Taxonomy, together with ontogeny and growing conditions, drives needleleaf species’ sensitivity to climate in boreal North America

Abstract

Currently, there is no consensus regarding the way that changes in climate will affect boreal forest growth, where warming is occurring faster than in other biomes. Some studies suggest negative effects due to drought‐induced stresses, while others provide evidence of increased growth rates due to a longer growing season. Studies focusing upon the effects of environmental conditions on growth‐climate relationships are usually limited to small sampling areas that do not encompass the full range of environmental conditions; therefore, they only provide a limited understanding of the processes at play. Here, we studied how environmental conditions and ontogeny modulated growth trends and growth‐climate relationships of black spruce (Picea mariana) and jack pine (Pinus banksiana) using an extensive data set from a forest inventory network. We quantified the long‐term growth trends at the stand scale, based upon analysis of the absolutely‐dated ring‐width measurements of 2266 trees. We assessed the relationship between annual growth rates and seasonal climatic variables, and evaluated the effects of various explanatory variables on long‐term growth trends and growth‐climate relationships. Both growth trends and growth‐climate relationships were species‐specific and spatially heterogeneous. While the growth of jack pine barely increased during the study period, we observed a growth decline for black spruce which was more pronounced for older stands. This decline was likely due to a negative balance between direct growth gains induced by improved photosynthesis during hotter‐than‐average growing conditions in early summers and the loss of growth occurring the following year due to the indirect effects of late‐summer heatwaves on accumulation of carbon reserves. For stands at the high end of our elevational gradient, frost damage during milder‐than‐average springs could act as an additional growth stressor. Competition and soil conditions also modified climate sensitivity, which suggests that effects of climate change will be highly heterogeneous across the boreal biome. This article is protected by copyright. All rights reserved.

Full text

DRIVERS OF BOREAL FOREST GROWTH 1
Taxonomy, together with ontogeny and growing conditions, drives needleleaf
1
species’ sensitivity to climate in boreal North America
2
Running head: DRIVERS OF BOREAL FOREST GROWTH
3
Marchand, W.1, 2, Girardin, M.P. 1, 2, Hartmann, H.3, Gauthier, S.1, 2, Bergeron, Y.2
4
1. Natural Resources Canada, Canadian Forest Service, Laurentian Forestry Centre,
5
1055 du P.E.P.S, P.O. Box 10380, Stn. Sainte-Foy, Québec, QC, G1V 4C7,
6
Canada
7
2. Centre d’étude de la forêt, Université du Québec à Montréal, C.P. 8888, succ.
8
Centre-ville, Montréal, QC, H3C 3P8, Canada; and Forest Research Institute,
9
Université du Québec en Abitibi-Témiscamingue, 445 boul. de l’Université,
10
Rouyn Noranda , QC, Canada, J9X 5E4
11
3. Max-Planck Institute for Biogeochemistry, Department of Biogeochemical
12
Processes, Hans-Knöll Str. 10, 07745 Jena, Germany
13
Corresponding author: William Marchand. Email: william.marchand@uqat.ca;
14
Tel.: (1) 418-809-4690
15
16
Paper type: Primary research article
17

DRIVERS OF BOREAL FOREST GROWTH 2
Abstract
18
Currently, there is no consensus regarding the way that changes in climate will affect boreal
19
forest growth, where warming is occurring faster than in other biomes. Some studies
20
suggest negative effects due to drought-induced stresses, while others provide evidence of
21
increased growth rates due to a longer growing season. Studies focusing upon the effects
22
of environmental conditions on growth-climate relationships are usually limited to small
23
sampling areas that do not encompass the full range of environmental conditions; therefore,
24
they only provide a limited understanding of the processes at play. Here, we studied how
25
environmental conditions and ontogeny modulated growth trends and growth-climate
26
relationships of black spruce (Picea mariana) and jack pine (Pinus banksiana) using an
27
extensive data set from a forest inventory network. We quantified the long-term growth
28
trends at the stand scale, based upon analysis of the absolutely-dated ring-width
29
measurements of 2266 trees. We assessed the relationship between annual growth rates and
30
seasonal climatic variables, and evaluated the effects of various explanatory variables on
31
long-term growth trends and growth-climate relationships. Both growth trends and growth-
32
climate relationships were species-specific and spatially heterogeneous. While the growth
33
of jack pine barely increased during the study period, we observed a growth decline for
34
black spruce which was more pronounced for older stands. This decline was likely due to
35
a negative balance between direct growth gains induced by improved photosynthesis
36
during hotter-than-average growing conditions in early summers and the loss of growth
37
occurring the following year due to the indirect effects of late-summer heatwaves on
38
accumulation of carbon reserves. For stands at the high end of our elevational gradient,
39
frost damage during milder-than-average springs could act as an additional growth stressor.
40

DRIVERS OF BOREAL FOREST GROWTH 3
Competition and soil conditions also modified climate sensitivity, which suggests that
41
effects of climate change will be highly heterogeneous across the boreal biome.
42
Keywords: boreal forest, Canada, climate change, climate-induced stress, dendroecology,
43
Quebec
44
Graphical Abstract
45
The figure displays the effects (red=negative, blue=positive) of explanatory variables on
46
tree sensitivity to climate, and the resulting 1970-2005 growth trends. Old-growth boreal
47
black spruce stands exhibited a more negative response to previous summer temperature,
48
identified as the primary climatic driver of growth trajectories for this species. This finding
49
suggests an exacerbated effect of heat-induced stresses, which resulted in more negative
50
long-term growth trends for old-growth stands, especially when combined with late-frost
51
damage. Other explanatory variables, such as regional climate, competition and soil
52
conditions, modified tree sensitivity to climate.
53

DRIVERS OF BOREAL FOREST GROWTH 4
1. INTRODUCTION
54
The boreal biome is warming faster than other regions of the world (IPCC, 2013). As a
55
result of a 35 % increase in atmospheric CO2 concentrations relative to pre-industrial
56
conditions, mean annual air temperature has risen by 0.5 to 3 °C in boreal North America
57
and an additional increase of 4-5 °C is expected by 2100 (Price et al., 2013). Climate
58
change threatens the ecological, social and economic services that boreal forests provide
59
(Gauthier, Bernier, Kuuluvainen, Shvidenko, & Schepaschenko, 2015). It is also unclear
60
whether boreal forests will continue to act as a carbon sink or will shift to become a carbon
61
source (Kurz et al., 2013), thereby renewing the scientific focus on these ecosystems and
62
on their ability to cope with, and to mitigate, global warming. As a proxy for tree vigour,
63
secondary growth can be used to study the response of trees to a changing climate and,
64
thus, to assess current and future trajectories of the boreal forest.
65
In the Northern Hemisphere, evidence of increased mortality rates and decreases in tree
66
growth and forest productivity is accumulating, not only for the interior of the boreal forest
67
(Cahoon et al., 2018; Girardin et al., 2016; Hember, Kurz, & Coops, 2016; Zhu et al., 2016),
68
but also at its northern edge (D’Arrigo et al., 2004). These ‘negative’ trends were linked,
69
amongst other factors, to heat and hydric stresses resulting from rising temperatures and
70
more frequent, longer-lasting, and harsher drought episodes (Barber, Juday, & Finney,
71
2000; Girardin et al., 2016; Juday & Alix, 2012; Nicklen et al., 2018; Trugman, Medvigy,
72
Anderegg, & Pacala, 2018; Zhang et al., 2008). In contrast, other studies provided strong
73
evidence for increased growth rates and higher stand productivity (Beck et al., 2011;
74
Girardin et al., 2011; Hember, Kurz, & Coops, 2017). These ‘positive’ trends, which were
75
observed mainly for the northernmost forested area, namely, the taiga, were likely due to
76

DRIVERS OF BOREAL FOREST GROWTH 5
higher rates of carbon assimilation and a longer growing season (Gennaretti, Arseneault,
77
Nicault, Perreault, & Bégin, 2014; Ju & Masek, 2016). These contrasting observations
78
demonstrate uncertainties regarding the persistence of the existing structure, composition
79
and function of the boreal biome under future warmer and dryer climatic conditions.
80
Tree sensitivity to climate is highly variable across the globe and is modulated by
81
environmental drivers that vary at local to global scales (Babst, Poulter, et al., 2012;
82
Girardin et al., 2016). Amongst these drivers, topography creates spatially heterogeneous
83
macroclimatic conditions and can modify the way that trees respond to changes in regional
84
climate (Hasler, Geertsema, Foord, Gruber, & Noetzli, 2015; Matías, Linares, Sánchez-
85
Miranda, & Jump, 2017). For example, in Central Europe, water-limited lowland forests
86
are more sensitive to drought, whereas forests in the upland portion of the elevational
87
gradient are primarily temperature-limited (Altman et al., 2017) and can benefit from
88
stronger and faster warming, which is expected for mountainous areas (Pepin et al., 2015).
89
More specifically, higher mean summer temperatures could improve the growth of trees at
90
the high end of the elevational gradient by increasing the number of wood cells that are
91
produced annually through improved xylogenetic processes and hydraulic performance
92
(Castagneri, Petit, & Carrer, 2015; Dulamsuren, Hauck, Kopp, Ruff, & Leuschner, 2017).
93
In contrast, some studies have observed decreased growth rates, even for trees growing in
94
mountainous sites in both central Europe and North America (Dittmar, Zech, & Elling,
95
2003; McLaughlin, Downing, Blasing, Cook, & Adams, 1987; Piovesan, Biondi, Filippo,
96
Alessandrini, & Maugeri, 2008), which questions the capacity of high-elevation forested
97
ecosystems to better cope with climate change (Austin & Niel, 2011).
98

DRIVERS OF BOREAL FOREST GROWTH 6
The annual growth performance of a tree is linked to its ability to access optimal
99
amounts of water, light and nutrients (Fritts, 1971), the availability of which is primarily
100
controlled by site-specific abiotic factors, such as soil conditions (e.g., Hember et al.,
101
2017). Soil structure, drainage and thickness of the organic layer determine soil water-
102
holding capacity (Minasny & McBratney, 2017) and drive nutrient cycling (e.g., Cavard,
103
Bergeron, Paré, Nilsson, & Wardle, 2018). In addition to its direct effects on tree growth,
104
soil moisture content alters microclimate and, thus, evapotranspiration rates, which may
105
modulate growth-climate relationships (Manrique-Alba et al., 2017). By modifying
106
resource availability, inter-individual competition can exacerbate tree sensitivity to harsh
107
climatic conditions (e.g. Buechling, Martin, & Canham, 2017; Ford et al., 2016; Gleason
108
et al., 2017; Jiang et al., 2018; Nicklen et al., 2018), or buffer growth gains from favourable
109
periods (Cortini, Comeau, & Bokalo, 2012). Ultimately, the capacity of a tree to efficiently
110
use resources will also dictate its response to climate (e.g., Carrer & Urbinati, 2004). Apart
111
from genotype-driven differences, ontogeny-related changes in a tree’s physiological needs
112
and in the efficiency of its hydraulic network (Ryan, Phillips, & Bond, 2006) can modify
113
its sensitivity to climate (e.g., Altman et al., 2017).
114
The high spatial variability in growing conditions that is encountered in boreal forests,
115
together with the multiplicity of interacting effects and feedbacks of environmental
116
variables that are present, hinder our understanding of the response of boreal forest trees to
117
climate. In regions with geographically limited and sparsely replicated sample networks
118
(Gewehr, Drobyshev, Berninger, & Bergeron, 2014), assessing climate effects on tree
119
growth is very difficult (but see Girardin et al., 2016), given that field-based measurements
120
do not cover the full range of variation in growing conditions. Some studies in western
121

