Embed Size (px)
Citation preview
Predicting flowering phenology in a subarctic plant community
Malie Lessard-Therrien a, b, e, Kjell Bolmgren b, c, and T. Jonathan Davies d
a Department of Biology, McGill University, 1205 Dr. Penfield, Montreal (Quebec), H3A
1B1, Canada
b Department of Ecology, Environment and Plant Sciences, Stockholm University, Lilla
Frescati, SE-106 91 Stockholm, Sweden
c Swedish University of Agricultural Sciences, Unit for Field-based Forest Research, SE-
360 30 Lammhult, Sweden
d Department of Biology, McGill University, 1205 Dr. Penfield, Montreal (Quebec), H3A
1B1, Canada
e Division of Conservation Biology, Institute of Ecology and Evolution, University of
Bern, Baltzerstrasse 6, 3012 Bern, Switzerland
* Corresponding author:
Malie Lessard-Therrien
Division of Conservation Biology, Institute of Ecology and Evolution, University of
Bern, Baltzerstrasse 6, 3012 Bern, SwitzerlandPhone: +41 (0)31. 631.31.53
Page 1 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
2
Abstract
Phenological studies are rarely reported from arctic and subarctic regions, but are
essential to evaluate species’ response to climate change in these rapidly warming
ecosystems. Here, we present a phylogenetic analysis of flowering phenology across an
elevational gradient in the Canadian subarctic. We found that the timing of first flower
was best explained by a combination of snowmelt, elevation and growing degree days.
We also show that early flowering species have demonstrated lower intraspecific
variability in their response to climate cues in comparison with late flowering species,
such that individual flowering times of early species are more closely tied to
environmental predictors. Previous work has suggested that early flowering species are
more variable in their phenology. However, these studies have mostly examined variation
in phenology over time, whereas we examined variation in phenology over space. We
suggest that both patterns can be explained by the tighter coupling between phenology
and climate cues for early flowering species. Thus, early flowering species have low
intraspecific variance in flowering times within a single growing season as individuals
respond more uniformly to a common set of cues in comparison to late flowering species.
However, these same species may show large variance between years reflecting inter-
annual variation in climate.
Key words First flowering day, variability, snowmelt, temperature sum, phylogenetics
Page 2 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
3
Introduction
Plants have responded to the harsh arctic and alpine environment with a high degree of
specialization in structure and function (Körner 2003). In the short growing season in
these ecosystems, plant species display a variety of flowering patterns related to the onset
and duration of flowering (Arroyo et al. 1981; Molau 1993; Jia et al. 2011). Because of
strong constraints on the timing of development due to a cold climate and a short growing
season, flowering phenological strategies have evolved to cope with and track these
climatic factors (Hülber et al. 2010; Cornelius et al. 2012). In arctic and alpine
landscapes, snow depth and duration are thought to be the most important determinants
for differentiation of tundra plant communities (Molau 1993; Kudo and Suzuki 1999).
Snowfall during winter varies from year to year, but the snow distribution pattern, with
snowbeds and snow-free ridges, is relatively constant over years at the landscape level,
determining the timing of plant phenology (Molau et al. 2005). Snow distribution also
determines water availability during summer, thus playing an important role in shaping
plant communities (Kudo 1991).
The flowering phenology of subarctic plants is thus predominantly determined by
the timing of snowmelt and subsequent temperature sums (Inouye and Wielgolaski 2003;
Molau et al. 2005; Iversen et al. 2009). In winter, snow provides protection for low plant
species from extreme cold temperatures, but can also compromise successful
reproduction if it persists through the growing season (Totland 1994). Temperature,
acting through cumulative heat sums above a threshold level, is an additional driver for
different life cycle events of plants, including flowering (Rathcke and Lacey 1985; Fitter
et al. 1995). Nonetheless, species can express very different degrees of phenological
Page 3 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
4
variability (Molau et al. 2005), even when cuing to similar environmental factors.
Variability in flowering time may confer a strong advantage in the highly seasonal
subarctic (Molau 1993). For early flowering species, high variability in day of year of
flowering might allow species to track the variable onset of spring among years, whereas
late flowering species might be less constrained in their flowering times, and thus
demonstrate less variability between years. In addition species with high inter-annual
variability in flowering dates may be better able to track the advancement of warm
temperature in spring with climate change (Arft et al. 1999; Fitter and Fitter 2002; Dunne
et al. 2003). Several studies have explored inter-annual variability in flowering
phenology at temperate latitudes (e.g. Menzel 2003; Studer et al. 2003; Van Wijk and
Williams 2003; Miller-Rushing and Inouye 2009), but it has been less well studied in
arctic and subarctic ecosystems.
Traditionally, phenological variability (among individuals or between years) is
quantified using the standard deviation (SD) of flowering times (e.g. Molau et al. 2005;
Hülber et al. 2010). However, this approach does not consider variability in
environmental cues experienced by individuals across space within a single growing
season. Here, we quantify variability in flowering time while controlling for variance in
environmental factors. Our study, designed across an elevational gradient, provides the
opportunity to characterise intraspecific variability using data collected in a single year,
as opposed to inter-annual variability. Elevational gradients naturally provide variation in
temperature scenarios and other abiotic factors. Importantly, patterns of variability in
flowering dates between years and variability in flowering dates across space within a
single growing season might show different trends but reflect similar adaptive responses.
