Content uploaded by John R. Butnor
Author content
All content in this area was uploaded by John R. Butnor on Jan 07, 2020
Content may be subject to copyright.
128
Phenotypic Variation in Climate-Associated Traits
of Red Spruce (Picea rubens Sarg.)
along Elevation Gradients in the Southern
Appalachian Mountains
John R. Butnor,1* Brittany M. Verrico,2 Kurt H. Johnsen,3 Christopher A. Maier,4
Victor Vankus,5 and Stephen R. Keller2
1USDA Forest Service, Southern Research Station, 81 Carrigan Drive,
Aiken Center, University of Vermont, Burlington, VT 05405
2University of Vermont, Department of Plant Biology, Burlington, VT
3USDA Forest Service, Southern Research Station, Bent Creek Exp. Forest, NC
4USDA Forest Service, Southern Research Station, RTP, NC
5USDA Forest Service, National Seed Laboratory, Dry Branch, GA
ABSTRACT
Red spruce (Picea rubens) is a long-lived tree species that thrives in cool, moist environs. Its ability
to adapt to rapidly changing climate is uncertain. In the southern Appalachian Mountains, red
spruce reaches its greatest abundance at high elevations, but can also occur across a range of
mid and lower elevations, suggesting the possibility of a correlation between genetic variation and
habitat. To assess clinal phenotypic variation in functional traits related to climate adaptation, we
collected seed from 82 maternal sib families located along replicated elevational gradients in the
Great Smoky Mountains National Park, TN (GSMNP) and Mount Mitchell State Park, NC (MMSP).
The percentage of filled seeds and seed mass increased with elevation, indicating that successful
pollination and seed development was greatest at the highest elevations. Seedlings sourced from
GSMNP displayed a strong relationship between elevation and bud set when grown under common
garden conditions. High elevation families set bud as many as 10 days earlier than low elevation
families, indicating adaptation to local climate. Across parks, no eect of elevation was noted for
bud flush. Our results demonstrate that red spruce in the southern Appalachian Mountains dis-
plays clinal variation in bud set that may reflect local adaptation to climate, although this varied
between the two parks sampled. We suggest that genetic adaption of red spruce to dierent climate
regimes, at both local and broad spatial scales, is in need of more intensive study, and should be
carefully considered when selecting seed sources for restoration.
Key words: adaptation, phenology, Picea rubens, red spruce, southern Appalachians
INTRODUCTION
Climate change is likely to pose significant threats to many endemic or regionally restricted species
in Appalachian forest ecosystems. Among the most vulnerable species are those inhabiting the high
elevation spruce-fir forests of the southern Appalachians (Virginia, North Carolina, and Tennessee),
as these communities form mountain-top “sky islands” where upslope migration in response to
climate change is limited. For this reason, migration on its own is unlikely to be a sucient response
for isolated high-elevation populations facing climate change, and a response to selection on existing
genetic variation in climate-adaptive traits will likely be necessary (Aitken et al. 2008, Vitt et al.
2010). Such genetic variation, if present, also has the potential to play an important role in resto-
ration eorts, and has historically been used in other regions to delineate “seed zones” used by
*email address: john.butnor@usda.gov
Received 6 February 2019; Accepted 12 June 2019
CASTANEA 84(2): 128–143 JUNE
Copyright 2019 Southern Appalachian Botanical Society
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 129
foresters and restoration ecologists in guiding climate-informed planting strategies (McKenney et
al. 2009, Bower et al. 2014, Thomas et al. 2014). Thus, there is a critical need to obtain estimates
of genetic variation in functional traits related to climate adaptation in foundational species within
the high-elevation spruce-fir ecosystem.
Red spruce (Picea rubens Sarg.) is an iconic conifer species of high elevation forests in the
southern Appalachians, where it provides critical habitat to a variety of other plant and animal
species (Rentch et al. 2007, Fortney et al. 2015, Diggins and Ford 2017, Walter et al. 2017). Red
spruce thrives in cool, moist environments; there are concerns that it may be unable to adapt to
land-use alteration and climate change (Iverson et al. 2008, Beane 2010, Andrews 2016). Pollen
records from southern Appalachian bog sediment cores show that Picea spp. became regionally
restricted to the higher elevations of the Cumberland and Allegheny Plateaus from 9,000 to 4,000
years ago as spruce migrated to higher latitudes and elevations in response to the warming climate
of the mid-Holocene (Delcourt and Delcourt 1984). The current populations of red spruce in the
southern Appalachians are locally isolated on high elevation ridgetops, disjunct from the more
abundant and well-connected northern populations in New York, New England, and the Canadian
Maritime provinces (Figure 1A). Heavy logging followed by severe fires in the early 20th century
resulted in >90% reduction in spruce-fir forests in the southern Appalachians (Korstian 1937).
As such, red spruce is the focus of multi-agency restoration eorts aimed at restoring red spruce
forests to high elevation landscapes in the central and southern Appalachians, i.e., the Central
Appalachian Spruce Restoration Initiative (CASRI; www.restoreredspruce.org) and Southern
Appalachian Spruce Restoration Initiative (SASRI; www.southernspruce.org).
Considering that the southern Appalachians were not glaciated and Picea species were long-
term occupants of glacial refugia in this region (Davis 1980), southern red spruce populations may
have evolved genetic variation in climate-adaptive traits associated with growth and phenology.
Southern populations also occur across a range of elevations (900 m to 2000 m) resulting in steep
climatic gradients in growing season length, sometimes over relatively short geographical distances
(Brown 1941, Crandall 1958, Schofield 1960, Cogbill and White 1991). As a result, red spruce located
along elevational gradients likely have experienced historic and ongoing divergent selection pressures
in response to varying local climate conditions, but it is unknown whether this has led to genetic and
phenotypic dierentiation in climate-adaptive traits. Identifying the potential genetic capacity of
red spruce to respond to climate change is critical to informing conservation strategies not only in
the southern Appalachians but also across its entire geographical range. Climate change has put
southern populations at risk of local extinction as upslope migration opportunities are limited.
