j ourna l h omepa ge: www.elsev ier .com/ locate /eco lmodel
ast-century decline in forest regeneration potential across aatitudinal and elevational gradient in Canada
dam Ericksona,∗, Craig Nitschkeb, Nicholas Coopsa, Steven Cummingc,ordon Stenhoused
Integrated Remote Sensing Studio, Department of Forest Resources Management, University of British Columbia, 2045-2424 Main Mall, Vancouver,ritish Columbia V6 T 1Z4, CanadaSchool of Ecosystem and Forest Sciences, University of Melbourne, 500 Yarra Boulevard, Richmond, Victoria 3121, AustraliaDepartment of Wood and Forest Sciences, Laval University, 2405 rue de la Terrasse, Quebec City, Quebec G1 V 0A6, CanadaFoothills Research Institute, Box 6330, Hinton, Alberta T7 V 1X6, Canada
r t i c l e i n f o
rticle history:eceived 21 April 2015eceived in revised form 16 June 2015ccepted 17 June 2015
eywords:rocess-based modelingree regenerationpecies distribution modelinghenology
a b s t r a c t
The regeneration niche of trees greatly narrows the fundamental niche and is sensitive to climatic change.Development from seed and phenology are regulated by biological and environmental controls, shap-ing forest successional pathways. We hypothesized that recent climate change is reducing regenerationsuitability in northern forests. We used a process-based ecophysiological model to examine changesin forest regeneration conditions across an elevational and latitudinal gradient in Alberta, Canada from1923 to 2012. We compared these results to a recent empirical study in the region to infer the recentdrivers of regeneration change in northern forests. Our results suggest that these forests are experiencingclimatically driven declines in conditions suitable for regeneration. Contrary to previous findings indi-cating poorer current conditions in low elevation forests, we found more stable regeneration potential
limate changeoil water balance
there, attributable to a relative abundance of soil moisture. Rocky soils resulted in modeled losses of soilmoisture at higher elevations, potentially preventing upslope migrations of species despite warming.We identify potential mechanisms driving unexpected tree regeneration patterns described in previousstudies. Our simulations suggest a delayed response of forest regeneration to warming throughout thepast 90 years.
Tree development and phenology are related to climate throughvolutionary controls, influencing the early niche space of trees,ith plasticity potentially providing a buffer to maintain fitness
Aitken et al., 2008; Vitasse et al., 2013). Important tree develop-ent and phenology events include germination, establishment,
ud burst, growth, bud set, leaf senescence, seed fall, and dor-ancy, among others (Richardson et al., 2013; Walck et al., 2011).
limatic change can uncouple the phasing of fine-scale seasonaleather variations with developmental processes and phenology
eyond the range of plasticity, reducing regeneration rates (Fridley,
012; Richardson et al., 2013). This phase uncoupling can alter theuration of important phenological processes and timing of phe-ological events.
The widespread adaptation of trees to local climatic con-ditions (Alberto et al., 2013) indicates that tree phenology isintricately tuned to optimize fitness for local environmental con-ditions through gene expression, posttranslational modification,and, genetic and epigenetic inheritance (Cooke et al., 2012; Liuet al., 2010; Matzke and Mosher, 2014). Environmental effects areestimated to exert greater influence on plasticity than genetics innorthern forests (Vitasse et al., 2013), while phenotypic variationreflecting phylogeographic origins (Alberto et al., 2013) is not nec-essarily adaptive (Duputié et al., 2015). Extreme weather events,such as frost or drought, occurring at critical times during treedevelopment can have strong demographic effects on forests. Giventhe importance of fine-scale climatic and phylogenetic variability,high temporal resolution climate data (Cook et al., 2010) alongwith a range of aggregate species tolerances can aid in the mod-
eling of these dynamics at the landscape scale, where individual-or population-level data is seldom attainable.
We hypothesized that warmer conditions combined withchanges in soil water balance (Dobrowski et al., 2013; Piedallu
A. Erickson et al. / Ecological Modelling 313 (2015) 94–102 95
Fig. 1. Study area overlaid on 90-m resolution NASA SRTM topography: (a) study area geographic context within North America; (b) biogeoclimatic subregions and weathers ; (d) S( eprest
e(aeMeba2feo2E
ebhtosstwco9
2
2
Er2(
tations; (c) biogeoclimatic regions with subregion outlines and weather stationsAWHC) classes are shown, the most sensitive edaphic model parameter, with red rhe figure legend, the reader is referred to the web version of this article.)
t al., 2013) and more rapid and severe extreme weather eventsAllen et al., 2010; Kamae et al., 2014; Trenberth et al., 2014) areltering regeneration patterns in northern forests. Recent empiricalvidence suggests that this shift is already occurring (Boisvert-arsh et al., 2014; Lenoir et al., 2009; Urbieta et al., 2011; Zhang
t al., 2015). However, direct measurement remains confoundedy forest turnover, which can increase the amount of space avail-ble for recruitment (Carvalhais et al., 2014; Park Williams et al.,013; Woodall et al., 2013; Zhu et al., 2014, 2012). Additional con-ounding factors include patterns of fine-scale climate (Dobrowskit al., 2013) and ontogenetic niche variation, whereby the nichesf species can change throughout development (Bertrand et al.,011a; Cavender-Bares and Bazzaz, 2000; Donohue et al., 2010;riksson, 2002; Niinemets, 2010; Urbieta et al., 2011).
We suggest that changes to tree regeneration throughout north-rn forests in recent decades have been driven by interactionsetween climatic change and local soil patterns. To test thisypothesis, we used a species-specific ecophysiological modelhat explicitly represents major tree regeneration processes, basedn forest gap models. We parameterized the model for treepecies and soil textural classes across a 25.2 million hectaretudy area in Alberta, Canada, encapsulating an important eleva-ional and latitudinal gradient. We used daily resolution historicaleather station data for three decadal periods over the last
entury, and for the most recent decade, to model the effectsf climatic change on forest regeneration throughout the past0 years.
