Climate variability affects the germination strategies exhibited by arid land plants

Abstract

Spatial and temporal environmental variability can lead to variation in selection pressures across a landscape. Strategies for coping with environmental heterogeneity range from specialized phenotypic responses to a narrow range of conditions to generalist strategies that function under a range of conditions. Here, we ask how mean climate and climate variation at individual sites and across a species’ range affect the specialist-generalist spectrum of germination strategies exhibited by 10 arid land forbs. We investigated these relationships using climate data for the western United States, occurrence records from herbaria, and germination trials with field-collected seeds, and predicted that generalist strategies would be most common in species that experience a high degree of climate variation or occur over a wide range of conditions. We used two metrics to describe variation in germination strategies: (a) selectivity (did seeds require specific cues to germinate?) and (b) population-level variation (did populations differ in their responses to germination cues?) in germination displayed by each species. Species exhibited distinct germination strategies, with some species demonstrating as much among-population variation as we observed among species. Modeling efforts suggested that generalist strategies evolve in response to higher spatial variation in actual evapotranspiration at a local scale and in available water in the spring and annual precipitation at a range-wide scale. Describing the conditions that lead to variation in early life-history traits is important for understanding the evolution of diversity in natural systems, as well as the possible responses of individual species to global climate change.

Full text

Vol.:(0123456789)
1 3
Oecologia
DOI 10.1007/s00442-017-3958-5
POPULATION ECOLOGY – ORIGINAL RESEARCH
Climate variability affects thegermination strategies exhibited
byarid land plants
SarahBarga1,2· ThomasE.Dilts1· ElizabethA.Leger1,2
Received: 9 January 2017 / Accepted: 8 September 2017
© Springer-Verlag GmbH Germany 2017
as we observed among species. Modeling efforts suggested
that generalist strategies evolve in response to higher spatial
variation in actual evapotranspiration at a local scale and
in available water in the spring and annual precipitation at
a range-wide scale. Describing the conditions that lead to
variation in early life-history traits is important for under-
standing the evolution of diversity in natural systems, as
well as the possible responses of individual species to global
climate change.
Keywords Seed dormancy· Great Basin· Intra-specific
variation· Population-level variation· Seed traits
Introduction
Across the range of many plant species, environmental
conditions vary spatially and temporally, resulting in vari-
ation in selection pressures that can affect their growth and
establishment (Lechowicz and Bell 1991; Levine and Rees
2004; Adler etal. 2006; Treurnicht etal. 2016). The first
interaction that a plant has with its environment occurs dur-
ing the critical process of seed germination, with the reli-
ance on environmental cues at this life-history stage acting
as a potential population bottleneck (Menges 1991). Thus,
climate plays a role in shaping the evolution of seed traits
(Cochrane etal. 2015; Rosbakh and Poschlod 2015), and
the interactions between seeds and climate determine the
subsequent conditions and selection pressures experienced
during plant growth and establishment (Donohue etal. 2010;
Poschlod etal. 2013; Fraaije etal. 2015; Mondoni etal.
2015; Jiménez-Alfaro etal. 2016). Given that climate var-
ies across space and through time, it has the potential to dif-
ferentially influence the life-history strategies of populations
across the geographic range of a species (Sher etal. 2004).
Abstract Spatial and temporal environmental variability
can lead to variation in selection pressures across a land-
scape. Strategies for coping with environmental heteroge-
neity range from specialized phenotypic responses to a nar-
row range of conditions to generalist strategies that function
under a range of conditions. Here, we ask how mean climate
and climate variation at individual sites and across a species’
range affect the specialist-generalist spectrum of germina-
tion strategies exhibited by 10 arid land forbs. We investi-
gated these relationships using climate data for the western
United States, occurrence records from herbaria, and ger-
mination trials with field-collected seeds, and predicted that
generalist strategies would be most common in species that
experience a high degree of climate variation or occur over
a wide range of conditions. We used two metrics to describe
variation in germination strategies: (a) selectivity (did seeds
require specific cues to germinate?) and (b) population-level
variation (did populations differ in their responses to ger-
mination cues?) in germination displayed by each species.
Species exhibited distinct germination strategies, with some
species demonstrating as much among-population variation
Communicated by Amy Freestone.
Electronic supplementary material The online version of this
article (doi:10.1007/s00442-017-3958-5) contains supplementary
material, which is available to authorized users.
* Sarah Barga
sarahcatherinebarga@gmail.com
1 Department ofNatural Resources andEnvironmental
Science, University ofNevada, Reno, 1664 N. Virginia
Street, MS 186, Reno, NV, USA
2 Program inEcology, Evolution, andConservation Biology,
University ofNevada, Reno, Reno, NV, USA

Oecologia
1 3
Both the type and scale of environmental variation can affect
the evolution of plant life-history strategies. For example,
divergent selection in highly contrasting environments can
lead to population differentiation (Kawecki and Ebert 2004;
Sambatti and Rice 2006; Leimu and Fischer 2008; Hereford
2009), whereas high levels of environmental stochasticity at
small spatial scales can lead to the development of character-
istics that would be beneficial under a variety of conditions
(Reboud and Bell 1997; Kassen 2002; Condon etal. 2014).
Strategies for coping with environmental heterogeneity
range from increased specialization to a narrow range of
conditions, i.e., producing a fixed phenotype, to develop-
ment of the ability to exploit a broader range of conditions
through a more generalist strategy, i.e., producing a range of
phenotypes under contrasting conditions. Specialization can
be particularly advantageous if the costs of being a general-
ist are high, e.g., if specialization allows for higher resource
use efficiency (Futuyma and Moreno 1988). In contrast, by
adopting a generalist strategy, some plants may be able to
change architectural, physiological, or phenological traits in
response to year-to-year changes in environmental indicators
of resource availability (Sultan 2000). This type of pheno-
typic plasticity is thought to be adaptive when it results in
higher fitness across a range of environmental conditions
(Bradshaw 1965; Sultan 1987). Both specialist and gen-
eralist strategies have widely been documented in natural
populations (Cook and Johnson 1968; Nagy and Rice 1997;
Kassen 2002; Heschel etal. 2004; Sambatti and Rice 2006).
In fact, given the variation in plant life histories and the
ubiquity of environmental heterogeneity, it is highly likely
that most plant species achieve some balance between spe-
cialization and phenotypic plasticity among individuals in
natural plant populations (Bell etal. 2000).
Establishment from seed is a key process in plant life
cycles, and many plants have developed some degree of seed
dormancy to cope with uncertainty in their environment at
this stage (Cohen 1966; Ellner 1985; Gremer etal. 2016).
In arid systems, for example, high levels of inter-annual cli-
matic variability, in addition to inherent water-limitations,
have a strong influence on germination and seedling sur-
vival (Clauss and Venable 2000; Chesson etal. 2004; Tor-
res-Martinez etal. 2016). Moisture and temperature cues
are the most common dormancy breaking mechanisms for
desert plants (Baskin and Baskin 2014). Seed dormancy
affects the seasonal timing of germination for many desert
plants (Baskin and Baskin 2014), and germination timing
influences the environmental conditions that seedlings will
experience and when and with whom they will compete for
resources (Freas and Kemp 1983; Weinig 2000; Chesson
etal. 2004; Kos and Poschlod 2007). In the Great Basin,
where our work is focused, germination generally occurs in
either fall/winter or the spring, with some species acting as
facultative winter germinators, meaning that if they do not
experience the appropriate conditions to stimulate germina-
tion in fall/winter, then they may delay germination until the
spring. Most seed germination is stimulated by pulsed rain
events that occur in the fall or winter; however, the timing
and quantity of precipitation events in arid systems is notori-
ously variable (Comstock and Ehleringer 1992; Schwinning
etal. 2004). Therefore, the evolution of seed dormancy in
these species is potentially related to the level of environ-
mental variability a species or population experiences, and
the environmental cues that indicate the level of resource
availability at different times of the year.
In general, it is predicted that for species that experi-
ence higher levels of variation across their range (spatial
variation) than year-to-year variation within populations
(temporal variation), natural selection would favor special-
ized, fixed life-history strategies, and greater differences
among populations (Kawecki and Ebert 2004); alternately,
for species that experience high levels of year-to-year vari-
ation in combination with reliable signals of future condi-
tions, natural selection would favor phenotypically plastic
responses, and possibly more similarity among populations
(Via etal. 1995; Gabriel etal. 2005; Valladares etal. 2007).
Meyer etal. (1995) demonstrated this pattern in Penstemon
species that vary in their niche breadth and have evolved
habitat specific germination strategies at locations across
their range, with species possessing broader niches or from
more unpredictable habitats exhibiting a broader range of
germination strategies. Germination strategies may also be
affected by variation at different spatial scales, due to dif-
ferences in local vs. range-wide dynamics. Range size, and
the associated breadth of habitats encompassed by larger
ranges, may influence both the overall germination strategy
of a species and the amount of population-level variation in
germination strategies exhibited by a species, with generalist
species typically having larger ranges (Brändle etal. 2003;
Luna etal. 2012). Thus, range-wide climate variability and
range size may also be predictive of the specialist–general-
ist spectrum of germination strategies exhibited by different
species.
Here, our goals were to examine overall differences in ger-
mination strategies among a suite of Great Basin forb species,
and to relate the relative degree of specialization in their ger-
mination strategies to environmental characteristics at both
local and range-wide scales. We used two metrics to quantify
the germination responses of these species: (a) the degree of
selectivity, describing whether species were able to germinate
across a wide variety of treatments or if they responded pri-
marily to specific cues, and (b) the amount of population-level
variation in germination strategies exhibited by each species.
This allowed us to describe species along a specialist–gen-
eralist spectrum, relative to the breadth of cues that resulted
in germination, and to describe among-population differ-
ences in these germination strategies. We next asked whether