Considering that red spruce achieves its greatest growth potential and stem quality in the southern
Appalachians (Korstian 1937, Nowacki et al. 2010), these populations could provide valuable seed
sources in light of continued climate change in the northern part of the range.
In this study, we test for phenotypic variation in seed and seedling traits related to climate
adaptation in southern Appalachian red spruce. We collected open-pollinated seed from maternal
half-sib families from multiple sites distributed along replicated elevation gradients in North
Carolina and Tennessee. Using elevation as a proxy for climate, we tested for a relationship between
source elevation and early life history traits related to seed quality, vegetative bud phenology and
growth that form important components of seedling performance under varying climate conditions.
Our results provide a first glimpse at the presence of climate-adaptive phenotypic variation along
elevation gradients in southern Appalachian red spruce, and call for further study of genetic vari-
ation and its potential for guiding seed selection for restoration of red spruce in this imperiled
ecosystem.
METHODS
Seed Collection and Analysis
Cones were collected from 82 red spruce trees growing naturally at elevations ranging from 1036 to
1988 m in North Carolina and Tennessee from 12–16 September, 2016 (Figures 1B, 1C, respectively;
130 Castanea, Vol. 84(2) 2019
Table 1). As cones were open-pollinated, all seeds from a given mother tree are considered maternal
families (hereafter, “families”) consisting of a potential mix of full and half-sibs with the same
mother but unknown fathers. Collections were made from trees thought to be naturally regenerated
using documentary evidence and cues such as uneven age stand structure and irregular spatial
distribution i.e. not row planted. Cone collections were centered on Mount Mitchell State Park
(MMSP) and Great Smoky Mountains National Park (GSMNP), as they are each populated with
red spruce across a broad elevation and climate gradient (Figures 1B, 1C; Table 1). The three sites
at MMSP were within 5 aerial km of each other. The high elevation Deep Gap Trail spans the ridge-
line from Mount Mitchell to Mount Craig and beyond to Big Tom. Disturbance history maps of
Mount Mitchell developed by Pyle and Schafale (1988) indicate that the west side of the ridge was
uncut as of 1916, while the east side was cut and burned. There was no evidence of planting in
this area and collections were made primarily from second growth trees. The Commissary Ridge
Trail that leads to Camp Alice was proximal (~1 km), but not immediately adjacent to areas that
were previously cut and planted with a variety of species in the 1920s (Minckler 1940). Cone col-
lections were primarily from naturally regenerated trees following harvest. The Mitchell trail from
1711 m to ~1400 m extends beyond the disturbance history maps by Pyle and Schafale (1988), but
the trees are markedly older and likely survived by being too small at the time of the destructive
Figure 1. Natural range of red spruce1 in eastern North America (A). Insets show location of mother trees and topo-
graphic features at Mount Mitchell State Park (B), and Great Smoky Mountains National Park (C).
1Public domain image. USGS Geosciences and Environmental Change Science Center: Digital Representations of Tree Species Range Maps
from “Atlas of United States Trees” by Elbert L. Little, Jr. 1971
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 131
Table 1. Collection of seed cones from red spruce naturally occurring along elevational gradients in the
southern Appalachians.
State Site Families Families Elevation MAT1 Latitude Longitude
Collected Propagated (m) (°C)
(#) (#)
NC Commissary Ridge Trail, 6 3 1736–1840 7.8–7.1 35.752 -82.276
MMSP2
NC Deep Gap Trail, MMSP 10 5 1897–1988 6.7–6.0 35.781 -82.260
NC Mitchell Trail, MMSP 19 8 1198–1711 11.5–7.9 35.754 -82.245
NC Andrew’s Bald Trail, 19 6 1734–1842 7.8–7.0 35.547 -83.494
GSMNP3
NC Heintooga Ridge Road, 10 7 1036–1627 12.6–8.5 35.575 -83.173
GSMNP
TN New Found Gap, 4 2 1314–1330 10.7–10.6 35.624 -83.430
GSMNP
TN Road Prong Trail, 14 7 1100–1617 12.2–8.0 35.615 -83.457
GSMNP
1Mean annual air temperature (MAT) was estimated using data (1998-2018) from four land-based stations:
Mount LeConte, TN (1979 m), Mount Mitchell, NC (1902 m), New Found Gap, TN (1536 m) and Cherokee, NC
(1036 m). After gap-filling, linear regression was used to predict temperature by elevation: MAT=19.75–0.0069 *
elevation (m); R2=0.99, p=0.001. Data were accessed from NOAA Climate Data Online, August 6, 2018 (https://
www.ncdc.noaa.gov/cdo-web/).
2Mount Mitchell State Park (MMSP)
3Great Smoky Mountains National Park (GSMNP)
harvesting. At GSMNP, three sites were within 8 aerial km, while the remaining site (Heintooga
Ridge Road) was, at its most distant point, 29 km away. Pyle and Schafale (1988) produced a de-
tailed map of the disturbance history near Clingman’s Dome and the vicinity of the Road Prong
trail. Our collections from the Andrews Bald trail were from naturally regenerated second growth
trees adjacent to an older uncut stand, to the north. Road Prong Trail had not been harvested and
featured very large, old trees. Collections from both the New Found Gap Road and Heintooga Ridge
Road were made from near the roadside. GSMNP had no documentary evidence that these areas
had red spruce plantings.
After collection, cones were dried and the seeds were extracted and processed at the USDA
Forest Service National Seed Laboratory in Dry Branch, Georgia. A subset of seeds from each family
was counted and weighed to yield mean seed mass prior to separating filled from unfilled seed,
and 200 seeds were imaged with a Faxitron Ultra Focus x-ray system (Faxitron Bioptics, LLC,
Tucson, Arizona) to estimate the percentage of filled seed out of the total seed count for each
family (Figure 2). Seeds are considered filled if the x-ray indicates they contain all tissues and
morphological features required for germination. Percentage of filled seed is an indirect indicator of
viability, but is not as definitive as a measure of germination, embryo growth, or metabolic activity
would be. The seed lot for each family was then cleaned with a blower to remove debris and light-
weight material such as unfilled seeds, followed by another subset of seeds per family counted and
weighed to yield clean seed mass. The seeds were then kept refrigerated at 4°C until planting.