. Materials and methods
.1. Study area
We applied the Tree And Climate Assessment Germination and
stablishment Model (TACA-GEM) across fourteen biogeoclimaticegions of western Alberta, Canada (Natural Regions Committee,006) (Fig. 1), coextensive with ecoregions in the United StatesRicketts, 1999). The study area comprises a transition zone from
oil regions with outlines and weather stations; available water holding capacityenting bare rock or effectively zero. (For interpretation of the references to color in
boreal forest at lower elevations to higher elevation Cordilleranfoothills and montane forests in the southern Canadian RockyMountains. We derived soil and climate parameters for thir-teen natural subregions, excluding the treeless alpine subregion.Regional soil properties reflect a recent glacial history, primarilyconsisting of morainal and glacio-lacustrine parent materials, withgray luvisols and black chernozems representing the dominant soiltypes (Natural Regions Committee, 2006). Luvisols are periodicallysaturated and depleted of oxygen, whereas Chernozems occurs insemiarid and subhumid climates, representing the dominant soil ofthe Canadian southern interior plains (Soil Classification WorkingGroup, 1998). The region consists primarily of well-drained uplandsoils.
Elevational and latitudinal gradients segment the study areabiogeoclimatically, with mean elevations ranging from 525 metersin the boreal to 2350 meters in the alpine. The study area cov-ers a latitudinal gradient from 49◦ at the U.S. border to 58◦ at thenorthernmost point (NAD83 datum). The heavily forested foothillsregion experiences higher levels of precipitation than surround-ing areas, supporting productive lodgepole pine (Pinus contortavar. latifolia) forests and an active timber industry. While mostCanadian provincial harvest levels remained stable over the pastfour decades, harvest increased approximately four-fold in Alberta(National Forestry Database, 2013), alongside a rise in oil, gas, andmineral extraction activities. Regionally abundant species includelodgepole pine (Pinus contorta), white spruce (Picea glauca), trem-bling aspen (Populus tremuloides), and black spruce (Picea mariana)(Natural Regions Committee, 2006; Zhang et al., 2015). Previousstudies show that this region became warmer and drier throughoutthe 20th century (Luo and Chen, 2013; Peng et al., 2011).
2.2. TACA-GEM model design
The latest version of the Tree And Climate Assessment Ger-mination and Establishment Model (TACA-GEM) presented herein(Fig. 2) builds on establishment-only TACA-EM (Nitschke and Innes,2008) and extends previous TACA-GEM versions (Nitschke et al.,
96 A. Erickson et al. / Ecological Modelling 313 (2015) 94–102
F iche me
2n(f(FeSwptlptmyoooiidp
r(qot
ig. 2. Modular design of the latest TACA-GEM model presented herein: (a) habitat nvents module.
012) with four improvements. First, establishment suitability iso longer governed by binary responses to growing degree daysGDD) and drought conditions for a given year. The GDD responseunctions from Zelig++ (Burton and Cumming, 1995), JABOWAnamed for Janak, Botkin, and Wallis) (Botkin et al., 1972) andORÊT (Shugart and West, 1977) are used to determine annualstablishment suitability as a probabilistic function of temperature.econd, drought is no longer represented as the portion of the yearhere water deficit occurs per the actual-to-potential evapotrans-iration ratio, but is now calculated based on the proportion ofhe year where soil water potential is equal or below the turgoross point (permanent wilting) for a given species. Third, soil waterotentials are calculated from soil water availability and soil tex-ure class, using a reformulation of the van Genuchten soil water
odel (Van Genuchten, 1980). Species suitability is equal to one inears with no water deficit and declines to zero if the proportionf the year under water deficit exceeds a species-specific thresh-ld. The fourth improvement to the model is the developmentf an extreme events module. The extreme events module mod-fies species regeneration by eliminating seedlings that regeneraten favorable years but are subjected to prolonged and/or extremerought or frost events, which result in mortality over decadal timeeriods.
Probabilistic germination functions interact with modifiersesulting from frost and drought events to model germination rates
Fig. A.1). The model utilizes species-specific submodels (linear,uadratic or cubic functions) that predict the timing and magnitudef germination on a daily time-step, as a function of tempera-ure, soil moisture, GDD, and environmental stratification related
to seed dormancy. Survival of germinants is determined by thetiming of germination in relation to frost events and drought con-ditions, or phasing. Seedling drought tolerance increases with age(Cavender-Bares and Bazzaz, 2000; Niinemets, 2010). Followinggermination, establishment suitability is calculated as specifiedabove. Importantly, TACA-GEM diverges from gap models in itsparameterization of species regeneration processes, utilizing mea-sured species responses rather than inferring them from speciesrange limits. Range-based estimates historically used by gap mod-els to infer species responses (e.g., growth response to temperature)have been shown to be ineffective for predicting future responses(Loehle and LeBlanc, 1996; Price et al., 2001). Here, we modelchanges in in situ regeneration based on measured climatic con-ditions and process-specific species responses.