Oecologia
1 3
there was evidence that mean climate characteristics, climate
variability, or range size plays a role in shaping the special-
ist–generalist spectrum of germination strategies exhibited by
these short-lived forbs, and which climate characteristics were
most strongly associated with different germination strategies
at different scales. We investigated these relationships using
climate data for the western United States from 1950 to 2014,
herbarium records to estimate the geographic and environ-
mental ranges of these species, and germination trials with
field-collected seeds of ten Great Basin forb species, includ-
ing: Agoseris grandiflora, Blepharipappus scaber, Chaenactis
douglasii, Collinsia parviflora, Crepis intermedia, Cryptantha
pterocarya, Gilia inconspicua, Mentzelia albicaulis, Microst‑
eris gracilis, and Phacelia hastata. Specifically, we asked the
following research questions:
1. Did species exhibit a variety of germination strategies?
2. Did species exhibit population-level differences in seed
germination?
We were also interested in whether our focal species
exhibited relationships between germination responses and
environmental characteristics, asking if there was a rela-
tionship between the following predictors and the degree
of either (a) selectivity or (b) population-level variation in
germination strategies for each species:
3. Spatial climate variability or climate mean values expe-
rienced at a local scale (i.e., differences in climate at
seed collection locations).
4. Spatial climate variability or climate mean values expe-
rienced at a range-wide scale.
5. Spatial and temporal (inter-annual) variation at a local
scale or across the range of a species.
We expected that species would differ in their germina-
tion responses, with species expressing a higher degree of
selectivity experiencing lower levels of inter-annual varia-
tion across their range. We also predicted that species with
less population-level variation in germination strategies
would experience higher levels of both spatial and inter-
annual climate variability. Finally, given the primacy of this
resource in the desert, we expected that water-related vari-
ables would be the most influential in shaping the germina-
tion strategies of our focal species.
Methods
Identifying germination strategies offocal species
We selected ten forbs that are commonly found co-occurring
in sagebrush steppe ecosystems in the western Great Basin,
and are of interest as part of the spring and summer flora
that provide forage and cover for wildlife in these systems.
Seeds were wild-collected (Supplemental Table1) from 3
populations of 9 species and 2 populations for 1 species (B.
scaber) from areas with 226–757mm of annual precipita-
tion, with a mean of 406mm, across the past 64years of
weather data. Collections were centered in Northern Nevada
for 9 of 10 species (Supplemental Fig.1). Sites were visited
weekly for the purpose of seed collection throughout the
reproductive window for each species, between February
2013 and September 2013 (Table1), and seeds were stored
in the dark at room temperature (~21°C) until germina-
tion trials began. Due to low regional availability, Phacelia
hastata seeds from the National Plant Germplasm System
(United States Department of Agriculture) collections were
used to supplement collections made in 2013; these col-
lections came from two areas in south-eastern Oregon. All
seeds from an individual site were a mixture from at least
50 maternal plants. Fifteen to forty seeds from each popula-
tion (based on seed availability) were sent to the Colorado
Seed Lab (http://seeds.agsci.colostate.edu/seedlab/home-2/)
for tetrazolium testing to determine seed viability (Table1).
Because seeds were wild-collected, there is the potential for
maternal effects to influence the outcome of our germina-
tion trials (Gutterman 2000; Baskin and Baskin 2014). We
attempted to limit this influence by collecting seeds con-
sistently throughout the reproductive window for these spe-
cies. In addition, the seeds of species that reproduced in the
spring were stored at room temperature for a longer period
of time before the start of the trial than the seeds of species
that reproduced in the late spring/summer (Table1). Longer
storage times may have reduced the seed dormancy of spe-
cies with non-deep physiological dormancy (Baskin etal.
2006); however, most species and populations produced
seed from late May to mid-June. In the case of P. hastata,
extended periods of time in cold storage, as often occurs in
seed preservation, may also have affected their response to
germination treatments (Baskin etal. 2006), despite the fact
that the seeds retained high viability.
Our germination methods loosely followed those of
Forbis (2010), with treatments varying after-ripening tem-
perature and length of cold stratification (Fig.1). For the
after-ripening treatment, seeds were placed in paper coin
envelopes and were exposed to one of two treatments for
4weeks, either a dark 40°C germination chamber or in the
dark at room temperature (~21°C), to test whether exposure
to summer conditions was a dormancy breaking require-
ment. Seeds were then tested for germination in response
to cold temperatures and moist conditions, indicative of a
requirement for exposure to fall or winter conditions to break
dormancy. After-ripened seeds were divided into four cold
stratification groups and placed in a dark growth chamber at
2°C for 2, 4, or 6weeks, and then transferred to a dark 15°C

Oecologia
1 3
chamber for the remainder of the study, these treatments
are hereafter referred to as 2C2, 2C4, and 2C6. The fourth
group of seeds was placed directly into the 15°C chamber
to test whether seeds would germinate in the absence of cold
stratification; this treatment is referred to as 15C. For each
population, equal numbers of seeds (5–20 seeds, based on
availability) (Table1), were placed on filter paper (Whatman
#597) in five replicate 90mm petri dishes and moistened
with deionized water. Dishes were checked weekly for ger-
mination and deionized water was added as needed; the cold
and warm chambers were switched every 2weeks to avoid
inadvertent chamber effects. Seeds germinated in both the
2°C the 15°C chambers. Germination experiments were
conducted from late September to late December 2013, at
the point where no seeds had germinated in any dish for over
2weeks. Total germination percentage was calculated for
each species as follows:
We determined differences in the total fraction of seeds
germinated in each treatment for all populations using one-
way analysis of variance (ANOVAs), using Program R (R
Development Core Team 2016).
We also analyzed germination data using survival anal-
ysis to distinguish differences in the timing of germina-
tion. We accounted for seed viability in this analysis by
Percent germination
=Number of germinated seeds
(Number of seeds per treatment ×Percent viability
)
×100.
Table 1 Species and methodological details for ten forbs native to the Intermountain West, collected from multiple populations
Information includes life form (annual or perennial), seed collection time frame (in months), % viability of a subset of seeds, seed mass
(mean±SD), and sample sizes per dish for germination treatments. See text within the Methods for details regarding seed viability testing. Seed
mass was estimated from a sample of ten seeds per population. Acronyms are as follows: Cool AR—cool after-ripening, Hot AR—hot after-
ripening
a Seeds for this population were procured from the National Plant Germplasm System
Species Life form Population Collection months Viability (%) Seed mass (mg) Seeds/dish
Agoseris grandiflora Perennial 1. Hunter Creek June–July 100 1.7±0.2 20
2. Peavine (~5000ft) June–July 97.5 1.6±0.2 20
3. Peavine (~7500ft) July–August 95 1.6±0.3 20
Blepharipappus scaber Annual 1. Hoge Road June–July 97.5 1.4±0.3 20
2. Hunter Creek June–July 92.5 1.5±0.3 20
Chaenactis douglasii Perennial 1. Thomas Creek June–July 97.5 3.1±0.4 20
2. Peavine (~5000ft) June–July 90 2.9±0.8 20
3. Peavine (~7400ft) July–August 90 2.9±0.5 20
Collinsia parviflora Annual 1. Keystone Canyon April–May 90 2.0±0.4 15
2. Peavine (~5100ft) April–May 77 1.4±0.4 15
3. Peavine (~7100ft) May–June 87.5 1.7±0.4 15
Crepis intermedia Perennial 1. Ball’s Canyon June–July 72.5 4.5±1.4 15
2. Keystone Canyon June–July 42.5 4.3±1.9 20
3. Yorkshire Road June–July 60 4.9±1.4 20
Cryptantha pterocarya Annual 1. Prison Hill June–July 85.7 0.4±0.1 Cool AR 10, Hot AR 15
2. Peavine (~5100ft) June–July 100 0.5±0.1 5
3. Yorkshire Road June–July 86.7 0.4±0.2 15
Gilia inconspicua Annual 1. Hoge Rd May–June 90 0.9±0.2 20
2. Yorkshire Road May–June 95 0.8±0.2 20
3. Washoe Valley May–June 87.5 0.6±0.2 Cool AR 20, Hot AR 10
Mentzelia albicaulis Annual 1. Red Rock Road June–July 95 0.4±0.1 20
2. Thomas Creek June–July 97.5 0.5±0.1 20
3. Yorkshire Road June–July 100 0.5±0.2 20
Microsteris gracilis Annual 1. Hoge Road April–June 100 1.9±0.3 10
2. Peavine (~6800ft) April–June 100 1.8±0.3 5
3. Thomas Creek June–July 100 2.0±0.3 5
Phacelia hastata Perennial 1. North Owyhee RiveraJune–July 95 1.1±0.3 15
2. South Owyhee Lakea95 1.0±0.2 15
3. Thomas Creek 80 0.9±0.1 5