Germination and Propagation
A subset of 38 red spruce families were selected for germination and propagation to maximize
the spatial distribution and elevation gradient at each park (Table 1). The dierence in latitude
between collection sites within each transect was minor (<0.25°). Twenty filled seeds from each
family were placed in deionized water to imbibe for germination. After 24 hours, the seeds were
drained and placed on moist blotter paper in Petri dishes and lightly sprinkled with sand to
aid wicking. The dishes were incubated in an Achieva precision tabletop light/dark germinator
132 Castanea, Vol. 84(2) 2019
(Seedburo Equipment Co., Des Plaines, Illinois) in darkness for 16 hours at 20°C followed by 30°C
in light for 8 hours (AOSA 2016). Germination was observed every 2–3 days, and recorded when
the emerging radicle was 2–3 mm in length. Seeds were germinated in two rounds beginning
on 17 January, 2017, and 10 February, 2017, as they became available from the National Seed
Laboratory. Germinated seeds were transplanted from the dish to Ray Leach conetainer pots
(model sc10) in a peat-perlite-vermiculite soilless media. The seedlings were maintained in a
Conviron PGR15 growth chamber with supplemental light (8 hours at 18°C dark followed by 16
hours at 25°C light) before being moved in May 2017 to an outdoor nursery under shade cloth (50%
light transmission) at the University of Vermont (UVM) greenhouse. While outside, the seedlings
were exposed to ambient temperatures for Burlington, Vermont, were watered daily and received
periodic water soluble fertilizer applications at a concentration of 150 ppm (19-3-18; N-P-K ratio).
After seedlings had naturally set bud outdoors in fall 2017, they were overwintered in a cold room
of the UVM greenhouse minimally heated to maintain above freezing temperatures with no supple-
mental lighting. In April 2018 they were again placed outdoors under shade cloth and allowed to
accumulate growing degree days leading up to spring bud flush.
Phenological and growth measurements
To quantify genetic variation in vegetative phenology and growth among families when exposed to
common environment conditions, we measured the onset of dormancy as bud set and the release
Figure 2. X-ray of red spruce seeds. Filled seeds appear white and details of embryonic structure are evident, while
unlled seeds remain dark and empty.
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 133
of spring dormancy and the start of active shoot growth as bud flush. Bud set was determined by
observing individual seedlings every 2–3 days during fall 2017 and recording the Julian date when
bud scales were clearly visible on the terminal shoot. After the terminal bud had set, the length of
the stem covered by live foliage was recorded as live crown. Bud flush was determined in spring
2018 by observing individual seedlings every 2–3 days, and recording the Julian date that bud
scales had broken and newly emerged needle tissue was evident.
Statistical Analyses
We used linear mixed-eects and generalized linear mixed-eects modeling fit by maximum likeli-
hood to test for significant variation in bud phenology, growth, filled seed, and germination as a
result of collection elevation, park, and the interaction of elevation and park. In each model, family
nested within park was included as a random eect to account for multiple seedlings per family.
Likelihood ratio tests were used to determine significance of the random eect when family was
removed from the model. In the analysis of seedling growth, we included the round in which seed-
lings were germinated as an additional fixed eect. Seed mass was collected at the family level,
hence we used linear models with collection elevation, park, and the interaction of elevation and
park as fixed eects. Pearson’s correlation coecient was used to define the bivariate correlation
between seed mass and germination percentage by park. Analyses were performed using R version
3.4.0 (R Core Team 2017) with packages lme4 (Bates et al. 2015) and lmerTest (Kuznetsova et al.
2017). Outlier detection was accomplished using the Cook’s Distance statistic, which combines in-
formation about the residuals with the degree of leverage that an observation has on the regression
equation (Cook 1977).
RESULTS
Seed properties
The mass of clean, filled seeds was highly variable (range 1.7 mg–5.3 mg per seed) and increased
significantly with elevation (Table 2, Figure 3A). The percentage of filled seed determined by radio-
graphic imaging showed a marginally significant positive trend with elevation (p=0.0703; Table 2).
When data from both parks were combined and analyzed with linear regression, 8% of the variation
in filled seed was explained by elevation (p=0.0092; Figure 3B). For both GSMNP and MMSP, the
very low percentages of filled seed at elevations <1400 m are driving the positive linear relationship
(Figure 3B). Germination percentages of cleaned, filled seed varied widely after 4 weeks (5%–95%;
mean across families = 45%). Significant eects of park (p<0.0001) and park*elevation (p=0.00028)
revealed a strong negative linear relationship between germination and elevation of seed origin
at MMSP, but not at GSMNP (Table 2, Figure 3C). There was no significant correlation between
seed mass and germination at either GSMNP (p=0.2840) or MMSP (p=0.8813). Significant variation
among families was found for filled seed (p<0.0001) and germination percentages (p<0.0001).
Phenology
Significant eects of elevation were identified for bud set (p=0.0108), but not bud flush (p=0.6724)
(Figure 4A, 4B; Table 2). When grown under common environmental conditions, high elevation
seedlings from GSMNP set bud as much as 10 days earlier than seedlings from low elevation
sources. Elevation explained 30% of the variation in fall bud set at GSMNP (Figure 4A). Two poten-
tial outliers at low elevation were observed (Cook’s Distances of 1.25 and 1.6), indicating they were
highly influential on the regression equation, whereas the remaining observations had minimal
leverage as outliers (Cook’s Distances of <0.2). When those two observations are removed, the re-
lationship between elevation and bud set strengthened considerably at GSMNP (Figure 4A), where
elevation accounted for 64% of the variation in bud set. It is notable that a strong relationship be-
tween elevation and bud set was not found at MMSP (Figure 4A). There was not a significant eect
of family for bud set (p=0.2471) or bud flush (p=0.6404), indicating a lack of significant genetic vari-
ation among families beyond what is already explained by the elevation gradient (e.g., for bud set).
