greatly exceed observed climate anomalies [Fan et al., 2014; Ning et al.,2015], it is important to explore how phenological trade-offs of NPP will develop over the next century.When studying the adaptation of forests to novel conditions (e.g., climate change), it is critical to incor-porate first principles of photosynthesis when mechanistic relationships derived from past conditionsmay exceed future conditions [Gustafson, 2013]. For example, temperature acclimation to respiration rela-tionships has been identified as an important process to include when simulating climate feedbacks tovegetation [Smith and Dukes, 2013]. New England’s diverse forest types, climate gradients, and land usehistory provide large spatial variability to observe simulated climate change affects.

Continued forest recovery from colonial era land use disturbances, are expected to drive large increases incarbon storage over the next century in New England [Duveneck et al., 2016]. However, the effects of cli-mate change on forest productivity are not uniform across the region. The seasonal phenology affectingthe spatial variation in net productivity under future climate change scenarios have been largely unex-plored. Using a forest landscape model [Scheller et al., 2007] coupled to a physiologically based ecosystemmodel [de Bruijn et al., 2014], we simulated the spatial variation in monthly net ecosystem productivityfrom 2010 to 2100 under four climate change scenarios. The parameterization of the model used in thisanalysis was described previously in Duveneck et al. [2016].

Projections of forests’ physiological response to future climate scenarios can improve our understanding ofthe phenological trade-offs associated with longer growing seasons and increased ecosystem respirationcosts to GPP. Our objective here was to quantify the variation in this trade-off among New England’s foresttypes and climate regions in terms of the changing NPP due to climate change.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 2

2. Methods2.1. Landscape Description

We simulated climate-induced trade-offs in productivity within the 18 million hectares of forest in NewEngland, a six-state region of the northeastern U.S. spanning 41° to 47° latitude (Figure 1). Mean annualtemperatures range from 3 to 10°C andmean annual precipitation ranges from 790 to 2550 mmwith warmertemperatures in the south and more precipitation at higher elevations [Daly and Gibson, 2002]. This regionincludes diverse forest types spanning a gradient of temperate oak-maple forests in the south to borealspruce and fir forests in the north [Duveneck et al., 2015]. Like much of the eastern U.S., New England forestsare in a protracted recovery from widespread agricultural and timber harvesting land use associated withEuropean colonization during the eighteenth and nineteenth centuries [Thompson et al., 2013]. While thereis considerable interannual variation, eddy flux measurements over the past two decades indicate that

Figure 1. Climate regions (n = 25) within New England delineated with an isocluster analysis to identify distinct regions ofmonthly temperature and precipitation. In general, lower numbered climate regions correspond to cooler averagetemperature. Stars denote locations of calibration sites at Harvard and Howland Forests. Inset shows study area (red) withinNorth American forests (green).

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 3

NPP is increasing throughout the region, including in spruce-fir forests in Maine [Hollinger et al., 2004],shallow soil northern hardwood forests in New Hampshire [Battles et al., 2014], hemlock forests inMassachusetts [Hadley and Schedlbauer, 2002], and northern hardwood forests in Massachusetts [Urbanskiet al., 2007]. Observed productivity increases in this region are largely attributed to continual aggradationof live biomass [Ollinger et al., 2002; Duveneck et al., 2016] and a period containing a high proportion ofwet years with above average moisture availability [Pederson et al., 2015].

2.2. Landscape Modeling Framework

We simulated 90 years of forest dynamics (2010–2100) at 5 year time steps across>2 million forested cells at250 m resolution. We used the LANDIS-II v6.1 spatially explicit forest landscape modeling framework [Schelleret al., 2007]. The LANDIS-II core model tracks tree species-age cohorts within raster spatial layer cells andsimulates seed dispersal frommature cohorts to surrounding cells with a declining exponential function withdistance from the source cell [Scheller et al., 2007]. To simulate forest ecosystem processes, including photo-synthesis and competition, we used the PnET-Succession v1.0 extension [Gustafson et al., 2014; de Bruijn et al.,2014] to LANDIS-II. PnET-Succession couples algorithms from the PnET-II ecophysiological model [Aber et al.,1995] with LANDIS-II to simulate forest processes including establishment, growth, and mortality of eachspecies-cohort at monthly time steps. LANDIS-II initialization is achieved by growing the current species-cohorts on the landscape from the establishment year of each cohort up to the starting year (2010). ThePnET models have demonstrated high reliability and accurate predictions of current and historic NPP withinforests in the northeastern USA [Aber and Federer, 1992; Aber et al., 1996; Ollinger et al., 1998, 2002; Ollingerand Smith, 2005] and within other regions and biomes [e.g., Pan et al., 2006, Thorn et al., 2015].

Climate variation is incorporated into PnET-Succession using monthly minimum andmaximum temperaturesand precipitation; PAR and carbon dioxide are also input at monthly time steps. When monthly temperaturesexceed species-specific minimum temperatures, photosynthesis begins. Photosynthesis peaks at species-specific optimal temperatures (Table 1) and declines at higher temperatures as conductance decreasesand vapor pressure deficit increases [Gustafson et al., 2017] allowing for a phenological response to changingmonthly temperature. Water is allocated to growth after accounting for precipitation, transpiration, evapora-tion, and runoff. Water in PnET-Succession is stored and released through spatially distinct soil types, knownto be important for accurately predicting NPP [e.g., Pan et al., 2006]. Additionally, photosynthesis is affectedby available light, foliar nitrogen content, and atmospheric carbon dioxide concentration. Increasing atmo-spheric carbon dioxide increases the photosynthetic rate by increasing water use efficiency. A doubling ofCO2 will result in nearly a 2 times increase in water use efficiency [see Aber et al., 1995]. A decline in growthoccurs as cohorts age and as temperatures depart from species-specific optimal temperatures. Respiration inPnET-Succession increases with increased temperature and is modified to account for respiration acclimationto increased temperature based on algorithms described in Wythers et al. [2013].

PnET-Succession simulates competition for light and water for photosynthesis by integrating PAR and wateracross multiple layers of the canopy. PnET-Succession partitions net photosynthesis to biomass pools ofwood, root, foliage, and nonstructural carbon reserves. Species establishment is calculated based on seedavailability, soil water, and subcanopy light to stochastically simulate regeneration of new cohorts. As treespecies have relatively long life spans, forest composition shifts are slow and depend on adequate seeddispersal, establishment, competition, and mortality.