The TACA-GEM model focuses on species-specific responsesthrough the use of logical submodels representing biological pro-cesses. The use of process-based species responses allows us tobridge the current knowledge gap related to fine-scale biolog-ical processes while minimizing cumulative model error. Suchprocess-based models are particularly useful for balancing modelspecificity and generalizability (Levins, 1966) to infer broad-scaleforest change, through a combination of biological appropriatenessand computational efficiency. This paper implements the previ-ously developed germination module (Mok et al., 2012) in NorthAmerica, parameterized for 21 regional species either currently
within the study area or with the potential to in-migrate fromadjacent regions, based on published literature on species germina-tion and seed ecology. We further provide an example of the threegermination functions utilized in the model (Fig. A.2).
al Mod
2
aTlcChtwediipt(Gsaistp
A(mGd2wisrWaaLt1dtrc
2
ooGGstwoaacpwstc
A. Erickson et al. / Ecologic
.3. Climate parameters
Most bioclimatic studies use monthly means of weather vari-bles averaged over a climate normal period, typically 30 years.he major factor influencing the selection of this temporal reso-ution is the widespread availability of monthly resolution generalirculation model projections (Intergovernmental Panel on Climatehange, 2014). However, monthly resolution data attenuates bothigh frequency variation within months and low frequency varia-ion at annual or decadal scales, and is therefore unlikely to coincideith vegetation phenology related to regeneration (Richardson
t al., 2013). Research has shown that the use of daily resolutionata markedly improves the results of phenological models used
n ecological forecasting (Cook et al., 2010). Although dynamicn nature (Cooke et al., 2012; Metlen et al., 2009), vegetationhenology has long been described at a daily resolution for bothhe timing of phenological events and related weather patternsMorin et al., 2009; Richardson et al., 2013). Accordingly, TACA-EM was designed to use daily resolution weather data to modeloil moisture conditions and species regeneration responses, along
decadal scale. Acquiring daily resolution climate projections typ-cally involves the use of stochastic weather generators or othertatistical downscaling approaches. However, daily resolution his-orical weather station measurements provide the most robusthenology model parameters.
We used data from the National Oceanic and Atmosphericdministration (NOAA) Global Historical Climate Network Daily
GHCN-D) version 3.11 to parameterize daily minimum andaximum temperature, and precipitation sum, in the TACA-EM model. The GHCN-D dataset is a global weather stationatabase subjected to uniform quality assurance (Menne et al.,012). Using the R programming language (R Core Team, 2014),e computed daily weather values for the median decades of
nterest within 30-year periods, averaged across each naturalubregion for each day, in order to provide model results compa-able to recent vegetation modeling studies (Wang et al., 2012).
e imputed missing values using a computationally efficientpproximation of the expectation-maximization bootstrappinglgorithm using the R FastImputation package (Honaker et al., 2011;ounici, 2012). Our GHCN-D functions are available as part ofhe rnoaa package. We applied this approach to the 1923–1952,953–1982, and 1983–2012 periods by modeling their medianecades (1933–1943, 1963–1972, and 1993–2002, respectively), ashe model is designed for decadal periods. We modeled the mostecent period (2003–2012) to offer the most accurate depiction ofurrent regeneration conditions.
.4. Soils parameters
We overlaid the biogeoclimatic natural regions and subregionsf Alberta (Natural Regions Committee, 2006) onto Soil Landscapesf Canada (SLC) v3.2 data (Soil Landscapes of Canada Workingroup, 2010) to generate soil textural class parameters for TACA-EM (Fig. A.3 and Table A.1). We characterized soils in each naturalubregion based on the dominant soil type. Soil texture, roo-ing zone depth, percentage of coarse fragment material, availableater storage capacity, and percolation rate were calculated based
n corresponding SLC values. We obtained soil moisture regimend mean elevation parameters from the natural subregion summ-ries (Natural Regions Committee, 2006). Traditional soil texturallasses were calculated using Agriculture and Agri-Food Canadaarticle size classes (Soil Classification Working Group, 1998). Soils
ere classified into textural groups based on SLC values for percent
and, silt, and clay, filtered by parent material texture. Percola-ion rates were calculated by subtracting available water holdingapacity from field capacity for each natural subregion, equal to
elling 313 (2015) 94–102 97
the soil permanent wilting point (Cassel and Nielsen Arnold, 1986).Organic soils were designated for one subregion, based on evidenceprovided in the biogeoclimatic region summaries (Natural RegionsCommittee, 2006).
2.5. Species parameters
Tree species modeled in the study included any presentlyextant or directly adjacent species, in order to account for poten-tial in-migrations (Table A.2). Tree species biophysical parameterswere derived from the literature and regional databases, fol-lowing methods applied previously (Nitschke and Innes, 2008;Nitschke et al., 2012). Sources for species biophysical param-eters used in the model are provided in the supplementaryinformation.
3. Results
We found that tree regeneration suitability, modeled as theprobability of reaching age ten, declined for most species in thestudy area (Figs. 3, 4, A.4 and A.6). The establishment of new cohortsfor most species was increasingly unlikely. Adding extreme climaticevents (i.e., drought and frost) further reduced regeneration con-ditions, which were poorest in recent decades. We estimate thatthe regeneration niches of extant and adjacent tree species arelargely out of equilibrium with climatic conditions and have beenfor decades, with regeneration conditions likely to worsen in thecoming years. The frequency and magnitude of drought followinggermination was the most limiting factor affecting regenerationconditions, due to reduced soil moisture.
We found that the most recent period modeled, 2003–2012,shows a slight deceleration in the rate of regeneration suitabil-ity change, likely attributable to a slowdown in warming (Kosakaand Xie, 2013). As multi-decadal warming continues unabated,modeled tree species are likely to fail to regenerate. An increasingmagnitude of climatic disequilibrium is likely to reduce for-est regeneration, which may be initiated through climate-drivenchanges to disturbance regimes (Magnani et al., 2007).
A significant mean decline in regeneration suitability was pre-dicted across the full study period. Compared to simulationswithout extreme events, including extreme events in the sim-ulations marginally decreased the probability of establishment(mean = 0.085, � = 0.220; mean = 0.059, � = 0.216) and the changein establishment across the full study period (mean = −0.138,� = 0.228; mean = −0.142, � = 0.248). More frequent drought,diminished germination success, and lengthened bud dormancydue to the failure to meet chilling requirements resulted in anoverall decline in regeneration suitability. Species regenerationalresponses varied across space and time, often responding simi-larly in direction to climatic and edaphic conditions within regionsand time periods. These trends (Fig. 3) are indicative of directionalclimate change.