Oecologia
1 3
multiplying the number of seeds in each treatment group
(after-ripening and cold stratification combination) by
the percent viability of each population and removing
the appropriate number of seeds from the data set. We
removed un-germinated seeds first, and when needed,
removed germinated seeds (selected randomly). We used
the Survival package (Therneau 2015) within Program
R (3.3.1) (R Development Core Team 2016) to model
germination timing using the Surv function with interval
censoring (type=interval2), enabling us to calculate the
survival function for each seed treatment and for each
population. Germination probabilities were calculated
using the function survfit and the resulting germina-
tion curves were compared with accelerated failure time
(AFT) regressions using the survreg function with a Wie-
bull distribution (Brown and Mayer 1988). We used the
scale parameter and the coefficient from the AFT model
to calculate the hazard ratio (HR) for these comparisons
using the following equation:
Here, the HR is a ratio of the rate of germination in one
treatment relative to a comparison treatment. For exam-
ple, if seeds experiencing the hot after-ripening treatment
Hazard ratio =exp (coefficient ×−1×(1∕scale)).
germinated at twice the rate of the cool after-ripened
seeds, then the HR for that comparison would be 2.
Describing variation ingermination strategies offocal
species
For each species, we calculated metrics to describe the
variability in total germination response to treatments,
including: differences in percent germination across pop-
ulations (population-level differences) and differences in
the percent germination for each population across all ger-
mination treatments (selectivity). Our focal species are
both annual and perennial forbs, and while we did not
aim to differentiate germination strategies between peren-
nial and annual species, we have organized our results to
allow qualitative inspection of differences between these
life-history strategies. We used the coefficient of variation
(CV) as our method for quantifying variability in percent
germination, generally calculated as the standard devia-
tion divided by the mean, with higher values indicating
a higher degree of variation. To account for differences
in sample sizes when calculating the CV across differ-
ent groups (e.g., one species had only two populations),
we calculated an unbiased CV using the methods of Abdi
(2010), as follows:
where N is the number of samples from the group being
measured.
We quantified the degree of population-level variation
for each species as the CVunbiased of the percent germina-
tion across populations in response to all treatments. For
this response, lower CV values indicate that all populations
of a species experienced similar values for total percent
germination, and may indicate either uniform levels of ger-
mination or uniform lack of germination across treatments.
Conversely, higher CV values indicate that populations dif-
fered in their response to the germination treatments.
We quantified the degree of selectivity of a popula-
tion to particular germination treatments by calculating
the CV for each population across all germination treat-
ments. For this measure, lower CV values indicate that
seeds from a particular population germinated in roughly
equal quantities in response to all germination treatments,
while higher CV values indicate that seeds of that popula-
tion experienced different levels of germination in differ-
ent treatments. We then calculated the mean CV across all
populations of a species to estimate the degree of selectiv-
ity of the species. Thus, if there was germination, a lower
mean CV indicates a more flexible germination strategy,
while a higher mean CV indicates a more specialist ger-
mination strategy.
CV
unbiased =(1+
1
4×N
)×
CV,
Fig. 1 Schematic of the experimental design for germination treat-
ments. All storage and treatments took place under dark conditions.
Arrow length indicates the relative amount of time of each treatment,
and arrow width indicates proportion of seeds in each treatment (sam-
ple sizes in Table2). Box color indicates relative temperature from
high (black) to low (white). Storage and after-ripening were per-
formed under dry conditions, while cold stratification and the 15°C
treatment were performed with seeds in petri dishes on moist filter
paper. Seeds germinated in both the 2 and 15°C treatments

Oecologia
1 3
Measuring local andrange‑wide environmental
characteristics foreach species
We obtained location information across the western United
States using herbarium records downloaded from three
websites: The Intermountain Region Herbarium Network
(http://intermountainbiota.org/portal/), The Consortium of
California Herbaria (http://ucjeps.berkeley.edu/consortium/),
and the Burke Museum at the University of Washington
(http://www.burkemuseum.org/research-and-collections/
botany-and-herbarium/collections/database/). We focused
the extent of our study area on the western United States, as
many of our focal species are confined to this region, and
limited points to those representing plants that were found
from 1950 to the present, due to frequent uncertainty about
the locations of older specimens. For each species, we per-
formed geographic filtering of the occurrence points col-
lected from the herbarium data, to reduce collection bias
(Kramer-Schadt etal. 2013; Boria etal. 2014). Specifically,
we used the SDM Toolbox for ArcGIS (Brown 2014) to
remove points if their occurrence was within a 20km buffer
of another point included in the data set, in an attempt to
limit spatial auto-correlation of measurements of the envi-
ronmental variables. We gathered environmental data in
two ways. First, we tabulated 29 biologically relevant vari-
ables for each point for use in data analysis (Supplemental
Table3). Environmental variables included precipitation
and temperature, as well as a suite of bioclimatic variables
(Booth etal. 1989) derived from monthly temperature and
precipitation data, obtained from the PRISM Climate Group
at the University of Oregon, for the western United States
from 1950 to 2014 (Daly etal. 2008), creating 64-year aver-
ages. We also calculated a suite of variables using a Thornth-
waite water balance approach, which considers the simulta-
neous availability of water and energy for plants (Stephenson
1998; Lutz etal. 2010). Many of the variables were derived
from measures of actual evapotranspiration (AET), poten-
tial evapotranspiration (PET), water supply (WS), soil water
balance (SWB), and climate water deficit (CWD); most of
these variables were calculated using the methods outlined
in Dilts etal. (2015). Values for each environmental variable
were extracted in ArcMap 10.1 for each point for each spe-
cies, including locations based on herbarium records (range-
wide points) and the seed collection locations (local points).
We then calculated the CVunbiased for each environmental
variable across locations for each species and used that as a
measure of spatial climate variability for a particular species.
These measures were used to describe spatial climate varia-
tion experienced at the local scale (i.e., differences in aver-
age climate between the specific seed collection locations,
Question 3) and across each species’ range (i.e., the amount
of variation in average climate between species occurrences
documented by herbarium collections, Question 4). They
were also used to examine how mean values for climate vari-
ables may influence germination patterns (Questions 3 and
4). We calculated the average value for each environmen-
tal variable across all points for each species from 1950 to
2014 and used that as a measure of the mean climate for a
particular species.
Second, for models examining inter-annual variation
in climate variables (Question 5), we extracted monthly
PRISM data for precipitation, minimum temperature, and
maximum temperature from both seed collection locations
(local) and herbarium record locations (range-wide) for each
species, from 1950 to 2014. Because these calculations were
more computationally intensive, we focused on a subset of
easily summarized variables, rather than calculating vari-
ation in composite and derived variables described above.
We summed the precipitation over the months of each sea-
son for each year and averaged the minimum and maximum
temperatures over the months of each season for each year.
We calculated spatial variation for each species by calculat-
ing the CVunbiased across all points for each season of each
year and taking the mean of these values across all years for
each season. We calculated the temporal variation for each
species by calculating the CVunbiased across all years for each
season at each point and taking the mean of these values
across all points. Finally, we estimated range size and niche
breadth for each species (Supplement 1). The Pearson’s cor-
relation between niche breadth and range size was 0.97, so
we chose to exclude niche breadth from these models and
retain range size, as this variable produces an intuitive meas-
ure in units that can be easily compared among species and
across studies.
Identifying relationships betweengermination
strategies andenvironmental characteristics atlocal
andrange‑wide scales
We used generalized linear models to determine relation-
ships between environmental characteristics, range size, and
seed germination strategies. We performed a Pearson’s cor-
relation analysis to determine which variables were highly
correlated (R>±0.7) across our collective species’ ranges,
and narrowed the focal set of variables down to a subset
of uncorrelated variables for each model (Supplemental
Table2). When selecting variables for each model, we
placed an emphasis on maintaining similar variables across
models. Once a group of variables was selected, Q–Q plots
were used to confirm a normal distribution within the data
parameters and plots of residual versus fitted values were
used to check for trends within the residuals for each of the
models, and values were transformed as needed.
To address our questions, we created three sets of general-
ized linear models. These included: models examining spa-
tial climate variability from 64-year averages of bioclimatic

