Edited by: Heikki Hänninen, University of Helsinki, Finland
Reviewed by: Glenn Thomas Howe, Oregon State University, USA; Outi Savolainen, University of Oulu, Finland
*Correspondence: Jason A. Holliday, Department of Forest Resources and Environmental Conservation, Virginia Polytechnic Institute and State University, 304 Cheatham Hall, Blacksburg, VA 24061, USA
This article was submitted to Functional Plant Ecology, a section of the journal Frontiers in Plant Science
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Local adaptation to climate in temperate forest trees involves the integration of multiple physiological, morphological, and phenological traits. Latitudinal clines are frequently observed for these traits, but environmental constraints also track longitude and altitude. We combined extensive phenotyping of 12 candidate adaptive traits, multivariate regression trees, quantitative genetics, and a genome-wide panel of SNP markers to better understand the interplay among geography, climate, and adaptation to abiotic factors in
Tree species and populations may respond to climate change through migration, phenotypic plasticity, evolutionary adaptation, or local extinction (Aitken et al.,
Understanding patterns of local adaptation across a species' range facilitates predictions of potential responses to climate change. Forest tree provenance trials that evaluate phenotypic differences among populations have identified traits under divergent selection and quantified their genetic bases (Sork et al.,
In 2010, branch cuttings of P. trichocarpa genotypes that span much of the species' range were collected and used to produce plantlets. The plantlets were grown for 6 months in a mist house and subsequently transplanted in May 2011 to an outdoor common garden located at the Reynolds Homestead Forest Resource Research Center located in Patrick County, Virginia (36°37′N and 80°09′W). Ramets of each genotype were planted in a randomized complete block design with 4 blocks. Annual climate variables (means of years 1981–2009) for the closest weather station (Patrick County, VA) showed a mean annual precipitation (MAP) of 1275 mm, a mean annual temperature (MAT) of 14.1°C, a mean warmest month temperature (MWMT) of 24.4°C, and a mean coldest month temperature (MCMT) of 4.2°C (obtained from the weatherbase.com). For this study, we selected 124 genotypes from the 789 available genotypes that covered a wide latitudinal and altitudinal gradient (37–58°N latitude and sea level through ~2300 m) (Figure
One branch-section from the current growth year was collected in March 2013 for each tree, oven-dried at 65°C until a constant weight was reached, and ground to 0.5 mm size. About 2 mg of each sample was combusted in an IsoPrime100 isotope ratio mass spectrometer (Isoprime Ltd, Cheadle Hulme, UK) located at the Virginia Tech Forest Soils and Biogeochemistry Laboratory. The carbon isotope ratio, 13C/12C (δ13C), was calculated relative to the international Pee Dee Belemnite standard (Farquhar et al.,
In March 2013, while the trees were in a dormant state, growth and branch characteristics (Figure
Bud stage was scored on a weekly basis during April 2012 (BF = bud flush) and September 2012 (BS = bud set), and the date of BF or BS was converted to the Julian day (the number of days from December 31). The bud was considered flushed when a leaf had emerged 1 cm from the bud, and set when dark red/brown scales covered the apex. Buds that set or flushed before the first week or after the last week of scoring were recorded as having occurred 1 week before or 1 week later, respectively. Fall cold injury (I−20) was assessed by measuring electrolytic leakage of branch sections collected on November 5, 2012 (Hannerz et al.,
Where
Data were visually inspected for normality and homoscedasticity, and NSyll was linearized by log transformation before analysis. To reduce the number of variables, and to deal with the high correlation between growth and branching variables, principal component analysis (PCA) was run on the growth and branch traits using the