DRIVERS OF BOREAL FOREST GROWTH 7
boreal North America and boreal Europe have examined variations in growth-climate
122
relationships along latitudinal and longitudinal gradients (Lloyd, Bunn, & Berner, 2011) or
123
between sites with different slope orientations (i.e. north vs south facing sites; Johnstone,
124
McIntire, Pedersen, King, & Pisaric, 2010; Walker & Johnstone, 2014) and moisture
125
conditions (Walker & Johnstone, 2014; Wilmking & Myers-Smith, 2008). However,
126
studies testing the effect of elevation gradient on the trees sensitivity to climate are lacking,
127
particularly in the eastern boreal North America. Furthermore, most past studies have
128
focused upon the direct effects of abiotic or biotic factors on tree growth, while the
129
feedback effects of environmental conditions on growth-climate relationships are still
130
rarely decribed (But see Nicklen et al., 2018; and Nicklen, Roland, Ruess, Schmidt, &
131
Lloyd, 2016 for the Pacific Coast of North America).
132
Here, we used an extensive and well-replicated provincial inventory network that
133
provides absolutely dated and annually resolved tree-growth data, as well as site-specific
134
environmental information for unmanaged forests in eastern boreal North America. This
135
network is located at the boundary between the interior boreal forest and the taiga, and
136
includes sample plots characterized by highly contrasting growing conditions. Our overall
137
objective was to examine whether the potential impacts of recent changes in climate varied
138
as functions of explanatory variables with respect to the growth of two needleleaf species
139
that are broadly distributed across North America, black spruce (Picea mariana (Miller)
140
B.S.P.) and jack pine (Pinus banksiana Lambert). We first quantified the recent growth
141
trends for the two species which, given the high variability in growing conditions, were
142
expected to be heterogeneous across the study zone. Then, we determined the climate
143
sensitivity of the two species, i.e., the relationship between inter-annual variation of
144

DRIVERS OF BOREAL FOREST GROWTH 8
secondary growth rates and fluctuations in seasonal values of mean temperature and total
145
precipitation over the period 1970-2005. We hypothesized that the growth of both species
146
would be negatively impacted by higher-than-average temperature during summer and
147
positively affected by higher-than-average temperature during spring and by higher-than-
148
average precipitation during summer. Finally, we assessed how explanatory variables (e.g.
149
climate, competition and soil conditions) affected spatial variability in growth-climate
150
relationships. We hypothesized that the negative effect of hotter- and dryer-than-average
151
summers, as well as the positive effect of high spring temperature on tree growth, would
152
be exacerbated in stands in the upper portion of the elevational gradient. We also
153
hypothesised that old stands, as well as trees growing in a highly competitive environment
154
and in well-drained sites, would respond more negatively to summer heatwaves.
155
2. MATERIALS AND METHODS
156
2.1 Sampling area
157
Our sampling network covered three degrees of latitude (50.25-53.25°N) and nearly
158
extended across the entire Province of Quebec (Canada) from east to west (57.5-78.25°W).
159
It was located in the boreal biome, which is characterised by needleleaf-dominated forests
160
(Robitaille, Saucier, Chabot, Côté, & Boudreault, 2015). Some regional patterns of climatic
161
conditions, dominant vegetation and natural disturbance regimes make it possible to divide
162
this wide biome into bioclimatic domains (Ansseau et al., 1997). In the north portion of the
163
region, the spruce-lichen bioclimatic domain is characterised by a harsh, cold and dry
164
climate, resulting in an open black spruce-dominated forest with a lichen mat, i.e., the taiga
165
vegetation subzone. South of the 52nd parallel, continuous boreal forest that is composed
166
mostly of pure black spruce stands covers the spruce-moss bioclimatic domain. The latter
167

DRIVERS OF BOREAL FOREST GROWTH 9
is subdivided into western and eastern zones based on precipitation patterns and fire cycles.
168
The western part is drier and, consequently, more prone to wildfire than the eastern zone
169
(Gouvernement du Québec, 2003). Within these three main bioclimatic domains, hereafter
170
referred to as “Boreal West,” “Boreal East” and “Taiga” (Figure 1), lower-level landscape
171
units are defined based upon the recurrent arrangements of the main permanent ecological
172
and vegetation features (48 landscape units are present in our sampling area), which in turn
173
are divided into ecological districts (284 ecological districts within our sampling area) that
174
are based upon their geological and physiographic features (Ansseau et al., 1997). Please
175
refer to the Figure 1 (B) for examples of geographical units mentioned throughout the
176
paper.
177

DRIVERS OF BOREAL FOREST GROWTH 10
FIGURE 1 (A) Forest inventory plot network. The pink squares and blue triangles
178
represent black spruce and jack pine temporary sample plots, respectively. The three main
179
bioclimatic domains encompassing the sample network are also delineated. The
180
background colour gradient represents the elevation gradient. (B) Geographical units
181
involved in statistical analyses, from the broader global scale of the province of Quebec to
182
the finer scale of the sample plot.
183
2.2 Tree-ring material
184

DRIVERS OF BOREAL FOREST GROWTH 11
The data that we used for this study were acquired from a sampling program of 400-m2
185
randomly distributed temporary circular sample plots (n = 875 plots), which was
186
established by the Ministère des Ressources naturelles et de la Faune du Québec (MRNFQ)
187
from 2005 to 2009 (Programme d’inventaire écoforestier nordique; Létourneau et al.,
188
2008). In each sample plot, the diameter at breast height (DBH, 1.3 m) of all living and
189
dead stems (DBH > 9 cm) was measured and environmental and stand-level conditions
190
were recorded. Disks were collected for stem analysis from one to three dominant or co-
191
dominant trees per species according to the provincial normative sampling protocol
192
(Ministère des Ressources Naturelles du Québec, 2008). We used only black spruce and
193
jack pine data since these species represented most (76 % and 15 %, respectively) of the
194
sampled trees. We selected 1-m-height stem-disks as a trade-off between basal ring
195
distortion and the number of visible rings (DesRochers & Gagnon, 1997). A total of 1914
196
black spruce and 352 jack pine disks with each having a minimum of 20 visible rings,
197
representing 812 sample plots, were retained for subsequent analyses. Sample disks were
198
processed using standard dendrochronological procedures for acquisition of ring-width
199
measurement series across four radii per disk (Ministère des Ressources Naturelles du
200
Québec, 2008). For each ring-width series, cross-dating and measurements were
201
statistically verified using the program COFECHA (Holmes, 1983). No major anomaly in
202
these tree-ring measurements was observed, and therefore all were retained for subsequent
203
analyses.
204
2.3 Climate data and explanatory variables
205
For each plot, daily maximum and minimum temperatures (°C) and total precipitation
206
(mm) were obtained for the period of 1970-2005 using thin plate spline smoothing
207

DRIVERS OF BOREAL FOREST GROWTH 12
algorithms (ANUSPLIN), which interpolate site-specific estimates at a 0.08° x 0.08°
208
spatial resolution from historical weather observations, as described in Price et al. (2011).
209
Mean temperatures, which were computed as the average of monthly maximum and
210
minimum temperature values, were averaged and precipitation was summed at a seasonal
211
scale (meteorological seasons according to Trenberth, 1983: spring = March-May, summer
212
= June-August, autumn = September-November, winter = December-February). Readers
213
should refer to Supporting Information S1 for an overview of the trends in seasonal climate
214
in the study area.
215
Stand and environmental conditions were acquired from the plot survey conducted
216
by the Ministère de la Faune, des Forêts et des Parcs du Québec (Table 1, MFFPQ;
217
Robitaille et al., 2015). We considered the proportion of clay, sand and silt in the mineral
218
soil, organic layer thickness (OLT) and hydrological conditions of the sample plot assessed
219
as drainage classes. Elevation and slope were extracted for our sample plots from the
220
SRTM 90m Digital Elevation Database v4.1 (Jarvis, Reuter, Nelson, & Guevara, 2008).
221
For stand-level demographic features, stand age was defined as the age of the oldest
222
sampled tree in the plot. Stand basal area (BA) was computed as the sum of basal areas of
223
all trees with DBH > 9 cm within the plot, on a per-hectare basis. A tree-level competition
224
index (CI) was computed as the number of trees that were taller than the focal tree, divided
225
by the total number of trees within the plot, to assess assymetric competition (Ford et al.,
226
2016), following Weber et al. (2008). To do so, the height of all trees within a plot was
227
estimated from DBH using the allometric equations of Fortin et al. (2009). Individual CI
228
values were averaged at the plot level to ensure consistency with the working scale of the
229
growth-climate analyses. Temperature and precipitation normals were computed for the
230

DRIVERS OF BOREAL FOREST GROWTH 13
1970-2005 period to account for the west-to-east (continental-to-oceanic) climate gradient.
231
For brevity’s sake, these plot-level characteristics will be refered hereafter to as
232
“explanatory variables”.
233
TABLE 1 Plot-level statistics for the studied explanatory variables, by bioclimatic domain
234
Boreal West
Boreal East
Taiga
mean
sd
range (min | max)
mean
sd
range (min | max)
mean
sd
range (min | max)
Clay (%)
6.69
13.32
0 | 79
4.76
2.86
0 | 18
4.96
6.34
0 | 47.9
Sand (%)
63.51
30.71
0 | 99.5
66.77
24.13
0 | 99.5
68.55
23.94
0 | 96.5
Silt (%)
12.11
11.95
0 | 52
18.75
11.31
0 | 53.9
17.93
12.29
0 | 72
OLT (cm)
21.23
25.45
1 | > 100
18.49
17.35
0 | > 100
15.93
19.66
0 | > 100
Drainage
(unitless)
3 (median
class)
-
1 | 6
3 (median
class)
-
1 | 6
3 (median
class)
-
1 | 6
Elevation (m
a.s.l.)
320.22
89.84
96 | 637
549.55
167.39
100 | 860
522.59
171.4
113 | 912
Slope (degree)
2.28
1.97
0.13 | 13.15
3.55
3.33
0.13 | 19.58
2.32
2.29
0.13 | 12.36
Age (years)
105.23
55.82
28 | 294
163.69
65.81
28 | 331
145.59
63.68
30 | 309
BA (m2 ha-1)
15.89
10.59
0.78 | 49.88
17.27
10.21
0.90 | 55.39
10.57
6.60
0.59 | 35.44
CI (unitless)
0.77
0.19
0.07 | 1.00
0.79
0.16
0.02 | 0.98
0.69
0.21
0.07 | 1
Prec. (mm)
807.11
59.07
685.56 | 927.26
956.09
106.87
775.83 | 1174.67
803.74
75.3
668.60 |955.34
Temp. (°C)
-1.71
0.61
-3.07 | -0.31
-2.21
1.28
-4.02 | 0.49
-2.28
1.03
-4.35 | -0.93
Note: Clay = percentage of clay within the soil; Sand = percentage of sand within the soil; Silt = percentage of silt within the soil; OLT = organic layer thickness;
Drainage = drainage classes: from 1: rapid drainage to 6: poor drainage ; Elevation = altitudinal gradient; Slope = terrain’s slope, in degrees; Age = stand age (age
of the oldest tree in a plot, computed as the number of years between the calendar year of the oldest ring and the calendar year of the most recent ring recorded for
a tree); BA = basal area; CI = competition index; Prec. = average annual precipitation over the 1970-2005 period; Temp. = average mean annual temperature over
the 1970-2005 period.
235