Page 4 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
5
A short vegetative and reproductive season, common in high altitude and latitude
ecosystems, may also constrain the evolution of flowering via selection on other plant
traits (Debussche et al. 2004; Bolmgren and Lönnberg 2005; Oberrath and Böhning-
Gaese 2002). As tundra species have adopted different strategies to survive in the harsh,
resource limited environment of the North (Jónsdóttir 2011), a comprehensive
understanding of arctic and subarctic flowering phenology must therefore consider plant
traits as well as environment. In this study, we explore key traits related to life form and
fruit development. Life form may influence the phenology of species if, for example,
different forms have different frost tolerance, nutrient storage capacity and reproductive
strategies in subarctic-alpine ecosystems (Billings and Mooney 1968; Kudo 1991; Chapin
et al. 1996). The type of fruit may impose a constraint on flowering time and duration
depending on its size, maturation time and nutrient content (Eriksson and Ehrlén 1991;
Bolmgren and Lönnberg 2005).
Here, we address three specific goals: (1) we describe the phenological strategy of
a subarctic plant community according to their life form and fruit type, (2) we relate the
first flowering day (FFD) to climatic factors and key traits to develop a single predictive
model of flowering phenology at the community level for herbs and dwarf shrub species
in a subarctic ecosystem, using a phylogenetic comparative approach to account for non-
independence among species, and (3) we explore intraspecific variability in flowering
phenology after accounting for variation in environment.
Material and Methods
Page 5 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
6
Our study was conducted across an elevational gradient ranging from 600 m to 800 m
above sea level, along the south west slope of Mount Irony (856 m), located
approximately 40 km northwest of Schefferville (Qc), Canada (54°43’N, 66°42’W). The
site provided a steep environmental gradient in a region with a strongly seasonal climate.
In 2012, air temperature ranged from 30°C (June 12th) to -42°C (February 12th) with an
annual mean of -2°C (Schefferville airport weather station
http://www.wunderground.com/history/). The total annual precipitation in 2012 was
462.65 mm.
The region is characterised by a mosaic of open boreal forest and tundra with
Betula glandulosa Michaux (shrub birch) as the most abundant species. At the bottom of
the south west slope is boreal forest with equally abundant Picea mariana (Miller)
Britton, Sterns and Poggenburgh (black spruce) and Picea glauca (Moench) Voss (white
spruce), which give way to dense shrub communities characterized by B. glandulosa and
Alnus viridis (Chaix) de Candolle (green alder) at intermediate elevations. At higher
elevations, there is a rocky tundra with dwarf shrubs including Vaccinium uliginosum
Linnaeus (alpine bilberry), Dryas integrifolia Vahl (entire-leaved mountain avens),
Arctous alpina (Linnaeus) Niedenzu (alpine bearberry), Empetrum nigrum Linnaeus
(black crowberry), and herbaceous species such as Bartsia alpina Linnaeus (alpine
bartsia) and Solidago multiradiata Aiton (multi-rayed goldenrod). The total elevational
range is relatively small (200 meters); however, this was sufficient to encompass large
differences in temperature and concomitant shifts in community structure at these high
latitudes.
Page 6 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
7
The subarctic elevational gradient studied here captures more than just
temperature differences. On the higher part of the elevational gradient, snow cover is thin
during winter due to strong winds, thus the soil surface is exposed early in the season and
the bare soil warms up quickly under the sun, although air temperatures are cooler over
summer. On the lower part of the gradient, snow cover is thick due to substantial
accumulation during the winter, a function of local topography. Further, at low
elevations, snowmelt occurs later because of increased shading under forest canopy. In
addition, the soil remains cool throughout the season because of the thick layer of moss in
the boreal forest.
Plant community data
We recorded phenology data on the south west slope of Mount Irony during the summer
of 2012 (May 27th until August 7th). A list of sampled species with additional
phenological information is found in Lessard-Therrien et al. (2013). We sampled 88 two-
meter diameter circular plots systematically spaced 100 meters apart across eight
elevation bands. Each plot was visited every three to six days during which we recorded
the first flowering day (FFD) for all species within each plot. To capture variation in
plant community types, we used the following elevation bands: 800 m (n=23 species),
775 m (n=19), 745 m (n=16), 710 m (n=11), 675 m (n=10), 645 m (n=18), 615 m (n=16)
and 600 m (n=13). In a few cases, plots were staggered to avoid bare rock. Plots at the
600 m and 615 m are located in the boreal forest, plots from 645 m to 745 m are
generally under the tree line, but contain a few, scattered trees of varying sizes, and plots
at 775 m and 800 m are devoid of trees. All plots had similar aspect (southwest-facing
Page 7 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
8
slope) and slope (between 0° and 18°). A total of 48 species were sampled and their
phylogenetic relationships were taken from Lessard-Therrien et al. (2013).
We recorded first flowering date (i.e. when the center of the flower was visible
between the petals) of all herbaceous and shrub species less than 50 cm in height.
Species’ identification was achieved with the Flore Laurentienne, 3ieme
édition (Marie-
Victorin 1995) and the Vascular plants of continental Northwest Territories, Canada
(Porsild and Cody 1980). Nomenclature was validated and updated using the Database of
Vascular Plants of Canada (VASCAN, Brouillet et al. 2012,
http://data.canadensys.net/vascan/search/).
Species were classified according to life form (herbaceous and shrub), using
Chapin et al.’s (1996) classification of functional types for arctic plant species, and fruit
type (fleshy or dry) following Bolmgren and Lönnberg (2005).
Environmental data
Temperature data were obtained from HOBO Pendant Temperature and Light Data
Loggers. These recording devices were buried in the summer of 2011 to a depth of 5cm
and placed one meter east of the edge of every plot to avoid community and soil
disturbance within the plots. The HOBOs recorded hourly temperature data through the
year.
Snowmelt date (SMD)
Because snow cover has strong insulating properties, sites under snow are
characterised by low, stable temperatures all winter regardless of the air temperature
Page 8 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
9
fluctuations. The soil temperature remains stable until snow cover becomes thin in the
spring and rises abruptly by three to four degrees once snow has melted. Following this
sudden rise in soil temperature, there is a more gradual increase as the growing season
advances, and increased daily fluctuation due to the removal of the insulating snow layer.