134 Castanea, Vol. 84(2) 2019
Table 2. Summary of general and mixed-effect linear models analyses describing the effect of elevation,
park, and the interaction of park*elevation on seed traits, seedling growth, and bud phenology. The fixed
effect “round” represents two different planting dates and was only included for the dependent variable
live crown. Family was treated as a random effect and significance was tested with likelihood ratio tests.
Some variables required re-scaling elevation to mean = 0, SD = 1 to achieve model convergence. p values
significant at the 0.05 level are indicated with bold font
Model Effects Estimate1 Std. Error t or χ2 p
Seed Mass
Elevation 1.63 E-6 3.70 E-7 4.436 <0.0001
Park 1.24 E-3 1.05 E-3 1.187 0.2392
Park*Elevation -6.10 E-7 6.20 E-7 -0.975 0.3330
Full Seed
Elevation2 0.2947 0.1628 1.810 0.0703
Park -0.3341 0.2508 -1.332 0.1828
Park*Elevation 0.2277 0.2546 0.894 0.3713
Family 1.1560 -- 2866.7 <0.0001
Germination
Elevation2 0.2794 0.1613 1.732 0.08319
Park 1.2392 0.2822 4.391 <0.0001
Park*Elevation -1.0795 0.2974 -3.629 0.00028
Family 0.3370 -- 18.885 <0.0001
Bud Set
Elevation -0.0089 0.0034 -2.638 0.0108
Park -15.5522 7.5912 -2.049 0.0489
Park*Elevation 0.0099 0.0047 2.119 0.0414
Family 1.9250 -- 1.340 0.2471
Bud Flush
Elevation 0.0010 0.0024 0.425 0.6724
Park 3.6345 5.0453 0.720 0.4786
Park*Elevation -0.0020 0.0031 -0.626 0.5369
Family 0.4812 -- 0.218 0.6404
Live Crown
Elevation2 -0.1145 0.1127 -1.015 0.3160
Round 1.6735 0.1864 8.980 <0.0001
Park 0.0260 0.1726 0.150 0.8820
Park*Elevation 0.2172 0.1604 1.353 0.1860
Family 0.1060 -- 6.960 0.0084
1For Elevation, Park, and Round, estimates are the fixed eects model parameters; for Family, estimates are the
standard deviation of the random eect.
2Test statistics report t-tests for fixed eects, and the χ2 likelihood ratio for random eects
Growth
The height of live crown was not significantly related to elevation (p=0.316) or park (p=0.882),
though planting round (p<0.0001) was highly significant (Figure 4C; Table 2). Variance among fami-
lies was highly significant for live crown (p=0.0084), suggesting a large amount of genetic variation
for early seedling growth. At GSMNP there was a significant linear relationship between bud set
and live crown length (p=0.0132); however this relationship was not observed for MMSP (Figure 5).
Considering that the extended length of the growing season prior to bud set is only 10 days, the dif-
ference in live crown may indicate slower growth rates of seedlings sourced from higher elevations.
DISCUSSION
We found significant variation in seed traits and bud phenology among red spruce families collected
along an elevational gradient in the southern Appalachians. These early life history traits are known
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 135
Figure 3. Mean family seed mass (n=73) (A), percentage of lled seed (n=80) (B), and percentage of seed germination
(n=38) (C) of red spruce collected in Great Smoky Mountains National Park (GSMNP) and Mount Mitchell State Park
(MMSP) by source elevation.
136 Castanea, Vol. 84(2) 2019
Figure 4. Mean bud set (A), bud ush (B) and live crown (C) by family (n=38) for seedlings grown in Burlington,
Vermont from seeds collected in Great Smoky Mountains National Park (GSMNP) and Mount Mitchell State Park
(MMSP) by source elevation. Arrows denote two outliers identied using Cook’s Distance statistic (A), the regression
parameters were re-calculated without the outliers for GSNMP.
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 137
to be important components of plant fitness (Westoby et al. 1996, Savolainen et al. 2007), and as
such may be under divergent selection in dierent environments. While our breadth of sampling is
moderately limited in regional scope, this study provides an important first report of potentially
adaptive phenotypic variation among neighboring southern Appalachian red spruce populations.
Both seed mass and percentage of filled seed exhibited positive clines with elevation. In many
montane plant species, seed size and number of seeds per plant are aected by elevation, and species
display both positive and negative relationships (Oleksyn et al. 1998, Liu et al. 2013, Olejniczak et
al. 2018). Oleksyn et al. (1998) found that Norway spruce (Picea abies (L.) H. Karst.) seed mass
declined and the percentage of unfilled seeds increased with elevation in the mountains of Poland
as the climate becomes colder. Similarly, black spruce seed mass and percentage of filled seed was
found to be negatively correlated with elevation across Canada (Liu et al. 2013). Further, across
Picea species, seed mass has been shown to decline with latitude (Miyazawa and Lechowicz 2004)
and within species black spruce and white spruce seed mass also decline with latitude (Liu et al.
2013). Thus, one might predict that colder, higher latitude or elevation sites with shorter growing
seasons would reduce seed quality. However, we observed the opposite trend with red spruce in
the southern Appalachians, with the percentage of filled seed and mean seed mass increasing with
elevation (Figure 3A, 3B).
The southern Appalachians lay at the trailing edge of the range of red spruce; it would be simple
to directly correlate seed mass to elevation with the rationale that low elevation sites have sub-
optimal climatic conditions while high elevation are considered optimal. High elevation sites may
experience reduced heat stress and reduced stress related to high vapor pressure deficit, but other
stresses such as air pollution (formerly), wind, rapid temperature swings, shorter growing season
as well as competition from Fraser fir (Abies fraseri, (Pursh) Poir.) may limit growth rate, longevity
and health. Mathias and Thomas (2018) found that red spruce at three sites in West Virginia (lat.