DRIVERS OF BOREAL FOREST GROWTH 14
2.4 Statistical procedures
236
To test our working hypotheses, we applied a 3-step statistical procedure involving
237
different spatio-temporal scales (see workflow diagram in Supporting Information S2).
238
Step 1 : Trend analysis
239
Ring-width measurements of the four radii were averaged (arithmetic mean statistics, see
240
Supporting Information S3 for descriptive statistics of the raw series), and the mean ring-
241
width series were converted into basal area increments (

) using the
242
function bai.out in the R-package dplr (Bunn, 2008). We assumed the cross-sections were
243
perfectly circular in shape, and used these as a proxy for secondary growth to provide an
244
accurate quantification of wood production with ever-increasing tree diameter (Biondi &
245
Qeadan, 2008). Rings that were formed during the first 10 years were then eliminated,
246
given that they usually exhibit an atypical response to environmental drivers compared
247
with more mature rings (Loader, McCarroll, Gagen, Robertson, & Jalkanen, 2007). Next,
248
BAI were detrended using Generalised Additive Mixed Models (GAMM) to remove the
249
remaining ontogeny-induced (i.e., tree age and size) trends. One model was constructed for
250
each species and ecological district (See Supporting Information S4 for information about
251
the BAI chronologies and diagnostic plots of the GAMM models). Organic layer thickness
252
was added as a fixed term to account for the spatially-heterogeneous and mostly time-
253
independent effect of site quality on tree growth (Lavoie, Harper, Paré, & Bergeron, 2007).
254
BAI values were log-transformed to improve the normality of their distributions. The
255
structure of the GAMM model is as follows:
256
     
257

DRIVERS OF BOREAL FOREST GROWTH 15
where i represents the individual tree, j represents the species, k represents the plot, l
258
represents the ecological district, and t represents the year. BAI is the basal area increment
259
of tree i at specific year t, BA is the basal area of tree i at specific year t (computed as the
260
sum of BAI of previous years), OLT is the organic layer thickness of plot k, and AgeC is
261
the cambial age (1-m height ring count) of tree i at year t. An autoregressive term, AR1
262
(autoregressive order p = 1, moving average order q = 0), was added to account for temporal
263
autocorrelation. We tested the significance of a nested random effect (tree nested in plot)
264
by conducting ANOVAs and likelihood ratio tests. Because it did not improve the model’s
265
fit and led to the same results (data not shown), we discarded the random term of the plot
266
from the final model and kept only the random effect of the tree (TreeID).
267
Annual Growth Changes (GC), which were expressed as the percent deviation from
268
predicted values of the GAMM model, were then computed following Girardin et al.
269
(2016). GC values were aggregated by year, plot and species using the median statistics for
270
computation of GCmedian chronologies (robust statistics; Huber, 2005). Because, for several
271
trees, the 2005 growth-ring was the last whole growth-ring, the upper temporal limit of the
272
analyses was fixed to 2005 to ensure consistency between chronologies. From the GCmedian
273
chronologies, growth trends were examined over two time periods: 1950-2005 and 1970-
274
2005. These periods were marked by significant increases in mean annual temperatures of
275
the area (Price et al., 2013) and characterized by the highest number of tree rings per
276
calendar year (i.e. the highest sample depth, see Supporting Information S3.2). Linear
277
regressions were applied (GCmedian ~ year), and the estimated regression slope was used as
278
a proxy for the long-term growth trend. Trend significance was assessed following the
279
statistical procedure described by Yue et al. (2004). This method corrects the p-value of
280

DRIVERS OF BOREAL FOREST GROWTH 16
the non-parametric Mann-Kendall trend test with the effective sample size of the time
281
series to reduce the influence of serial correlation (function mkTrend in the R-package
282
fume; Santander Meteorology Group, 2012). Even if there were trend reversals for a few
283
plots (Figure 2), the 1970-2005 and 1950-2005 trends were globally highly correlated (see
284
Supporting Information S5.1). For the purposes of comparison, 1970-2005 GAMM-based
285
trends were compared with trends that were estimated from the application of two other
286
commonly used detrending methods, namely modified negative exponential models and
287
regional curve standardisation (See Supporting Information S5.2).
288
Step 2 : Growth-climate relationships
289
Since weather station data availability and, therefore, climate data accuracy, is better for
290
the most recent time periods (Ols, Girardin, Hofgaard, Bergeron, & Drobyshev, 2017), we
291
decided to retain data from the shorter and most recent period, i.e., 1970-2005, for climate-
292
growth analyses. Linear mixed models were fitted by plot and species, which included
293
residuals of GAMM-detrended BAI as response variables, together with the set of
294
seasonally aggregated climatic variables as fixed terms, and tree identity as a random term.
295
Mean seasonal temperature and total precipitation of the year of ring formation were
296
considered as explanatory variables. Since trees can allocate carbohydrates that were
297
acquired in the previous growing season to the biomass production of the year of ring
298
formation (Granda & Camarero, 2017), climate data from summer and autumn of the
299
previous year were also considered as fixed terms, leading to a total of ten climatic
300
variables (please refer to the Supporting Information S6.1 for the list of climate variables
301
used in linear mixed models). The structure of the global model is as follows:
302

DRIVERS OF BOREAL FOREST GROWTH 17
     


303
where i represents the tree, j represents the species, k represents the plot and t represents
304
the year. (TreeID) is a random term that accounts for the variability between individual
305
trees. An error term with an AR1 (p = 1, q = 0) correlation structure was added to the model
306
which accounts for the serial correlation. Collinearity amongst climatic variables was low,
307
with the mean of pairwise Pearson correlations between variables below a stringent
308
threshold value of 0.4 (Supporting Information S6.1; maximum value of |0.37|; Dormann
309
et al., 2013). Multi-model selection based upon the Akaike information criterion corrected
310
for small sample size (AICc), was performed for this global model using the package
311
MuMIn (Bartoń, 2018). A 95% confidence set of models was selected for multi-model
312
inference as models whose cumulative Akaike weight is below 0.95 (Burnham &
313
Anderson, 2002). Readers can consult Supporting Information S6.3 for AICc values of all
314
of the 1024 evaluated models, along with Akaike weights of the best model and the number
315
of models used for multi-model inferences. Weighted averages of parameter estimates were
316
derived from this set of best approximating models, and Student’s t-statistics were
317
computed as the ratio between the average model estimate and its corresponding standard
318
error. These statistics provide both the general direction of the growth-climate relationship
319
(i.e., negative or positive slope), and the strength of this relationship (the farther from zero
320
the t-value is, the stronger is the effect), weighted by the model’s predictive capacity and
321
based upon the selected climatic variables. The 95 % adjusted confidence intervals of the
322
t-statistics were also computed, together with Pearson correlations between residuals from
323
the GAMM models and predicted values from the climate models (Supporting Information
324

DRIVERS OF BOREAL FOREST GROWTH 18
S6.2) as an additional means of assessing the model’s predictive capacity. Results of
325
growth-climate analyses that were based upon residuals from the two additional detrending
326
methods are provided in Supporting Information S7.
327
Slopes from the linear regressions and t-statistics from the climate-growth mixed
328
models were interpolated across the whole area using the Empirical Bayesian Kriging
329
algorithm of the Geostatistical Analyst tool in ArcGIS v.10.4 (input options: empirical
330
transformation of the data, semi-variogram model = exponential-type, search radius = 1°,
331
smoothing factor = 0.2). The output raster maps permitted visual examination of
332
geographical patterns in long-term growth trends and climate sensitivity.
333
Step 3 : Link with explanatory variables
334
The relationships between explanatory variables (listed in Table 1) and tree sensitivity to
335
climate were assessed by conducting redundancy analyses (RDA) using Canoco software
336
v.4.5 (ter Braak & Smilauer, 2009). Because tree sensitivity to climate and environmental
337
conditions are highly variable from east to west (see Figure 4 and Table 1), site conditions
338
might affect growth-climate relationships depending upon the location of the plot (Wu et
339
al., 2018). If averaged over the whole gradient, the effect of these conditions could cancel
340
each other out. Consequently, one RDA was conducted per bioclimatic domain as a trade-
341
off between data aggregation and ecological relevance, as recommended by Ols et al.
342
(2018). The t-statistics from the climate mixed models were considered as response
343
variables (i.e., the “species” data matrix) and explanatory variables were considered as
344
independent variables (i.e., the “environment” matrix). Climate normals were also included
345
as independent variables, together with a dummy variable accounting for the species
346

DRIVERS OF BOREAL FOREST GROWTH 19
identity of the sampled tree, i.e., the difference in sensitivity to climate between jack pine
347
(the reference level) and black spruce. Please refer to Supporting Information S8.3 for the
348
list of independent variables considered in RDA analyses. Latitude, longitude and the
349
average distance to the four nearest weather stations (ranging from 3.8 km to 153.1 km, see
350
Supporting Information S8.1) were added as conditioning variables to remove the effects
351
of spatial non-independence of the plots and the potential lack of accuracy in the climate
352
data set prior to analysis. Independent variables were transformed to improve the normality
353
of their distributions, and then standardised prior to analysis (R package rcompanion;
354
Mangiafico, 2017; Tukey’s ladder of powers; Tukey, 1977). Forward selection of
355
independent variables was done using Monte-Carlo permutation tests (n = 9999
356
permutations under the reduced model; α = 0.05). Growth trends were included passively
357
in the RDA in order to examine these in context with climate-environmental relationships
358
(such supplementary ‘passive’ variables do not influence the ordination). To summarise
359
the information that was displayed by the ordination plots (Supporting Information S8.2),
360
modified t-tests accounting for spatial autocorrelation were conducted between each of the
361
RDA-selected independent variables and response variables (i.e., tree sensitivity to
362
climate). The function modified.ttest of the R package SpatialPack was used (Osorio,
363
Vallejos, Cuevas, & Mancilla, 2018); α = 0.05).
364
Significant variables were grouped into six sets according to the ecological process they
365
represent: stand maturity, competition, altitudinal gradient, soil conditions, regional
366
climate, and species identity (also see Supporting Information S8.3). Variation partitioning
367
was then conducted to identify common and unique contributions to the total percentage
368
of variation in the matrix of response variables (adjusted R2) explained by the model and
369