The abrupt increase in temperature is a reliable cue to estimate snowmelt date (SMD),
matching light records when those are available (D. Inouye, personal communication).
SMD was therefore estimated at each site as the date of the first abrupt rise in
temperature (>4°Celcius) from the start of the calendar year (Fig. 1).
Temperature sums measurements
We adapted the temperature sum method to subarctic ecosystems following Molau and
Mølgaard (1996). We use 0°Celsius as the threshold base temperature, soil thawing date
as starting date, and soil temperature instead of air temperature. Base temperature is the
threshold necessary for plant growth and varies according to the ecosystem. In arctic and
subarctic environments, a base temperature of 0°Celsius is the most relevant (Molau and
Mølgaard 1996). Thawing degree days (TDD) represent the accumulation of the daily
temperature sum from the thawing date. The daily temperature was calculated as the
mean over 24 hours. TDD captures heat accumulation even under snow pack as long as
temperatures are above 0°Celsius. Growing degree days (GDD) were calculated as the
summation of daily mean temperature when daily temperatures were above 0°Celsius,
starting from snow melt. Finally, thawing degree hours (TDH) were obtained by
summing the total number of hours for which the temperature recordings were above
0°Celsius from the initial thawing date.
Page 9 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
10
Statistical analyses
To identify predictors of FFD at the community level, we included data from all
individuals across species. We used phylogenetic generalized least square (PGLS) with
the maximum likelihood value of λ, as implemented in the caper R-package (Orme 2012)
using R 2.15.2 (http://cran.r-project.org/), to evaluate alternative models including abiotic
and biotic variables as predictors of FFD. PGLS includes the phylogenetic structure of
the data as a covariance matrix in a linear model, and thereby controls for phylogenetic
non-independence among species (for details on phylogeny reconstruction, see Lessard-
Therrien et al. 2013).To account for within-species variation, we included individuals in
our phylogeny as a terminal polytomy. However, because the PGLS model fails with zero
branch lengths, we randomly resolved polytomies by adding arbitrarily small branch
lengths of 0.001, repeating each of the starting models using 1000 random resolutions to
assess sensitivity of the model to alternative resolutions. Start models were constructed
including either all biotic (life form and fruit type) or all abiotic (SMD, GDD, TDH,
TDD, and elevation band as a continuous variable) predictors. Because we were able to
show that parameter estimates were stable between models using different phylogenetic
resolutions, we selected one random tree for further analysis.
Model selection was based on the lowest value of the Akaike Information
Criterion (AIC) (Akaike 1974), a measure of the inverse of the model’s log-likelihood
(Burnham and Anderson 2002). AIC is a tool for model selection as it provides a measure
of the relative quality of a statistical model specific to a given set of data. Because
explanatory variables derived from temperature (i.e. thawing degree hours (TDH),
growing degree days (GDD) and thawing degree days (TDD)) measured the same
Page 10 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
11
variable in different ways, we tested for co-linearity between them using the variance
inflation factor (VIF) function in the HH R-package (Heiberger 2009). Highly co-linear
variables were not included in the same model.
We analysed intra-specific variability in FFD using a subset of species with five
or more records during the growing season in 2012. We estimated variability in two
ways. First, we used standard deviation (SD) of FFD, as conventionally used in the
literature. The SD computes the variance per species, accounting for differences in the
number of individuals among species. Second, we estimated the SD from the residuals
(SDres) of the best model explaining FFD. We used this latter approach to quantify
variability after correcting for variation in environmental conditions that cue FFD.
Last, we tested for phylogenetic conservatism in species flowering variability
using Blomberg’s K (Blomberg et al. 2003) and evaluated the relationship between mean
elevation, life form, fruit type, mean FFD and variability in FFD using PGLS. Here,
model selection was based on the lowest second-order Akaike Information Criterion
(AICc), a measure corrected for small sample size (Burnham and Anderson 2002), with
the AICcmodavg R-package (Mazerolle 2012).
Results
General description of climate and flowering patterns on Mount Irony
We found a significant negative correlation between snowmelt date (SMD) and elevation.
However, the variance explained in the model was low (R2=0.05, p<0.001) (Fig. 1),
Page 11 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
12
indicating that other factors not included in our model are also important in determining
SMD. There was no correlation between temperature and elevation.
Species differed in flowering strategy according to their traits. Dwarf shrubs
tended to flower on average 10 days earlier than herbs (ANOVA df=1, F value=109.8,
p<0.001) (Fig. 2 a)). The mean FFD for shrubs was DOY 168 (June 16) and DOY 178
(June 26) for herbaceous species. Species with fleshy fruits flowered on average four
days earlier than species with dry fruits (ANOVA df=1, F value=13.13, p<0.001) (Fig. 2
b)). The mean FFD for fleshy fruit species was DOY 171 (June 19) and DOY 175 (June
23) for dry fruit species. Accordingly, to reach first bloom, shrubs tended to have lower
GDD requirements than herbs (160 vs 258 mean GDD respectively) (ANOVA df=1, F
value=73.7, p<0.001), and fleshy fruit species tended to have lower GDD requirements
than dry fruit species (178 vs 231 mean GDD respectively) (ANOVA df=1, F
value=18.91, p<0.001). The correlation between life form and fruit type was quite high
(polychoric correlation ρ=0.596) meaning that most shrubs are fleshy fruit species and
most herbs are dry fruit species. Life form was also strongly correlated with elevation
(Fig. 2 c) (ANOVA F value=13.03, df=1, p<0.001).