The boreal forest and foothills regions showed a transitionin the regeneration niche of species toward deciduous poplarspecies, with improved suitability for Engelmann spruce (Piceaengelmannii), a high altitude montane species. Other regions indi-cated regenerational improvements for pine species that are moreresilient to drought. The grassland region, which has the lowestregeneration suitability overall due to high hydraulic conductiv-ity, showed a transition toward fir, poplar, and pine. The parklandregion showed the strongest net improvement in regeneration con-
ditions overall, transitioning toward fir and pine species. The RockyMountain region showed the most widespread regenerationaldecline, symptomatic of relatively severe soil moisture limita-tions. Overall, areas of higher elevation experienced the greatest
98 A. Erickson et al. / Ecological Modelling 313 (2015) 94–102
F s 192c ed Grs olor in
rl
idtfpgai(cCS
iH
ig. 3. Mean change in species regeneration probability (Delta) between the periodonsistently failed to regenerate appear unchanged, as in the soil moisture-constrainolid red = −1.0; none = 0; solid green = 1.0 (For interpretation of the references to c
egenerational decline, due to a soil water balance shift towardower elevations (Fig. A.5).
The impact of directional warming on species-specific phys-ological drought frequency varied considerably across regions,ue to interactions with regional soil properties, precipitation pat-erns, and snowmelt timing. Results indicate that in the borealorest, grassland, and Rocky Mountain regions, the 1983–2003eriod crossed species drought tolerance thresholds with thereatest frequency, while showing an increased incidence of frostfter bud flush, indicative of increased temperature variabil-ty. Modeled regeneration suitability for the 2003–2012 periodFigs. 4 and A.6) reflects a temporary slowdown in warmingombined with increased precipitation, in agreement with thelausius–Clapeyron equation and previous observations (Allan and
oden, 2008; Trenberth, 2011; Wentz et al., 2007).
The biophysical niche space of species accounted for a signif-cant amount of variation in regeneration responses (p ≤ 0.001).owever, species often responded with similar directionality to
3–1952 and 1983–2012 including extreme events for the five regions; species thatassland region; a value of 1.0 represents a 100% change in regeneration probability;
the figure legend, the reader is referred to the web version of this article.).
climatic and edaphic changes, particularly for drought and turgorloss. While inter-specific variability was clearly present (Fig. 5d;mean � = 0.179), changes to regeneration suitability were betterexplained by spatiotemporal variation (p ≤ 0.001; mean � = 0.198;Fig. 5a and b). In terms of regeneration, gymnosperms fared simi-larly to angiosperms across all periods and regions (gymnospermmean = 0.18, � = 0.225; angiosperm mean = 0.14, � = 0.198), whilethe difference diminished with the inclusion of extreme effects(gymnosperm, mean = 0.14, � = 0.221; angiosperm, mean = 0.12,� = 0.197). Species, regions, and time periods showed greater dis-tributional heterogeneity than differences between conifers andangiosperms (Fig. 5).
A fixed-effects analysis of variance (ANOVA) shows a signif-icant correlation between variance in regeneration probabilities
and species, period, region, subregion, growing degree days, killingfrosts, drought frequency, germination frequency, germinationevents, and stratification (p ≤ 0.001), and a less significant rela-tionship with spermatophyte taxon, ordinal date of bud break
A. Erickson et al. / Ecological Modelling 313 (2015) 94–102 99
Fig. 4. Means species regeneration probabilities averaged across thirteen biogeoclimatic subregions; blue = 1923–1952; light blue = 1953–1982; light red = 1983–2012;r f the
p ≤ 0.01), and physiological dormancy (p ≤ 0.05). Chilling require-ents met, frost frequency, frost days, frost events, and turgor loss
oint frequency did not show a significant relationship with speciesegeneration probabilities (p > 0.05). The frequency of germinationuccess for species varied across the study period, while generallyeclining. Ponderosa pine (Pinus ponderosa), an arid fire-adaptedpecialist, experienced the greatest reduction (Fig. A.7).
We conducted a regression analysis to determine the relativemportance of different mechanisms in explaining regenerationalues, filtering covariates at a threshold of |r| > 0.7 (Dormannt al., 2013), before filtering variables at a significance thresh-ld of p ≤ 0.05. We utilized R2 partitioned by averaging overrderings of regressors (LMG) and proportional marginal varianceecomposition (PMVD) with the R relaimpo package (Grömping,006) to determine predictor variable relative importance in lin-ar regression. With extreme events for all species and periods,oth metrics indicated that germination frequency (LMG = 0.28;MVD = 0.31), drought frequency (LMG = 0.08; PMVD = 0.16), num-er of growing degree days (LMG = 0.06; PMVD = 0.07), and turgor
oss point frequency (LMG = 0.05; PMVD = 0.001) were the mostmportant mechanisms determining regeneration responses, pro-iding details that were absent in the previously described ANOVA.rom these metrics, a multi-decadal climatic warming signal is evi-ent, with heterogeneous regional implications.
The modeled frequency of reaching species’ turgor loss pointnd exceeding physiological drought tolerances increased for mostpecies in most regions over the full time period. Turgor-loss-pointrequency was relatively homogeneous across species, exhibiting
generalized model response of turgor loss point and leaf waterotential at 50% loss of hydraulic conductivity to climate changemeasures of leaf vulnerability to cavitation that are functionsf xylem structure) (Bartlett et al., 2012). Species physiological
references to color in the figure legend, the reader is referred to the web version of
drought frequency was heterogeneous across regions and homoge-neous within regions. Specific regeneration probabilities fluctuatedprimarily in response to climatic change relative to soil mois-ture conditions. Black cottonwood (Populus trichocarpa) and balsampoplar (Populus balsamifera) exhibited the greatest sensitivity todrought, as shown in empirical studies (Nitschke et al., 2012). Firand pine species were the most tolerant of increasingly frequentand severe drought conditions.