38°N) were now growing at a faster rate than at the prior peak in the early 1960’s before air pollu-
tion led to sharp growth declines. This renewed growth under the present warmer atmospheric
conditions as well as recent downslope migration of montane ecotones (Foster and D’Amato 2015)
indicates that red spruce may tolerate warmer conditions and occupy greater “potential range”
than previously thought. In addition, compared to related Picea species, the relationship between
Figure 5. Relation between bud set and total live crown at dormancy in fall 2017.
138 Castanea, Vol. 84(2) 2019
elevation and seed traits in red spruce suggest mechanisms other than climate, possibly reflecting
Allee eects (Allee and Bowen 1932, Stephens et al. 1999) of a larger, more diverse mating population
growing in more optimal environments at higher elevations. For example, some of the variation in
seed traits may reflect population dierences in pollination eciency or resource availability with
elevation. Empty or non-viable seeds may result from self-fertilization, inbreeding, or lack of pollen
when the female cone is receptive to fertilization. In black spruce, the percentage of filled seed is
directly related to the quantity of pollen grains at a given location (Caron and Powell 1989) and
this would also be the expectation for other Picea species. While we can only speculate as to the
cause of the elevation gradient on seed quality in the current study, possible explanations include:
1) increased population density at higher elevations increasing the size and diversity of the pollen
pool and osetting pollen limitation and the probability of inbreeding depression (Mosseler et al.
2000, Rajora et al. 2000); 2) better phenological synchrony of available mating partners at higher
elevations (LaMontagne and Boutin 2007), 3) better pollination eciency due to favorable wind
characteristics on mountain tops and ridgelines (Fall 1992), or 4) increased drought or heat stress
during the seed provisioning stage at lower elevations. While it is not possible to identify specific
causes for the elevation eect on seed quality from our results, it does seem clear that seed quality
is impacted at lower elevations.
Germination percentages of filled red spruce seed were much lower than reported in the liter-
ature for other regions, e.g. New Hampshire, 62–77% (Baldwin 1934), New Brunswick and Nova
Scotia, >80%, more distant populations in Ontario, >70% (Major et al. 2003), and New Brunswick
>95% (Butnor et al. 2018). Surprisingly, there was no relationship between germination and seed
mass, but we found significant variance among families. The low mean germination percentage
from the southern Appalachians (45% in the present study) could represent higher levels of inbreed-
ing depression in the more fragmented, southern part of the range compared to the more intercon-
nected northern part of the range. In a study comparing old-growth remnant red spruce stands in
Ontario, Canada, Mosseler et al. (2003) found tall stands (with height as an indicator of fitness)
acted as potential reservoirs of genetic diversity and reproductive fitness, where mean stand height
was positively related to molecular measures of genetic diversity (mean number of alleles per locus
and percent of polymorphic loci), and inversely related to the frequency of rare and possibly dele-
terious alleles. Given a lack of recent land use history in the old-growth stands studied by Mosseler
et al. (2003), the height-diversity relationship suggests that genetically diverse populations have
favorable growth attributes compared to less diverse stands, perhaps because they are better able to
avoid the negative eects of inbreeding depression (Mosseler et al. 2003). In our study, germination
was highest at Mitchell Trail (71%) and Deep Gap Trail (58%), but all other populations were below
50%. The Road Prong population in GSMNP was primarily original forest with some of the largest
mother trees in our study, yet mean germination was only 28%. Additional research into the genetic
diversity and rate of inbreeding in red spruce using molecular markers and common garden studies
is underway, and will improve our understanding of the causes of fitness trait variation in these
remnant populations.
Common garden growing conditions in our study revealed a strong elevational cline in bud set,
with earlier bud set of GSMNP red spruce seedlings from high elevation families compared to low
elevation families. Because these dierences exist when all seedlings were experiencing the same
day length and temperature regimes, these dierences likely reflect underlying genetic variation for
bud set along the elevational gradient. However, maternal environmental eects or epigenetic influ-
ences on seedling traits may also be present, and could contribute to dierences in early seedling
growth or phenology among families collected at dierent elevations (Herman and Sultan 2011).
Earlier bud set for trees adapted to shorter growing seasons has been observed in many other tree
species, and is consistent with bud set evolving in response to local selection pressures (Mergen
1963, Johnsen et al. 1988, Oleksyn et al. 1992, Johnsen et al. 1996, Chmura 2006). Early bud set at
higher elevations may be advantageous in order to terminate growth and acquire cold hardiness
before freezing conditions arrive.
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 139
It is interesting that red spruce trees in close geographic proximity at MMSP, that experienced
the same photoperiod and similar elevation gradient (800 m) as GSMNP, did not demonstrate
strong bud set variation with elevation. One possibility is that this represents the eects of prior
land use disrupting local adaptation. In the early 1900’s the majority of merchantable red spruce at
MMSP were cut, followed by experimental reforestation with planted red spruce among many other
species (Korstian 1937, Minckler 1940, Minckler 1945, Wahlenberg 1951, Speers 1975). Dr. Clarence
Korstian, who played a central role in designing the planting trial was a strong advocate for planting
local seed sources (Korstian 1937) and later documentation of the experiment indicate that the red
spruce were propagated from local sources (Speers 1975). Despite being “local”, it is uncertain
whether precise pairing between seed source and planting elevation occurred. The relatively recent
introduction of red spruce genotypes from a variety of elevations could explain the lack or dilution
of clinal variation along the elevational gradient at Mount Mitchell.