Phylogenetic resolution
We incorporated intraspecific variation by adding tips to the phylogenetic tree as a
terminal polytomy, we therefore first analyzed how randomly resolving these polytomies
affected the model estimates with PGLS. The results from the 1000 PGLS replicates,
randomly resolving polytomies using short branch lengths, for the models with abiotic
and biotic explanatory variables were largely consistent across alternative resolutions
Page 12 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
13
with small variation for the estimated model statistics (Table S1). Therefore, we
arbitrarily chose a single resolved tree topology for subsequent analyses.
Models of flowering phenology
There was strong co-linearity between growing degree days (GDD) and thawing degree
days (TDD) (Variance Inflation Factor of 1189.16 and 1269.76 respectively). In addition,
thawing degree hours (TDH) provides another measure of heat sums. We therefore
compared models including each derived temperature index (TDH, GDD, TDD) in turn
along with snowmelt date (SMD), elevation and key traits (see Table 1).
Model selection by AIC identified four equally supported models explaining
variation in FFD, based on delta AIC <3 (Burnham and Anderson 2002) (Table 1). All
four models included GDD, SMD and elevation. In addition to these abiotic variables,
one model included fruit type, another included its interaction with GDD and the most
complex included all traits: fruit type and life form (Table 1). All abiotic variables were
highly significant (p<0.001), but traits were not significant within the model (all p>0.05).
The best model was able to explain 64% of the variation in FFD (Table 2). As
expected, FFD was positively correlated with GDD and SMD – later snow melt delayed
FFD. As shown in (Lessard-Therrien et al. 2013, -), plant species also bloomed earlier at
higher elevations. The high value of λ (0.902) indicates a strong influence of
phylogenetic non-independence in the model, a possible consequence of including
multiple individuals within the same species.
Phenological variability
Page 13 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
14
There was no phylogenetic signal in variability (K value<0.001, p= 0.573) measured as
the standard deviation (SD) in FFD. However, species with earlier FFD were less
variable (PGLS R2=0.45, p<0.001). Two models were equivalent based on model
selection by AICc (Table 3), but no environmental variable or trait other than FFD was
significant in the model. We obtained a similar relationship between FFD and variability
measured as the SDres from the best model explaining FFD (PGLS R2=0.46, p<0.001;
Fig. 3), which controls for covariance between FFD and environmental cues. An ML
value of λ=0 in both analyses indicates that the evolutionary relationship among species
did not add information to the explained phenological variability.
Discussion
We showed that first flowering day (FFD) in a subarctic plant community is best
explained by snowmelt date (SMD), elevation and growing degree days (GDD),
confirming the findings of Molau (1993), Kudo and Suzuki (1999), Molau et al. (2005),
Inouye and Wielgolaski (2003) and Hülber et al. (2010).. Day of first flower is earlier at
higher elevations, and delayed with later SMD. The positive correlation between
elevation and FFD might seem surprising, but can be explained by a thinner snowpack
towards the summit. The single best predictor of FFD is GDD, indicating that most
species cue flowering to temperature sums. There was no phylogenetic signal in
phenological variability; however, earlier flowering species showed significantly less
intraspecific variability in their flowering times.
Predicting first flowering dates of plant communities on Mont Irony
Page 14 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
15
We observed significant interspecific differences in flowering phenology. Dwarf shrubs
tended to bloom earlier than herbaceous plants, and have lower GDD requirements. Our
result is comparable to the findings of Molau et al. (2005), who recorded TDD
requirements as a temperature sum measure in northern Sweden, another subarctic–
tundra landscape. The mechanisms responsible for phenological differences between life
forms may be partly based on vegetative tissue renewal (Klady et al. 2011). Shrubs
produce persistent woody structures that increase nutrient storage capacity. Unlike
herbaceous species, there is no need for shrubs to regrow their structural support each
year. This storage effect reduces the influence of annual climate variability on the start of
flowering time. As shown in Figure 2 a), shrubs mostly bloom during the first half of the
growing season, while herbs have a more extended blooming period using most of the
growing season. As many shrubs produce fleshy fruits, it is an advantageous strategy to
bloom early and to have a short flowering duration to allow enough time for fruit
development and ripening before the end of the growing season (Primack 1985; Oberrath
and Böhning-Gaese 2002; Bolmgren and Lönnberg 2005). However, because life form is
strongly correlated with elevation, the relationship between life form and FFD might
alternatively reflect the co-variation of both with environment.
Across all individuals in the community, we found that FFD was best explained
by a combination of SMD, elevation and GDD. SMD initiates the growing season,
determining the moment when plants can begin receiving the light energy which they
need to trigger growth, leaf out and flowering phenology (Inouye 2008). The gradient in
elevation determines a wide range of environmental conditions that regulates community
assembly (Cornelius et al. 2013). For example, the distribution of shrubs is highly
Page 15 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
16
variable across the gradient, tall shrubs being more abundant at mid elevations, and dwarf
shrubs becoming more dominant towards the summit, where growing conditions are
harsh (Kudo and Suzuki 2002). The accumulation of energy, expressed in the form of
GDD, is the most biologically relevant cue for plants, and seems to be the main driver of
the onset of flowering in high altitude/latitude environments (Molau et al. 2005; Hülber et
al. 2010). Furthermore, GDD (measured from SMD) is a better predictor than TDD
(measured from ground temperature rising above 0° Celsius). Following Molau and
Mølgaard (1996), we assumed 0° Celsius as the most relevant base temperature for arctic
and subarctic plant species. Life processes in plants begin at temperatures as low as 0°
Celsius in the arctic and subarctic as a result of evolutionary adaptations to survive in
environments with a short growing season. However, the heat sum relevant to flowering
seems to matter only once the snow has melted. It is also noticeable that when included in
the model, the effect of fruit type or its interaction with GDD was not significant,
emphasising the overarching importance of temperature sums in cuing phenology.