Previous research with LANDIS-II coupled to PnET-Succession [Gustafson et al., 2017] identified sensitivities ofannual NPP, biomass, and nonstructural carbon reserves (NSC) to precipitation (�40%, 0%, and +40%),temperature (+0°C, +3°C, and +6°C), PAR (�10%, 0%, and +10%), and soil water capacity (60.5 mm,107.5 mm, and 150.8 mm). Within experimental sites in Wisconsin, USA, Gustafson et al. [2017] found that pre-cipitation had the largest effect to response variables with increases in NPP, biomass, and NSC with increasedprecipitation. Temperature had a varied effect: annual NPP increased at +3°C, but declined at +6°C; biomassand NSC decreased at +3°C and at +6°C. PAR and soil water capacity had a positive effect to all responsevariables. Additional results including landscape scale response and the interactions between explanatoryvariables were also explored with LANDIS-II coupled to PnET-Succession [Gustafson et al., 2017]. It is impor-tant to note that our New England landscape has greater moisture available (i.e., 790 to 2550 mm annual)compared to the baseline climate simulated in the Wisconsin experiments (i.e., 521 to 967 mm annual)[Gustafson et al., 2017].

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 4

2.3. Climate Scenarios

We simulated a continuation of recent climate conditions as our “current” climate scenario and used this as abaseline for comparison to four different climate change scenarios. Our current climate scenario utilizedhistoric climate data from 1981 to 2015 provided by the Parameter-elevation Relationships onIndependent Slopes Model (PRISM) [Daly and Gibson, 2002]. For each simulation year spanning 2010 to2100, we imputed a randomly selected year from the historical time series to populate the current climatescenario. Therefore, the current climate scenario results characterize future successional dynamics withclimate represented by contemporary observations.

We simulated climate change forced by a high-emission representative concentration pathway (RCP 8.5) pro-vided by the International Panel on Climate Change (IPCC). We selected the single RCP 8.5 emission futurebecause we wanted to focus on a departure from current climate. This emission scenario assumes a futurewith high population growth and nonclimate policy scenarios [Riahi et al., 2011]. This RCP represents thehighest emissions scenario published by the IPCC; however, current realized carbon emissions alreadyexceed the RCP 8.5 trajectory [Peters et al., 2012].

Using the RCP 8.5 emission future, we simulated climate change using four independent global circulationmodels (GCMs): (1) the Community Climate System Model v4.0 (CCSM4), (2) Community Earth SystemModel v1-Community Atmosphere Model v5 (CESM1), (3) Hadley Global Environment Model v2-EarthSystem (HADGE), and (4) the Max Planck Institute Earth System Model-Low Resolution (MPIMLR). TheseGCMs have performed well against historical climate in the northeastern U.S. [Sillmann et al., 2013; Ninget al., 2015]. For each GCM used, we accessed 12 km monthly downscaled projections of maximum

Table 1. Tree Species and Parameters Simulated Within Species Group Forest Typesa

Genus Species Forest Type FOL_N SLW_MAX HALF_SAT PSN_MIN PSN_OPT

Betula lenta aspen-birch 2.26 80 519 3 21Betula papyrifera aspen-birch 2.3 80 519 3 21Betula populifolia aspen-birch 2.26 80 519 3 21Populus balsamifera aspen-birch 2.4 115 600 0 21Populus grandidentata aspen-birch 2.5 115 600 2 22Populus tremuloides aspen-birch 2.5 115 600 2 22Larix laricina hemlock-tamarack-cedar 2.3 110 600 1 20Thuja occidentalis hemlock-tamarack-cedar 1.3 245 437 3 20Tsuga canadensis hemlock-tamarack-cedar 1.1 195 230 3 21Acer rubrum northern-hardwood 2.6 80 437 4 26Acer saccharum northern-hardwood 2.1 70 275 2 23Betula alleghaniensis northern-hardwood 2.2 80 356 3 21Carya glabra northern-hardwood 2.6 90 519 2 23Fagus grandifolia northern-hardwood 1.8 80 275 2 23Fraxinus americana northern-hardwood 2.5 85 437 3 25Fraxinus nigra northern-hardwood 2.7 85 519 3 23Ostrya virginiana northern-hardwood 2.02 100 250 3 23Prunus serotina northern-hardwood 2.8 90 519 3 25Tilia americana northern-hardwood 2.6 75 356 3 23Ulmus americana northern-hardwood 2.3 85 437 3 23Quercus alba oak 2.5 100 519 2 26Quercus coccinea oak 2 95 519 2 26Quercus prinus oak 2.39 90 437 2 26Quercus rubra oak 2.5 85 437 2 24Quercus velutina oak 2.7 85 437 2 24Pinus resinosa pine 1.7 250 519 3 21Pinus rigida pine 2.3 250 437 3 21Pinus strobus pine 2 235 437 3 21Abies balsamea spruce-fir 1.4 215 356 2 19Picea glauca spruce-fir 1.4 250 437 2 20Picea mariana spruce-fir 1.2 250 437 2 20Picea rubens spruce-fir 1.2 250 437 2 20

aFOL_N = Percent foliar nitrogen content. SLW_MAX = Maximum specific leaf weight (g m�2). HALF_SAT = Half saturation light level for photosynthesis (umolm�2 s�1). PSN_MIN = Minimum temperature for photosynthesis (°C). PSN_OPT = Optimal temperature for photosynthesis (°C).

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 5

temperature, minimum temperature, and precipitation obtained from the USGS Geo Data Portal [Stoner et al.,2013]. Within the climate change scenarios, we used the projected increase in carbon dioxide concentrationprovided by the RCP 8.5 emission future [IPCC, 2013]. Within the current climate scenario, we held carbondioxide constant at 390 ppm. LANDIS-II requires users to delineate zones of homogeneous climate. We aggre-gated climate input data within 25 delineated “climate regions” (Figure 1). Climate regions were delineatedusing an isocluster analysis designed to optimally parse distinct regions based on within-region similarity ofminimum and maximum monthly temperature and precipitation from historical observations (PRISM) [Dalyand Gibson, 2002] (Figure 1).