At higher elevations and latitudes, modeled frost events follow-ing both bud flush and germination declined in the most recentdecades. However, frost events more frequently exceeded specificfrost tolerance threshold parameters, indicative of warming andgreater temperature variability. More frequent and severe modeledphysiological drought conditions were the main factor limitingmodeled establishment values in the boreal, grassland, and park-land regions, with modeled soil moisture particularly meagre in thelatter two regions due to the high hydraulic conductivity of glacialand aeolian deposits. The inclusion of extreme weather eventsparticularly reduced modeled establishment conditions for fir andspruce species with lower drought tolerance – but similar frost tol-erance – than pines. The modeled number of growing degree daysand probability of germination success also declined. In essence,a phase decoupling of climatic patterns and species phenology intime and space seems likely from our model results for multipleregenerational processes.
4. Discussion
Based on our simulations, we conclude that warmer and morevariable climatic conditions are diminishing the conditions forextant and adjacent tree species regeneration in Alberta, Canada.Some studies indicate that forest regeneration conditions should
100 A. Erickson et al. / Ecological Modelling 313 (2015) 94–102
F (a) timt inter
bL(miif2dld2i2iptUe
pfsse2t
ig. 5. Boxplots for species regeneration probability including extreme events by:hese boxplots represent the median, first and third quartiles, and a 95% confidence
e improving at higher elevations and latitudes (Brubaker, 1986;enoir et al., 2009) and declining in lower elevation forestsBertrand et al., 2011b; Loarie et al., 2009), while others provide
ixed results potentially related to changes in human activ-ty (Boisvert-Marsh et al., 2014). Our findings support a relativemprovement in regeneration conditions in low-elevation northernorests (Crimmins et al., 2011; Dobrowski et al., 2013; Zhu et al.,014, 2012). We found that changes to soil moisture conditionsrove species regeneration niches toward the foothills and park-
and regions, indicating that changes to soil water balance mayrive future species migrations under warming (Crimmins et al.,011; Piedallu et al., 2013). The inclusion of soil water balance
s particularly important in mountain watersheds (Hwang et al.,014) and in the Canadian boreal forest (Barnett et al., 2005), as
t is the key limiting factor driven by climate. Our model resultsrovide a potential explanation of complex tree regeneration pat-erns observed in previous studies (Boisvert-Marsh et al., 2014;rbieta et al., 2011; Woodall et al., 2013; Zhang et al., 2015; Zhut al., 2014, 2012), providing direction for future empirical work.
A recent study (Zhang et al., 2015) provides empirical sup-ort for our modeling results. Using permanent sample plot dataor western Canada, their study shows that competition plays atronger role in local forest dynamics than climate, long demon-
trated in gap and hybrid model studies based on theoretical andmpirical formulations (Deutschman et al., 1997; Schumacher et al.,004; Shugart, 1984). The Zhang et al. (2015) study also modeledrends in recruitment rates for western Canadian provinces, based
e period; (b) biogeoclimatic region; (c) conifers versus angiosperms; (d) species;val of the median.
on empirical data. The study shows an overall decline in recruit-ment across regions, with a linear decline exhibited in Alberta. Inagreement with our results and contrary to findings for the SierraNevada range (Dolanc et al., 2013), the most dramatic reduction inrecruitment and growth rates occurred in the high-elevation mon-tane Cordillera region (Zhang et al., 2015). Our study indicates thatthis reduction in recruitment is likely due to changes in soil waterbalance driven by the interaction of warming and soil textural prop-erties.
In the coming years, increases in the frequency of anthropogenicdisturbances (Kurz et al., 2008; Park Williams et al., 2013) mayaccelerate the currently delayed regenerational response of forestsby increasing the number of sites available for recruitment. Con-currently, warmer and wetter conditions projected for northernforests (Trenberth, 2011) may accelerate recruitment by increasingthe rate of forest turnover (Carvalhais et al., 2014; Zhu et al., 2014).Due to the long-lived nature and relatively rapid dispersal abilityof trees (Clark et al., 1998), future compositional changes will likelyoccur in pulses as climatic change intensifies. Directional changes tothe region’s forests may occur through rare long-distance migrationevents (Clark et al., 1998) by species better adapted to low soil mois-ture, producing no-analog communities. Future empirical studiesshould investigate evidence of regenerational change in northern
forests with ground plot data in connection with directly measuredlocal climate data. Future modeling studies should incorporateimportant forest dynamics, such as competition, dispersal, and dis-turbance by fusing theoretical and empirical formulations with
al Mod
dumo
A
zHhatm
A
t0
R
A
A
A
A
B
B
B
B
B
B
B
B
C
C
C
C
A. Erickson et al. / Ecologic
etailed remote sensing structural measurements. An improvednderstanding of forest regeneration may help forest managers toeet multiple-use goals, while providing a more complete picture
f biospheric climate feedbacks.
cknowledgements
Funding for this research was generously provided by the Griz-ly Bear Program of the Foothills Research Institute located ininton, Alberta, Canada, with additional information available atttp://foothillsresearchinstitute.ca. Funding was also provided byn NSERC grant (RGPIN 311926-13) to N.C.C. We would like tohank Sally Aitken, Robert Guy, and Scott Nielsen for providing
anuscript feedback.
