Patterns of forest tree growth at landscape and regional scales are often affected by changes in broad scale site quality and climate [1
]. However, at local scales heterogeneous environments generally have more variable impacts on individual growth [2
]. Many characteristic growth patterns in trees, such as leaf size [4
], shape [5
], height [6
] and radial growth [7
] are linked to environmental factors like temperature [8
], moisture availability [4
], competitive interactions [10
], and stress [12
]. Various morphological and physiological changes can occur when trees are subjected to environmental stress. For instance, changes in foliar growth, leaf/needle length, root growth and the production of defense compounds can be altered across heterogeneous sites [13
]. Alpine treelines are unique site that are spatially heterogeneous boundaries between sub-alpine forest and alpine tundra. At global scales treeline position is temperature controlled [14
], but at local scales it is structured by microclimatic, topographic, edaphic, and biotic variation [15
]. Alpine treeline thus represents a harsh environment with strong influence on spatial patterns of radial growth [7
]. Despite the fact that patterns of growth are influenced by various biotic and abiotic processes, no consensus has been reached regarding the degree to which the stability of growth, or the overall reduction in year-to-year variation in growth, in plants is related to individual genetic diversity. That is, if plants have higher genetic diversity (individual heterozygosity), do they also have more consistent and stable growth patterns (homeostasis) relative to environmental variation [18
] like that found at the alpine treeline?
Understanding how individual trees will respond to changing environmental conditions is important. Conifer trees have high levels of within population genetic diversity and low between population genetic differentiation [20
]. In addition, conifers have high degrees of phenotypic plasticity [22
] allowing them to respond differentially to changing environmental conditions. Climate change has the potential to alter geographic patterns of plant ranges. Often, the most rapid response for a plant is to shift its range to higher latitudes or elevations through seed dispersal [23
]. However, in the case of slow growing long lived species, like conifers, the ability to disperse over long distance to escape changing conditions may be curtailed by the length of time it takes for the species to germinate and grow to reproductive maturity resulting in an adaptational lag [24
]. One alternate response to dispersal and range shift is to rely on their plastic responses to withstand variable environmental changes. To this end, individual trees that have less variable radial growth across heterogeneous environments have a greater fitness advantage and may be the subject of natural selection and local adaptation. This idea, known as developmental or genetic homeostasis, is not new [18
]. Positive relationships between heterozygosity and reduced morphological variance have been identified in several organisms [28
]. Despite the aforementioned relationship, previous results in forest trees, and conifers in particular, have been less consistent. For instance positive associations between heterozygosity and basal growth were identified in quaking aspen (Populus tremuloides
] and height in Douglas fir (Pseudotsuga menziesii
(Mirb.) Franco) [30
], but in lodgepole pine (Pinus contorta
Dougl. ex Loud.), and ponderosa pine (Pinus ponderosa
Dougl. ex Laws.) no significant relationship was found [31
], while mixed results were identified in pitch pine (Pinus rigida
Mill.), where some stands had a positive growth relationship with heterozygosity while others had a negative association [32
]. Most of these earlier studies relied on allozyme loci. Because of the inconsistencies in earlier results the subject is worth exploring in more detail under the lens of new genomic approaches which allows for a high resolution investigation into the patterns of individual genetic variation and growth.
A recent study by Babushkina et al. [33
] introduced the idea of correlating individual heterozygosity to growth parameters using dendrochronological techniques in Siberian larch (Larix sibirica
Ledeb.). The authors of the study compared average individual tree-ring width (AvTRW), variance (VarTRW) and individual heterozygosity (IndHet), using eight highly polymorphic microsatellite markers, also known as simple sequence repeats (SSRs), with the assumption that individual trees with higher heterozygosity would positively correlate with heterosis and have more stable homeostasis reflecting less developmental dependence on their environment. With this in mind, Babushkina et al. [33
] tested the hypotheses that there would be a negative correlation between IndHet and VarTRW and a positive correlation between IndHet and AvTRW. Ultimately, their findings were inconclusive. They found non-significant positive correlations between IndHet and both VarTRW and AvTRW. They also found that AvTRW and VarTRW were significantly and positively correlated with each other. Babushkina et al. [33
] suggested that perhaps under unfavorable growing conditions most fast growing trees grow unstably regardless of the level of genetic diversity. However, they posit that with only eight microsatellite markers the resolution of their analysis may be limited and recommended an analysis using genome wide genotyping with thousands of molecular markers.
In this study, we tested the association between tree growth at the alpine treeline, measured as an individual average tree-ring width (AvTRW) along an elevational gradient, and individual genetic diversity, measured as an observed individual heterozygosity (IndHet) averaged over 4665 genomic single nucleotide polymorphism (SNP) loci. We assessed the patterns of individual growth and genetic diversity in a treeline formed by a conifer species, mountain hemlock (Tsuga mertensiana
Bong Carr.) on the Kenai Peninsula, Alaska, USA. We followed the approach outlined in Babushkina et al. [33
] to clarify our understanding of the molecular basis underlying stability and growth in forest species with a focus on the question: do individual trees with higher heterozygosity, approaching their altitudinal limit, exhibit more stable patterns of growth? We answered our research question by testing two hypotheses: (1) IndHet will be negatively correlated with VarTRW; and (2) IndHet will be positively correlated with AvTRW.