DRIVERS OF BOREAL FOREST GROWTH 20
better address the question of relative influences of the six sets of independent variables
370
that were considered in the model (Peres-Neto, Legendre, Dray, & Borcard, 2006). The
371
forward selection procedure used in the RDA led to up to five sets of independent variables
372
by bioclimatic domain. The variation partitioning algorithm (varpart) in the R-package
373
vegan was used (9999 permutations, partitions computed from adjusted R2; Oksanen et al.,
374
2018), which only allows a maximum of four sets of independent variables to be considered
375
in a same analysis. To overcome this limitation, we determined the unique and common
376
contributions of stand maturity, competition, altitudinal gradient, soil conditions and
377
regional climate. Next, we determined the contribution of the taxonomic identity of the tree
378
(selected in each of the three bioclimatic domains) by comparing it to the contribution of
379
all other independent variables grouped together. The dummy species variable in RDAs
380
allowed the quantification of the variability in growth-climate relationships resulting from
381
the difference between the two species without splitting the data by species, which would
382
have lowered the number of sample plot by analysis and consequently the statistical power,
383
i.e. the likelihood to correctly reject the null hypothesis. Analyses by species were also
384
tested and results of these analyses are provided as Supporting Information S8.4.
385
3. RESULTS
386
3.1 Growth trends are spatially heterogeneous and species-specific
387
When averaged over the sample plots, dissimilar long-term growth trends appeared
388
between species (Supporting Information S9). Growth rates of black spruce decreased, with
389
a trend estimated at -0.35 % y-1 ± (std) 1.61 % y-1 from 1950 to 2005. For the 1970-2005
390
period, the trend in the annual growth rate was -0.14 % y-1 ± 2.44 % y-1. For jack pine, both
391
the 1950-2005 and 1970-2005 periods were characterised by an annual increase in growth
392

DRIVERS OF BOREAL FOREST GROWTH 21
of 0.21 % y-1 ± 3.31 % y-1 and 0.21 % y-1 ± 3.37 % y-1, respectively. However, species-
393
specific growth trajectories were not homogeneous across the study region (Figures 2 and
394
3; Supporting Information S9). Growth of black spruce increased in the western part of the
395
Boreal West and in the central part of the Boreal East between 1970 and 2005, but
396
decreased elsewhere (Figure 2). Growth of jack pine increased regardless of bioclimatic
397
domain between 1950 and 2005, but decreased in the western parts of Boreal West and
398
Taiga between 1970 and 2005 (Figures 2 and 3).
399
FIGURE 2 (A) Growth trends for black spruce and jack pine, for the 1950-2005 and 1970-
400
2005 periods, shown as slope coefficients of the plot-scaled regression models of detrended
401
BAI values against calendar years. Empirical Bayesian kriging was applied to interpolate
402
plot-based trends across the entire area. Dots highlight significant trends (α = 0.1). The
403

DRIVERS OF BOREAL FOREST GROWTH 22
proportion of significant trends is shown at the bottom of each map. (B) Distributions of
404
growth trend slopes by species and bioclimatic domain (boxplots). Black dots represent the
405
mean value for the specific species and bioclimatic domain. Black lines inside the boxplots
406
are median values, and error bars represent the lower and upper whiskers (representing the
407
variability outside the upper and lower quartiles). The dotted line represents a value of zero,
408
i.e., no trend in long-term growth.
409
FIGURE 3 Median chronologies (red curves) of black spruce (left panels) and jack pine
410
(right panels) detrended BAI (growth change) per bioclimatic domain (upper row: Boreal
411
West; middle row: Boreal East; lower row: Taiga). Yellow shading and dotted lines delimit
412
the bootstrapped 95 % confidence intervals, with LOESS smoothing shown by the blue
413
lines (span = 0.2). Violet box and blue shading highlight the two time intervals (1950-2005
414
and 1970-2005, respectively). Black dashed lines denote a zero effect, i.e., no deviation
415
compared to the value predicted by the GAMM model.
416
3.2 Sensitivity to climate is dissimilar across the landscape
417

DRIVERS OF BOREAL FOREST GROWTH 23
Growth–climate response patterns were estimated for the two tree species to identify the
418
key climate factors that were driving observed variability in growth (Figure 4). Summer
419
temperature of the year preceding growth and spring precipitation in the year of growth
420
had significant negative relationships with black spruce growth, while winter precipitation
421
and winter temperature had a positive influence. The importance of these variables was not
422
limited to particular regions but extended across vast areas (Figure 5). Black spruce tree
423
sensitivity to other climate variables was more spatially heterogeneous (Figures 4 and 5).
424
A high level of precipitation during previous-year summers had a significant positive effect
425
upon the growth of black spruce within the Boreal East and Boreal West; this effect was
426
not statistically significant in the Taiga (Figure 4). Excess-heat and high precipitation
427
during previous-year autumns negatively affected spruce growth in the Boreal West and
428
Taiga but had no significant effect in the Boreal East. Within the Boreal East and Taiga,
429
the growth of black spruce was increased by hotter-than-average summers occurring during
430
the year of ring formation and was decreased by milder-than-average springs. These
431
relationships were mostly the opposite of what was observed within the Boreal West.
432
The response of jack pine to climate was less statistically significant and often
433
opposite to that of black spruce. Regardless of bioclimatic domain, jack pine growth was
434
increased by previous-year warm autumns and current-year summer warmth, but it was
435
decreased by high winter precipitation (Figures 4 and 5). Current- and previous-year wet
436
summers significantly increased the growth of jack pine within the Boreal East and Boreal
437
West (Figure 4). Jack pine growth was positively correlated with mild and wet springs
438
within the Taiga and with mild winters within the Boreal West, but was negatively
439
impacted by wet springs within the Boreal East (Figure 4).
440

DRIVERS OF BOREAL FOREST GROWTH 24
FIGURE 4 Arithmetic means (black dots) and bootstrapped 95 % confidence intervals
441
(rectangles, R=10000 replications) of t-statistic values per bioclimatic domain for black
442
spruce and jack pine, for each of the seasonal climatic variables. “T” and “P” at the
443
beginning of a variable’s name denote temperature and precipitation, respectively.
444
Uppercase letters denote climatic variables for the current growing season (winter, spring,
445

DRIVERS OF BOREAL FOREST GROWTH 25
summer), and lowercase letters denote climatic variables of the previous growing season
446
(previous summer, previous autumn). Blue and red rectangles indicate a significant (95 %
447
confidence interval excluding zeroes) positive and negative effect, respectively, of the
448
climatic variable at the scale of the bioclimatic domain, and grey rectangles are for non-
449
significant values.
450

DRIVERS OF BOREAL FOREST GROWTH 26
451
452
453
454
455
456
457
458
459
FIGURE 5 Kriging-interpolated growth response significance (based on t-statistics, 1970-2005) to seasonal climatic variable (left:
460
temperature, right: precipitation) for black spruce (left panel) and jack pine (right panel). Green-to-blue colours denote a negative effect
461
of the climate variable on tree growth, yellow colour means no impact of the climate variable on tree growth, and orange-to-red colours
462

DRIVERS OF BOREAL FOREST GROWTH 27
denote a positive effect. Dots display significant values, i.e., plots for which the 95 % confidence interval of the t-statistics excludes
463
zero.
464

DRIVERS OF BOREAL FOREST GROWTH 28
3.3 Plot-level features had low but significant effects on growth-climate relationships
465
Sensitivity to climate differed between the two species, especially within the Boreal West,
466
where species identity of the sampled trees alone accounted for 15 % of variation in
467
growth-climate relationships (Figure 6). Such taxononomic variability in growth sensitivity
468
to climate can be readily noted in Figures 4 and 5. Contributions of the sets of explanatory
469
variables stand maturity, competition, altitudinal gradient, soil conditions, and regional
470
climate to the climate sensitivity variance were much lower. The elevational gradient
471
explained the highest proportion of variation in tree response to climate within the Boreal
472
East and Taiga (5 % and 9 %, respectively; Figure 6). Stand maturity, alone or in
473
combination with other explanatory variables, accounted for 7 %, 1 % and 5 % of the
474
variation in growth-climate relationships within the Boreal West, Boreal East and Taiga,
475
respectively. For competition, these values were respectively 3 %, 2 % and 2 %.
476
Stands that were composed mainly of old black spruce trees exhibited growth that
477
was more negatively correlated with previous-year summer and autumn temperatures, but
478
more positively correlated with winter precipitation compared to recently regenerated
479
stands (Figures 6 and 7, and Supporting Information S10.1). These old-grown black spruce
480
stands also exhibited the steepest declines in growth rates during 1970-2005 (Supporting
481
Information S11). The positive effect of warmer-than-average autumns, winter and springs
482
on the growth of jack pine was lower for stands that were composed of old trees in
483
comparison with more recently regenerated stands (Supporting Information S8.4). Snowy
484
and mild winters increased the growth of black spruce more than that of jack pine, but
485
black spruce growth was more negatively correlated with wet and warm springs and with
486

DRIVERS OF BOREAL FOREST GROWTH 29
excess-heat during autumns of the previous years than that of jack pine. Previous-year wet
487
summers and current-year mild springs decreased the growth of stands in the upper portion
488
of the elevational gradient (i.e., above 500 m a.s.l.), while excessively high temperatures
489
during current-year summers increased their growth more strongly than for stands at lower
490
elevations (Figures 6 and 7, and Supporting Information S10.1 and S10.2). Similarly,
491
growth in stands that were composed of taller trees (higher CI) was more negatively
492
affected by excess-heat during previous-year summers than those stands that were
493
composed of smaller-sized trees. Tree growth in more densely populated stands (higher
494
BA) was also more positively correlated with winter temperature, but less positively
495
correlated (within Boreal East and Taiga) or more negatively correlated (within Boreal
496
West) with current-year summer precipitation than stands of lower densities (Figures 6 and
497
7 and Supporting Information S9.1).
498
The effect of other explanatory variables on tree sensitivity to climate was restricted
499
to a specific region, such as soil conditions within the Boreal West and the continental-to-
500
oceanic climate gradient within the Boreal East, which accounted for 3 % and 5 % of the
501
variation in growth-climate relationships, respectively (Figure 6).
502

DRIVERS OF BOREAL FOREST GROWTH 30
FIGURE 6 Left: The sets of independent variables used in variation partitioning (Please
503
refer to Table 1 for variable ranges). Middle: Effect of each explanatory variable on tree
504
sensitivity to climate, based on autocorrelation-corrected Pearson correlations. The
505
relationship between long-term growth trends and climatic variables is also shown
506
(uppercase letters: current year; lowercase letters: previous year). Red and blue shadings
507
are for negative and positive relationships, respectively, that are significant at α = 0.05.
508
Gray shadings denote non-significant relationships. Significant relationships common to at
509
least two bioclimatic domains are emphasised with a dot. Right: Proportion of variance
510