ppendix A. Supplementary data
Supplementary data associated with this article can be found, inhe online version, at http://dx.doi.org/10.1016/j.ecolmodel.2015.6.027
eferences
itken, S.N., Yeaman, S., Holliday, J.a., Wang, T., Curtis-McLane, S., 2008. Adaptation,migration or extirpation: climate change outcomes for tree populations. Evol.Appl. 1, 95–111, http://dx.doi.org/10.1111/j.1752-4571.2007.00013.x
lberto, F.J., Aitken, S.N., Alía, R., González-Martínez, S.C., Hänninen, H., Kremer, A.,Lefèvre, F., Lenormand, T., Yeaman, S., Whetten, R., Savolainen, O., 2013. Potentialfor evolutionary responses to climate change – evidence from tree populations.Glob. Change Biol. 19, 1645–1661, http://dx.doi.org/10.1111/gcb.12181
llan, R.P., Soden, B.J., 2008. Atmospheric warming and the amplification of pre-cipitation extremes. Science 321, 1481–1484, http://dx.doi.org/10.1126/science.1160787
llen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier,M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H. (Ted), Gonzalez, P., Fen-sham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G., Running, S.W.,Semerci, A., Cobb, N., 2010. A global overview of drought and heat-induced treemortality reveals emerging climate change risks for forests. For. Ecol. Manage.259, 660–684, http://dx.doi.org/10.1016/j.foreco.2009.09.001
arnett, T.P., Adam, J.C., Lettenmaier, D.P., 2005. Potential impacts of a warmingclimate on water availability in snow-dominated regions. Nature 438, 303–309.
artlett, M.K., Scoffoni, C., Sack, L., 2012. The determinants of leaf turgor loss pointand prediction of drought tolerance of species and biomes: a global meta-analysis. Ecol. Lett. 15, 393–405, http://dx.doi.org/10.1111/j.1461-0248.2012.01751.x
ertrand, R., Gégout, J.-C., Bontemps, J.-D., 2011a. Niches of temperate treespecies converge towards nutrient-richer conditions over ontogeny. Oikos 120,1479–1488, http://dx.doi.org/10.1111/j.1600-0706.2011.19582.x
ertrand, R., Lenoir, J., Piedallu, C., Riofrío-Dillon, G., de Ruffray, P., Vidal, C., Pierrat,J.-C., Gégout, J.-C., 2011b. Changes in plant community composition lag behindclimate warming in lowland forests. Nature 479, 517–520, http://dx.doi.org/10.1038/nature10548
oisvert-Marsh, L., Périé, C., de Blois, S., 2014. Shifting with climate? Evidence forrecent changes in tree species distribution at high latitudes. Ecosphere 5, art83,http://dx.doi.org/10.1890/ES14-00111.1
otkin, D.B., Janak, J.F., Wallis, J.R., 1972. Some ecological consequences of a com-puter model of forest growth. J. Ecol. 60, 849–872.
rubaker, L.B., 1986. Responses of tree populations to climatic change. Vegetatio 67,119–130.
urton, P.J., Cumming, S.G., 1995. Potential effects of climatic change on some west-ern Canadian forests, based on phenological enhancements to a patch model offorest succession. Water Air Soil Pollut. 82, 401–414, http://dx.doi.org/10.1007/BF01182850
arvalhais, N., Forkel, M., Khomik, M., Bellarby, J., Jung, M., Migliavacca, M.,[Mgr]u, M., Saatchi, S., Santoro, M., Thurner, M., Weber, U., Ahrens, B., Beer,C., Cescatti, A., Randerson, J.T., Reichstein, M., 2014. Global covariation ofcarbon turnover times with climate in terrestrial ecosystems. Nature 514,213–217.
assel, D.K., Nielsen Arnold, D.R.E.-K., 1986. Field capacity and available water capac-ity. In: Methods of Soil Analysis. Part 1. Physical and Mineralogical Methods., pp.901–926.
avender-Bares, J., Bazzaz, F.A., 2000. Changes in drought response strategies withontogeny in Quercus rubra: implications for scaling from seedlings to mature
lark, J.S., Fastie, C., Hurtt, G., Jackson, S.T., Johnson, C., King, G.A., Lewis, M., Lynch,J., Pacala, S., Prentice, C., Schupp, E.W., Thompson Webb, I.I.I., Wyckoff, P., 1998.Reid’s paradox of rapid plant migration. Bioscience 48, 13–24, http://dx.doi.org/10.2307/1313224, CR – Copyright © University of Ca.
elling 313 (2015) 94–102 101
Cook, B.I., Terando, A., Steiner, A., 2010. Ecological forecasting under climatic datauncertainty: a case study in phenological modeling. Environ. Res. Lett. 5, 44014.