2.4. Model Evaluation and Experimental Design

The initial distribution of tree species-age cohorts, including 32 species (Table 1), was built based on an impu-tation of tree inventory plots at 250 m resolution [Duveneck et al., 2015]. We calibrated simulated annual bio-mass of single sites representing observed biomass at six sites throughout New England over multipledecades. Specifically, we compared simulated AGB at 1 ha sites to long-term observations of biomass mea-sured over multiple decades at the Hemlock tower [Hadley and Schedlbauer, 2002], EMS tower [Urbanskiet al., 2007], and Lyford plots [Eisen and Plotkin, 2015] at the Harvard Forest in Massachusetts; and theHowland Forest in Maine [Hollinger et al., 2004], Black Rock Forest in New York [Schuster et al., 2008], andHubbard Brook Forest in New Hampshire [Battles et al., 2014]. At the landscape scale, we validated simulatedaboveground biomass (AGB) at 2010 to the estimated biomass [Bechtold and Patterson, 2005] reported fromimputed forest inventory plots within 181 distinct “ecoregions” where relatively homogeneous climate andsoils were present. Across ecoregions we found a root-mean-square error (RMSE) of 1.76 kg m�2 andPearson’s correlation coefficient (r) of 0.82. We validated simulated monthly net ecosystem exchange (NEE)at a spruce forest (Howland Forest), a hemlock forest (Harvard Forest), and a deciduous forest (HarvardForest) with eddy covariance flux instruments (Figure 1). Details of the net ecosystem exchange validationwas described previously [Duveneck et al., 2016]. Validation at flux towers showed good agreement in the tim-ing and magnitude of monthly NEE. Specifically, validation of monthly NEE report RMSE of 40.88 g m�2,54.92 g m�2, and 71.07 g m�2, and r of 0.57, 0.63, and 0.66 for the spruce, hemlock, and deciduous forests,respectively [Duveneck et al., 2016].

We simulated low-frequency gap-scale disturbances as described inDuveneck et al. [2016] but excluded othernatural and anthropogenic disturbance regimes in order to isolate climate change effects. We replicated eachscenario 5 times; however, due to the low variation among landscape-scale response variables amongreplicates, we selected a single replicate from each scenario for further analysis. We evaluated monthlyGPP, ecosystem respiration, and NPP across 10 year time steps and all climate scenarios. In addition, weaggregated monthly NPP within climate regions and forest types based on species groups (Table 1).Within growing season months, we assessed the relationship between NPP and maximum temperature,minimum temperature, and precipitation across climate scenarios and climate regions.

3. Results3.1. Monthly Gross Primary Productivity, Ecosystem Respiration, and Net Primary Productivity

At the full regional scale, our simulations indicate that increases in GPP due to climate change exceed lossesincurred due to greater respiration, resulting in higher annual NPP. Compared to the current climate scenario,climate change scenarios in 2100 included greater maximum temperature (Figure 2a), and varied monthlyprecipitation (Figure 2b). In turn, an increase in simulated ecosystem respiration was observed in all monthsof the year with the greatest increase in ecosystem respiration occurring in April, May, and June (Figure 2d).Compared to the current climate scenario, climate change at year 2100 resulted in greater simulated GPP(Figure 2c) and NPP (Figure 2e) in the spring and fall months. In summer months, however, climate changescenarios resulted in less GPP and less NPP compared to the current climate scenario. Spatially, monthlyNPP varied across New England and across all climate scenarios with the south resulting in longer growingseasons with an earlier increase and later decrease in NPP (Figure 3).

3.2. Annual Net Primary Productivity

The trade-off in monthly NPP resulted in greater annual NPP under climate change. Annual NPP varied spa-tially; however, a majority of sites (between 74% and 98%, depending on GCM) experienced greater annual

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 6

Figure 2. Differences in year 2100 between climate change models and current climate in terms of (a) maximum monthlytemperature, (b) precipitation, (c) gross primary productivity (GPP), (d) ecosystem respiration, and (e) net primaryproductivity (NPP). Monthly values were averaged from all sites throughout New England.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 7

NPP under climate change (Figure 4). Within New England, annual NPP under the current climate scenarioaveraged 383.5 g m�2 yr�1 compared to 491.5 g m�2 yr�1 averaged over all four GCMs under climatechange at year 2100. Most of the increase in NPP attributable to climate change came during the springmonths of March, April, and May (between 81% and 92%, depending on GCM); a smaller increase wasobserved in the autumn months of September, October, and November (Figure 2e). The increase in NPPunder climate change during spring months resulted from both an earlier start and a greater rate ofproductivity. Early in the simulation decreased summer NPP under climate change exceeded the enhancedspring and autumn NPP resulting in less annual NPP under climate change. For example, at simulationyear 2030, climate change scenarios averaged 310.2 g m�2 yr�1 compared to 346.8 g m�2 yr�1 under thecurrent climate scenario. By simulation year 2060, however, the magnitude of growth enhancement inspring and autumn offset the summer decrease under climate change.

3.3. Spatial Variation in Monthly Net Primary Productivity

Overall, the maximum monthly rate of NPP was greater under climate change compared to current climate.Larger maximum monthly NPP rates were observed throughout New England under all climate scenarios(Figure 5). The effect of climate change was more pronounced in the north where a greater differencebetween current climate and climate change was observed. In northern regions, difference (i.e., increase)in maximum monthly NPP between climate change and current climate averaged 48.3 g m�2 month�1,whereas central and southern region differences averaged 46.0 g m�2 month�1, 41.0 g m�2 month�1,respectively. In northern New England, June experienced the maximummonthly NPP when using the currentclimate scenario; under climate change, maximum monthly NPP shifted 1 month earlier to May. In southernNew England, maximum NPP rates occurred in May, under the current climate and climate change scenarios,though climate change scenarios sustained high NPP in both April and May.