Oecologia
1 3
variables (Questions 3 and 4), models examining mean val-
ues of bioclimatic variables (Questions 3 and 4), and mod-
els examining spatial and temporal climate variability from
monthly measures of precipitation and temperature (Ques-
tion 5). For each set of questions, we ran separate models
that used either (a) selectivity or (b) population-level differ-
ences in germination strategies as response variables.
We analyzed these generalized linear models using multi-
model inference, performed using the package MuMIn for
program R (Barton 2016). We performed model selection
using the dredge function to generate a set of candidate mod-
els, each containing no more than five terms, using combina-
tions of the variables from the previously described global
models for each question. We then performed model aver-
aging across all models produced by the model selection
process. This allows us to obtain estimates of the regression
coefficients that are averaged across all models, with each
value weighted by the corrected Akaike information crite-
rion (AICc) scores for the models that contained it. We used
both zero averaging (ZA; assigns a parameter estimate of
zero to predictor variables that are excluded from a particu-
lar model and includes those zero values when performing
model averaging) and natural averaging (NA, only averages
across models that contain that particular predictor variable)
to estimate individual parameters. The ZA approach is bet-
ter for assessing the relative importance of all parameters
from the global model, whereas the NA approach is better
for determining the importance of an individual parameter
(Burnham and Anderson 2002; Grueber etal. 2011). Model
averaging also provides an estimate of parameter importance
(IMP) for each of the predictor variables, which is based on
the proportion of highly predictive models that contain the
focal parameter. Higher IMP values indicate that a parameter
was either included in more models and/or was included in
highly predictive models.
Results
Did focal species exhibit avariety ofgermination
strategies (Question 1)?
Overall, focal species exhibited distinct differences in the
total number of seeds that germinated in response to treat-
ments (Fig.2). Four species (A. grandiflora, C. parviflora, C.
pterocarya, and M. gracilis) appeared to possess generalist
germination strategies and experienced very high levels of
germination in response to all treatments. Two other species,
P. hastata and M. albicaulis, experienced very low levels
of germination in all treatments, indicating that they may
require additional cues to completely break dormancy. The
remaining species fell somewhere in-between. Two species
preferred longer periods of cold stratification (C. intermedia
and C. douglasii), one species preferred no cold stratification
(G. inconspicua), and one species achieved a moderate level
of germination in all treatments (B. scaber).
Our results indicated that hot temperatures were more
likely to affect the timing of germination, rather than the
total quantity of germinated seeds. While hot after-ripening
stimulated faster germination in A. grandiflora, M. graci‑
lis, and P. hastata, the total number of germinated seeds
did not differ between after-ripening treatments (Table2).
For C. douglasii, cool after-ripening resulted in both faster
germination and higher total number of germinated seeds
(Table2).
Our results also indicated that winter and spring condi-
tions have the potential to affect both the rate of germination
and the total number of germinated seeds, a result consistent
with other germination research (Meyer etal. 1995; Forbis
2010; Baskin and Baskin 2014). Species with high levels
of germination during cold stratification (potential winter
germinators) included A. grandiflora, B. scaber, C. parvi‑
flora, C. pterocarya, and M. gracilis. None of these species
experienced significant differences in total germination in
response to the cold stratification treatments, although they
did show differences in their rates of germination (Table2).
A. grandiflora germinated more quickly in the 15°C treat-
ment, B. scaber experienced faster germination in the cold
stratification treatments, and C. pterocarya experienced a
high initial rate of germination in 15°C, with a little dif-
ference between the rates of germination in the cold strati-
fication treatments (Table2). Some species germinated
more quickly in warmer temperatures after exposure to cold
stratification (Table2): C. douglasii, C. intermedia, and P.
hastata. These species also exhibited significant differences
in total germination in response to the cold stratification
treatments, with C. douglasii experiencing higher germina-
tion with longer periods of cold stratification, and C. inter‑
media and P. hastata experiencing higher germination in all
cold stratification treatments (Table2). Only G. inconspicua
exhibited a higher rate of germination coupled with higher
overall seed germination when placed directly into 15°C
(Table2). Although M. albicaulis experienced a relatively
high rate of germination in the 15°C treatment, it exhibited
very low levels of germination overall, with no significant
difference in total germination between cold stratification
treatments (Table2).
Did focal species exhibit population‑level variation
inseed germination (Question 2)?
All but one species, C. parviflora, experienced significant
population-level differences in the rate of germination and/
or the total seeds germinated (Table2; Fig.3). Two species
with generalist strategies, A. grandiflora and C. pterocarya,
exhibited significant differences in the rates of germination

Oecologia
1 3
between populations without exhibiting differences in the
total number of seeds germinated (Table2). Another species
with a generalist strategy, M. gracilis, exhibited both popu-
lation-level differences in germination rates and in the total
number of seeds germinated (Table2). The most dramatic
population-level differences were exhibited by C. inter‑
media in both germination rate and total germinated seeds
(Table2). B. scaber exhibited population-level differences
in both the rate and the total seeds germinated (Table2).
Two other species, C. douglasii and G. inconspicua, had one
of their populations which germinate at a much faster rate
than the other two populations, as well as a higher level of
germination (Table2). Finally, the two species that exhibited
low levels of germination, P. hastata and M. albicaulis, still
showed population-level differences in germination rates and
total seeds germinated, but may have required additional
cues to completely break dormancy (Table2; Fig.3). We
quantified both species and population-level variation in the
fraction of seeds germinated (Table3).
Are there relationships betweengermination responses
andenvironmental characteristics?
Spatial climate variability andmean climate experienced
atalocal scale (Question 3)
At the local scale, the natural average (NA) of spatial varia-
tion in cumulative AET was negatively correlated with both
0.00
0.25
0.50
0.75
1.00
Species
Mean fraction germinated
Treatment
Cool/15C
Cool/2C2
Cool/2C4
Cool/2C6
Hot/15C
Hot/2C2
Hot/2C4
Hot/2C6
0.00
0.25
0.50
0.75
1.00
AGGR CHDO CRIN PHHA
BLSC COPA CRPT GIIN MEAL MIGR
Species
Mean fraction germinate
d
(b)
(a)
Fig. 2 Fraction of germinated seeds for each species/treatment
combination (mean±SE across all populations of each species) for
a perennial species and b annual species, after accounting for seed
viability for each population. Treatments included either hot or cool
after-ripening, followed by cold stratification for 0 (15C), 2 (2C2),
4 (2C4), or 6 (2C6) weeks. Species acronyms indicate the following
species: AGGR, Agoseris grandiflora; CHDO, Chaenactis douglasii;
CRIN, Crepis intermedia; PHHA, Phacelia hastata; BLSC, Blephari‑
pappus scaber; COPA, Collinsia parviflora; CRPT, Cryptantha
pterocarya; GIIN, Gilia inconspicua; MEAL, Mentzelia albicaulis;
and MIGR, Microsteris gracilis

Oecologia
1 3
Table 2 Statistical results for differences in rate and total germination in response to treatments and among populations for (a) perennials and (b) annuals
Hazard ratios (HR) from the survival analysis indicate pair-wise differences between the rates of germination, with numbers greater than one indicating faster germination and numbers less than
one indicating slower germination for the treatment in the numerator of the comparison. For total germination, F-statistics for ANOVAs are shown, with degrees of freedom indicated as follows:
1F(1,118), 2F(3,116), 3F(2,117), and 4F (1,78)
Significance levels are indicated as follows: t P<0.10, *P<0.05, **P<0.01, and ***P<0.001
Arrows indicate whether treatments increased or decreased germination. The total germination column in the population differences analysis indicates when populations (Pop) differed in total
germination, with population numbers as described in Table1
A “–” indicates no differences among treatments or populations
Treatments Population differences
After-ripening Cold stratification
Rate (HR) Rate (HR) Rate (HR)
Species Hot/cool Total germination 2C2/15C 2C4/15C 2C6/15C Total germination Pop 2/Pop 1 Pop 3/Pop 1 Pop 3/Pop 2 Total germination
(a)
Agoseris grandiflora 1.15*** – 0.36*** 0.44*** 0.40*** – 1.18** 0.82*** 0.69*** –
Chaenactis douglasii 0.68*** ↑cool 4.041* 1.00 4.24*** 6.57*** ↓15C 65.342*** 1.10 1.85*** 1.69*** ↑Pop3—4.603*
Crepis intermedia 0.94 – 2.16*** 4.58*** 3.70*** ↓15C 22.752*** 3.43*** 5.93*** 1.73*** ↓Pop1 ↑Pop3 27.243***
Phacelia hastata 1.36* – 3.31*** 4.24*** 5.89*** ↓15C 8.122*** 0.56** 2.18*** 3.92*** ↓Pop2 ↑Pop3 14.523***
(b)
Blepharipappus scaber 1.03 – 1.33** 1.45*** 1.37** – 0.50*** N/A N/A ↓Pop1 81.664***
Collinsia parviflora 1.00 – 0.85* 0.87t0.78** – 0.92 1.00 1.08 –
Cryptantha pterocarya 0.99 – 0.31*** 0.29*** 0.30*** – 1.52*** 0.50*** 0.33*** ↓Pop3 ↑Pop2 2.933t
Gilia inconspicua 0.87 – 0.07*** 0.07*** 0.14*** ↑15C 62.222*** 1.03 3.05*** 2.95*** ↑Pop3 12.723***
Mentzelia albicaulis 1.02 – 0.42** 0.40*** 0.69t↑15C 2.422t 0.17*** 0.26*** 1.49 ↑Pop1 21.713***
Microsteris gracilis 1.18* – 0.36*** 0.38*** 0.45*** – 1.57*** 1.69*** 1.07 ↓Pop1 3.923***