Variability in first flowering dates
Our study design allowed us to characterize intraspecific variability in phenology by
using information on flowering times of individuals within a species growing in different
microhabitats and across the elevational gradient. The absence of phylogenetic signal for
phenological variability indicated that there was little evolutionary conservatism for this
trait. Similar variability was found in flowering times among both closely and less closely
related species.
Page 16 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
17
Our data were collected over a single growing season, but individuals experience
different temperature regimes both among microhabitats and across the elevational
gradient. It is therefore possible that individual flowering times may have varied while
GDD requirements remained constant. We found that FFD itself is a significant predictor
of variation in FFD, with later flowering species demonstrating more variability, but such
a relationship might also arise if species with later FFD occurred in more variable
environments. To standardise for environment, we therefore re-estimated variability in
flowering times using the standard deviation of the residuals (SDres) of the best model
explaining FFD (described above). Here, residuals represent individual differences after
controlling for variation in environmental cues. For example, species comprised of
individuals that all flowered earlier than predicted from our model for FFD would have
low variance in residuals, whereas species comprised of individuals that flower both
earlier and later than predicted from our model would have high variance in residuals.
This measure of variability within species allowed us to better explore intrinsic variability
in flowering time. Our results using SDres matched closely the results using SD of FFD:
species that flower later in the season were more variable in timing of flowering and
GDD requirements.
In contrast to our study, most previous work on phenological variability has
focused on inter-annual variation, finding that early flowering species are the most
responsive to variation in temperature (Wolkovich et al. 2012; Bolmgren et al. 2013) and,
hence, show higher variability in flowering dates between years (Fitter et al. 1995; Fitter
and Fitter 2002; Menzel et al. 2006; Mazer et al. 2013, but see Molau et al. 2005 and
Miller-Rushing and Primack 2008 for the opposite pattern). The contrast between inter-
Page 17 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
18
annual and intra-specific phenological variation can be understood if we assume that
early flowering species cue more closely to climatic factors, such as temperature (see
Bolmgren et al. 2013), than late flowering species. If individuals within early-flowering
species trigger their flowering times tightly to these specific cues, this would result in low
intraspecific variability for flowering time within a single growing season, and high
variability between years, reflecting inter-annual variation in climate. Thus variability in
FFD between seasons over several years demonstrates an opposite trend with FFD to
variability in FFD over a single growing season, but the underlying mechanisms are the
same. By looking at SDres, we evaluated variation in flowering times relative to
phenological cues, and we again showed that early flowering species demonstrated lower
variability than later flowering species, supporting our interpretation of more tightly
evolved tracking of environmental cues by early flowering species.
In contrast with early flowering species, phenological variability in late flowering
species may have less significant impact on their fitness if they have a longer flowering
duration (see Lessard-Therrien et al. 2013). If so, late flowering species would have
experienced less evolutionary pressure to decrease variability in FFD because their
flowering duration is longer than early flowering species, reducing their risk of
phenological mismatches and helping avoid potential competition for pollinators
(Mosquin 1971; Heinrich 1976). However, initial attempts to analyse phenological
change and demographic or fitness effects have found that the better a species tracks
climate change phenologically, the better it fares from a demographic or fitness
perspective (Willis et al. 2008; Cleland et al. 2012).
Methodological considerations
Page 18 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
19
Our method provides a useful approach for exploring FFD at the community level,
correcting for shared evolutionary history among species to ensure independence in the
data. Indeed, shared evolutionary history has been proven to affect phylogenetic patterns
of flowering time (Willis et al. 2008; Davis et al. 2010; Davies et al. 2013; Lessard-
Therrien et al. 2013; Mazer et al. 2013), and therefore we included phylogenetic
relationships within our analysis of flowering phenology. Here, we added individuals as
terminal taxa in order to consider within-species variation. Standard phylogenetic
comparative methods assume a single species value, although new methods where
measurement error can be incorporated, for example as variance for species' mean values,
have been developed recently (e.g. Ives et al. 2007; Felsenstein 2008). However, methods
that explicitly include intraspecific variation in models within a phylogenetic framework
are still lacking. Our method treats individuals as species separated by short phylogenetic
branch lengths. This may not be the best representation of evolutionary relationships
among individuals, which might be better depicted as a reticulate network. Nonetheless,
we believe our approach is conservative and is superior to considering individuals and
species as independent.
Conclusion
We show that dwarf shrubs and fleshy fruited species flower earlier than
herbaceous and dry fruited species, and have lower heat sum requirements. However, our
results suggest that after accounting for environment, biological traits do not significantly
predict flowering phenology. First flowering day (FFD) of plants along the gradient is
Page 19 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
20
primarily determined by snow melt, elevation and temperature sums (following snow
melt). In addition, we found lower intraspecific variability in FFD for early flowering
species, a presumed consequence of close tracking of abiotic cues. If later flowering
species have a longer flowering period, then the closer tracking of flowering time to
spring temperatures by earlier flowering species may compensate under climate warming.
Page 20 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
21
Acknowledgments
The study was supported by the Natural Sciences and Engineering Research Council, the
Fonds québécois de la recherche sur la nature et les technologies, the Northern Scientific
Training Program and the Quebec Centre for Biodiversity Science. We also thank McGill
Subarctic Research Station and Hardy B. Granberg for providing infrastructure and
Claudel Fournier for his precious help with field work.
Page 21 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
22
References
Akaike, H. 1974. A new look at the statistical model identification. IEEE Transactions
on Automatic Control, 19(6), 716‑723.