The spring and autumn increase in GPP due to climate change offset increased respiratory demands in thesummer within every forest type (Figure 6). The pine forest type resulted in the largest maximum monthlyNPP which averaged 195.9 g m�2 month�1 across climate change GCMs compared to 148.3 g m�2 month�1

within the current climate scenario at year 2100. The spruce-fir forest type had the lowest maximummonthly

Figure 3. Monthly net primary productivity (g m�2 month�1) at year 2100 under current climate and four climate change models.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 8

NPP, which averaged 104.2 gm�2 month�1 across climate change scenarios and 77.5 g m�2 month�1 withinthe current climate scenario at year 2100. Although the pine forest type experienced the largest overall NPPrate under climate change, the 32.1% increase compared to current climate was the lowest increase of anyforest type. In contrast, the northern hardwoods forest type resulted in a 52.3% increase (the largestincrease in maximum NPP of any forest type), which at year 2100 averaged 154.4 g m�2 month�1 acrossclimate change scenarios compared to 104.2 g m�2 month�1 within the current climate scenario.

Under current climate, May had the highest NPP within northern hardwoods, aspen-birch, pine, and oakforest types. In the north, where more boreal species dominate, the largest NPP under current climate inspruce-fir, and hemlock-tamarack-cedar forest types occurred in June. Under climate change, the largestNPP occurred in May under all forest types at year 2100. The climate change summer decline in NPP at year2100 occurred in all forest types (Figure 6). The maximum difference in NPP of any month when currentclimate exceeded climate change (average of GCMs) was 29.5 g m�2 month�1 for spruce-fir forest types atyear 2100. Alternatively, more productive northern hardwood and pine forest types experienced a maximumdifference in NPP of 33.4 and 42.5 g m�2 month�1, respectively. The sum of the difference in monthly NPP ofmonths when current climate exceeded climate change (average of GCMs) was 54.7 g m�2 month�1 forspruce-fir forest types and 7.2, and 103.0 g m�2 month�1 for northern hardwood and pine foresttypes, respectively.

3.4. Effect of Temperature and Precipitation on Net Primary Productivity

There was a large variation between the GCMs’ representations of future monthly temperature and precipi-tation for the RCP 8.5 emission scenario (Figure 7). The timing and, to a lesser extent, the magnitude ofchange in NPP was driven by the choice of GCM. For example, The CCSM4 model largely matched the

Figure 4. Difference in annual net primary productivity (g m�2) between simulations using current climactic conditionsand climate change scenarios at year 2100. Smoothed lines (Lowess) show percent of landscape represented by thedifference in net primary productivity between current climate and each corresponding climate model. Positive changevalues represent areas where annual net productivity of a climate change model was larger than the net productivity ofthe current climate.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 9

magnitude of peak production of the other climate change models in May, but lagged behind HADGE andMPIMLR GCMs in April resulting in less annual net production (Figures 2, 3, and 7). No single GCMconsistently encapsulated the greatest change in temperature or precipitation across months and climateregions, nor did any GCM consistently result in the highest monthly NPP (Figure 7).

Spatially, there was a positive relationship between warmer minimum and maximum monthly temperaturesand NPP in the spring and autumn months (i.e., March, April, May, September, and October) but a negativerelationship in summer months (i.e., June, July, and August) (Figure 7). Across climate regions, northernregions were cooler and relatively less productive than southern regions in autumn and early spring months;however, these regions became relatively more productive in May, June, and July. Within individual GCMs,the relationship between temperature and NPP shifted in April and May when warmer regions in thesouthern part of the landscape resulted in less productivity than those in the north. Across GCMs,

Figure 5. Regional monthly net primary productivity in New England at year 2100 using a continuation of current climate conditions and four climate changemodels. Plots represent mean NPP within climate regions (shown in Figure 1) from the northern (climate regions 1 and 2), central (climate regions 12 and 12),and southern (climate regions 24 and 25) regions of New England.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 10

however, the relationship between temperature and NPP did not shift until June when warmer GCMsresulted in less productivity.

There were no clear patterns of changing NPP due to changes in precipitation. In some months (generally inthe spring season) there was a positive relationship between precipitation and NPP at lower precipitationlevels, but this pattern diminished or became negative where there was greater precipitation (e.g., June).In April and May, the warmest GCM (i.e., HADGE (Figure 2) produced a positive relationship between precipi-tation and NPP, although the relationship was not consistent across climate regions (Figure 7).

4. Discussion

Conceptual models of changing patterns in seasonal growth have been presented and defended previously[Richardson et al., 2010], which suggested that greater annual NPP results when increased summer respirationis offset and exceeded by greater earlier spring GPP under years with longer growing seasons [Richardsonet al., 2010]. This conceptual model holds up to our mechanistic simulations of a high emissions future.

Figure 6. Simulated monthly net primary productivity within forest types in New England at year 2100 within current climate and four climate change models.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 11

Figure 7. Net primary productivity for select growing seasonmonths plotted against climate variables: maximum tempera-ture (Tmax), minimum temperature (Tmin), and precipitation (Prec) within each of the 25 climate regions at year 2100 foreach climate scenario. Darkness of symbol color is proportional to climate regions delineated in Figure 1 (i.e., darkersymbols represent southern climate regions; lighter symbols represent more northern climate regions).

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 12

Indeed, all of the GCMs we examined resulted in increased ecosystem respiration that was more than offsetby increased GPP during the spring throughout New England. This finding is also consistent with more recentphenological measurements of carbon uptake [Keenan et al., 2014; Wolf et al., 2016] and an Earth systemmodeling study across the Northern Hemisphere using coarser spatial resolution, finer temporal resolution,and detailed incorporation of photoperiod [Chen et al., 2016]. Compared to current climate, the summerdecline in NPP under climate change is due to increased respiration under warmer scenarios.

