Spatial modeling of climate-growth relationships of eastern North American trees

Background/Question/Methods

The principle of ecological amplitude is a long-held paradigm in the field of dendrochronology that contends tree species near their geographic range edges will be subject to greater environmental stress due to lower suitability of habitat.  Greater environmental stress should lead to decreased growth rates, higher sensitivity to climate variability, and may be an explanatory mechanism of species distributional limits.  Here, we investigated the principle of ecological amplitude by analyzing the pattern of climate sensitivity in annual growth of eastern hemlock (Tsuga canadensis) and white oak (Quercus alba) two widely distributed, long-lived species that have been extensively used for dendrochronological research in eastern North America.  We used a generalized additive model (GAM) to investigate the pattern of annual growth variability across the geographic ranges of both tree species using existing tree-ring records.  We examined the influence of distance from range edge, as well as latitudinal and longitudinal position within each species’ range.  We hypothesized that variability of annual growth (mean sensitivity) would be higher near the edges of species’ ranges and lower at interior range locations.  Additionally, we predicted that eastern hemlock sensitivity would be less variable than white oak because hemlock has a highly restricted topographic distribution. 

Results/Conclusions

Despite the fact that a very flexible model was fit to the tree-ring data, distance from range edge was not a significant predictor of average mean sensitivity for either tree species and did not influence the pattern of sensitivity in the GAM predictions.  Model output predicted a broad-scale trend of increasing sensitivity along a southeast to northwest gradient through each species’ range.  Additionally, trends in ring-width sensitivity are significantly correlated with climate variables associated with climate continentality.  Mean sensitivity of eastern hemlock growth was significantly correlated with annual temperature range (r = 0.56, p < 0.05) and mean annual precipitation (r = –0.46, p < 0.05).  White oak mean sensitivity was significantly correlated with annual temperature range (r = 0.34, p < 0.05) and mean annual precipitation (r = –0.42, p < 0.05), but the strength of association was weaker than eastern hemlock.  These results suggest that spatial variability in annual growth does not follow a distance-decay pattern from range center to range edge, but rather growth variability changes along a monotonic gradient of climate continentality.  These findings have implications for research attempting to model geographic range dynamics under novel climate conditions.