Oecologia
1 3
selectivity (Question 3a) and population-level differences in
germination (Question 3b) (Table4a), indicating that species
collected from environments with greater spatial variation in
annual productivity had more generalist germination strate-
gies and smaller differences among populations. In contrast,
mean values of these bioclimatic variables had no associa-
tion with either selectivity (Question 3a) or population-level
variation (Question 3b) at the local scale (see Supplemental
Table3a).
Spatial climate variability andmean climate experienced
atarange‑wide scale (Question 4)
At a range-wide scale, variation in available water in the
spring and annual precipitation were negatively correlated
with selectivity (Question 4a) (Table4b), indicating that
species that experienced higher spatial variability in the
amount of water available for runoff and deep percolation in
the spring and in total annual precipitation across their range
were more likely to have generalist strategies. At the range-
wide scale, none of the variables from the global model were
significantly correlated with population-level variation in
germination (Question 4b). Mean values of these bioclimatic
variables had no association with either selectivity (Question
4a) or population-level variation (Question 4b) at the range-
wide scale (see Supplemental Table3b).
Spatial andtemporal (inter‑annual) variation atalocal
scale andrange‑wide scale (Question 5)
Models did not indicate a significant relationship between
the degree of spatial or temporal (inter-annual) variation in
seasonal variables (precipitation, minimum temperature, or
maximum temperature) and either selectivity or population-
level differences at either a local or a range-wide scale (see
Supplemental Table4).
Discussion
Seed germination is a key element of a plant’s response to
its environment, and variation in seed germination strategies
is commonly observed among species (Meyer etal. 1995;
Petru and Tielborger 2008; Forbis 2010; Baskin and Baskin
2014). Consistent with these observations, we found that our
focal species exhibited a variety of germination strategies,
encompassing both generalist and specialist germination
traits. Much less work has been done describing differences
among populations, though this type of variation may be
important for species persistence in response to climate vari-
ability (Cochrane etal. 2015). We found evidence for popu-
lation-level differences in germination strategies for nine out
of ten species, all except for C. parviflora. Given that these
(d)
Phacelia hastata
North Owyhee
South Owyhee
Thomas Creek
(g)Cryptantha pterocarya
Prison Hill
Peavine Low
Yorkshire
(f) Collinsia parviflora
Keystone
Peavine Low
Peavine High
Probability of germinating
0.0
0.2
0.4
0.6
0.8
1.0
(e) Blepharipappus scaber
Hoge Road
Hunter Creek
Time (d)
(i) Mentzelia albicaulis
Red Rock
Thomas Creek
Yorkshire
(c)
Crepis intermedia
Ball's Canyon
Keystone
Yorkshire
(b)
Chaenactis douglasii
Thomas Creek
Peavine Low
Peavine High
0.0
0.2
0.4
0.6
0.8
1.0
(a)
Agoseris grandiflora
Hunter Creek
Peavine Low
Peavine High
7213549637791721 35 49 63 77 91 7213549637791
0.0
0.2
0.4
0.6
0.8
1.0
(h)Gilia inconspicua
Hoge Road
Yorkshire
Washoe Valley
(j) Microsteris gracilis
Hoge Road
Peavine
Thomas Creek
Fig. 3 Time-to-event plots of seed germination for each population of (a–d) perennial and (e–j) annual species across all treatments. Ninety-five
percent confidence intervals are indicated by the colored area around the curves for each population

Oecologia
1 3
seeds were collected over a small area, relative to the poten-
tial ranges of these species, it is interesting to discover that
populations can exhibit dramatic differences in germination
strategies at this small spatial scale. Our results support the
findings of other research reporting population-level differ-
ences in plant traits (Sambatti and Rice 2006; Becker etal.
2008; Banta etal. 2012; Granado-Yela etal. 2013; Prendev-
ille etal. 2013; Torres-Martinez etal. 2016), and there is
increasing interest in research focusing on population-level
variation in plant early life-history traits (Cochrane etal.
2015; Jiménez-Alfaro etal. 2016).
As we predicted, there were relationships between germi-
nation strategies and climate variability at different spatial
scales, primarily related to water availability and the simul-
taneous availability of water and energy (AET). We found
evidence that spatial variation in AET at a local scale influ-
enced both the selectivity in germination response and the
degree of population-level differences in germination exhib-
ited by our focal species, with increases in spatial variation
in AET associated with decreases in both selectivity and
population-level differences. Given that AET is a proxy
for productivity, this supports the idea that natural selec-
tion would favor a generalist strategy in populations that
experience high spatial variability in resource availability.
This may be due to population-level variation in competitive
pressures (Kadmon and Shmida 1990), but may also be due
to other local characteristics, such as edaphic factors (Wright
etal. 2006) or other biotic and abiotic factors (Linhart and
Grant 1996).
At the range-wide scale, species that experienced higher
spatial variation in available water in the spring and annual
precipitation across their range had more generalist germina-
tion strategies, with seeds ready to germinate in response to
available moisture, rather than waiting for specific tempera-
ture/moisture combinations. The fact that different variables
were important at local and range-wide scales supports the
idea that resource availability, mediated by factors such as
competition and edaphic characteristics, is generally more
Table 3 Coefficient of variation (CV) of fraction of total germinated seeds for (a) perennial and (b) annual forbs, with population numbers as
described in Table1
See text within the "Methods" section for details regarding calculating the values within this table. Higher CV values indicate higher variation in
total seed germination among treatments or populations
(a) Agoseris
grandiflora Chaenactis
douglasii Crepis
intermedia Phacelia
hastata
CV by species across treatments: 0.010 0.773 0.455 0.788
CV by population across treatments:
1 0.009 0.675 0.912 0.620
2 0.010 1.056 0.281 0.705
3 0.011 0.687 0.116 0.554
Selectivity: 0.010 0.806 0.436 0.626
Fraction germinated by population:
1 0.989 0.273 0.480 0.143
2 0.994 0.307 0.843 0.083
3 0.986 0.450 0.944 0.281
CV by species across populations: 0.004 0.296 0.350 0.652
(b) Blepharipappus
scaber Collinsia
parviflora Cryptantha
pterocarya Gilia
inconspicua Mentzelia
albicaulis Microsteris
gracilis
CV by species across treatments: 0.349 0.000 0.017 0.957 1.147 0.013
CV by population across treatments:
1 0.396 0.000 0.000 1.392 0.721 0.018
2 0.094 0.000 0.000 0.601 0.914 0.000
3 0.000 0.027 1.137 0.447 0.000
Selectivity: 0.245 0.000 0.009 1.044 0.694 0.006
Fraction germinated by population:
1 0.371 1.000 1.000 0.231 0.113 0.983
2 0.639 1.000 1.000 0.239 0.020 1.000
3 1.000 0.983 0.526 0.030 1.000
CV by species across populations: 0.307 0.000 0.011 0.549 1.007 0.011