Although we did not explicitly examine the effect of precipitation under climate change, water limitation mayalso limit summer growth under our climate change scenarios. During a 2012 drought that affected much ofthe continental U.S., an earlier spring that year largely compensated for the reduced summer carbon uptakedue to the drought [Wolf et al., 2016]. These results mirror our NPP results under future climate warming inthe northeast. In addition, compared to the current climate scenario, climate change also resulted in negativeanomalies of GPP in June, July, and August (Figure 2c). This suggests a midsummer drought effect to GPP thatin addition to increased ecosystem respiration, also contributes to the negative anomalies of summer NPP. Adecrease in summer water availability may, in part, be due to greater transpiration in the spring. Indeed, inthe 2012 drought, greater carbon uptake in the spring also contributed to summer water limitations [Wolfet al., 2016]. While, PnET-Succession does not provide a feedback from water balance to the climate tempera-ture input, soil water availability is accounted for through precipitation, evaporation, transpiration, and run-off. Seasonally, greater NPP is expected in spring compared to autumn due to longer day-lengths [Way andMontgomery, 2015], although others have observed and predicted an increase in NPP due to a later autumngrowing season [e.g., Dragoni et al., 2011; Jeong and Medvigy, 2014; Xie et al., 2015]. Following a summerdecline in productivity under climate change, our results indicate a small increase in autumn NPP across for-est types (Figure 6). Although, nondeciduous species can photosynthesize during warm periods, during win-ter months [Hadley and Schedlbauer, 2002; Kelly and Goulden, 2016] we may have overestimated the lateautumn increase for the reason that most photosynthesis stops at the end of autumn leaf senescence.

The most productive areas simulated in New England were largely in the south central region of the land-scape which is consistent with previous analysis of the region [e.g., Ollinger et al., 1998]. Northern regions,however, resulted in the greatest increases in NPP resulting from climate change. These results corroborateprevious analysis of the entire Northern Hemisphere, where Chen et al. [2016] found that climate change innorthern latitudes resulted in a greater difference in earlier spring onset than southern latitudes.Alternatively, other research using historic satellite imagery of New England found that phenology was moresensitive to temperatures at lower latitudes [Xie et al., 2015]. The magnitude of climate variation under theGCMs that we used compared to historic observations may explain these differences. The two warmest cli-mate scenarios (HADGE and MPIMLR) resulted in the greatest increase in annual NPP; however, the spatialpattern was different. The greatest annual growth in the HADGE scenario occurred in the warmer southernpart of the landscape but included greater precipitation. The MPIMLR scenario, which included less precipita-tion, resulted in more increased productivity in the cooler north suggesting a temperature-precipitationinteraction to increased growth. Likewise, central New England experienced less of a summer decline inNPP under climate change. Greater NPP in these regions may be due to observed wetter late spring monthsand consistently warmer temperatures in this region (Figure 7). Southern regions experienced substantiallyless precipitation under climate change across most months. Less precipitation resulted in lower NPP inthe southern region of the landscape and is consistent with others predicting greater heat stress in southerncoastal New England resulting in relatively less productivity [Xie et al., 2015]. In northern regions, somemonths had a lower increase in precipitation, which may have likewise resulted in a lower increase in NPP;however, the pattern of decreased precipitation in the north was not consistent across months. In addition,the south central region of the landscape where higher relative NPP was found is encompassed by the“tension zone” where a shift of species is found along a north-south gradient [Cogbill et al., 2002]. Within thisregion, a large diversity of tree species and forest types are found [Duveneck et al., 2015] and may havecontributed to the higher relative NPP.

More complex interactions between varying precipitation, cooler temperatures, and species compositionmay further explain the variation in NPP. For example, a positive relationship between precipitation andNPP may depend on a threshold of increased temperature that must be met before additional water canbe utilized for photosynthesis [Cook and Johnson, 1989, Luyssaert et al., 2007, Wolf et al., 2016]; and anon-linear interaction between photoperiod, temperature, and moisture may affect heat stress and

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 13

consequently growing season length [Ollinger et al., 1998; Xie et al., 2015]. Furthermore, withinmixed forestedstands, deciduous and coniferous tree species may respond differently to changing climate [Ollinger et al.,1998, 2008].

Although eastern North America may experience water-dependent growth limitation under drier future con-ditions [Martin-Benito and Pederson, 2015], regional climate change projections indicate greater precipitationin the region, especially during winter, autumn, and spring along with rising temperatures year-round [Fanet al., 2014]. Indeed, evidence suggests that eastern North America is poised to experience climate favorableconditions that will result in increased forest NPP [Keenan et al., 2014; Gauthier et al., 2015; D’Orangeville et al.,2016]. Our study provides a first mechanistic look at regional variation in the productivity respiration trade-offs imposed by climate warming. Nonetheless, several limitations and uncertainties remain. Importantly,the projections of future climate are largely uncertain, as evidenced by the large between-GCM variationobserved in this study. Our simulations did not consider, for example, warmer or drier conditions thanprojected. More severe climate change may result in a decline in annual NPP. In addition, uncertainty existsin tree species response to climate change. For example, while this is the first application of a temperatureacclimating respiration function [Wythers et al., 2005] within a forest landscape model that we are awareof, we did not vary the pattern of acclimation spatially, among species, foliar nitrogen concentration, lightavailability (PAR), canopy position, or developmental stage. This variation may further affect the dynamicnature of respiration response to increased temperature [Atkin et al., 2005; Reich et al., 2016]. However,accounting for precise respiration acclimation may have limited effects on global NPP and may be regionallymore important north and south of New England [Atkin et al., 2008]. In addition, an evolutionary trade-offmay cause species to balance a later spring to avoid frost damage with an earlier spring to increase capacity[Bennie et al., 2010]. Beyond what we simulated, climate extremes may drastically change NPP [e.g.,Ma et al.,2015]. Finally, greater NPP under climate change may depend on the largely uncertain effect of increasedcarbon dioxide fertilization. As more carbon dioxide increases water use efficiency [Keenan et al., 2013], adecline in nitrogen and other nutrients may result in a depletion of resources and a decrease in NPP[Elmore et al., 2016].

Phenology trade-offs have been identified as important factors to be included in future analyses of netecosystem production [Richardson et al., 2013; Chen et al., 2016]. This work builds upon previous spatialmodeling of NPP in the region [Ollinger and Smith, 2005; Pan et al., 2006; Ollinger et al., 2008] by contributinga discussion of the spatial variation in phenology trade-offs under climate change. In addition, our modelingframework built largely around first principles of plant physiology and forest succession includes the pro-cesses of seed dispersal, growth, and mortality of individual tree species under climate change. The climatescenarios we simulated were based on a high emission future. Socioeconomic factors, that are difficult topredict, may ultimately determine our carbon emission future. Ultimately, greater NPP under climate changewill depend on additional factors, (e.g., adequate water threatened by drought [Cook et al., 2015], forest coverthreatened by land use change [Olofsson et al., 2016], nutrient availability threatened by pollution exposurelevels [Ollinger et al., 2002; Templer et al., 2012; Elmore et al., 2016], and future pests and pathogens [Albaniet al., 2010]). Nevertheless, our results suggest that barring other limitations, greater growth in spring willoffset increased respiratory demand and contribute to net increases in productivity.

