Tree neighborhood and climate effects on radial growth in an eastern hemlock stand
Climate and local stand density (i.e., neighborhood) are major factors determining growth of individual trees, yet disentangling their relative importance is challenging because most dendroclimatological approaches use ring-width indices that remove growth variation resulting from stand dynamics. We sought to estimate the relative influence of target tree size, neighborhood configuration, and climate on growth in an old-growth eastern hemlock forest in the Helderberg Mountains of west-central New York. The stand has been the subject of a three-decade study on forest dynamics and five maps of tree positions measured at 8 year intervals were available for estimating neighborhood configuration. Increment cores were obtained from 60 to 80 individuals of hemlock, sugar maple, and yellow birch (only hemlock is reported here); these were prepared and measured using standard methods. Annual precipitation and mean winter air temperature were obtained from gridded topoclimatic datasets. A neighbor competition index was estimated from sums of distance-weighted diameters of trees < 10 m from target trees. Target tree diameter at each year was estimated by subtracting cumulative increases in ring width from radial dimension in the year when samples were taken. Radial growth was modeled as a function of size, neighborhood, and climate using likelihood methods.
Median age of sampled trees was 130 yrs, but several hemlocks were 200 yr old and maximum age was >300 yr. Size, neighborhood configuration and climate accounted for ca. 27% of observed growth in hemlocks. Size (diameter) had the largest influence on tree growth; radial growth peaked at 44 cm diameter then declined gradually (sampling did not capture the early rapid stages of ring growth). Neighbor competition was the next-most important factor. Neighboring conifers had a much weaker competitive effect than deciduous trees (per unit of cross-sectional trunk area) against hemlocks. Incorporating neighbor effects increased R2 by 0.04-0.08. Climate effects, although judged important by information criteria (AICc), produced a weaker increase in R2 (0.01-0.04%) than size or neighborhood. The data suggested a growth peak at mean maximum winter temperatures of 1.1 °C, a winter temperature which will be reached in a decade in the forest studied if the present increase rate of 0.4°C decade continues. Using a combination of neighborhood methods and traditional dendrological methods provides multi-dimensional insight into controls on individual tree growth.