Oecologia
1 3
influential on plant fitness at smaller spatial scales (Turk-
ington and Harper 1979; Snaydon and Davies 1982; Becker
etal. 2008), while climate factors are generally more influ-
ential at larger spatial scales (Santamaria etal. 2003; Macel
etal. 2007, but see Carta etal. 2016). We did not find evi-
dence linking population-level variation in germination
response with range-wide climate variability; this may be
due to the fact that these arid land species may cue into dif-
ferent aspects of climate (Chesson etal. 2004), leading to
individualized responses that make it difficult to find general
patterns between specific climate variables and variation in
germination strategies. This emphasizes the importance of
studying the unique natural histories and adaptations of indi-
vidual species (Macel etal. 2007), in addition to searching
for large-scale patterns in life-history strategies.
Germination strategies can provide a means for tracking
suitable conditions through time by delaying seed germi-
nation until conditions improve (Gremer etal. 2016). Bet
hedging is a germination strategy where plants sacrifice
their mean fitness in a single year to increase their long-
term fitness across years (Cohen 1966; Venable 2007). With
this strategy, a plant produces seeds that can be separated
into different groups, or seed fractions, that each germinate
in response to different cues, enabling the plant to spread
germination across several years and the risk of seedling
failure through time. This strategy is thought to be an adap-
tive response to environmental variability (Nevoux etal.
2010), and is well documented in desert annuals (Cohen
1966; Venable 2007; Gremer etal. 2016). It is possible that
some germination strategies may integrate elements of both
bet hedging and phenotypic plasticity (Simons 2014; Botero
etal. 2015). Thus, some species may have displayed popu-
lation-level variation in their germination strategies due to
bet hedging; most notably, B. scaber exhibited a moderate
level of germination in response to most of our germination
treatments, a pattern that would be consistent with a bet
Table 4 Effects of (a) local and (b) range-wide spatial climate vari-
ability on the degree of selectivity in response to treatments and the
population-level differences in seed germination demonstrated by ten
forb species native to the Intermountain West, determined using a
model averaging approach
Standardized parameter estimates from the naturally averaged model (NA) and the zero-averaged model (ZA) are shown
See text for details regarding the relative parameter importance (IMP) and methods for measuring climate variability
Acronyms for climate variables are defined as follows: AET actual evapotranspiration, SWB soil water balance, CWD climate water deficit, PPT
precipitation
Values in bold indicate a significant relationship (P<0.05)
Selectivity Population-level difference
NA ZA IMP NA ZA IMP
(a)
Range size 3.38E−04 6.20E−05 0.18 3.40E−04 7.74E−05 0.23
Fraction of AET from PPT 9.02E−01 4.18E−02 0.05 8.80E−01 4.02E−02 0.05
Available water in the spring 9.69E−01 1.38E−01 0.14 1.01E+00 3.29E−01 0.32
SWB −1.37E−01 −7.58E−03 0.06 3.55E−01 2.42E−02 0.07
Minimum temperature 9.70E+01 8.41E+00 0.09 −1.30E+02 −1.12E+01 0.09
Summer PPT 1.71E+00 1.82E−01 0.11 3.03E−01 1.15E−02 0.04
Annual PPT −1.49E+00 −2.46E−01 0.17 −4.07E−01 −3.16E−02 0.08
AET −2.15E+00 −1.41E+00 0.66 −1.79E+00 −1.11E+00 0.62
Selectivity Population-level difference
NA ZA IMP NA ZA IMP
(b)
Range size 2.10E−04 1.32E−05 0.06 Range size 3.04E−04 3.71E−05 0.12
Fraction of AET from PPT −1.62E+00 −1.17−01 0.07 Fraction of AET from PPT −2.43E−01 −1.58E−02 0.07
Available water in the spring −3.61E+00 −2.38E+00 0.66 Available water in the spring −8.38E−01 −5.61E−02 0.07
SWB −8.68E−02 −4.38E−03 0.05 SWB −1.80E−01 −1.90E−02 0.11
Minimum temperature −3.72E+01 −2.63E+00 0.07 Minimum temperature −1.77E+01 −1.33E+00 0.07
Summer PPT −4.87E−01 −2.38E−02 0.05 log (summer PPT) −6.21E−01 −4.36E−02 0.07
log (annual PPT) −2.35E+00 −1.05E+00 0.45 Annual PPT −1.37E+00 −2.63E−01 0.19
AET:CWD −2.81E−02 −2.09E−03 0.07 AET:CWD −2.29E−02 −2.02E−03 0.09
Relative drought 9.61E−01 5.05E−02 0.05 Relative drought 1.88E+00 4.63E−01 0.25

Oecologia
1 3
hedging strategy. Further research on the survival of germi-
nating seeds in contrasting habitats could be used to model
overall success of the generalist/specialist strategies which
we observed among these species.
Finally, we know that different germination strategies
influence the timing of germination, and that this can affect
seedling success (Rathcke and Lacey 1985; Pake and Ven-
able 1996; Donohue etal. 2005) and have lasting conse-
quences over the lifetime of a plant (Rathcke and Lacey
1985). In general, facultative winter germination and winter
germination prioritize appearing early in the growing sea-
son; these are strategies adopted by species that are highly
competitive (Raynal and Bazzaz 1975; Winsor 1983) or that
grow in areas where competition for resources is inherently
low. In contrast, species with spring germination may be
better at tolerating harsh and summer conditions, and may
benefit from the lower level of competition for resources
presented later in the growing season. At shorter timescales,
differences in timing of days to weeks can also affect over-
all plant survival and fitness, both in general (Baskin and
Baskin 1972; Warwick and Briggs 1978; Marks and Prince
1981) and in arid systems in particular (Leger etal. 2009;
Kulpa and Leger 2013). There is also evidence that the order
of emergence may be more important than the emergence
date in determining seedling success for some species (War-
wick and Briggs 1978; Weaver and Cavers 1979). Thus, spe-
cies may partition resources in space and time by expressing
different germination strategies, allowing for a diversity of
plants to persist in resource limited systems (Chesson etal.
2004; Moreira etal. 2012).
In summary, our research demonstrates a link between
climate variability and generalist life-history strategies, and
demonstrates how climate may influence intra-specific vari-
ability in seed germination. As expected, species experienc-
ing higher levels of environmental variation exhibited more
generalist strategies, and variation in water-related variables
wasan important predictor of where species occurred along
the specialist–generalist spectrum of life-history strategies.
We also observed that co-occurring species can possess dis-
tinct germination strategies, and that populations can also
vary in their germination strategies as much or more than
the strategies of different species. Because of the key role
that early life-history characteristics play in a species’ inter-
actions with its environment and the influence of germina-
tion timing on plant species persistence, knowledge of these
strategies will become increasingly important in the face of
climate change (Cochrane etal. 2015; Mondoni etal. 2015;
Jiménez-Alfaro etal. 2016; Doherty etal. 2017).
Acknowledgements We would like to thank the Great Basin Native
Plant Project for their generous funding and the Germplasm Research
Information Network/National Plant Germplasm System (GRIN/
NPGS) for providing me with hand-collected seeds for my work with
Phacelia hastata. Brittany Trimble, Lyndsey Boyer, Travis Allen, Vicki
Thill, and Brianna Koorman provided valuable assistance with seed
germination monitoring.
Author contribution statement SCB and EAL conceived and
designed the experiments. SCB performed the germination experi-
ments. TED generated climate data and assisted with analyses using
geospatial information and tools. SCB analyzed the data. SCB and EAL
wrote the manuscript, and TED edited the manuscript.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict
of interest.
Human and animal rights statement This article does not contain
any studies with human participants or animals performed by any of
the authors.
References
Abdi H (2010) Coefficient of variation. In: Salkind N (ed) Encyclope-
dia of research design. Sage, Thousand Oaks, pp 169–171
Adler PB, Hillerislambers J, Kyriakidis PC, Guan Q, Levine JM, Til-
man GD (2006) Climate variability has a stabilizing effect on the
coexistence of prairie grasses. Proc Natl Acadamy Sci 103:12793–
12798. doi:10.1073/pnas.0600599103
Banta JA, Ehrenreich IM, Gerard S, Chou L, Wilczek A, Schmitt J,
Kover PX, Purugganan MD (2012) Climate envelope modelling
reveals intraspecific relationships among flowering phenology,
niche breadth and potential range size in Arabidopsis thaliana.
Ecol Lett 15:769–777. doi:10.1111/j.1461-0248.2012.01796.x
Barton K (2016) MuMIn: multi-model inference, v. 1.15.6. https://
cran.r-project.org/package=MuMIn
Baskin JM, Baskin CC (1972) Influence of germination date on sur-
vival and seed production in a natural population of Leavenwor‑
thia stylosa. Am Midl Nat 88:318–323. doi:10.2307/2424357
Baskin CC, Baskin JM (2014) Seeds: ecology, biogeography, and evo-
lution of dormancy and germination, 2nd edn. Academic Press,
San Diego
Baskin CC, Thompson K, Baskin JM (2006) Mistakes in germina-
tion ecology and how to avoid them. Seed Sci Res 16:165–168.
doi:10.1079/SSR2006247
Becker U, Dostal P, Jorritsma-Wienk LD, Matthies D (2008) The
spatial scale of adaptive population differentiation in a wide-
spread, well-dispersed plant species. Oikos 117:1865–1873.
doi:10.1111/j.0030-1299.2008.16939.x
Bell G, Lechowicz MJ, Waterway MJ (2000) Environmental hetero-
geneity and species diversity of forest sedges. J Ecol 88:67–87.
doi:10.1046/j.1365-2745.2000.00427.x
Booth TH, Searle SD, Boland DJ (1989) Bioclimatic analysis to assist
provenance selection for trials. New For 3:225–234. doi:10.1007/
BF00028930
Boria RA, Olson LE, Goodman SM, Anderson RP (2014) Spatial fil-
tering to reduce sampling bias can improve the performance of
ecological niche models. Ecol Modell 275:73–77. doi:10.1016/j.
ecolmodel.2013.12.012
Botero CA, Weissing FJ, Wright J, Rubenstein DR, Pacala SW
(2015) Evolutionary tipping points in the capacity to adapt to
environmental change. Proc Natl Acadamy Sci 112:184–189.
doi:10.1073/pnas.1408589111