ReferencesAber, J. D., and C. A. Federer (1992), A generalized, lumped-parameter model of photosynthesis, evapotranspiration and net primary

production in temperate and boreal forest ecosystems, Oecologia, 92, 463–474.Ollinger, S. V., J. Aber, and P. B. Reich (2002), Interactive effects of nitrogen deposition, tropospheric ozone, elevated CO2 and land use history

on the carbon dynamics of northern hardwood forests, Global Change Biol., 8, 545–562.Aber, J. D., S. V. Ollinger, C. A. Federer, P. B. Reich, M. L. Goulden, D. W. Kicklighter, J. M. Melillo, and R. G. Lathrop (1995), Predicting the effects

of climate change on water yield and forest production in the northeastern United States, Clim. Res., 5, 207–222.Aber, J. D., P. B. Reich, and M. L. Goulden (1996), Extrapolating leaf CO2 exchange to the canopy: A generalized model of forest

photosynthesis compared with measurements by eddy correlation, Oecologia, 106, 257–265.Albani, M., P. R. Moorcroft, A. M. Ellison, D. A. Orwig, and D. R. Foster (2010), Predicting the impact of hemlock woolly adelgid on carbon

dynamics of eastern United States forests, Can. J. For. Res., 40, 119–133.Atkin, O. K., D. Bruhn, V. M. Hurry, and M. G. Tjoelker (2005), The hot and the cold: Unravelling the variable response of plant respiration to

temperature, Funct. Plant Biol., 32, 87–105.Atkin, O. K., L. J. Atkinson, R. A. Fisher, C. D. Campbell, J. Zaragoza-Castells, J. W. Pitchford, F. I. Woodward, and V. Hurry (2008), Using

temperature-dependent changes in leaf scaling relationships to quantitatively account for thermal acclimation of respiration in a coupledglobal climate-vegetation model, Global Change Biol., 14, 2709–2726.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 14

AcknowledgmentsThis research was supported in part bythe National Science FoundationHarvard Forest Long Term EcologicalResearch Program (grant NSF-DEB12-37491) and the Scenarios Societyand Solutions Research CoordinationNetwork (grant NSF-DEB-13-38809). Weappreciate the Harvard Forest lab groupand Scott Ollinger for providingproductive feedback on an early versionof this manuscript. We also thank AxiosReview and three additionalanonymous reviewers for valuablefeedback. We thank Eric Gustafson,Brian Miranda, and Arjan de Bruijn formodifying PnET-Succession in LANDIS-IIto provide the ability to track monthlyoutputs. All MODEL input layers andparameter files to generate results forthis study are available on Github:https://github.com/mduveneck/Seasonal-Phenology-Tradeoffs.

Battles, J. J., T. J. Fahey, C. T. Driscoll, J. D. Blum, and C. E. Johnson (2014), Restoring soil calcium reverses forest decline, Environ. Sci. Technol.Lett., 1, 15–19.

Bechtold, W. A., and P. L. Patterson. 2005. The enhanced forest inventory and analysis program—National sampling design and estimationprocedures. Gen. Tech. Rep. SRS-80, U.S. Department of Agriculture, Forest Service, Southern Research Station, Ashville, N. C.

Bennie, J., E. Kubin, A. Wiltshire, B. Huntley, and R. Baxter (2010), Predicting spatial and temporal patterns of bud-burst and spring frost risk innorth-west Europe: The implications of local adaptation to climate, Global Change Biol., 16, 1503–1514.

Buermann, W., P. R. Bikash, M. Jung, D. H. Burn, and M. Reichstein (2013), Earlier springs decrease peak summer productivity in NorthAmerican boreal forests, Environ. Res. Lett., 8(2).

Chen, M., E. K. Melaas, J. M. Gray, M. A. Friedl, and A. D. Richardson (2016), A new seasonal-deciduous spring phenology submodel in thecommunity land model 4.5: Impacts on carbon and water cycling under future climate scenarios, Global Change Biol., 22, 3675–3688.

Cogbill, C. V., J. Burk, and G. Motzkin (2002), The forests of presettlement New England, USA: Spatial and compositional patterns based ontown proprietor surveys, J. Biogeogr., 29, 1279–1304.

Cook, E. R., and A. H. Johnson (1989), Climate change and forest decline—A review of the red spruce case,Water Air Soil Pollut., 48, 127–140.Cook, B. I., T. R. Ault, and J. E. Smerdon (2015), Unprecedented 21st century drought risk in the American southwest and Central Plains, Sci.

Adv., 1, 1–7.Cramer, W., et al. (2001), Global response of terrestrial ecosystem structure and function to CO2 and climate change: Results from six dynamic

global vegetation models, Global Change Biol., 7, 357–373.D’Orangeville, L., L. Duchesne, D. Houle, D. Kneeshaw, B. Côté, and N. Pederson (2016), Northeastern North America as a potential refugium

for boreal forests in a warming climate, Science, 352, 1452–1455.Daly, C., and W. Gibson (2002), 103-Year High-Resolution Temperature Climate Data set for the Conterminous United States, The PRISM Climate