Arft, A. M., Walker, M. D., Gurevitch, J., Alatalo, J. M., Bret-Harte, M. S., Dale, M., …
Wookey, P. A. 1999. Responses of tundra plants to experimental warming:meta-
analysis of the International Tundra Experiment. Ecological Monographs, 69(4),
491‑511.
Arroyo, M, T, K., Armesto, J. J, and Villiignin. C 1981. Plant phenological patterns in the
high Andean cordillera of central Chile. Journal of. Ecology, 69, 205 223.
Billings, W. D., and Mooney, H. A. 1968. The ecology of arctic and alpine plants.
Biological Reviews, 43(4), 481‑529.
Blomberg, S. P., Garland, T., and Ives, A. R. 2003. Testing for phylogenetic signal in
comparative data: behavioral traits are more labile. Evolution, 57(4), 717‑745.
Bolmgren, K., and Lönnberg, K. 2005. Herbarium data reveal an association between
fleshy fruit type and earlier flowering time. International Journal of Plant
Sciences, 166(4), 663‑670.
Bolmgren, K., A. Miller-Rushing and D. Vanhoenacker 2013. One man, 73 years, and 25
species. Evaluating phenological responses using a lifelong study of first
flowering dates. International Journal of Biometeorology 57, 367-75.
Page 22 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
23
Brouillet, L., F. Coursol, S.J. Meades, M. Favreau, M. Anions, P. Bélisle and P. Desmet.
2012. VASCAN, la Base de données des plantes vasculaires du Canada.
http://data.canadensys.net/vascan/.
Burnham, K.P., and Anderson, D.R. 2002. Model selection and multimodel inference: a
practical information theoretic approach, Second edition. Springer-Verlag, New
York.
Cleland, E. E., Allen, J. M., Crimmins, T. M., Dunne, J. A., Pau, S., Travers, S. E., ... and
Wolkovich, E. M. 2012. Phenological tracking enables positive species responses
to climate change. Ecology, 93, 1765-1771.
Chapin, F. S., Bret-Harte, M. S., Hobbie, S. E., and Zhong, H. 1996. Plant functional
types as predictors of transient responses of arctic vegetation to global change.
Journal of Vegetation Science, 7(3), 347‑358.
Cornelius, C., Estrella, N., Franz, H., and Menzel, A. 2013. Linking altitudinal gradients
and temperature responses of plant phenology in the Bavarian Alps. Plant
Biology, 15, 57‑69.
Cornelius, C., Leingartner, A., Hoiss, B., Krauss, J., Steffan-Dewenter, I., and Menzel, A.
2012. Phenological response of grassland species to manipulative snowmelt and
drought along an altitudinal gradient. Journal of Experimental Botany, 64(1), 241
‑251.
Page 23 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
24
Davies, T. J., Wolkovich, E. M., Kraft, N. J. B., Salamin, N., Allen, J. M., Ault, T. R.,
Betancourt, J. L., Bolmgren, K., Cleland, E. E., Cook, B. I., Crimmins, T. M.,
Mazer, S. J., McCabe, G., J., Pau, S., Regetz, J., Schwartz, M. D. and Travers, S.
E. 2013. Phylogenetic conservatism in plant phenology. Journal of Ecology ,
101(6), 1520-1530.
Davis, C. C., Willis, C. G., Primack, R. B., & Miller-Rushing, A. J. (2010). The
importance of phylogeny to the study of phenological response to global climate
change. Philosophical Transactions of the Royal Society B: Biological Sciences,
365, 3201‑3213.
Debussche, M., Garnier, E., and Thompson, J. D. 2004. Exploring the causes of variation
in phenology and morphology in Mediterranean geophytes: a genus-wide study of
Cyclamen. Botanical Journal of the Linnean Society, 145(4), 469‑484.
Dunne, J. A., Harte, J., and Taylor, K. J. 2003. Subalpine meadow flowering phenology
responses to climate change: integrating experimental and gradient methods.
Ecological Monographs, 73(1), 69‑86.
Eriksson, O., and Ehrlén, J. 1991. Phenological variation in fruit characteristics in
vertebrate-dispersed plants. Oecologia, 86, 463-470.
Felsenstein, J. 2008. Comparative methods with sampling error and within‐species
variation: contrasts revisited and revised. The American Naturalist, 171(6), 713‑
725.
Page 24 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
25
Fitter, A. H. and Fitter S. R. 2002. Rapid changes in flowering time in British plants.
Science, 296, 1689‑1691.
Fitter, A. H., Fitter, S. R., Harris, I. T. B., and Williamson, M. H. 1995. Relationships
between first flowering date and temperature in the flora of a locality in central
England. Functional Ecology, 55-60.
Heiberger, R. M. 2009. HH: Statistical analysis and data display: Heiberger and
Holland. R package version, 2-1.
Heinrich, B. 1976. Flowering phenologies: bog, woodland, and disturbed habitats.
Ecology, 890-899.
Hülber, K., Winkler, M., and Grabherr, G. 2010. Intraseasonal climate and habitat-
specific variability controls the flowering phenology of high alpine plant species.
Functional Ecology, 24(2), 245‑252.
Inouye, D. W. 2008. Effects of climate change on phenology, frost damage, and floral
abundance of montane wildflowers. Ecology, 89, 353‑362.
Inouye, D.W. and Wielgolaski, F.E. 2003. High altitude climates. In: Schwartz, M.D.
(ed.) Phenology: an integrative ecological science. pp. 195–214. Kluwer
Academic Publishers, Dordrecht.
Iversen, M., Bråthen, K. A., Yoccoz, N. G., and Ims, R. A. 2009. Predictors of plant
phenology in a diverse high-latitude alpine landscape: growth forms and
topography. Journal of Vegetation Science, 20(5), 903‑915.
