Login / Signup

Applying space-for-time substitution to infer the growth response to climate may lead to overestimation of tree maladaptation: Evidence from the North American White Spruce Network.

Fang WuYuan JiangShoudong ZhaoYan WenWenqing LiMuyi Kang
Published in: Global change biology (2022)
Under climate change circumstances, increasing studies have reported the temporal instability of tree growth responses to climate, which poses a major challenge to linearly extrapolating past climate and future growth dynamics using tree-ring data. Space-for-time substitution (SFTS) is a potential solution to this problem that is widely used in the dendrochronology field to project past or future temporal growth response trajectories from contemporary spatial patterns. However, the projected accuracy of the SFTS in the climate effects on tree growth remains uncertain. Here, we empirically test the SFTS method by comparing the effect of spatial and temporal climate variations on climate responses of white spruce (Picea glauca), which has a transcontinental range in North America. We first applied a response surface regression model to capture the variations in growth responses along the spatial climate gradients. The results showed that the relationships between growth and June temperature varied along spatial climate gradients in a predictable way. And their relationships varied mainly along with local temperate condition. Then, the projected correlation coefficients between growth and climate using SFTS were compared against the observed. We found that the growth response changes caused by spatial versus temporal climate variations showed opposite trends. Moreover, the projected correlation coefficients using the SFTS were significantly lower than the observed. This finding suggests that applying the SFTS to project the growth response of white spruce might lead to an overestimation of the degree of tree maladaptation in future climate scenarios. And the overestimation is likely to get weaker from Alaska and Yukon Territory in the west to Quebec in the east. Although this is only a case study of the SFTS method for projecting tree growth response, our findings suggest that direct application of the SFTS method may not be applicable to all regions and all tree species.
Keyphrases
  • climate change
  • machine learning
  • depressive symptoms
  • electronic health record