Group, Oregon State Univ., Corvallis, Ore.de Bruijn, A., E. J. Gustafson, B. R. Sturtevant, J. R. Foster, B. R. Miranda, N. I. Lichti, and D. F. Jacobs (2014), Toward more robust projections of

forest landscape dynamics under novel environmental conditions: Embedding PnET within LANDIS-II, Ecol. Modell., 287, 44–57.De Kauwe, M. G., et al. (2013), Forest water use and water use efficiency at elevated CO2: A model-data intercomparison at two contrasting

temperate forest FACE sites, Global Change Biol., 19, 1759–1779.Dragoni, D., H. P. Schmid, C. A.Wayson, H. Potter, C. S. B. Grimmond, and J. C. Randolph (2011), Evidence of increased net ecosystemproductivity

associated with a longer vegetated season in a deciduous forest in south-central Indiana, USA, Global Change Biol., 17, 886–897.Duveneck, M. J., J. R. Thompson, and B. T. Wilson (2015), An imputed forest composition map for New England screened by species range

boundaries, For. Ecol. Manage., 347, 107–115.Duveneck, M. J., J. R. Thompson, E. J. Gustafson, Y. Liang, and A. M. G. de Bruijn (2016), Recovery dynamics and climate change effects to

future New England forests, Landsc. Ecol., 1–13.Dymond, C. C., S. Beukema, C. R. Nitschke, K. D. Coates, and R. M. Scheller (2016), Carbon sequestration in managed temperate coniferous

forests under climate change, Biogeosci. Discuss., 12, 20283–20321.Eisen, K., and A. B. Plotkin (2015), Forty years of forest measurements support steadily increasing aboveground biomass in a maturing,

Quercus-dominant northeastern forest, J. Torrey Bot. Soc., 142, 97–112.Elmore, A. J., D. M. Nelson, and J. M. Craine (2016), Earlier springs are causing reduced nitrogen availability in north American eastern

deciduous forests, Nature Plants, 2, 16133.Fan, F., R. S. Bradley, and M. A. Rawlins (2014), Climate change in the northeastern US: Regional climate model validation and climate change

projections, Clim. Dyn., 43, 145–161.Franks, P. J., et al. (2013), Tansley review, Sensitivity of plants to changing atmospheric CO2 concentration: From the geological past to the

next century, New Phytol., 197, 1077–1094.Gauthier, S., P. Bernier, T. Kuuluvainen, A. Z. Shvidenko, and D. G. Schepaschenko (2015), Boreal forest health and global change, Science, 349,

819–822.Gustafson, E. J. (2013), When relationships estimated in the past cannot be used to predict the future: Using mechanistic models to predict

landscape ecological dynamics in a changing world, Landsc. Ecol., 28, 1429–1437.Gustafson, E. J., A. M. G. De Bruijn, R. E. Pangle, J.-M. Limousin, N. G. McDowell, W. T. Pockman, B. R. Sturtevant, J. D. Muss, and M. E. Kubiske

(2014), Integrating ecophysiology and forest landscape models to improve projections of drought effects under climate change, GlobalChange Biol., 21, 843–856.

Gustafson, E. J., A. M. G. De Bruijn, B. R. Miranda, B. R. Sturtevant, and M. E. Kubiske (2017), Do rising temperatures always increase forestproductivity? Interacting effects of temperature, precipitation, cloudiness and soil texture on tree species growth and competition, inEnvironmental Modelling and Software, vol. 97, pp. 171–183.

Hadley, J. L., and J. L. Schedlbauer (2002), Carbon exchange of an old-growth eastern hemlock (Tsuga Canadensis) forest in central NewEngland, Tree Physiol., 22, 1079–1092.

Hollinger, D. Y., et al. (2004), Spatial and temporal variability in forest-atmosphere CO2 exchange, Global Change Biol., 10, 1689–1706.IPCC (2013), Climate change 2013: The physical science basis, Working Group I contribution to the Fifth Assessment Report of the

Intergovernmental Panel On Climate Change, summary for policymakers, doi:10.1017/CBO9781107415324.Jeong, S., and D. Medvigy (2014), Macroscale prediction of autumn leaf coloration throughout the continental United States, Global Ecol.

Biogeogr., 23, 1245–1254, doi:10.1111/geb.12206.Jeong, S. J., C. H. Ho, H. J. Gim, and M. E. Brown (2011), Phenology shifts at start vs. end of growing season in temperate vegetation over the

northern hemisphere for the period 1982-2008, Global Change Biol., 17, 2385–2399.Keenan, T., J. Maria Serra, F. Lloret, M. Ninyerola, and S. Sabate (2011), Predicting the future of forests in the Mediterranean under climate

change, with niche- and process-based models: CO2 matters!, Global Change Biol., 17, 565–579, doi:10.1111/j.1365-2486.2010.02254.x.Keenan, T. F., D. Y. Hollinger, G. Bohrer, D. Dragoni, J. W. Munger, H. P. Schmid, and A. D. Richardson (2013), Increase in forest water-use

efficiency as atmospheric carbon dioxide concentrations rise, Nature, 499, 324–327.Keenan, T., J. Gray, andM. Friedl (2014), Net carbon uptake has increased through warming-induced changes in temperate forest phenology,

Nat. Clim. Change, 4, 598–604.Kelly, A. E., and M. L. Goulden (2016), A montane Mediterranean climate supports year-round photosynthesis and high forest biomass, Tree

Physiol., 36, 459–468.Luyssaert, S., et al. (2007), CO2 balance of boreal, temperate, and tropical forests derived from a global database, Global Change Biol., 13,

2509–2537.Ma, X., A. Huete, S. Moran, G. Ponce-Campos, and D. Eamus (2015), Abrupt shifts in phenology and vegetation productivity under climate

extremes, J. Geophys. Res. Biogeosci, 120, 2036–2052, doi:10.1002/2015JG003144.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 15

Martin-Benito, D., and N. Pederson (2015), Convergence in drought stress, but a divergence of climatic drivers across a latitudinal gradient ina temperate broadleaf forest, J. Biogeogr., 42, 925–937.

Ning, L., E. E. Riddle, and R. S. Bradley (2015), Projected changes in climate extremes over the northeastern United States, J. Clim., 28,3289–3310.

Ollinger, S. V., and M.-L. Smith (2005), Net primary production and canopy nitrogen in a temperate Forest landscape: An analysis usingimaging spectroscopy, modeling and field data, Ecosystems, 8, 760–778.

Ollinger, S. V., J. D. Aber, and A. Federer (1998), Estimating regional forest productivity and water yield using an ecosystem model linked to aGIS, Landsc. Ecol., 13, 323–334.

Ollinger, S. V., C. L. Goodale, K. Hayhoe, and J. P. Jenkins (2008), Potential effects of climate change and rising CO2 on ecosystem processes innortheastern U.S. forests, Mitig. Adapt. Strateg. Global Change, 13, 467–485.