Oecologia
1 3
Bradshaw A (1965) The evolutionary significance of phenotypic
plasticity in plants. Adv Genet 13:115–155. doi:10.1016/
s0065-2660(08)60048-6
Brändle M, Stadler J, Klotz S, Brandl R (2003) Distributional range size
of weedy plant species is correlated to germination patterns. Ecol-
ogy 84:136–144. doi:10.1890/0012-9658(2003)084[0136:DRSO
WP]2.0.CO;2
Brown JL (2014) SDMtoolbox: a python-based GIS toolkit
for landscape genetic, biogeographic and species distri-
bution model analyses. Methods Ecol Evol 5:694–700.
doi:10.1111/2041-210X.12200
Brown R, Mayer D (1988) Representing cumulative germination. 2.
The use of the Weibull function and other empirically derived
curves. Ann Bot 61:127–138. doi:10.1093/oxfordjournals.aob.
a087535
Burnham K, Anderson D (2002) Model selection and multi-model
inference: a practical information-theoretic approach. Springer,
New York
Carta A, Probert R, Puglia G, Peruzzi L, Bedini G (2016) Local cli-
mate explains degree of seed dormancy in Hypericum elodes L.
(Hypericaceae). Plant Biol 18:76–82. doi:10.1111/plb.12310
Chesson P, Gebauer RLE, Schwinning S, Huntly N, Wiegand K, Ernest
MSK, Sher A, Novoplansky A, Weltzin JF (2004) Resource
pulses, species interactions, and diversity maintenance in arid and
semi-arid environments. Oecologia 141:236–253. doi:10.1007/
s00442-004-1551-1
Clauss MJ, Venable DL (2000) Seed germination in desert annuals:
an empirical test of adaptive bet hedging. Am Nat 155:168–186.
doi:10.1086/303314
Cochrane A, Yates CJ, Hoyle GL, Nicotra AB (2015) Will among-
population variation in seed traits improve the chance of species
persistence under climate change? Glob Ecol Biogeogr 24:12–24.
doi:10.1111/geb.12234
Cohen D (1966) Optimizing reproduction in a randomly
varying environment. J Theor Biol 12:119–129.
doi:10.1016/0022-5193(66)90188-3
Comstock JP, Ehleringer JR (1992) Plant adaptations in the Great Basin
and Colorado Plateau. Gt Basin Nat 52:195–215
Condon C, Cooper BS, Yeaman S, Angilletta MJ (2014) Temporal vari-
ation favors the evolution of generalists in experimental popula-
tions of Drosophila melanogaster. Evolution (N Y) 68:720–728.
doi:10.1111/evo.12296
Cook SA, Johnson MP (1968) Adaptation to heterogeneous environ-
ments. I. Variation in heterophylly in Ranunculus flammula L.
Evolution (N Y) 22:496–516. doi:10.1111/j.1558-5646.1968.
tb03988.x
Daly C, Halbleib M, Smith JI, Gibson WP, Doggett MK, Taylor GH,
Curtis J, Pasteris PP (2008) Physiographically sensitive mapping
of climatological temperature and precipitation across the conter-
minous United States. Int J Climatol 28:2031–2064. doi:10.1002/
joc.1688
Dilts TE, Weisberg PJ, Dencker CM, Chambers JC (2015) Functionally
relevant climate variables for arid lands: a climatic water deficit
approach for modelling desert shrub distributions. J Biogeogr
42:1986–1997. doi:10.1111/jbi.12561
Doherty KD, Butterfield BJ, Wood TE (2017) Matching seed to site by
climate similarity: techniques to prioritize plant materials devel-
opment and use in restoration. Ecol Appl. doi:10.1002/eap.1505
Donohue K, Dorn L, Griffith C, Kim E, Aguilera A, Polisetty
CR, Schmitt J (2005) The evolutionary ecology of seed ger-
mination of Arabidopsis thaliana: variable natural selec-
tion on germination timing. Evolution (N Y) 59:758–770.
doi:10.1111/j.0014-3820.2005.tb01751.x
Donohue K, Rubio De Casas R, Burghardt L, Kovach K, Wil-
lis CG (2010) Germination, postgermination adaptation, and
species ecological ranges. Annu Rev Ecol Evol Syst 41:293–319.
doi:10.1146/annurev-ecolsys-102209-144715
Ellner S (1985) ESS germination strategies in randomly varying envi-
ronments. I. Logistic-type models. Theor Popul Biol 28:50–79.
doi:10.1016/0040-5809(85)90022-x
Forbis TA (2010) Germination phenology of some Great Basin
native annual forb species. Plant Species Biol 25:221–230.
doi:10.1111/j.1442-1984.2010.00289.x
Fraaije RGA, ter Braak CJF, Verduyn B, Breeman LBS, Verhoeven JTA,
Soons MB (2015) Early plant recruitment stages set the template
for the development of vegetation patterns along a hydrological
gradient. Funct Ecol 29:971–980. doi:10.1111/1365-2435.12441
Freas KE, Kemp PR (1983) Some relationships between environmen-
tal reliability and seed dormancy in desert annual plants. J Ecol
71:211–217. doi:10.2307/2259973
Futuyma DJ, Moreno G (1988) The evolution of ecological speciali-
zation. Annu Rev Ecol Syst 19:207–233. doi:10.1146/annurev.
ecolsys.19.1.207
Gabriel W, Luttbeg B, Sih A, Tollrian R (2005) Environmental tol-
erance, heterogeneity, and the evolution of reversible plastic
responses. Am Nat 166:339–353. doi:10.1086/432558
Granado-Yela C, Balaguer L, García-Verdugo C, Carrillo K, Méndez
M (2013) Thriving at the limit: differential reproductive perfor-
mance in range-edge populations of a Mediterranean sclerophyll
(Olea europaea). Acta Oecologica 52:29–37. doi:10.1016/j.
actao.2013.07.002
Gremer JR, Kimball S, Venable DL, Pannell J (2016) Within-and
among-year germination in Sonoran Desert winter annuals: bet
hedging and predictive germination in a variable environment.
Ecol Lett 19:1209–1218. doi:10.1111/ele.12655
Grueber CE, Nakagawa S, Laws RJ, Jamieson IG (2011) Multimodel
inference in ecology and evolution: challenges and solutions. J
Evol Biol 24:699–711. doi:10.1111/j.1420-9101.2010.02210.x
Gutterman Y (2000) Maternal effects on seeds during development.
In: Fenner M (ed) Seeds: the ecology of regeneration in plant
communities, 2nd edn. CABI Publishing, Wallingford, pp 59–84
Hereford J (2009) A quantitative survey of local adaptation and fitness
trade-offs. Am Nat 173:579–588. doi:10.1086/597611
Heschel MS, Sultan SE, Glover S, Sloan D (2004) Population differ-
entiation and plastic responses to drought stress in the general-
ist annual Polygonum persicaria. Int J Plant Sci 165:817–824.
doi:10.1086/421477
Jiménez-Alfaro B, Silveira FAO, Fidelis A, Poschlod P, Commander
LE (2016) Seed germination traits can contribute better to plant
community ecology. J Veg Sci 27:637–645. doi:10.1111/jvs.12375
Kadmon R, Shmida A (1990) Competition in a variable environment:
an experimental study in a desert annual plant population. Isr J
Bot 39:403–412
Kassen R (2002) The experimental evolution of specialists, general-
ists, and the maintenance of diversity. J Evol Biol 15:173–190.
doi:10.1046/j.1420-9101.2002.00377.x
Kawecki TJ, Ebert D (2004) Conceptual issues in local adaptation.
Ecol Lett 7:1225–1241. doi:10.1111/j.1461-0248.2004.00684.x
Kos M, Poschlod P (2007) Seeds use temperature cues to ensure ger-
mination under nurse-plant shade in xeric Kalahari savannah. Ann
Bot 99:667–675. doi:10.1093/aob/mcl293
Kramer-Schadt S, Niedballa J, Pilgrim JD, Schröder B, Lindenborn
J, Reinfelder V, Stillfried M, Heckmann I, Scharf AK, Augeri
DM, Cheyne SM, Hearn AJ, Ross J, Macdonald DW, Mathai J,
Eaton J, Marshall AJ, Semiadi G, Rustam R, Bernard H, Alfred
R, Samejima H, Duckworth JW, Breitenmoser-Wuersten C, Belant
JL, Hofer H, Wilting A (2013) The importance of correcting for
sampling bias in MaxEnt species distribution models. Divers Dis-
trib 19:1366–1379. doi:10.1111/ddi.12096