There were no significant relationships between elevation and bud flush from either the GSMNP
or MMSP sources, despite notable dierences in mean annual air temperature (MAT). The eleva-
tional cline in bud set but not flush likely reflects greater genetic determination of bud set whereas
bud flush may be more environmentally plastic. Given our results, a greater selective pressure
seems to exist for bud set versus bud flush along fine-scale environmental gradients. Both traits
evolve as bet-hedging strategies between maximizing the period available for carbon acquisition
with the need to avoid tissue damage caused by early or late freeze events. In many temperate and
boreal trees, shoot elongation stops and dormant buds form in response to decreasing day length,
prior to the arrival of potentially damaging cold temperatures; hence bud set is largely controlled
by photoperiod, while bud flush is stimulated by the accumulation of warm temperatures in spring
(Wareing 1956, Olsen et al. 2014). At the fine spatial scale of our sampling, red spruce families
collected from dierent elevations experience highly similar photoperiods, so dierences in bud set
must reflect response to a dierent underlying cue, possibly temperature.
Similar results as ours were also found in a study of Norway spruce phenology in the Beskidy
Mountains of Poland; 23 high elevation seed sources within a 12 by 18 km area experienced early
onset of dormancy (800 m max) compared to low elevation (540 m min), though there were no
dierences in bud flush (Chmura 2006). In contrast, Johnsen et al. (1996) found that black spruce
from colder, high latitude (63°) sources initiated growth earlier than those from low latitude (45°)
sources when grown in common gardens, as well as exhibited dierences in photosynthetic rates.
Rossi and Isabel (2017) used MAT to characterize populations within more narrow latitude ranges,
finding black spruce sources from colder areas initiate growth earlier than locations with higher
MAT when grown together in common gardens.
Red spruce has been generally described as having low genetic diversity that could limit its
ability to respond to changes in climate (Dehayes and Hawley 1992). Past assessments based on
broad geographic sampling have revealed little clinal variation in morphometric traits (Gordon 1976)
or in growth and survival (Fowler et al. 1988). In GSMNP, we found strong clinal variation in bud
set consistent with an adaptive response to climate dierences across an elevation gradient.
Phenological adaptation of red spruce from low elevations in the southern Appalachians could be-
come beneficial in future climate scenarios. Specifically, later bud set to delay growth cessation
would be advantageous in a warmer climate. Eriksson et al. (1978) found strong genetic control of
bud phenology in Norway spruce and through inter-provenance breeding programs, progeny could
be prescribed for specific environmental conditions. Considering that past logging has left a signif-
icant amount of otherwise suitable red spruce land area vacant in the southeastern United States
(Walter et al. 2017), there are still opportunities for breeding and restoration despite negative
climate predictions for the species’ range (Iverson et al. 2008, Koo et al. 2015). A thorough under-
standing of genetic variation for climate-adaptive functional traits should constitute a key compo-
nent of this eort. Our study provides a first step towards this goal for southern Appalachian red
spruce.
140 Castanea, Vol. 84(2) 2019
CONCLUSIONS
Despite occupying a relatively small geographical area, red spruce in the southern Appalachians
display clinal variation in seed mass and bud set phenology related to climate, along with signifi-
cant variation in filled seed and seedling germination. Faced with a warming climate, an important
question is whether red spruce populations in the southern Appalachians have the capacity to
respond to rapid changes in climate, as there are no natural migration routes in the present warm-
ing scenario that would permit a large-scale northward range shift. However, red spruce currently
exists over a wide elevational range in the region, from roughly 900 to 2000 m, indicating that there
are opportunities for taking advantage of naturally occurring genetic variation for restoration, de-
pending on micro-climate and selection of appropriate stock. Our results support locally adaptive
clines in seed and bud set traits, but the extent to which red spruce shows fine-scale adaptive
genetic variation in other traits needs further exploration. New studies to quantify the extent of
genetic variation in additional climate-adaptive traits across local environmental gradients in the
southern Appalachians are imperative to better inform restorative eorts and decisions regarding
appropriate seed sources to plant in a changing climate. Ideally, this would include reciprocal
transplant experiments of high and low elevation seed sources into each respective environment to
better understand local adaptation in physiological traits such as carbon allocation, photosynthesis,
water use eciency, and tolerance of temperature extremes. It may also be that the portfolio of
genetic diversity and climate adaptation in the southern Appalachian red spruce will prove useful
in maintaining resilient populations further north or at higher elevations, representing a form of
assisted migration.
ACKNOWLEDGMENTS
Support for collecting and cleaning seeds was provided by USFS employees: Jill Barbour, Brandy
Benz, Joel Burley, Tom Christensen, Robert Eaton, Shelly Hooke, Carol Maddox, Loree McCranie,
Katie Morgan, Karen Sarsony and Marcus Wind. We thank Jeremy Weiland for assistance in carrying
out the germination experiments in the laboratory. Personnel at the North Carolina Division of
Parks and Recreation and Mount Mitchell State Park were very helpful with permitting and access
logistics. We also thank Dr. Paul Super, Science Coordinator at Great Smoky Mountains National
Park for guidance during the permitting process and sharing knowledge of park history. The USDA
Forest Service, National Seed Laboratory provided assistance with seed processing and other tech-
nical support. This work was funded jointly by the USDA Forest Service (J.B., K.J., C.M. and V.V)
and USDA Hatch and National Science foundation grants (S.K.).
LITERATURE CITED
Aitken, S.N., S. Yeaman, J.A. Holliday, T.L. Wang, and S. Curtis-McLane. 2008. Adaptation, migration
or extirpation: climate change outcomes for tree populations. Evol. Appl. 1:95–111.
Allee, W.C., and E.S. Bowen. 1932. Studies in animal aggregations: Mass protection against colloidal
silver among goldfishes. J. Exp. Zool. 61:185–207.
Andrews, C. 2016. Modeling and Forecasting the Influence of Current and Future Climate on
Eastern North American Spruce-Fir (Picea-Abies) Forests. M.S. Thesis, University of Maine,
Orono, Maine.
AOSA. 2016. Vol. 1. Principles and procedures. Association of Ocial Seed Analysts., Washington,
D.C.
Baldwin, H.I. 1934. Germination of the red spruce. Pl. Physiol. 9:491–532.