Olofsson, P., C. E. Holden, E. L. Bullock, and C. E. Woodcock (2016), Time series analysis of satellite data reveals continuous deforestation ofNew England since the 1980s, Environ. Res. Lett., 11, 1–8.

Pan, Y., R. Birdsey, J. Hom, K. McCullough, and K. Clark (2006), Improved estimates of net primary productivity from MODIS satellite data atregional and local scales, Ecol. Appl., 16, 125–132.

Pederson, N., et al. (2015), Climate remains an important driver of post-European vegetation change in the eastern United States, GlobalChange Biol., 21, 2105–2110.

Peters, G. P., R. M. Andrew, T. Boden, J. G. Canadell, P. Ciais, C. Le Quéré, G. Marland, M. R. Raupach, and C. Wilson (2012), The challenge tokeep global warming below 2°C, Nat. Clim. Change, 3, 4–6.

Reich, P. B., K. M. Sendall, A. Stefanski, X. Wei, R. L. Rich, and R. A. Montgomery (2016), Boreal and temperate trees show strong acclimation ofrespiration to warming, Nature, 531, 633–636.

Riahi, K., S. Rao, V. Krey, C. Cho, V. Chirkov, G. Fischer, G. Kindermann, N. Nakicenovic, and P. Rafaj (2011), RCP 8.5—A scenario ofcomparatively high greenhouse gas emissions, Clim. Change, 109, 33–57.

Richardson, A. D., et al. (2010), Influence of spring and autumn phenological transitions on forest ecosystem productivity, Philos. Trans. R. Soc.London, Ser. B, 365, 3227–3246.

Richardson, A. D., T. F. Keenan, M. Migliavacca, Y. Ryu, O. Sonnentag, and M. Toomey (2013), Climate change, phenology, and phenologicalcontrol of vegetation feedbacks to the climate system, Agric. For. Meteorol., 169, 156–173.

Sage, R. F., D. A. Way, and D. S. Kubien (2008), Rubisco, rubisco activase, and global climate change, J. Exp. Bot., 59, 1581–1595.Scheller, R. M., J. B. Domingo, B. R. Sturtevant, J. S. Williams, A. Rudy, E. J. Gustafson, and D. J. Mladenoff (2007), Design, development, and

application of LANDIS-II, a spatial landscape simulation model with flexible temporal and spatial resolution, Ecol. Modell., 201, 409–419.Schuster, W. S. F., K. L. Griffin, H. Roth, M. H. Turnbull, D. Whitehead, and D. T. Tissue (2008), Changes in composition, structure and

aboveground biomass over seventy-six years (1930-2006) in the black rock Forest, Hudson highlands, southeastern New York state, TreePhysiol., 28, 537–549.

Schwartz, M. D., R. Ahas, and A. Aasa (2006), Onset of spring starting earlier across the northern hemisphere, Global Change Biol., 12, 343–351.Sillmann, J., V. V. Kharin, X. Zhang, F. W. Zwiers, and D. Bronaugh (2013), Climate extremes indices in the CMIP5 multimodel ensemble: Part 1.

Model evaluation in the present climate, J. Geophys. Res. Atmos., 118, 1716–1733.Smith, N. G., and J. S. Dukes (2013), Plant respiration and photosynthesis in global-scale models: Incorporating acclimation to temperature

and CO2, Global Change Biol., 19, 45–63.Stoner, A. M. K., K. Hayhoe, X. Yang, and D. J. Wuebbles (2013), An asynchronous regional regression model for statistical downscaling of daily

climate variables, Int. J. Climatol., 33, 2473–2494.Templer, P. H., R. W. Pinder, and C. L. Goodale (2012), Effects of nitrogen deposition on greenhouse-gas fluxes for forests and grasslands of

North America, Front. Ecol. Environ., 10, 547–553.Thompson, J. R., D. N. Carpenter, C. V. Cogbill, and D. R. Foster (2013), Four centuries of change in northeastern United States forests, PLoS

One, 8, e72540.Thorn, A. M., J. Xiao, and S. V. Ollinger (2015), Generalization and evaluation of the process-based forest ecosystem model PnET-CN for other

biomes, Ecosphere, 6(3), 43, doi:10.1890/ES14-00542.1.Urbanski, S., C. Barford, S. Wofsy, C. Kucharik, E. Pyle, J. Budney, K. McKain, D. Fitzjarrald, M. Czikowsky, and J. W. Munger (2007), Factors

controlling CO2 exchange on timescales from hourly to decadal at Harvard Forest, J. Geophys. Res., 112, G02020, doi:10.1029/2006JG000293.

Walker, X. J., M. C. Mack, and J. F. Johnstone (2015), Stable carbon isotope analysis reveals widespread drought stress in boreal black spruceforests, Global Change Biol., 21, 3102–3113.

Way, D. A., and R. A. Montgomery (2015), Photoperiod constraints on tree phenology, performance and migration in a warming world, PlantCell Environ., 38, 1725–1736.

Wolf, S., et al. (2016), Warm spring reduced carbon cycle impact of the 2012 US summer drought, Proc. Natl. Acad. Sci. U.S.A., 113, 5880–5885.Wythers, K. R., P. B. Reich, M. G. Tjoelker, and P. B. Bolstad (2005), Foliar respiration acclimation to temperature and temperature variable Q10

alter ecosystem carbon balance, Global Change Biol., 11, 435–449.Wythers, K. R., P. B. Reich, and J. B. Bradford (2013), Incorporating temperature-sensitive Q10 and foliar respiration acclimation algorithms

modifies modeled ecosystem responses to global change, J. Geophys. Res. Biogeosci., 118, 77–90, doi:10.1029/2011JG001897.Xie, Y., X. Wang, and J. A. Silander (2015), Deciduous forest responses to temperature, precipitation, and drought imply complex climate

change impacts, Proc. Natl. Acad. Sci. U.S.A., 112, 13585–13590.

Journal of Geophysical Research: Biogeosciences 10.1002/2017JG004025

DUVENECK AND THOMPSON PHENOLOGICAL TRADE-OFFS ON FORESTS 16