Page 25 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
26
Ives, A. R., Midford, P. E., and Garland, T. 2007. Within-species variation and
measurement error in phylogenetic comparative methods. Systematic Biology,
56(2), 252‑270.
Jia, P., Bayaerta, T., Li, X., and Du, G. 2011. Relationships between flowering phenology
and functional traits in eastern Tibet alpine meadow. Arctic, Antarctic, and Alpine
Research, 43(4), 585‑592.
Jónsdóttir, I. S. 2011. Diversity of plant life histories in the Arctic. Preslia, 83: 281-300.
Klady, R. A., Henry, G. H. R., and Lemay, V. 2011. Changes in high arctic tundra plant
reproduction in response to long-term experimental warming. Global Change
Biology, 17(4), 1611‑1624.
Körner, C. 2003. Alpine plant life: functional plant ecology of high mountain
ecosystems; with 47 Tables. Springer.
Kudo, G. 1991. Effects of snow-free period on the phenology of alpine plants inhabiting
snow patches. Arctic and Alpine Research, 436-443.
Kudo, G., and Suzuki, S. 1999. Flowering phenology of alpine plant communities along a
gradient of snowmelt timing. Polar Bioscience, 12, 100-113.
Kudo, G., and Suzuki, S. 2002. Relationships between flowering phenology and fruit-set
of dwarf shrubs in alpine fellfields in northern Japan: a comparison with a
subarctic heathland in northern Sweden. Arctic, Antarctic, and Alpine Research,
185-190.
Page 26 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
27
Lessard-Therrien, M., Davies T. J., and Bolmgren K. 2013. A phylogenetic comparative
study of flowering phenology along an elevational gradient in the Canadian
subarctic. International Journal of Biometeorology, 1-8.
Marie-Victorin, F. 1995. Flore laurentienne. Third edition updated by L. Brouillet, S. G.
Hay and I. Goulet in collaboration with M. Blondeau, J. Cayouette and J.
Labrecque. Edition: Les Presses de l'Université de Montréal, Montréal.
Mazer, S. M., Travers, S. E., Cook, B. I., Davies T. J., Bolmgren, K., Kraft, N. J.,
Salomin, N. and Inouye, D. I. 2013. Flowering date of taxonomic families predicts
phenological sensitivity to temperature: implications for forecasting the effects of
climate change on unstudied taxa. American Journal of Botany.
Mazerolle, M. J. 2012. AICcmodavg: Model Selection and Multimodel Inference Based
on (Q)AIC(c). R package version 1.24.
http://cran.rproject.org/web/packages/AICcmodavg/index.html.
Menzel, A. 2003. Plant phenological anomalies in Germany and their relation to air
temperature and NAO. Climatic Change, 57, 243-263.
Menzel, A., Sparks, T.H., Estrella, N., and Roy, D.B. 2006. Altered geographic and
temporal variability in phenology in response to climate change. Global Ecology
and Biogeography, 15, 498-504.
Miller-Rushing, A. J., and Inouye, D. W. 2009. Variation in the impact of climate change
on flowering phenology and abundance: an examination of two pairs of closely
related wildflower species. American Journal of Botany, 96, 1821-1829.
Page 27 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
28
Miller-Rushing, A. J., and Primack, R. B. 2008. Global warming and flowering times in
Thoreau's Concord: a community perspective. Ecology, 89, 332-341.
Molau, U. 1993. Relationships between flowering phenology and life history strategies in
tundra plants. Arctic and Alpine Research, 391-402.
Molau, U., and Mølgaard, P. 1996. International Tundra Experiment (ITEX) Manual. 2nd
ed., 6–10. Danish Polar Center Denmark, Copenhagen.
Molau, U., Nordenhall, U., and Eriksen, B. 2005. Onset of flowering and climate
variability in an alpine landscape: a 10-year study from Swedish Lapland.
American Journal of Botany, 92(3), 422‑431.
Mosquin, T. 1971. Competition for pollinators as a stimulus for the evolution of
flowering time. Oikos, 398-402.
Oberrath, R., and Böhning-Gaese, K. 2002. Phenological adaptation of ant-dispersed
plants to seasonal variation in ant activity. Ecology, 83(5), 1412‑1420.
Orme, C. D. L., Freckleton, R., Thomas, G., Petzoldt, T., Fritz, S. and Isaac, N. C. 2012.
Comparative analysis of phylogenetics and evolution in R. R package version 0.4.
URL: http://CRAN. R-project. org/package= caper.
Porsild, A. E., and W. J. Cody. 1980. Vascular plants of continental Northwest
Territories, Canada. National Museum of Natural Sciences, Ottawa.
Page 28 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
29
Primack, R. B. 1985. Patterns of flowering phenology in communities, populations,
individuals and single flowers. In: White J (ed) The Population Structure of
Vegetation, Junk, Dordrecht, 571–593.
Rathcke, B., and Lacey, E. P. 1985. Phenological patterns of terrestrial plants.Annual
Review of Ecology and Systematics, 16, 179-214.
Studer, S., Appenzeller, C., and Defila, C. 2005. Inter-annual variability and decadal
trends in alpine spring phenology: a multivariate analysis approach. Climatic
Change, 73, 395-414.
Totland, O. 1994. Intraseasonal variation in pollination intensity and seed set in an alpine
population of Ranunculus acris in southwestern Norway. Ecography, 17(2), 159‑
165.
Van Wijk, M. T., and Williams, M. 2003. Interannual variability of plant phenology in
tussock tundra: modelling interactions of plant productivity, plant phenology,
snowmelt and soil thaw. Global Change Biology, 9, 743-758.
Willis, C. G., Ruhfel, B., Primack, R. B., Miller-Rushing, A. J., and Davis, C. C. 2008.