DRIVERS OF BOREAL FOREST GROWTH 31
explained by each set of independent variables, alone or in combination with other sets
511
(Venn diagrams), by bioclimatic domain. The proportion of variance explained only by the
512
species (pine or spruce) and the proportion of variance unexplained by the selected
513
variables are shown below the diagrams.
514
515
FIGURE 7 Effect of topographic position on sensitivity to (A) spring temperature, (B)
516
summer temperature, (C) previous sumer precipitation, and (D) effect of the age of the
517
stand on the sensitivity to previous summer temperature. Blue and orange dots are observed
518
values for jack pine and black spruce, respectively. Also shown are the Spearman’s rho
519
coefficient (r) and p-value of the modified t-test corrected for the effect of spatial
520
correlation.
521
4. DISCUSSION
522

DRIVERS OF BOREAL FOREST GROWTH 32
Using a dendroecological dataset from a randomly distributed forest inventory network
523
that consisted of 812 plots and 2266 trees, we provided an overview of the response of two
524
major boreal needleleaf species to recent climate change, across explanatory variables that
525
include stand maturity, competition, elevational gradient, soil conditions, and regional
526
climate within eastern boreal North America. Our results highlighted spatial heterogeneity
527
in long-term growth trends across the studied forest: in some areas growth rates decreased,
528
while in others growth increased over the last few decades. Tree sensitivity to climate was
529
also highly spatially heterogeneous. Our study underscores the utility of employing broadly
530
distributed datasets for assessing the complexity of climate change effects on a forest
531
ecosystem (Klesse et al., 2018; Nicklen et al., 2018).
532
The species identity of the tree explained a greater proportion of variation in
533
growth-climate relationships than did all other explanatory variables in the Boreal West.
534
Further, we observed contrasting growth trends between the two species. Our analyses
535
suggest that sensitivity to climate is determined primarily by a species-specific set of
536
functional traits. Black spruce and jack pine usually occupy sites with different soil
537
structures (Balland, Bhatti, Errington, Castonguay, & Arp, 2006) and have very different
538
root system architectures and physiological efficiencies (Blake & Li, 2003; Strong & La
539
Roi, 1983), which could explain differences in climate sensitivity.
540
We identified previous-year summer temperature and, to a lesser extent, previous-
541
autumn temperature as two of the climatic factors with the greatest negative effects on
542
annual growth rates of black spruce trees. Excess-heat during late summer and autumn may
543
lead to declines in carbohydrate reserve accumulation at the end of the growing season,
544
thereby negatively affecting spring growth, as was previously observed in both boreal
545

DRIVERS OF BOREAL FOREST GROWTH 33
North America and northern Europe (Girardin et al., 2016; Ols et al., 2018). Decreased
546
reserve formation of heat-stressed trees can result from accelerated respiration, which leads
547
to higher and more rapid use of photosynthates that otherwise would be available for
548
storage (Anderegg & Anderegg, 2013; Granda & Camarero, 2017; Sala, Woodruff, &
549
Meinzer, 2012). The steepest declines in observed growth for individuals that were the
550
most greatly affected by above-average temperature during summer of the year prior to
551
ring formation suggest that hot extremes are one of the primary determinants of growth
552
trajectories for boreal black spruce forests, as has been observed for white spruce seedlings
553
in plantations by Benomar et al. (2018). The deeper root system of jack pine trees could
554
have allowed them to access additional water resource in deeper soil layers, and their
555
greater resource use efficiency could have prevented them from an overuse of carbon
556
reserves during heatwaves, potentially leading to an uninterrupted carbohydrates storage
557
during hotter-than-average late growing seasons. The resulting higher amount of
558
photosynthates available the following spring could explain the positive correlation
559
between jack pine growth and previous autumn temperature. Our results also suggest that
560
in addition to the effect of species-specific traits, some variation originated from spatially
561
varying site features like stand maturity, position along the elevational gradient, regional
562
climate and soil conditions. Yet, these site-level features explained a lower proportion of
563
the variance in patterns of growth-climate relationships within bioclimatic domains.
564
As predicted from our main working hypothesis, position along the elevational
565
gradient explained a low but significant proportion of the variation in tree response to
566
spring and summer temperature and to previous summer precipitation. This finding
567
illustrates diverging climatic constraints, from water-limited trees at low elevation to trees
568

DRIVERS OF BOREAL FOREST GROWTH 34
constrained by cold temperatures during the growing season at the high end of the
569
elevational gradient. Contrary to our expectations, growth of black spruce was more
570
negatively affected by mild springs when located in the upper portion of our elevational
571
gradient. This counter-intuitive effect of mild springs was recently observed elsewhere
572
(Babst, Carrer, et al., 2012; Ols et al., 2017) and could result from earlier onset of
573
physiological activity and growth in response to warming (Gu et al., 2008; Richardson et
574
al., 2018; Vitasse, Signarbieux, & Fu, 2017). Late-frost events generally occur more
575
frequently at higher elevations and in cold regions such as the Boreal East and Taiga than
576
at lower elevation or in relatively warmer regions such as the Boreal West. These events
577
can damage early formed shoots and roots, thereby reducing total seasonal growth (Kidd,
578
Copenheaver, & Zink-Sharp, 2014; Montwé, Isaac-Renton, Hamann, & Spiecker, 2018).
579
Summer warmth had a contrasting effect on tree growth, depending upon its
580
occurrence. While previous-year high temperatures had negative effects on growth during
581
the following growing season, hotter-than-average summers had an immediate and positive
582
effect on growth rates in the year of occurrence. The latter relationship, which had been
583
observed in the Boreal East and Taiga, was more pronounced for stands at upper elevation
584
sites than at lower elevations and could be linked to a decrease in the risk of late-frost
585
damage and faster snowmelt in early summer (Vitasse et al., 2017). Hot summers are also
586
correlated with high solar radiation and, consequently, with higher rates of photosynthesis,
587
especially in sites where water is not a factor limiting to tree growth, such as stands at the
588
high end of our elevational gradient in the Boreal East and Taiga. Temperature generally
589
decreases with elevation, so an increase in summer temperature can lead to a greater net
590
beneficial effect on tree growth at higher elevations (see Supporting Information S12).
591

DRIVERS OF BOREAL FOREST GROWTH 35
However, the resulting growth gain would have been outweighted by the growth decline
592
due to late frost damage, which could explain that growth trends in stands within the central
593
hilly area were more negative than in the westernmost stands.
594
Effects of other variables on climate sensitivity, such as stand maturity and
595
competition, were generally consistent accross regions. Excess-heat during previous-year
596
summers and autumns had a significantly greater negative impact on black spruce growth
597
in older stands compared to more recently regenerated stands. The increase in climate
598
sensitivity with age has been extensively documented (e.g., Altman et al., 2017; Schuster
599
& Oberhuber, 2013), and was linked to ontogeny-related morphophysiological changes
600
(Ryan et al., 2006) leading to a decrease in hydraulic conductance (Magnani, Mencuccini,
601
& Grace, 2000). During drought, hydraulic conductance may decrease more strongly in old
602
and tall trees because of greater path resistance (Ryan & Yoder, 1997); the resulting
603
decreases in stomatal conductance and photosynthesis may entail, along with greater
604
metabolic demand in tall trees (Hartmann, 2011), depletion of carbohydrate reserves in
605
older stands. This response is a potential explanation for the negative relationship between
606
1970-2005 growth trends and stand age (Supporting Information S11; see also Chen et al.
607
(2016); Girardin et al. (2014)).
608
Competition pressure also significantly modulated the growth-climate
609
relationships. Growth of trees in densely vegetated (high BA) stands that were composed
610
of taller individuals (high CI) was more negatively correlated with excess-heat during
611
previous-year summers and less positively correlated to current-year wet summers than in
612
a less competitive environment. These relationships could have originated from lower
613
carbon assimilation and carbohydrate reserve formation originating from reduced water
614

DRIVERS OF BOREAL FOREST GROWTH 36
availability (Gleason et al., 2017). In contrast, black spruce trees responded more positively
615
to mild winters in densely vegetated compared to more sparsely populated stands, which
616
may be due to the stabilizing effect of a dense canopy on local-scale hydrothermal
617
properties (Gu et al., 2008; Vaganov, Hughes, Kirdyanov, Schweingruber, & Silkin, 1999),
618
similar to the effect of high structural diversity (Aussenac et al., 2017).
619
Soil conditions accounted for a significant proportion of the variation in growth-
620
climate relationships, but this was true only for black spruce in the Boreal West. This region
621
is characterised by a contrasting physiography spanning comparatively flat landcapes with
622
a high proportion of peatlands in the west to hilly terrain with sandy-loam soils in the east
623
(Robitaille et al., 2015; also see Figure 1 and Table 1), together with resulting differences
624
in soil hydrology. During hot summers, the water table of soils with a high proportion of
625
organic material is lowered, and in combination with the high degree of dessication of the
626
peat substrate (Gewehr et al., 2014; Voortman et al., 2013), may have reduced water
627
availability and exacerbated summer heat stress, particularly for trees with shallow rooting
628
systems such as black spruce. Paradoxically, an excess of water during consecutive wet
629
springs and summers also reduced growth of black spruce in Boreal West (Figure 5), most
630
likely because of hypoxic stresses resulting from elevation of the water table in poorly
631
drained sites (Zobel, 1990). A positive correlation between growth of western black spruce
632
trees and the annual area burned, which is a proxy for litter and deep organic layer dryness
633
(Molinari et al., 2018), adds credibility to the assumption that tree sensitivity to
634
precipitation was strongly modulated by soil hydrology in the Boreal West (Supporting
635
Information S13).
636

DRIVERS OF BOREAL FOREST GROWTH 37
Overall, we identified mostly negative growth trends for black spruce and only
637
barely positive trends for jack pine during the 1970-2005 period, which confirms the
638
absence of climatically-induced stimulation of tree vigour that was previously observed for
639
the boreal forest (Girardin et al., 2016; Hember et al., 2016; Ju & Masek, 2016; Zhu et al.,
640
2016). However, forest growth trends were spatially heterogeneous, and the productivity
641
of some areas increased over the last few decades. Variability in growth-climate
642
relationships that was explained by the set of variables considered in our analysis remained
643
low (< 25 %), as is the case in many studies focusing on ecological processes. Our random
644
sampling strategy implies that many factors, which are potentially involved in growth-
645
climate relationships, were not considered and could not be controlled for, such as the
646
effect of non-tree vegetation, insect epidemics, or nutrient cycling. In addition, the genetic
647
diversity of the species under study surely induced different responses to climate between
648
populations (Housset et al., 2016; but see Avanzi et al., 2019). Based upon our results, we
649
suggest that the warming threshold beyond which the productivity of the boreal forest will
650
shift from positive to negative (~ +2 °C; D’Orangeville et al., 2018) is likely very
651
heterogeneous across the boreal biome, but may already have been reached in many of our
652
black spruce stands.
653
ACKNOWLEDGEMENTS
654
This research was conducted as part of the International Research Group on Cold Forests.
655
This study was made possible thanks to the financial support that was provided by the
656
Strategic and Discovery programs of NSERC (Natural Sciences and Engineering Research
657
Council of Canada), and a MITACS scholarship co-funded by NSERC and Ouranos.
658
Additional financial support was provided by the Canadian Forest Service and the UQAM
659