Cooke, J.E.K., Eriksson, M.E., Junttila, O., 2012. The dynamic nature of bud dormancyin trees: environmental control and molecular mechanisms. Plant Cell Environ.35, 1707–1728, http://dx.doi.org/10.1111/j.1365-3040.2012.02552.x
Crimmins, S.M., Dobrowski, S.Z., Greenberg, J.a, Abatzoglou, J.T., Mynsberge, A.R.,2011. Changes in climatic water balance drive downhill shifts in plant species’optimum elevations. Science 331, 324–327, http://dx.doi.org/10.1126/science.1199040
Deutschman, D.H., Levin, S.A., Devine, C., Buttel, L.A., 1997. Scaling from trees toforests: analysis of a complex simulation model. Science 277 (80), 1684, http://dx.doi.org/10.1126/science.277.5332.1684b
Dobrowski, S.Z., Abatzoglou, J., Swanson, A.K., Greenberg, J.A., Mynsberge, A.R.,Holden, Z.A., Schwartz, M.K., 2013. The climate velocity of the contiguous UnitedStates during the 20th century. Glob. Change Biol. 19, 241–251, http://dx.doi.org/10.1111/gcb.12026
Dolanc, C.R., Thorne, J.H., Safford, H.D., 2013. Widespread shifts in the demographicstructure of subalpine forests in the Sierra Nevada, California, 1934 to 2007.Glob. Ecol. Biogeogr. 22, 264–276, http://dx.doi.org/10.1111/j.1466-8238.2011.00748.x
Donohue, K., Rubio de Casas, R., Burghardt, L., Kovach, K., Willis, C.G., 2010. Ger-mination postgermination adaptation, and species ecological ranges. Annu.Rev. Ecol. Evol. Syst. 41, 293–319, http://dx.doi.org/10.1146/annurev-ecolsys-102209-144715
Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G.,Gruber, B., Lafourcade, B., Leitão, P.J., Münkemüller, T., McClean, C., Osborne,P.E., Reineking, B., Schröder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2013.Collinearity: a review of methods to deal with it and a simulation study evaluat-ing their performance. Ecography (Cop.) 36, 27–46, http://dx.doi.org/10.1111/j.1600-0587.2012.07348.x
Duputié, A., Rutschmann, A., Ronce, O., Chuine, I., 2015. Phenological plasticity willnot help all species adapt to climate change. Glob. Change Biol., http://dx.doi.org/10.1111/gcb.12914
Eriksson, O., 2002. Ontogenetic niche shifts and their implications for recruitmentin three clonal Vaccinium shrubs: Vaccinium myrtillus, Vaccinium vitis-idaea, andVaccinium oxycoccos. Can. J. Bot. 80, 635–641, http://dx.doi.org/10.1139/b02-044
Fridley, J.D., 2012. Extended leaf phenology and the autumn niche in deciduousforest invasions. Nature 485, 359–362, http://dx.doi.org/10.1038/nature11056
Grömping, U., 2006. Relative importance for linear regression in R: the packagerelaimpo. J. Stat. Softw., 17.
Honaker, J., King, G., Blackwell, M., 2011. Amelia II: a program for missing data. J.Stat. Softw., 45.
Intergovernmental Panel on Climate Change, 2014. Climate Change 2013: The Phys-ical Science Basis.
Kamae, Y., Shiogama, H., Watanabe, M., Kimoto, M., 2014. Attributing the increase inNorthern Hemisphere hot summers since the late 20th century. Geophys. Res.Lett., http://dx.doi.org/10.1002/2014GL061062
Kurz, W.A., Dymond, C.C., Stinson, G., Rampley, G.J., Neilson, E.T., Carroll, A.L., Ebata,T., Safranyik, L., 2008. Mountain pine beetle and forest carbon feedback to cli-mate change. Nature 452, 987–990.
Lenoir, J., Gégout, J.-C., Pierrat, J.-C., Bontemps, J.-D., Dhôte, J.-F., 2009. Differencesbetween tree species seedling and adult altitudinal distribution in mountainforests during the recent warm period (1986–2006). Ecography (Cop.) 32,765–777, http://dx.doi.org/10.1111/j.1600-0587.2009.05791.x
Levins, R., 1966. The strategy of model building in population biology. Am. Sci. 54,421–431.
Loehle, C., LeBlanc, D., 1996. Model-based assessments of climate change effects onforests: a critical review. Ecol. Modell. 90, 1–31, http://dx.doi.org/10.1016/0304-3800(96)83709-4
Lounici, K., eprint arXiv:1201.2577.Luo, Y., Chen, H.Y.H., 2013. Observations from old forests underestimate climate
change effects on tree mortality. Nat. Commun. 4, 1655.Magnani, F., Mencuccini, M., Borghetti, M., Berbigier, P., Berninger, F., Delzon, S.,
Grelle, A., Hari, P., Jarvis, P.G., Kolari, P., Kowalski, A.S., Lankreijer, H., Law,B.E., Lindroth, A., Loustau, D., Manca, G., Moncrieff, J.B., Rayment, M., Tedeschi,V., Valentini, R., Grace, J., 2007. The human footprint in the carbon cycle oftemperate and boreal forests. Nature 447, 848–850, http://dx.doi.org/10.1038/nature05847
Matzke, M.A., Mosher, R.A., 2014. RNA-directed DNA methylation: an epigeneticpathway of increasing complexity. Nat. Rev. Genet. 15, 394–408.
Menne, M.J., Durre, I., Vose, R.S., Gleason, B.E., Houston, T.G., 2012. An overview of theglobal historical climatology network-daily database. J. Atmos. Ocean. Technol.29, 897–910, http://dx.doi.org/10.1175/JTECH-D-11-00103.1
ok, H.-F., Arndt, S.K., Nitschke, C.R., 2012. Modelling the potential impact of climatevariability and change on species regeneration potential in the temperate forestsof South-Eastern Australia. Glob. Change Biol. 18, 1053–1072, http://dx.doi.org/10.1111/j.1365-2486.2011.02591.x
orin, X., Lechowicz, M.J., Augspurger, C., O’Keefe, J., Viner, D., Chuine, I., 2009. Leafphenology in 22 North American tree species during the 21st century. Glob.Change Biol. 15, 961–975, http://dx.doi.org/10.1111/j.1365-2486.2008.01735.x
ational Forestry Database, 2013. Silviculture Statistics by Province/Territory1975–2011. Ottawa, Ontario.
atural Regions Committee, 2006. Natural Regions and Subregions of Alberta.Edmonton, Alberta.