Oecologia
1 3
Kulpa SM, Leger EA (2013) Strong natural selection during plant
restoration favors an unexpected suite of plant traits. Evol Appl
6:510–523. doi:10.1111/eva.12038
Lechowicz MJ, Bell G (1991) The ecology and genetics of fitness in
forest plants. II. Microspatial heterogeneity of the edaphic envi-
ronment. J Ecol 79:687–696. doi:10.2307/2260661
Leger EA, Espeland EK, Merrill KR, Meyer SE (2009) Genetic vari-
ation and local adaptation at a cheatgrass (Bromus tectorum)
invasion edge in western Nevada. Mol Ecol 18:4366–4379.
doi:10.1111/j.1365-294X.2009.04357.x
Leimu R, Fischer M (2008) A meta-analysis of local adaptation in
plants. PLoS ONE 3:1–8. doi:10.1371/journal.pone.0004010
Levine JM, Rees M (2004) Effects of temporal variability on rare
plant persistence in annual systems. Am Nat 164:350–363.
doi:10.1086/422859
Linhart YB, Grant MC (1996) Evolutionary significance of local
genetic differentiation in p lants. Annu Rev Ecol Syst 27:237–277.
doi:10.1146/annurev.ecolsys.27.1.237
Luna B, Pérez B, Torres I, Moreno JM (2012) Effects of Incubation
temperature on seed germination of Mediterranean plants with
different geographical distribution ranges. Folia Geobot 47:17–27.
doi:10.1007/s12224-011-9110-0
Lutz JA, Van Wagtendonk JW, Franklin JF (2010) Climatic water deficit,
tree species ranges, and climate change in Yosemite National Park.
J Biogeogr 37:936–950. doi:10.1111/j.1365-2699.2009.02268.x
Macel M, Lawson CS, Mortimer SR, Šmilauerova M, Bischoff A,
Crémieux L, Doleźal J, Edwards AR, Lanta V, Bezemer TM,
Van Der Putten WH, Igual JM, Rodriguez-Barrueco C, Müller-
Schärer H, Steinger T (2007) Climate vs. soil factors in local
adaptation of two common plant species. Ecology 88:424–433.
doi:10.1890/0012-9658(2007)88[424:CVSFIL]2.0.CO;2
Marks M, Prince S (1981) Influence of germination date on survival
and fecundity in wild lettuce Lactuca serriola. Oikos 36:326–330.
doi:10.2307/3544630
Menges ES (1991) Seed germination percentage increases with popula-
tion size in a fragmented prairie species. Conserv Biol 5:158–164.
doi:10.1111/j.1523-1739.1991.tb00120.x
Meyer SE, Kitchen SG, Carlson SL (1995) Seed germination timing
patterns in Intermountain Penstemon (Scrophulariaceae). Am J
Bot 82:377–389. doi:10.2307/2445584
Mondoni A, Pedrini S, Bernareggi G, Rossi G, Abeli T, Probert RJ,
Ghitti M, Bonomi C, Orsenigo S (2015) Climate warming could
increase recruitment success in glacier foreland plants. Ann Bot
116:907–916. doi:10.1093/aob/mcv101
Moreira B, Tavsanoglu C, Pausas J (2012) Local versus regional
intraspecific variability in regeneration traits. Oecologia 168:671–
677. doi:10.1007/S00442-01
Nagy ES, Rice KJ (1997) Local adaptation in two subspecies of an
annual plant: implications for migration and gene flow. Evolution
(N Y) 51:1079–1089. doi:10.2307/2411037
Nevoux M, Forcada J, Barbraud C, Croxall J, Weimerskirch H (2010)
Bet-hedging response to environmental variability, an intraspecific
comparison. Ecology 91:2416–2427. doi:10.1890/09-0143.1
Pake CE, Venable DL (1996) Seed banks in desert annuals: implica-
tions for persistence and coexistence in variable environments.
Ecology 77:1427–1435. doi:10.2307/2265540
Petru M, Tielborger K (2008) Germination behaviour of annual plants
under changing climatic conditions: separating local and regional
environmental effects. Oecologia 155:717–728. doi:10.1007/
s00442-007-0955-0
Poschlod P, Abedi M, Bartelheimer M, Drobnik J, Rosbakh S, Saat-
kamp A (2013) Seed ecology and assembly rules in plant commu-
nities. In: van der Maarel E, Franklin J (eds) Vegetation Ecology,
2nd edn. Wiley, Chichester, pp 164–202
Prendeville HR, Barnard-Kubow K, Dai C, Barringer BC, Galloway
LF (2013) Clinal variation for only some phenological traits
across a species range. Oecologia 173:421–430. doi:10.1007/
s00442-013-2630-y
R Development Core Team (2016) R: A language and environment for
statistical computing. R http://www.r-project.org/. Vienna, Austria
Rathcke B, Lacey EP (1985) Phenological patterns of terrestrial
plants. Annu Rev Ecol Syst 16:179–214. doi:10.1146/annurev.
es.16.110185.001143
Raynal DJ, Bazzaz FA (1975) Interference of winter annuals with
Ambrosia artemisiifolia in early successional fields. Ecology
56:35–49. doi:10.2307/1935298
Reboud X, Bell G (1997) Experimental evolution in Chlamydomonas.
Ill. Evolution of specialist and generalist types in environments
that vary in space and time. Heredity (Edinb) 78:507–514.
doi:10.1038/sj.hdy.6881210
Rosbakh S, Poschlod P (2015) Initial temperature of seed germination
as related to species occurrence along a temperature gradient.
Funct Ecol 29:5–14. doi:10.1111/1365-2435.12304
Sambatti JBM, Rice KJ (2006) Local adaptation, patterns of selection,
and gene flow in the California serpentine sunflower (Helianthus
exilis). Evolution (N Y) 60:696–710. doi:10.1554/05-479.1
Santamaria L, Figuerola J, Pilon JJ, Mjelde M, Green AJ, De Boer T,
King RA, Gornall RJ (2003) Plant performance across latitude:
the role of plasticity and local adaptation in an aquatic plant. Ecol-
ogy 84:2454–2461. doi:10.1890/02-0431
Schwinning S, Sala OE, Loik ME, Ehleringer JR (2004) Thresholds,
memory, and seasonality: understanding pulse dynamics in arid/
semi-arid ecosystems. Oecologia 141:191–193. doi:10.1007/
s00442-004-1683-3
Sher AA, Goldberg DE, Novoplansky A (2004) The effect of mean
and variance in resource supply on survival of annuals from
Mediterranean and desert environments. Oecologia 141:353–362.
doi:10.1007/s00442-003-1435-9
Simons AM (2014) Playing smart vs. playing safe: the joint expres-
sion of phenotypic plasticity and potential bet hedging across
and within thermal environments. J Evol Biol 27:1047–1056.
doi:10.1111/jeb.12378
Snaydon RW, Davies TM (1982) Rapid divergence of plant popula-
tions in response to recent changes in soil conditions. Source Evol
36:289–297. doi:10.2307/2408047
Stephenson NL (1998) Actual evapotranspiration and defi-
cit: biologically meaningful correlates of vegetation dis-
tribution across spatial scales. J Biogeogr 25:855–870.
doi:10.1046/j.1365-2699.1998.00233.x
Sultan SE (1987) DNA evolutionary implications of phenotypic plastic-
ity in plants. In: Hecht MK, Wallace B, Prance GT (eds) Evolu-
tionary biology, vol 21. Springer, New York, pp 127–178
Sultan SE (2000) Phenotypic plasticity for plant development, func-
tion, and life history. Trends Plant Sci 5:537–542. doi:10.1016/
s1360-1385(00)01797-0
Therneau TM (2015) A package for survival analysis in S, v. 2.38.
https://cran.r-project.org/package=survival
Torres-Martinez L, Weldy P, Levy M, Emery NC (2016) Spatiotem-
poral heterogeneity in precipitation patterns explain population-
level germination strategies in an edaphic specialist. Ann Bot
1–13:253–265. doi:10.1093/aob/mcw161
Treurnicht M, Pagel J, Esler KJ, Schutte-Vlok A, Nottebrock H, Kraaij
T, Rebelo AG, Schurr FM (2016) Environmental drivers of demo-
graphic variation across the global geographical range of 26 plant
species. J Ecol 104:331–342. doi:10.1111/1365-2745.12508
Turkington R, Harper JL (1979) The growth, distribution and neigh-
bour relationships of Trifolium repens in a permanent pas-
ture: IV. Fine-scale biotic differentiation. J Ecol 67:245–254.
doi:10.2307/2259348

Oecologia
1 3
Valladares F, Gianoli E, Gomez JM (2007) Ecological lim-
its to plant phenotypic plasticity. New Phytol 176:749–763.
doi:10.1111/j.1469-8137.2007.02275.x
Venable DL (2007) Bet hedging in a guild of desert annuals. Ecology
88:1086–1090. doi:10.1890/06-1495
Via S, Gomulkiewicz R, De Jong G, Scheiner SM, Schlichting CD,
Van Tienderen PH (1995) Adaptive phenotypic plasticity: consen-
sus and controversy. Trends Ecol Evol 10:212–217. doi:10.1016/
s0169-5347(00)89061-8
Warwick SI, Briggs D (1978) The genecology of lawn weeds: II. Evi-
dence for disruptive selection in Poa annua L. in a mosaic envi-
ronment of bowling green lawns and flower bedsI. New Phytol
81:725–737. doi:10.1111/j.1469-8137.1978.tb01647.x
Weaver SE, Cavers PB (1979) The effects of date of emergence and
emergence order on seedling survival rates in Rumex crispus and
R. obtusifolius. Can J Bot 57:730–738. doi:10.1139/b79-092
Weinig C (2000) Differing selection in alternative competition environ-
ments: shade-avoidance responses and germination timing. Evolu-
tion (N Y) 54:124–136. doi:10.1111/j.0014-3820.2000.tb00013.x
Winsor J (1983) Persistence by habitat dominance in the annual
Impatiens capensis (Balsaminaceae). J Ecol 71:451–466.
doi:10.2307/2259727
Wright JW, Stanton ML, Scherson R (2006) Local adaptation to ser-
pentine and non-serpentine soils in Collinsia sparsiflora. Evol
Ecol Res 8:1–21