DRIVERS OF BOREAL FOREST GROWTH 38
Foundation (De Sève Foundation fellowship and TEMBEC forest ecology fellowship). We
660
thank Dan McKenney and Pia Papadopol (Canadian Forest Service) for providing the
661
ANUSPLIN climate data. Many thanks to XiaoJing Guo (Canadian Forest Service) for the
662
initial version of the R-scripts and helpful advice during statistical analyses, to Claire
663
Depardieu for relevant comments on the manuscript, and to Isabelle Lamarre and W. F. J.
664
Parsons for language editing.
665
REFERENCES
666
Altman, J., Fibich, P., Santruckova, H., Dolezal, J., Stepanek, P., Kopacek, J., … Cienciala, E. (2017).
667
Environmental factors exert strong control over the climate-growth relationships of
668
Picea abies in Central Europe. Science of the Total Environment, 609, 506–516.
669
https://doi.org/10.1016/j.scitotenv.2017.07.134
670
Anderegg, W. R. L., & Anderegg, L. D. L. (2013). Hydraulic and carbohydrate changes in
671
experimental drought-induced mortality of saplings in two conifer species. Tree
672
Physiology, 33(3), 252–260. https://doi.org/10.1093/treephys/tpt016
673
Ansseau, C., Bélanger, L., Bergeron, J. F., Bouchard, A., Brisson, J., De Grandpré, L., … Vasseur, L.
674
(1997). Écologie forestière. In Manuel de foresterie (1st Edition, pp. 134–279). Sainte-
675
Foy, Québec: Presses de l’Université Laval.
676
Aussenac, R., Bergeron, Y., Mekontchou, C. G., Gravel, D., Pilch, K., & Drobyshev, I. (2017).
677
Intraspecific variability in growth response to environmental fluctuations modulates the
678
stabilizing effect of species diversity on forest growth. Journal of Ecology, 105(4), 1010–
679
1020. https://doi.org/10.1111/1365-2745.12728
680
Austin, M. P., & Niel, K. P. V. (2011). Improving species distribution models for climate change
681
studies: variable selection and scale. Journal of Biogeography, 38(1), 1–8.
682
https://doi.org/10.1111/j.1365-2699.2010.02416.x
683

DRIVERS OF BOREAL FOREST GROWTH 39
Avanzi, C., Piermattei, A., Piotti, A., Büntgen, U., Heer, K., Opgenoorth, L., … Leonardi, S. (2019).
684
Disentangling the effects of spatial proximity and genetic similarity on individual growth
685
performances in Norway spruce natural populations. Science of the Total Environment,
686
650, 493–504. https://doi.org/10.1016/j.scitotenv.2018.08.348
687
Babst, F., Carrer, M., Poulter, B., Urbinati, C., Neuwirth, B., & Frank, D. (2012). 500 years of
688
regional forest growth variability and links to climatic extreme events in Europe.
689
Environmental Research Letters, 7(4), 045705.
690
Babst, F., Poulter, B., Trouet, V., Tan, K., Neuwirth, B., Wilson, R., … Frank, D. (2012). Site- and
691
species-specific responses of forest growth to climate across the European continent.
692
Global Ecology and Biogeography, 22(6), 706–717. https://doi.org/10.1111/geb.12023
693
Balland, V., Bhatti, J., Errington, R., Castonguay, M., & Arp, P. A. (2006). Modeling snowpack and
694
soil temperature and moisture conditions in a jack pine, black spruce and aspen forest
695
stand in central Saskatchewan (BOREAS SSA). Canadian Journal of Soil Science,
696
86(Special Issue), 203–217. https://doi.org/10.4141/S05-088
697
Barber, V. A., Juday, G. P., & Finney, B. P. (2000). Reduced growth of Alaskan white spruce in the
698
twentieth century from temperature-induced drought stress. Nature, 405(6787), 668–
699
673. https://doi.org/10.1038/35015049
700
Bartoń, K. (2018). MuMIn: Multi-Model Inference (Version 1.40.4). Retrieved from
701
https://CRAN.R-project.org/package=MuMIn
702
Beck, P. S. A., Juday, G. P., Alix, C., Barber, V. A., Winslow, S. E., Sousa, E. E., … Goetz, S. J. (2011).
703
Changes in forest productivity across Alaska consistent with biome shift. Ecology Letters,
704
14(4), 373–379. https://doi.org/10.1111/j.1461-0248.2011.01598.x
705
Benomar, L., Lamhamedi, M. S., Pepin, S., Rainville, A., Lambert, M.-C., Margolis, H. A., …
706
Beaulieu, J. (2018). Thermal acclimation of photosynthesis and respiration of southern
707

DRIVERS OF BOREAL FOREST GROWTH 40
and northern white spruce seed sources tested along a regional climatic gradient
708
indicates limited potential to cope with temperature warming. Annals of Botany, 121(3),
709
443–457. https://doi.org/10.1093/aob/mcx174
710
Biondi, F., & Qeadan, F. (2008). A theory-driven approach to tree-ring standardization: defining
711
the biological trend from expected basal area increment. Tree-Ring Research, 64(2), 81–
712
96. https://doi.org/10.3959/2008-6.1
713
Blake, T. J., & Li, J. (2003). Hydraulic adjustment in jack pine and black spruce seedlings under
714
controlled cycles of dehydration and rehydration. Physiologia Plantarum, 117(4), 532–
715
539.
716
Buechling, A., Martin, P. H., & Canham, C. D. (2017). Climate and competition effects on tree
717
growth in Rocky Mountain forests. Journal of Ecology, 105(6), 1636–1647.
718
https://doi.org/10.1111/1365-2745.12782
719
Bunn, A. G. (2008). A dendrochronology program library in R (dplR). Dendrochronologia, 26(2),
720
115–124. https://doi.org/10.1016/j.dendro.2008.01.002
721
Burnham, K. P., & Anderson, D. R. (2002). Model Selection and Multimodel Inference: A Practical
722
Information-Theoretic Approach (2nd ed.). Retrieved from
723
https://www.springer.com/us/book/9780387953649
724
Cahoon, S. M. P., Sullivan, P. F., Brownlee, A. H., Pattison, R. R., Andersen, H.-E., Legner, K., &
725
Hollingsworth, T. N. (2018). Contrasting drivers and trends of coniferous and deciduous
726
tree growth in interior Alaska. Ecology, 99(6), 1284–1295.
727
https://doi.org/10.1002/ecy.2223
728
Carrer, M., & Urbinati, C. (2004). Age‐dependent tree‐ring growth responses to climate in Larix
729
decidua and Pinus cembra. Ecology, 85(3), 730–740. https://doi.org/10.1890/02-0478
730

DRIVERS OF BOREAL FOREST GROWTH 41
Castagneri, D., Petit, G., & Carrer, M. (2015). Divergent climate response on hydraulic-related
731
xylem anatomical traits of Picea abies along a 900-m altitudinal gradient. Tree
732
Physiology, 35(12), 1378–1387. https://doi.org/10.1093/treephys/tpv085
733
Cavard, X., Bergeron, Y., Paré, D., Nilsson, M.-C., & Wardle, D. A. (2018). Disentangling effects of
734
time since fire, overstory composition and organic layer thickness on nutrient
735
availability in Canadian boreal forest. Ecosystems, 1–16.
736
https://doi.org/10.1007/s10021-018-0251-3
737
Chen, H. Y. H., Luo, Y., Reich, P. B., Searle, E. B., & Biswas, S. R. (2016). Climate change-
738
associated trends in net biomass change are age dependent in western boreal forests of
739
Canada. Ecology Letters, 19(9), 1150–1158. https://doi.org/10.1111/ele.12653
740
Cortini, F., Comeau, P. G., & Bokalo, M. (2012). Trembling aspen competition and climate effects
741
on white spruce growth in boreal mixtures of Western Canada. Forest Ecology and
742
Management, 277, 67–73. https://doi.org/10.1016/j.foreco.2012.04.022
743
D’Arrigo, R. D., Kaufmann, R. K., Davi, N., Jacoby, G. C., Laskowski, C., Myneni, R. B., & Cherubini,
744
P. (2004). Thresholds for warming-induced growth decline at elevational tree line in the
745
Yukon Territory, Canada. Global Biogeochemical Cycles, 18(3).
746
https://doi.org/10.1029/2004GB002249
747
DesRochers, A., & Gagnon, R. (1997). Is ring count at ground level a good estimation of black
748
spruce age? Canadian Journal of Forest Research, 27(8), 1263–1267.
749
https://doi.org/10.1139/x97-086
750
Dittmar, C., Zech, W., & Elling, W. (2003). Growth variations of Common beech (Fagus sylvatica
751
L.) under different climatic and environmental conditions in Europe—a dendroecological
752
study. Forest Ecology and Management, 173(1), 63–78. https://doi.org/10.1016/S0378-
753
1127(01)00816-7
754

DRIVERS OF BOREAL FOREST GROWTH 42
D’Orangeville, L., Houle, D., Duchesne, L., Phillips, R. P., Bergeron, Y., & Kneeshaw, D. (2018).
755
Beneficial effects of climate warming on boreal tree growth may be transitory. Nature
756
Communications, 9(1), 3213. https://doi.org/10.1038/s41467-018-05705-4
757
Dormann, C. F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., … Lautenbach, S. (2013).
758
Collinearity: a review of methods to deal with it and a simulation study evaluating their
759
performance. Ecography, 36(1), 27–46. https://doi.org/10.1111/j.1600-
760
0587.2012.07348.x
761
Dulamsuren, C., Hauck, M., Kopp, G., Ruff, M., & Leuschner, C. (2017). European beech responds
762
to climate change with growth decline at lower, and growth increase at higher
763
elevations in the center of its distribution range (SW Germany). Trees, 31(2), 673–686.
764
https://doi.org/10.1007/s00468-016-1499-x
765
Ford, K. R., Breckheimer, I. K., Franklin, J. F., Freund, J. A., Kroiss, S. J., Larson, A. J., …
766
HilleRisLambers, J. (2016). Competition alters tree growth responses to climate at
767
individual and stand scales. Canadian Journal of Forest Research, 47(1), 53–62.
768
https://doi.org/10.1139/cjfr-2016-0188
769
Fortin, M., Bernier, S., Saucier, J.-P., & Labbé, F. (2009). Une relation hauteur-diamètre tenant
770
compte de l’influence de la station et du climate pour 20 espèces commerciales du
771
Québec (No. 153; p. 41). Québec (Province): Direction de la Recherche Forestière.
772
Fritts, H. C. (1971). Dendroclimatology and Dendroecology. Quaternary Research, 1(4), 419–449.
773
https://doi.org/10.1016/0033-5894(71)90057-3
774
Gauthier, S., Bernier, P., Kuuluvainen, T., Shvidenko, A. Z., & Schepaschenko, D. G. (2015). Boreal
775
forest health and global change. Science, 349(6250), 819–822.
776
https://doi.org/10.1126/science.aaa9092
777