Bates, D., M. Machler, B.M. Bolker, and S.C. Walker. 2015. Fitting Linear Mixed-Eects Models Using
lme4. J. Stat. Softw. 67:1–48.
Beane, N.R. 2010. Using environmental and site-specific variables to model current and future dis-
tribution of red spruce (Picea rubens Sarg.) forest habitat in West Virginia. M.S. thesis, West
Virginia University, Morgantown, West Virginia.
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 141
Bower, A.D., B. St Clair, and V. Erickson. 2014. Generalized provisional seed zones for native plants.
Ecol. Appl. 24:913–919.
Brown, D.M. 1941. Vegetation of Roan Mountain: A phytosociological and successional study. Ecol.
Monogr. 11:61–97.
Butnor, J.R., B.M. Verrico, V. Vankus, and S.R. Keller. 2018. Ethanol exposure can inhibit red spruce
(Picea rubens) seed germination. Seed Sci. Technol. 46:259–265.
Caron, G.E., and G.R. Powell. 1989. Cone size and seed yield in young Picea mariana trees. Can. J.
For. Res. 19:351–358.
Chmura, D.J. 2006. Phenology diers among Norway spruce populations in relation to local varia-
tion in altitude of maternal stands in the Beskidy Mountains. New For. 32:21–31.
Cogbill, C.V., and P.S. White. 1991. The latitude-elevation relationship for spruce-fir forest and
treeline along the Appalachian mountain chain. Vegetatio 94:153–175.
Cook, R.D. 1977. Detection of influential observation in linear regression. Technometrics 19:15–18.
Crandall, D.L. 1958. Ground vegetation patterns of the spruce-fir area of the Great Smoky
Mountains National Park. Ecol. Monogr. 28:337–360.
Davis, M.B., R.W. Spear, and L.C.K. Shane. 1980. Holocene climate of New England. Quaternary
Research 14:240–250.
Dehayes, D.H., and G.J. Hawley. 1992. Genetic implications in the decline of red spruce. Water Air
Soil Pollut. 62:233–248.
Delcourt, H.R., and P.A. Delcourt. 1984. Late-quaternary history of the spruce-fir on ecosystem
in the southern Appalachian mountain region. p 22–35 In: White, P.S. (ed.). The southern
Appalachian spruce-fir ecosystem: Its biology and threats, USDI Research/Resource Management
Report SER 71. Atlanta, Georgia.
Diggins, C.A., and W.M. Ford. 2017. Microhabitat selection of the Virginia northern flying squirrel
(Glaucomys sabrinus fuscus Miller) in the central Appalachians. Northeast. Nat. 24:173–190.
Eriksson, G., I. Ekberg, I. Dormling, B. Matérn, and D. von Wettstein. 1978. Inheritance of bud-set
and bud-flushing in Picea abies (L.) Karst. Theor. Appl. Genet. 52:3–19.
Fall, P.L. 1992. Spatial patterns of atmospheric pollen dispersal in the Colorado Rocky Mountains,
USA. Rev. Palaeobot. Palynol. 74:293–313.
Fortney, R.H., S.L. Stephenson, and J.S. Rentch. 2015. Rare plant communities in Canaan Valley,
West Virginia. Southeast. Nat. 14:121–135.
Foster, J.R., and A.W. D’Amato. 2015. Montane forest ecotones moved downslope in northeastern
USA in spite of warming between 1984 and 2011. Global Change Biol. 21:4497–4507.
Fowler, D.P., Y.S. Park, and A.G. Gordon. 1988. Genetic variation of red spruce in the Maritimes.
Can. J. For. Res. 18:703–709.
Gordon, A.G. 1976. Taxonomy and genetics of Picea rubens and its relationship to Picea mariana.
Can. J. Bot. 54:781–813.
Herman, J., and S. Sultan. 2011. Adaptive transgenerational plasticity in plants: case studies, mech-
anisms, and implications for natural populations. Front. Plant Sci. 2:102.
Iverson, L.R., A.M. Prasad, S.N. Matthews, and M. Peters. 2008. Estimating potential habitat for 134
eastern US tree species under six climate scenarios. For. Ecol. Manag. 254:390–406.
Johnsen, K.H., P.P. Feret, and J.R. Seiler. 1988. Root growth potential and shoot activity of northern
and southern provenances of 1-0 eastern white pine seedlings grown in a Virginia nursery. Can.
J. For. Res. 18:610–614.
Johnsen, K.H., J.R. Seiler, and J.E. Major. 1996. Growth, shoot phenology and physiology of diverse
seed sources of black spruce: II. 23-year-old field trees. Tree Physiol. 16:375–380.
Koo, A.K., C.B. Patten, and M. Madden. 2015. Predicting eects of climate change on habitat suit-
ability of red spruce (Picea rubens Sarg.) in the southern Appalachian Mountains of the USA:
understanding complex systems mechanisms through modeling. Forests 6:1208–1226.
Korstian, C.F. 1937. Perpetuation of spruce on cut-over and burned lands in the higher southern
Appalachian Mountains. Ecol. Monogr. 7:125–167.
142 Castanea, Vol. 84(2) 2019
Kuznetsova, A., P.B. Brockho, and R.H.B. Christensen. 2017. lmerTest Package: Tests in linear
mixed eects models. J. Stat. Softw. 82:1–26.
LaMontagne, J.M., and S. Boutin. 2007. Local-scale synchrony and variability in mast seed produc-
tion patterns of Picea glauca. J. Ecol. 95:991–1000.
Liu, J.S., Y.G. Bai, E.G. Lamb, D. Simpson, G.F. Liu, Y.S. Wei, D.L. Wang, D.W. McKenney, and P.
Papadopol. 2013. Patterns of cross-continental variation in tree seed mass in the Canadian
boreal forest. PLoS ONE 8(4):e61060.
Major, J.E., A. Mosseler, D.C. Barsi, M. Campbell, and O.P. Rajora. 2003. Morphometric, allometric,
and developmentally adaptive traits in red spruce and black spruce. I. Species and seed-source
variation. Can. J. For. Res. 33:885–896.