Our analysis identified no significant relationships between AvTRW, VarTRW and IndHet. We hypothesized a negative trend between VarTRW and IndHet as a function of growth and development being less dependent on the environment at higher levels of heterozygosity (homeostasis). Our results revealed a non-significant weak positive trend in the data. The direction of the trend was opposite of the expectations we laid out for the study (Table 2
, Figure 4
). AvTRW had a non-significant weak positive trend with IndHet matching the expectation that if growth were to be considered an adaptive trait, it would be positively correlated with individual genomic diversity due to heterosis. One of the most surprising findings was the significant positive relationship between AvTRW and VarTRW which, as Babushkina et al. [33
] suggested, supports the idea that under poor environmental conditions fast growing trees grow unstably regardless of the level of heterozygosity. Though it is difficult to draw sound mechanistic explanations about this finding, we can speculate on the causes of this phenomenon. Alpine treelines are temperature limited and generally have short growing seasons constituting a poor growing environment. Because of this trees will generally grow more slowly regardless of the level of IndHet. This growth pattern is often confirmed at treeline where older trees will have small diameters relative to trees of the same age in more productive systems in the continuous subalpine forest [16
]. However, when relatively favorable conditions do occur at the treeline, trees with higher heterozygosity should respond more favorably by increasing radial growth leading to a positive association between growth and variance. Moreover, the positive association between AvTRW and VarTRW could be the result of tradeoffs in reproductive output [31
]. This hypothesis argues that if seed production (reproduction) and growth are negatively correlated, a plausible tradeoff at alpine treeline, increased growth would not by itself increase the fitness of individuals there. This finding would mean that greater variance for the growth trait would occur due to more fluctuations in cone production in fast growing trees. We cannot add additional insight to this hypothesis here, and we encourage further research into this potential relationship.
In their study, Babushkina et al. [33
] identified the same three trends using eight highly polymorphic SSR markers across two populations of L. sibirica.
Both AvTRW (r
= 0.146, p
= 0.147) and VarTRW (r
= 0.122, p
= 0.225) had non-significant and positive trends with IndHet, as well as a significant positive correlation between AvTRW and VarTRW (r
= 0.726, p
< 0.001). The authors of the aforementioned study concluded that the relationships between heterozygosity and growth were clearly complex and non-linear. The lack of correlation between individual heterozygosity, derived from the SSRs, and growth parameters could have been a result of the markers failure to accurately reflect underlying genomic diversity. Because of this Babushkina et al. [33
] recommended the use of genome wide sequence data as a means to better capture underlying individual genomic diversity. Despite their recommendation, using more than 4000 SNPs, our study failed to identify a significant relationship between growth and genomic diversity. Even more surprising was the consistency in results between the two studies suggesting that even when using high resolution genomic sequence data patterns of homeostasis may be difficult to detect and may be non-linear.
In terms of climate and growth as measured by the ring width index (RWI), our chronology statistics were qualitatively similar to earlier studies carried out in T. mertensiana
on the Keani Peninsula and in the region broadly [58
]. Response function analyses of the ring-width chronology identified significant correlations with both July precipitation in the year preceding growth and June temperature in the year of ring formation, as the dominant climate factors influencing radial growth in T. mertensiana
(unpublished data). Peterson and Peterson [17
] established that at high elevations in the Pacific Northwest radial growth in T. mertensiana
was positively correlated with growth-year summer temperature and negatively correlated with spring snowpack.
It must be noted, the results of our study should be viewed as a case study and care should be taken when extrapolating the findings to other systems. Our study was based on a limited number of individuals, though representative of the local population. Future work should focus on a more robust set of samples from several tree species.
Growth and Stability at Alpine Treeline
The alpine treeline is typically characterized by poor growing conditions [14
]. At a global scale temperature is the primary limiting factor affecting tree growth at both the altitudinal and latitudinal limit [14
]. However, micro-site conditions, including growing season length, moisture availability, regeneration limitation, slope position and geomorphic limitations to growth, ultimately determine site suitability and tree growth potential [63
It is tempting to suggest that individual trees with higher genomic diversity should be more resilient under the poor growing conditions found at alpine treeline. Moreover, under future climate change, it is important to begin to untangle the alternative strategies that forest trees employ in response to rapid changes in their environment. Previous studies of heterosis and genetic homeostasis have shown that in some systems there is a relationships between higher individual genetic diversity and stable growth patterns [26
]. In fact, both our study and that of Babushkina [33
] have been unable to find support for this hypothesis in two different conifer species using two different numbers and types of genetic markers, thereby calling into question the validity of the hypotheses. The mountain hemlock treeline on the Kenai Peninsula, Alaska has been characterized by high pollen and seed dispersal into the ecotone [34
]. Perhaps, the high degree of gene flow, particularly from pollen, contributes to the lack of association between IndHet and AvTRW/VarTRW. Genetic diversity is often structured according to the center-periphery (central-peripheral or central-marginal) model. In this model higher gene flow from large central populations into small peripheral ones, such as the altitudinal range limit constituting the alpine treeline, maintains genetic diversity [76
]. This phenomenon is often associated with reduced adaptive potential of individuals at the range edge. This occurs because the influx of genes adapted to the center of the range counters the impact of selection for traits suitable to the surrounding environment (e.g., gene swamping) [77
]. If gene swamping is occurring at the alpine treeline, then the patterns of growth may not accurately reflect the underlying genetic diversity represented in the local gene pool, and this may ultimately limit the adaptability of trees at their range limits.
The degree to which the level of heterozygosity contributes to homeostasis needs to be further examined using a more extensive spatial sampling across an expanded geographic extent among a variety of treeline types. We recommend that analysis incorporate both selectively neutral and putatively adaptive genomic markers to more accurately assess levels of heterozygosity. The use of dendrochronological techniques in combination with genomic sequence data shows promise, and the development of longer chronologies may help to untangle some of the climatic variability allowing a longer temporal scale assessment of growth.