DRIVERS OF BOREAL FOREST GROWTH 43
Gennaretti, F., Arseneault, D., Nicault, A., Perreault, L., & Bégin, Y. (2014). Volcano-induced
778
regime shifts in millennial tree-ring chronologies from northeastern North America.
779
Proceedings of the National Academy of Sciences of the United States of America,
780
111(28), 10077–10082. https://doi.org/10.1073/pnas.1324220111
781
Gewehr, S., Drobyshev, I., Berninger, F., & Bergeron, Y. (2014). Soil characteristics mediate the
782
distribution and response of boreal trees to climatic variability. Canadian Journal of
783
Forest Research, 44, 487–498. https://doi.org/10.1139/cjfr-2013-0481
784
Girardin, M. P., Bernier, P. Y., Raulier, F., Tardif, J. C., Conciatori, F., & Guo, X. J. (2011). Testing
785
for a CO2 fertilization effect on growth of Canadian boreal forests. Journal of
786
Geophysical Research: Biogeosciences, 116(G1), G01012.
787
https://doi.org/10.1029/2010JG001287
788
Girardin, M. P., Bouriaud, O., Hogg, E. H., Kurz, W., Zimmermann, N. E., Metsaranta, J. M., …
789
Bhatti, J. (2016). No growth stimulation of Canada’s boreal forest under half-century of
790
combined warming and CO2 fertilization. Proceedings of the National Academy of
791
Sciences of the United States of America, 113(52), E8406–E8414.
792
https://doi.org/10.1073/pnas.1610156113
793
Girardin, M. P., Guo, X. J., De Jong, R., Kinnard, C., Bernier, P., & Raulier, F. (2014). Unusual
794
forest growth decline in boreal North America covaries with the retreat of Arctic sea ice.
795
Global Change Biology, 20(3), 851–866. https://doi.org/10.1111/gcb.12400
796
Gleason, K. E., Bradford, J. B., Bottero, A., D’Amato, A. W., Fraver, S., Palik, B. J., … Kern, C. C.
797
(2017). Competition amplifies drought stress in forests across broad climatic and
798
compositional gradients. Ecosphere, 8(7), e01849. https://doi.org/10.1002/ecs2.1849
799
Gouvernement du Québec. (2003). Vegetation Zones and Bioclimatic Domains in Québec.
800

DRIVERS OF BOREAL FOREST GROWTH 44
Granda, E., & Camarero, J. J. (2017). Drought reduces growth and stimulates sugar
801
accumulation: new evidence of environmentally driven non-structural carbohydrate use.
802
Tree Physiology, 37(8), 997–1000. https://doi.org/10.1093/treephys/tpx097
803
Gu, L., Hanson, P. J., Post, W. M., Kaiser, D. P., Yang, B., Nemani, R., … Meyers, T. (2008). The
804
2007 Eastern US spring freeze: Increased cold damage in a warming world. BioScience,
805
58(3), 253–262. https://doi.org/10.1641/B580311
806
Hartmann, H. (2011). Will a 385 million year-struggle for light become a struggle for water and
807
for carbon? – How trees may cope with more frequent climate change-type drought
808
events. Global Change Biology, 17(1), 642–655. https://doi.org/10.1111/j.1365-
809
2486.2010.02248.x
810
Hasler, A., Geertsema, M., Foord, V., Gruber, S., & Noetzli, J. (2015). The influence of surface
811
characteristics, topography and continentality on mountain permafrost in British
812
Columbia. The Cryosphere, 9(3), 1025–1038. https://doi.org/10.5194/tc-9-1025-2015
813
Hember, R. A., Kurz, W. A., & Coops, N. C. (2016). Relationships between individual-tree
814
mortality and water-balance variables indicate positive trends in water stress-induced
815
tree mortality across North America. Global Change Biology, 23(4), 1691–1710.
816
https://doi.org/10.1111/gcb.13428
817
Hember, R. A., Kurz, W. A., & Coops, N. C. (2017). Increasing net ecosystem biomass production
818
of Canada’s boreal and temperate forests despite decline in dry climates. Global
819
Biogeochemical Cycles, 31(1), 2016GB005459. https://doi.org/10.1002/2016GB005459
820
Holmes, R. L. (1983). Computer assisted quality control in tree-ring dating and measurement.
821
Retrieved from https://www.scienceopen.com/document?vid=a110a27a-36fa-41dd-
822
8536-434bb4d62436
823

DRIVERS OF BOREAL FOREST GROWTH 45
Housset, J. M., Carcaillet, C., Girardin, M. P., Xu, H., Tremblay, F., & Bergeron, Y. (2016). In situ
824
comparison of tree-ring responses to climate and population genetics: The need to
825
control for local climate and site variables. Frontiers in Ecology and Evolution, 4.
826
https://doi.org/10.3389/fevo.2016.00123
827
Huber, P., J. (2005). Generalities. In Robust Statistics (pp. 1–19).
828
https://doi.org/10.1002/0471725250.ch1
829
IPCC. (2013). Climate Change 2013 : The physical science basis. Contribution of Working Group I
830
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. (p.
831
1535).
832
Jarvis, A., Reuter, H., Nelson, A., & Guevara, E. (2008, January 1). Hole-filled seamless SRTM data
833
v4. Retrieved from http://srtm.csi.cgiar.org
834
Jiang, X., Huang, J.-G., Cheng, J., Dawson, A., Stadt, K. J., Comeau, P. G., & Chen, H. Y. H. (2018).
835
Interspecific variation in growth responses to tree size, competition and climate of
836
western Canadian boreal mixed forests. Science of The Total Environment, 631–632,
837
1070–1078. https://doi.org/10.1016/j.scitotenv.2018.03.099
838
Johnstone, J. F., McIntire, E. J. B., Pedersen, E. J., King, G., & Pisaric, M. J. F. (2010). A sensitive
839
slope: estimating landscape patterns of forest resilience in a changing climate.
840
Ecosphere, 1(6), art14. https://doi.org/10.1890/ES10-00102.1
841
Ju, J., & Masek, J. G. (2016). The vegetation greenness trend in Canada and US Alaska from
842
1984–2012 Landsat data. Remote Sensing of Environment, 176, 1–16.
843
https://doi.org/10.1016/j.rse.2016.01.001
844
Juday, G. P., & Alix, C. (2012). Consistent negative temperature sensitivity and positive influence
845
of precipitation on growth of floodplain Picea glauca in Interior Alaska. Canadian Journal
846
of Forest Research, 42(3), 561–573. https://doi.org/10.1139/x2012-008
847

DRIVERS OF BOREAL FOREST GROWTH 46
Kidd, K. R., Copenheaver, C. A., & Zink-Sharp, A. (2014). Frequency and factors of earlywood
848
frost ring formation in jack pine (Pinus banksiana) across northern lower Michigan.
849
Écoscience, 21(2), 157–167. https://doi.org/10.2980/21-2-3708
850
Klesse, S., DeRose, R. J., Guiterman, C. H., Lynch, A. M., O’Connor, C. D., Shaw, J. D., & Evans, M.
851
E. K. (2018). Sampling bias overestimates climate change impacts on forest growth in
852
the southwestern United States. Nature Communications, 9(1), 5336.
853
https://doi.org/10.1038/s41467-018-07800-y
854
Kurz, W. A., Shaw, C. H., Boisvenue, C., Stinson, G., Metsaranta, J., Leckie, D., … Neilson, E. T.
855
(2013). Carbon in Canada’s boreal forest — A synthesis. Environmental Reviews, 21(4),
856
260–292. https://doi.org/10.1139/er-2013-0041
857
Lavoie, M., Harper, K., Paré, D., & Bergeron, Y. (2007). Spatial pattern in the organic layer and
858
tree growth: A case study from regenerating Picea mariana stands prone to
859
paludification. Journal of Vegetation Science, 18(2), 213–222.
860
https://doi.org/10.1111/j.1654-1103.2007.tb02532.x
861
Létourneau, J. P., Matejek, S., Morneau, C., Robitaille, A., Roméo, T., Brunelle, J., & Leboeuf, A.
862
(2008). Norme de cartographie écoforestière du Programme d’inventaire écoforestier
863
nordique. Québec: Ministère des Ressources naturelles et de la Faune du Québec.
864
Lloyd, A. H., Bunn, A. G., & Berner, L. (2011). A latitudinal gradient in tree growth response to
865
climate warming in the Siberian taiga. Global Change Biology, 17(5), 1935–1945.
866
https://doi.org/10.1111/j.1365-2486.2010.02360.x
867
Loader, N. J., McCarroll, D., Gagen, M., Robertson, I., & Jalkanen, R. (2007). Extracting climatic
868
information from stable isotopes in tree rings. In Stable Isotopes as Indicators of
869
Ecological Change: Vol. 1. Terrestrial Ecology (pp. 25–48).
870
https://doi.org/10.1016/S1936-7961(07)01003-2
871

DRIVERS OF BOREAL FOREST GROWTH 47
Magnani, F., Mencuccini, M., & Grace, J. (2000). Age-related decline in stand productivity: the
872
role of structural acclimation under hydraulic constraints. Plant, Cell & Environment,
873
23(3), 251–263. https://doi.org/10.1046/j.1365-3040.2000.00537.x
874
Mangiafico, S. (2017). rcompanion: Functions to Support Extension Education Program
875
Evaluation (Version 1.11.1). Retrieved from https://cran.r-
876
project.org/web/packages/rcompanion/index.html
877
Manrique-Alba, À., Ruiz-Yanetti, S., Moutahir, H., Novak, K., De Luis, M., & Bellot, J. (2017). Soil
878
moisture and its role in growth-climate relationships across an aridity gradient in
879
semiarid Pinus halepensis forests. Science of The Total Environment, 574, 982–990.
880
https://doi.org/10.1016/j.scitotenv.2016.09.123
881
Matías, L., Linares, J. C., Sánchez-Miranda, Á., & Jump, A. S. (2017). Contrasting growth forecasts
882
across the geographical range of Scots pine due to altitudinal and latitudinal differences
883
in climatic sensitivity. Global Change Biology, 23(10), 4106–4116.
884
https://doi.org/10.1111/gcb.13627
885
McLaughlin, S. B., Downing, D. J., Blasing, T. J., Cook, E. R., & Adams, H. S. (1987). An analysis of
886
climate and competition as contributors to decline of red spruce in high elevation
887
Appalachian forests of the Eastern United states. Oecologia, 72(4), 487–501.
888
https://doi.org/10.1007/BF00378973
889
Minasny, B., & McBratney, A. B. (2017). Limited effect of organic matter on soil available water
890
capacity. European Journal of Soil Science, 69(1), 39–47.
891
https://doi.org/10.1111/ejss.12475
892
Ministère des Ressources Naturelles du Québec. (2008). Norme d’inventaire écodendrométrique
893
nordique (p. 245). Direction des inventaires forestiers.
894