iinemets, Ü., 2010. Responses of forest trees to single and multiple environmentalstresses from seedlings to mature plants: past stress history, stress interactions,tolerance and acclimation. For. Ecol. Manage. 260, 1623–1639, http://dx.doi.org/10.1016/j.foreco.2010.07.054
itschke, C.R., Innes, J.L., 2008. A tree and climate assessment tool for modellingecosystem response to climate change. Ecol. Modell. 210, 263–277, http://dx.doi.org/10.1016/j.ecolmodel.2007.07.026
itschke, C.R., Amoroso, M., Coates, K.D., Astrup, R., 2012. The influence of cli-mate change, site type, and disturbance on stand dynamics in northwest BritishColumbia, Canada. Ecosphere 3, art11, http://dx.doi.org/10.1890/ES11-00282.1
ark Williams, A., Allen, C.D., Macalady, A.K., Griffin, D., Woodhouse, C.A., Meko,D.M., Swetnam, T.W., Rauscher, S.A., Seager, R., Grissino-Mayer, H.D., Dean, J.S.,Cook, E.R., Gangodagamage, C., Cai, M., McDowell, N.G., 2013. Temperature asa potent driver of regional forest drought stress and tree mortality. Nat. Clim.Change 3, 292–297.
eng, C., Ma, Z., Lei, X., Zhu, Q., Chen, H., Wang, W., Liu, S., Li, W., Fang, X., Zhou, X.,2011. A drought-induced pervasive increase in tree mortality across Canada’sboreal forests. Nat. Clim. Change 1, 467–471.
iedallu, C., Gégout, J.-C., Perez, V., Lebourgeois, F., 2013. Soil water balance performsbetter than climatic water variables in tree species distribution modelling. Glob.Ecol. Biogeogr. 22, 470–482, http://dx.doi.org/10.1111/geb.12012
rice, D., Zimmermann, N., van der Meer, P., Lexer, M., Leadley, P., Jorritsma, I.M.,Schaber, J., Clark, D., Lasch, P., McNulty, S., Wu, J., Smith, B., 2001. Regenerationin gap models: priority issues for studying forest responses to climate change.Clim. Change 51, 475–508, http://dx.doi.org/10.1023/A:1012579107129
Core Team, 2014. R: A Language for Statistical Computing.ichardson, A.D., Keenan, T.F., Migliavacca, M., Ryu, Y., Sonnentag, O., Toomey, M.,
2013. Climate change, phenology, and phenological control of vegetation feed-backs to the climate system. Agric. For. Meteorol. 169, 156–173, http://dx.doi.org/10.1016/j.agrformet.2012.09.012
icketts, T.H., 1999. Terrestrial Ecoregions of North America: A Conservation Assess-ment. Island Press, Washington, DC.
chumacher, S., Bugmann, H., Mladenoff, D.J., 2004. Improving the formulation oftree growth and succession in a spatially explicit landscape model. Ecol. Modell.180, 175–194, http://dx.doi.org/10.1016/j.ecolmodel.2003.12.055
elling 313 (2015) 94–102
Shugart, H.H., 1984. A Theory of Forest Dynamics: The Ecological Implications ofForest Succession Models. Springer-Verlag, New York.
Shugart, H.H., West, D.C., 1977. Development of an Appalachian deciduous forestsuccession model and its application to assessment of the impact of the ChestnutBlight. J. Environ. Manage.
Soil Classification Working Group, 1998. The Canadian System of Soil Classification:Third Edition.
Soil Landscapes of Canada Working Group, 2010. Soil Landscapes of Canada Version3.2., pp. 2.
Trenberth, K.E., 2011. Changes in precipitation with climate change. Clim. Res. 47,123–138.
Trenberth, K.E., Dai, A., van der Schrier, G., Jones, P.D., Barichivich, J., Briffa, K.R.,Sheffield, J., 2014. Global warming and changes in drought. Nat. Clim. Change 4,17–22.
Urbieta, I.R., García, L.V., Zavala, M.A., Maranón, T., 2011. Mediterranean pine andoak distribution in southern Spain: IWs there a mismatch between regenerationand adult distribution? J. Veg. Sci. 22, 18–31, http://dx.doi.org/10.1111/j.1654-1103.2010.01222.x
Van Genuchten, M.T., 1980. A closed-form equation for predicting the hydraulicconductivity of unsaturated soils.
Vitasse, Y., Hoch, G., Randin, C., Lenz, A., Kollas, C., Scheepens, J.F., Körner, C., 2013.Elevational adaptation and plasticity in seedling phenology of temperate decid-uous tree species. Oecologia 171, 663–678, http://dx.doi.org/10.1007/s00442-012-2580-9
Walck, J.L., Hidayati, S.N., Dixon, K.W., Thompson, K., Poschlod, P., 2011. Climatechange and plant regeneration from seed. Glob. Change Biol. 17, 2145–2161,http://dx.doi.org/10.1111/j.1365-2486.2010.02368.x
Wentz, F.J., Ricciardulli, L., Hilburn, K., Mears, C., 2007. How much more rain willglobal warming bring? Sci. 317, 233–235, http://dx.doi.org/10.1126/science.1140746
Woodall, C.W., Zhu, K., Westfall, J.A., Oswalt, C.M., D’Amato, A.W., Walters, B.F., Lintz,H.E., 2013. Assessing the stability of tree ranges and influence of disturbance ineastern US forests. For. Ecol. Manage. 291, 172–180, http://dx.doi.org/10.1016/j.foreco.2012.11.047
Zhang, J., Huang, S., He, F., 2015. Half-century evidence from western Canada showsforest dynamics are primarily driven by competition followed by climate. Proc.Natl. Acad. Sci. 112, 4009–4014, http://dx.doi.org/10.1073/pnas.1420844112
Zhu, K., Woodall, C.W., Clark, J.S., 2012. Failure to migrate: lack of tree range expan-
climate change: forest migration and turnover through life history. Glob. ChangeBiol. 20, 251–264, http://dx.doi.org/10.1111/gcb.12382.