Phylogenetic patterns of species loss in Thoreau's woods are driven by climate
change. Proceedings of the National Academy of Sciences, 105(44), 17029-17033.
Wolkovich, E.M., Cook, B.I., Allen, J.M., Crimmins, T.M., Betancourt, J.L., Travers,
S.E., Pau, S., Regetz, J., Davies, T.J., Kraft, N.J.B., Ault, T.R., Bolmgren, K.,
Mazer, S.J., McCabe, G.J., McGill, B.J., Parmesan, C., Salamin, N., Schwartz,
Page 29 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
30
M.D., and Cleland, E.E. 2012. Warming experiments underpredict plant
phenological responses to climate change. Nature 485(7399), 494–497.
Page 30 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
31
Tables captions
Table 1. Alternative PGLS models with biotic and abiotic variables explaining first
flower day (FFD) (n=39 species, 396 observations) using a single phylogeny with
intraspecific polytomies randomly resolved
Table 2. Coefficients from the models of first flowering day (FFD), from Table 1
Table 3. Selection of models with biotic and abiotic variables to explain variability in
first flower day (FFD) (n=25 species with five or more observations)
Page 31 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
32
Table 1.
# Modelsa dfb AIC ∆i
1 TDH + SMD + elevation 4 2312.35 280.41
2 TDD + SMD + elevation 4 2036.25 4.31
3 GDD + SMD + elevation 4 2033.76 1.82
4 GDD + SMD + elevation+ fruit type 5 2031.94 0.00
5 GDD * fruit type + SMD + elevation 6 2033.63 1.69
6 GDD+SMD+ elevation + life form + fruit type 6 2032.30 0.36
a: TDH; thawing degree hours, SMD; snowmelt date, GDD; growing degree days, TDD; thawing degree
days, life form and fruit type are categorical variables respectively divided in shrub/herb, fleshy/dry
b: df; degree of freedom, AIC; Akaike information criterion, ∆i; delta AIC
Page 32 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
33
Table 2.
Explanatory variables’ statistics
β SE t value p-value
(Intercept) 139.522 5.733 24.336 < 0.001 ***
GDD 0.055 0.002 25.131 < 0.001 ***
SMD 0.242 0.021 11.768 < 0.001 ***
elevation -0.018 0.003 -6.322 < 0.001 ***
Model’s statistics λ 95.0% CI Adjusted R2 p-value
0.902 0.813, 0.947 0.639 < 0.001 ***
Estimated coefficients (β), standard errors (SE), PGLS’ lambda (λ), 95% confidence intervals (CI) and
other statistic showed were calculated using phylogenetic approaches (PGLS model).
Page 33 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
34
Table 3.
# Modelsa Kb AICc ∆i
1 Mean (FFD) + mean (elevation) + life form + fruit type 6 99.71 8.48
2 Mean (FFD) + mean (elevation) + life form 5 96.97 5.74
3 Mean (FFD) + mean (elevation) 4 94.09 2.86
4 Mean (FFD) 3 91.23 0.00
a: FFD; first flowering day b: K; number of parameters, AICc; second order Akaike information criterion, ∆i; delta AICc
Page 34 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
35
Figures captions
Figure 1. Snowmelt date (SMD) along the elevational gradient on Mount Irony, Canada
during summer 2012. The earlier SMD at 710 m can be explained by the presence of a
large scree slope. Dots represent sampled communities.
Figure 2. Violin plot illustrating a) first flowering day (FFD) for herbs and shrubs, b)
first flowering day (FFD) of dry and fleshy fruit species, c) life form distribution of
angiosperms along the elevational gradient on Mount Irony, Labrador, Canada. Dwarf
shrubs are mostly found at higher elevations, whereas herbs have a fairly even
distribution along the gradient with a slight increase at lower elevations. Violin width
represents number of individuals
Figure 3. Scatter plot and line of best fit for the relationship between mean of first flower
day (FFD) and variability in FFD. Early flowering species show less variability in FFD
than late flowering species (analysis on species with five or more observations, n=25)
Page 35 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
130
135
140
145
150
155
600 650 700 750 800Elevation (m)
Sno
wm
elt d
ate
(DO
Y)
Page 36 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
a)
160
170
180
190
200
210
herb shrubLife form
Firs
t flo
wer
day
(D
OY
)
b)
160
170
180
190
200
210
dry fleshyFruit type
Firs
t flo
wer
day
(D
OY
)
c)
600
650
700
750
800
herb shrubLife form
Ele
vatio
n (m
)Page 37 of 38
Bot
any
Dow
nloa
ded
from
ww
w.n
rcre
sear
chpr
ess.
com
by
Uni
vers
ity o
f O
tago
on
09/0
9/14
For
pers
onal
use
onl
y. T
his
Just
-IN
man
uscr
ipt i
s th
e ac
cept
ed m
anus
crip
t pri
or to
cop
y ed
iting
and
pag
e co
mpo
sitio
n. I
t may
dif
fer
from
the
fina
l off
icia
l ver
sion
of
reco
rd.
2
3
4
5
160 170 180 190Mean first flower day (DOY)
Var
iabi
lity
in F
FD
Page 38 of 38B
otan
y D
ownl
oade
d fr
om w
ww
.nrc
rese
arch
pres
s.co
m b
y U
nive
rsity
of
Ota
go o
n 09
/09/
14Fo
r pe
rson
al u
se o
nly.
Thi
s Ju
st-I
N m
anus
crip
t is
the
acce
pted
man
uscr
ipt p
rior
to c
opy
editi
ng a
nd p
age
com
posi
tion.
It m
ay d
iffe
r fr
om th
e fi
nal o
ffic
ial v
ersi
on o
f re
cord
.