DRIVERS OF BOREAL FOREST GROWTH 48
Molinari, C., Lehsten, V., Blarquez, O., Carcaillet, C., Davis, B. A. S., Kaplan, J. O., … Bradshaw, R.
895
H. W. (2018). The climate, the fuel and the land use: Long-term regional variability of
896
biomass burning in boreal forests. Global Change Biology, 0(0).
897
https://doi.org/10.1111/gcb.14380
898
Montwé, D., Isaac-Renton, M., Hamann, A., & Spiecker, H. (2018). Cold adaptation recorded in
899
tree rings highlights risks associated with climate change and assisted migration. Nature
900
Communications, 9(1), 1574. https://doi.org/10.1038/s41467-018-04039-5
901
Nicklen, E. F., Roland, C. A., Csank, A. Z., Wilmking, M., Ruess, R. W., & Muldoon, L. A. (2018).
902
Stand basal area and solar radiation amplify white spruce climate sensitivity in interior
903
Alaska: Evidence from carbon isotopes and tree rings. Global Change Biology, 0(0), 1–16.
904
https://doi.org/10.1111/gcb.14511
905
Nicklen, E. F., Roland, C. A., Ruess, R. W., Schmidt, J. H., & Lloyd, A. H. (2016). Local site
906
conditions drive climate–growth responses of Picea mariana and Picea glauca in interior
907
Alaska. Ecosphere, 7(10), e01507. https://doi.org/10.1002/ecs2.1507
908
Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., … Wagner, H.
909
(2018). vegan: Community Ecology Package (Version 2.5-2). Retrieved from
910
https://CRAN.R-project.org/package=vegan
911
Ols, C., Girardin, M. P., Hofgaard, A., Bergeron, Y., & Drobyshev, I. (2017). Monitoring climate
912
sensitivity shifts in tree-rings of eastern boreal North America using model-data
913
comparison. Ecosystems, 1–16. https://doi.org/10.1007/s10021-017-0203-3
914
Ols, C., Trouet, V., Girardin, M. P., Hofgaard, A., Bergeron, Y., & Drobyshev, I. (2018). Post-1980
915
shifts in the sensitivity of boreal tree growth to North Atlantic Ocean dynamics and
916
seasonal climate. Global and Planetary Change, 165, 1–12.
917
https://doi.org/10.1016/j.gloplacha.2018.03.006
918

DRIVERS OF BOREAL FOREST GROWTH 49
Osorio, F., Vallejos, R., Cuevas, F., & Mancilla, D. (2018). SpatialPack: Tools for assessment of the
919
association between two spatial processes (Version 0.3). Retrieved from
920
https://CRAN.R-project.org/package=SpatialPack
921
Pepin, N., Bradley, R. S., Diaz, H. F., Caceres, E. B., Forsythe, N., Fowler, H., … Yang, D. Q. (2015).
922
Elevation-dependent warming in mountain regions of the world. Nature Climate
923
Change, 5, 424–430.
924
Peres-Neto, P. R., Legendre, P., Dray, S., & Borcard, D. (2006). Variation partitionning of species
925
data matrices: estimation and comparison of fractions. Ecology, 87(10), 2614–2625.
926
https://doi.org/10.1890/0012-9658(2006)87[2614:VPOSDM]2.0.CO;2
927
Piovesan, G., Biondi, F., Filippo, A. D., Alessandrini, A., & Maugeri, M. (2008). Drought-driven
928
growth reduction in old beech (Fagus sylvatica L.) forests of the central Apennines, Italy.
929
Global Change Biology, 14(6), 1265–1281. https://doi.org/10.1111/j.1365-
930
2486.2008.01570.x
931
Price, D. T., Alfaro, R. I., Brown, K. J., Flannigan, M. D., Fleming, R. A., Hogg, E. H., … Venier, L. A.
932
(2013). Anticipating the consequences of climate change for Canada’s boreal forest
933
ecosystems. Environmental Reviews, 21(4), 322–365. https://doi.org/10.1139/er-2013-
934
0042
935
Price, D. T., McKenney, D. W., Joyce, L. A., Siltanen, R. M., Papadopol, P., & Lawrence, K. (2011).
936
High-resolution interpolation of climate scenarios for Canada derived from general
937
circulation model simulations. Information Report NOR-X-421. Edmonton, AB.: Natural
938
Resources Canada, Canadian Forest Service, Northern Forestry Centre. 104 P. Retrieved
939
from https://www.fs.usda.gov/treesearch/pubs/39890
940
Richardson, A. D., Hufkens, K., Milliman, T., Aubrecht, D. M., Furze, M. E., Seyednasrollah, B., …
941
Hanson, P. J. (2018). Ecosystem warming extends vegetation activity but heightens
942

DRIVERS OF BOREAL FOREST GROWTH 50
vulnerability to cold temperatures. Nature, 560(7718), 368–371.
943
https://doi.org/10.1038/s41586-018-0399-1
944
Robitaille, A., Saucier, J.-P., Chabot, M., Côté, D., & Boudreault, C. (2015). An approach for
945
assessing suitability for forest management based on constraints of the physical
946
environment at a regional scale. Canadian Journal of Forest Research, 45(5), 529–539.
947
https://doi.org/10.1139/cjfr-2014-0338
948
Ryan, M. G., Phillips, N., & Bond, B. J. (2006). The hydraulic limitation hypothesis revisited. Plant,
949
Cell & Environment, 29(3), 367–381. https://doi.org/10.1111/j.1365-3040.2005.01478.x
950
Ryan, M. G., & Yoder, B. J. (1997). Hydraulic Limits to Tree Height and Tree Growth. BioScience,
951
47(4), 235–242. https://doi.org/10.2307/1313077
952
Sala, A., Woodruff, D. R., & Meinzer, F. C. (2012). Carbon dynamics in trees: feast or famine?
953
Tree Physiology, 32(6), 764–775. https://doi.org/10.1093/treephys/tpr143
954
Santander Meteorology Group. (2012). fume: FUME package (Version R package version 1.0).
955
Retrieved from https://CRAN.R-project.org/package=fume
956
Schuster, R., & Oberhuber, W. (2013). Age-dependent climate-growth relationships and
957
regeneration of Picea abies in a drought-prone mixed-coniferous forest in the Alps.
958
Canadian Journal of Forest Research, 43, 609–618. https://doi.org/10.1139/cjfr-2012-
959
0426
960
Strong, W. L., & La Roi, G. H. (1983). Root-system morphology of common boreal forest trees in
961
Alberta, Canada. Canadian Journal of Forest Research, 13(6), 1164–1173.
962
https://doi.org/10.1139/x83-155
963
ter Braak, C., & Smilauer, P. (2009). Canoco for Windows Version 4.56. Biometris - Plant
964
Research International.
965

DRIVERS OF BOREAL FOREST GROWTH 51
Trenberth, K. E. (1983). What are the Seasons? Bulletin of the American Meteorological Society,
966
64(11), 1276–1282. https://doi.org/10.1175/1520-0477(1983)064<1276:WATS>2.0.CO;2
967
Trugman, A. T., Medvigy, D., Anderegg, W. R. L., & Pacala, S. W. (2018). Differential declines in
968
Alaskan boreal forest vitality related to climate and competition. Global Change Biology,
969
24(3), 1097–1107. https://doi.org/10.1111/gcb.13952
970
Tukey, J. W. (1977). Exploratory data analysis (Addison-Wesley). Reading, MA.
971
Vaganov, E. A., Hughes, M. K., Kirdyanov, A. V., Schweingruber, F. H., & Silkin, P. P. (1999).
972
Influence of snowfall and melt timing on tree growth in subarctic Eurasia. Nature,
973
400(6740), 149–151. https://doi.org/10.1038/22087
974
Vitasse, Y., Signarbieux, C., & Fu, Y. H. (2017). Global warming leads to more uniform spring
975
phenology across elevations. Proceedings of the National Academy of Sciences of the
976
United States of America, 201717342. https://doi.org/10.1073/pnas.1717342115
977
Voortman, B. R., Bartholomeus, R. P., Bodegom, P. M. van, Gooren, H., Zee, S. E. A. T. M. van
978
der, & Witte, J.-P. M. (2013). Unsaturated hydraulic properties of xerophilous mosses:
979
towards implementation of moss covered soils in hydrological models. Hydrological
980
Processes, 28(26), 6251–6264. https://doi.org/10.1002/hyp.10111
981
Walker, X. J., & Johnstone, J. F. (2014). Widespread negative correlations between black spruce
982
growth and temperature across topographic moisture gradients in the boreal forest.
983
Environmental Research Letters, 9(6), 064016. https://doi.org/10.1088/1748-
984
9326/9/6/064016
985
Weber, P., Bugmann, H., Fonti, P., & Rigling, A. (2008). Using a retrospective dynamic
986
competition index to reconstruct forest succession. Forest Ecology and Management,
987
254(1), 96–106. https://doi.org/10.1016/j.foreco.2007.07.031
988

DRIVERS OF BOREAL FOREST GROWTH 52
Wilmking, M., & Myers-Smith, I. (2008). Changing climate sensitivity of black spruce (Picea
989
mariana Mill.) in a peatland–forest landscape in Interior Alaska. Dendrochronologia,
990
25(3), 167–175. https://doi.org/10.1016/j.dendro.2007.04.003
991
Wu, X., Li, X., Liu, H., Ciais, P., Li, Y., Xu, C., … Zhang, C. (2018). Uneven winter snow influence on
992
tree growth across temperate China. Global Change Biology.
993
https://doi.org/10.1111/gcb.14464
994
Yue, S., & Wang, C. (2004). The Mann-Kendall test modified by effective sample size to detect
995
trend in serially correlated hydrological series. Water Resources Management, 18(3),
996
201–218. https://doi.org/10.1023/B:WARM.0000043140.61082.60
997
Zhang, K., Kimball, J. S., Hogg, E. H., Zhao, M., Oechel, W. C., Cassano, J. J., & Running, S. W.
998
(2008). Satellite-based model detection of recent climate-driven changes in northern
999
high-latitude vegetation productivity. Journal of Geophysical Research: Biogeosciences,
1000
113(G3), G03033. https://doi.org/10.1029/2007JG000621
1001
Zhu, Z., Piao, S., Myneni, R. B., Huang, M., Zeng, Z., Canadell, J. G., … Zeng, N. (2016). Greening
1002
of the Earth and its drivers. Nature Climate Change, 6(8), 791–795.
1003
https://doi.org/10.1038/nclimate3004
1004
Zobel, M. (1990). Soil oxygen conditions in paludifying boreal forest sites. Suo, 41(4–5), 81–89.
1005
1006