Samples were grouped into 20 populations according to their geographic origins (latitude, longitude, and altitude) for computation of population differentiation in quantitative traits (QST) and broad-sense heritability (H2). For each trait, the clonal best linear unbiased predictors (BLUPs) and variance components were estimated using restricted maximum likelihood (REML) with block as a fixed factor and genotype and population as random factors according to the following model:
Where σ2
Finally, genetic correlations between traits were obtained by calculating correlations among the respective BLUPs, and environmental correlations were estimated according to Searle (
Where R, r, and r′ are respectively the phenotypic, genetic, and environmental correlations, and
We previously reported using sequence capture to retrieve and sequence much of the exome for 48
Correlations were computed for each pair of variables to evaluate the relationships among traits, geography, and climate. Among geographical variables, latitude had the highest number of significant correlations with climate variables (20/22 significant and 10/22 with |r| > 0.6) (Table
Mean annual standard precipitation evapotranspiration index (SPEI12) | 0.12 | 0.33 |
−0.06 |
Mean annual temperature (MAT) | 0.67 |
−0.18 | −0.43 |
Mean warmest month temperature (MWMT) | 0.74 |
−0.43 |
−0.14 |
Mean coldest month temperature (MCMT) | 0.58 |
0.01 | −0.47 |
Temperature difference between MWMT and MCMT (TD) | 0.14 | −0.39 |
0.51 |
Degree-days below 0°C (DD_0) | 0.49 |
0.03 | 0.50 |
Degree-days above 5°C (DD_5) | 0.71 |
−0.25 |
−0.35 |
Degree-days below 18°C (DD_18) | 0.65 |
0.16 | 0.46 |
Degree-days above 18°C (DD18) | 0.72 |
−0.31 |
−0.13 |
Number of frost-free days (NFFD) | 0.44 |
0.03 | −0.63 |
Julian date on which FFP begins (bFFP) | 0.51 |
0.02 | 0.57 |
Julian date on which FFP ends (eFFP) | 0.48 |
0.01 | −0.57 |
Frost free period (FFP) | 0.50 |
−0.01 | −0.57 |
Extreme minimum temperature over 30 years (EMT) | 0.40 |
0.03 | −0.63 |
Extreme minimum temperature (EXT) | 0.76 |
−0.53 |
−0.04 |
Precipitation as snow (PAS) | 0.41 |
−0.03 | 0.47 |
Mean annual precipitation (MAP) | 0.30 |
0.22 |
−0.22 |
Mean annual summer precipitation (MSP) | 0.60 |
0.42 |
−0.14 |
Annual heat:moisture index (AHM) | 0.43 |
−0.22 |
0.10 |
Summer heat:moisture index (SHM) | 0.77 |
−0.30 |
0.08 |
Hargreaves reference evaporation (Eref) | 0.89 |
−0.50 |
0.11 |
Hargreaves climatic moisture deficit (CMD) | 0.80 |
−0.40 |
0.19 |
All traits except RNB had a significant (
H | −0.33 |
−0.20 |
−0.23 |
D | −0.26 |
−0.16 | −0.25 |
CD | −0.35 |
−0.19 |
−0.20 |
NB | −0.32 |
−0.21 |
−0.18 |
NSyll | −0.33 |
−0.15 | −0.13 |
RNB | −0.17 | −0.14 | 0.03 |
RCD | −0.18 |
−0.19 |
−0.08 |
VI | −0.28 |
−0.12ns | −0.24 |
δ13C | 0.20 |
−0.04ns | −0.46 |
BF | 0.45 |
0.47 |
−0.09 |
BS | −0.79 |
−0.39 |
−0.03 |
I−20 | −0.59 |
−0.20 |
−0.11 |
H | 0.90 |
0.87 |
0.88 |
0.73 |
0.59 |
0.06 | 0.49 |
0.25 |
−0.48 |
0.71 |
0.83 |
|
D | 0.88 |
0.96 |
0.92 |
0.81 |
0.67 |
0.20 |
0.54 |
0.23 |
−0.43 |
0.61 |
0.74 |
|
VI | 0.82 |
0.93 |
0.89 |
0.83 |
0.71 |
0.24 |
0.54 |
0.19 |
−0.36 |
0.61 |
0.71 |
|
CD | 0.81 |
0.89 |
0.80 |
0.82 |
0.67 |
0.29 |
0.56 |
0.18 |
−0.49 |
0.70 |
0.81 |
|
NB | 0.65 |
0.75 |
0.69 |
0.76 |
0.87 |
0.68 |
0.75 |
0.13 | −0.45 |
0.63 |
0.63 |
|
Nsyll | 0.43 |
0.56 |
0.56 |
0.53 |
0.83 |
0.57 |
0.61 |
0.01 | −0.39 |
0.62 |
0.58 |
|
RNB | 0.21 |
0.40 |
0.31 |
0.47 |
0.81 |
0.66 |
0.67 |
−0.12 | −0.23 |
0.21 |
0.17 | |
RCD | 0.38 |
0.50 |
0.41 |
0.51 |
0.74 |
0.59 |
0.77 |
0.04 | −0.31 |
0.37 |
0.39 |
|
δ 13C | 0.05 | −0.01 | 0.02 | 0.05 | 0.06 | −0.04 | 0.02 | −0.04 | 0.12 | −0.14 | −0.08 | |
BF | −0.32 |
−0.36 |
−0.27 |
−0.38 |
−0.27 |
−0.19 |
−0.17 | −0.12 | 0.17 | −0.61 |
−0.40 |
|
BS | 0.39 |
0.34 |
0.33 |
0.32 |
0.30 |
0.33 |
0.13 | 0.25 |
0.01 | −0.05 | 0.74 |
|
I–20 | 0.26 |
0.20 |
0.17 | 0.26 |
0.21 |
0.13 | 0.1 | 0.24 |
−0.01 | −0.05 | 0.53 |
Prior to MRT analysis, growth and branching traits were subjected to PCA to remove their colinearity. The first two principal components were retained because they explained most of the variation in growth and branching traits (88%). The loadings of the traits (Supplemental Table
MRTs using climate variables as predictors also split the 124 genotypes in five groups, which explained 39% of variation in the measured traits (Figure