Mathias, J.M., and R.B. Thomas. 2018. Disentangling the eects of acidic air pollution, atmospheric
CO2, and climate change on recent growth of red spruce trees in the Central Appalachian
Mountains. Global Change Biol. 24:3938–3953.
McKenney, D., J. Pedlar, and G. O’Neill. 2009. Climate change and forest seed zones: Past trends,
future prospects and challenges to ponder. Forest. Chron. 85:258–266.
Mergen, F. 1963. Ecotypic variation in Pinus strobus L. Ecology 44:716–727.
Minckler, L.S. 1940. Early planting experiments in the spruce-fir type of the southern Appalachians.
J. Forest. 38:651–654.
Minckler, L.S. 1945. Reforestation in the Spruce Type in the Southern Appalachians. J. Forest.
43:349–356.
Miyazawa, K., and M.J. Lechowicz. 2004. Comparative seedling ecology of eight north American
spruce (Picea) species in relation to their geographic ranges. Ann. Bot. 94:635–644.
Mosseler, A., J.E. Major, and O.P. Rajora. 2003. Old-growth red spruce forests as reservoirs of
genetic diversity and reproductive fitness. Theor. Appl. Genet. 106:931–937.
Mosseler, A., J.E. Major, J.D. Simpson, B. Daigle, K. Lange, Y.S. Park, K.H. Johnsen, and O.P. Rajora.
2000. Indicators of population viability in red spruce, Picea rubens. I. Reproductive traits and
fecundity. Can. J. Bot. 78:928–940.
Nowacki, G., R. Carr, and M. Van Dyck. 2010. The current status of red spruce in the eastern
United States: distribution, population trends, and environmental drivers. p. 140–162 In: J.S.
Rentch and T.M. Schuler, editors. Proceedings from the conference on the ecology and man-
agement of high-elevation forests in the central and southern Appalachian Mountains, 2009
May 14–15; Slatyfork, WV. Gen. Tech. Rep. NRS-P-64. Newtown Square, PA: U.S. Department of
Agriculture, Forest Service, Northern Research Station.
Olejniczak, P., M. Czarnoleski, A. Delimat, B.M. Majcher, and K. Szczepka. 2018. Seed size in
mountain herbaceous plants changes with elevation in a species-specific manner. PLoS ONE
13(6):e0199224.
Oleksyn, J., J. Modrzýnski, M.G. Tjoelker, R. Z·ytkowiak, P.B. Reich, and P. Karolewski. 1998.
Growth and physiology of Picea abies populations from elevational transects: common garden
evidence for altitudinal ecotypes and cold adaptation. Funct. Ecol. 12:573–590.
Oleksyn, J., M.G. Tjoelker, and P.B. Reich. 1992. partitioning of populations of European Pinus
sylvestris L under simulated 50° and 60°N daylengths: evidence for photoperiodic ecotypes. New
Phytol. 120:561–574.
Olsen, J.E., Y. Lee, and O. Junttila. 2014. Eect of alternating day and night temperature on short
day-induced bud set and subsequent bud burst in long days in Norway spruce. Front. Plant. Sci.
5:691.
Pyle, C. and M.P. Schafale. 1988. Land use history of three spruce-fir forest sites in southern
Appalachia. Forest & Conservation History 32:4–21.
R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. (https://www.R-project.org/, 10 June 2019).
Butnor et.al., Phenotypic variation in climate-associated traits of red spruce 143
Rajora, O.P., A. Mosseler, and J.E. Major. 2000. Indicators of population viability in red spruce,
Picea rubens. II. Genetic diversity, population structure, and mating behavior. Can. J. Bot.
78:941–956.
Rentch, J.S., T.M. Schuler, W.M. Ford, and G.J. Nowacki. 2007. Red spruce stand dynamics, simula-
tions, and restoration opportunities in the central Appalachians. Restor. Ecol. 15:440–452.
Rossi, S., and N. Isabel. 2017. Bud break responds more strongly to daytime than night-time tempera-
ture under asymmetric experimental warming. Global Change Biol. 23:446–454.
Savolainen, O., T. Pyhajarvi, and T. Knurr. 2007. Gene flow and local adaptation in trees. Annu. Rev.
Ecol. Evol. Syst. 38:595–619.
Schofield, W.B. 1960. The ecotone between spruce-fir and deciduous forest in the Great Smoky
Mountains. Ph.D. dissertation, Duke University, Durham, North Carolina.
Speers, C.F. 1975. Experimental planting in cutover spruce-fir in the southern Appalachians: 50
year results. Research Note SE-219. Southeastern Forest Experiment Station Asheville, North
Carolina, 5 p.
Stephens, P.A., W.J. Sutherland, and R.P. Freckleton. 1999. What Is the Allee Eect? Oikos
87:185–190.
Thomas, E., R. Jalonen, J. Loo, D. Boshier, L. Gallo, S. Cavers, S. Bordacs, P. Smith, and M. Bozzano.
2014. Genetic considerations in ecosystem restoration using native tree species. For. Ecol. Manag.
333:66–75.
Vitt, P., K. Havens, A.T. Kramer, D. Sollenberger, and E. Yates. 2010. Assisted migration of plants:
Changes in latitudes, changes in attitudes. Biol. Conserv. 143:18–27.
Wahlenberg, W.G. 1951. Planting in the Appalachian Spruce-Fir Type. J. Forest. 49:569–571.
Walter, J.A., J.C. Neblett, J.W. Atkins, and H.E. Epstein. 2017. Regional- and watershed-scale analysis
of red spruce habitat in the southeastern United States: implications for future restoration
eorts. Plant Ecol. 218:305–316.
Wareing, P.F. 1956. Photoperiodism in woody plants. Annu. Rev. Plant Physiol. Plant Mol. Biol.
7:191–214.
Westoby, M., M. Leishman, and J. Lord. 1996. Comparative ecology of seed size and dispersal.
Philos. Trans., Ser. B 351:1309–1317.