BS and I–20 had the highest among-population differentiation, with QST ≥ 0.5. Growth traits (H, D and VI) and δ13C had moderate values (0.32 ≤ QST ≤ 0.44), while differentiation for branch traits (NB, NSyll, RNB and RCD) and BF was low (QST < 0.25) (Table
H | 0.44 | 0.22 | 0.61 | 0.24 | 0.13 | 0.35 | 0.44 | 0.15 | 0.61 |
D | 0.41 | 0.29 | 0.58 | 0.19 | 0.08 | 0.29 | 0.36 | 0.09 | 0.5 |
VI | 0.43 | 0.25 | 0.59 | 0.13 | 0.02 | 0.24 | 0.26 | 0.01 | 0.39 |
CD | 0.3 | 0.15 | 0.5 | 0.25 | 0.14 | 0.33 | 0.38 | 0.14 | 0.5 |
NB | 0.22 | 0.09 | 0.34 | 0.32 | 0.2 | 0.43 | 0.43 | 0.23 | 0.52 |
Nsyll | 0.18 | 0.05 | 0.31 | 0.29 | 0.19 | 0.43 | 0.37 | 0.17 | 0.48 |
RNB | 0.04 | −0.16 | 0.17 | 0.25 | 0.13 | 0.37 | 0.26 | 0.12 | 0.38 |
RCD | 0.1 | −0.08 | 0.24 | 0.21 | 0.1 | 0.33 | 0.25 | 0.11 | 0.36 |
δ13C | 0.34 | 0.14 | 0.49 | 0.23 | 0.11 | 0.38 | 0.37 | 0.14 | 0.49 |
BF | 0.24 | 0.1 | 0.4 | 0.3 | 0.18 | 0.4 | 0.41 | 0.17 | 0.57 |
BS | 0.67 | 0.46 | 0.79 | 0.22 | 0.1 | 0.35 | 0.58 | 0.22 | 0.72 |
I−20 | 0.54 | 0.34 | 0.68 | 0.29 | 0.16 | 0.38 | 0.55 | 0.26 | 0.63 |
In spite of high gene flow among populations, temperate and boreal trees often display substantial adaptive genetic differentiation due to strong climatic gradients across their ranges (Farmer,
Genotypes from the northern periphery of our sampling area had the lowest growth, which probably reflects adaptation to a short growing season (shorter FFP) in their native environment. For many tree species, common garden experiments have revealed that high latitude populations achieve less height growth even when they display higher assimilation rates than low latitude populations (Burtt,
The center of the range (southern BC, Oregon, Washington) was partitioned into three populations. Genotypes from southern BC, which is characterized by moderate temperatures, high precipitation, and low evaporative demand, were differentiated into two groups according to continentality: the coastal group, which originates from a mild climate (MWMT > 16.85) with a long growing season, displayed better growth (slightly above the overall mean), and was more susceptible to fall cold damage than the interior group. The interior population in southern BC was merged with the northern groups in the MRT using climate predictors, as the temperature variables of these two population locations were similar. Genotypes spanning the coast of Oregon and Washington were the tallest, and had the highest number of branches in the common garden. This area is below 400 m altitude and has a mild climate (highest MAT, MWMT, EXT and low TD) that is similar to the common garden in terms of temperature variables. The climate variable-based MRT showed that these genotypes could be further differentiated according to MAP in spite of closely overlapping geography. This local spatial heterogeneity in climate could help these populations respond to climate change, facilitating migration by reducing dispersal distance required to find an appropriate climatic niche (Ackerly et al.,
Plants adapted to arid environments generally have higher δ13C, an indirect proxy for water use efficiency (WUE), than plants adapted to wet environments (Passioura,
Branch characteristics are closely related to productivity as they determine the quantity of light interception and CO2 assimilation (Halle et al.,
Our data suggest that
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We would like to thank Kyle Peer, Clay Sawyers, and Deborah Bird (Virginia Tech Reynold's Homestead Forestry Research Station) for assistance with plant propagation, establishment, and maintenance of the common garden, as well as the following organizations and individuals for providing genetic material for this study: Drs. Chang-Yi Xie and Alvin Yanchuk (British Columbia Ministry of Forests, Lands, and Natural Resource Operations), Dr. Brian Stanton (Greenwood Resources), Dr. Brad St. Clair [United States Forest Service (USFS)], and Dr. Dennis Ringes (USFS, retired). Finally, we thank two reviewers whose insightful comments greatly improved this manuscript. This work was supported by the National Science Foundation Plant Genome Research Program (IOS: 1054444) (grant to JH) and a Fulbright Institute of International Education Fellowship to RWO.
The Supplementary Material for this article can be found online at: