Harvard Forest Research

Effects of warming on tree species’ recruitment in deciduous forests of the eastern United States

Principal Investigator: Jerry Melillo
Marine Biological Laboratories: Jan 01 2008 - Jan 01 2012:

Abstract:
Climate change will restructure the forests of the United States over the coming century,
although the details of this restructuring remain uncertain. Projected increases in surface
temperatures of between 2 and 8oC and associated changes in soil moisture by 2100 will affect
various aspects of tree recruitment, including germination, growth and mortality. It has become
increasingly apparent over the last decade that seedling recruitment most often determines tree
species’ abundances and distributions. In contrast to mature trees, which have carbohydrate and
water reserves that buffer against environmental variation as well as extensive root systems to
forage broadly, seedlings can be highly sensitive to short-term changes in temperature and soil
moisture status. Surprisingly, few studies examine recruitment impacts of climate variation and
change in relevant settings. We propose to determine the effects of air and soil warming on
recruitment in mixed deciduous forests in southern New England and in the Piedmont region of
North Carolina.
Our proposal builds on two extensive, long-term experimental and observational studies
of recruitment effects of climate variation. These include a 4-yr experiment of soil warming at
Harvard Forest in central Massachusetts (Melillo et al. 2007, Mohan et al. 2007a) and 6 to 14-yr
studies exploiting experimental planting, natural gradients in temperature and soil moisture, and
canopy manipulations in North Carolina (Clark et al. 2003, Ibáñez et al. 2007). In our continuing
experiment on the effects of a 5oC soil warming on ecosystem structure and function of a midsuccessional
forest in central Massachusetts, we have observed dramatic juvenile tree response
during four years of treatment. Soil warming has caused the largest growth increases in slowgrowing
species. Fast-growing species exhibited diminished growth and increased mortality. We
attribute part, but not all, of the enhanced growth for the slow-growing species to increased
nitrogen availability in the warmed plots associated with more rapid decay of soil organic matter.
In North Carolina, analysis of 14,000 tree seedlings responding to spatio-temporal variation in
climate showed that warm spring temperatures and summer drought are the most important
predictors of germination, seedling growth, and survival. Ibáñez et al. (2006, 2007) point out
that the demographic responses we observe to temperature and soil moisture in these
experimental and observation studies are not consonant with predictions of climate envelop
models for the region (e.g., Iverson and Prasad 1998, Iverson et al. 1999, Iverson and Prasad
2002). Together these studies motivate warming experiments to assess the full complement of
species responses in these two ecosystems.
Two questions organize our proposed research: (1) Might temperate tree species near
the “warm” end of their range in the eastern United States decline in abundance during the
coming century due to projected warming? and (2) Might trees near the “cool” end of their
range in the eastern United States increase in abundance, or extend their range, during the
coming 100 years because of projected warming?
1
To explore these questions, we will conduct air and soil warming experiments in two
eastern deciduous forest sites; one at the Harvard Forest in central Massachusetts and the other at
the Duke Forest in the Piedmont region of North Carolina. We will work with a set of tree
species selected to represent taxa common to both Harvard and Duke Forests (such as red, black,
and white oaks), those near northern range limits (black oak, flowing dogwood, tulip poplar), and
those near southern limits (yellow birch, sugar maple, Virginia pine). Our extensive research at
both sites with all of the dominant species provides us with a sound basis for this manipulation
(Harvard Forest, Melillo and Mohan; Duke Forest, Clark and Mohan).
At each site, we will plant seeds and seedlings of selected tree species in common
gardens established in temperature-controlled, open-top chambers. The experimental design is
replicated (n=3) and fully factorial and involves three temperature regimes (ambient, +3oC and
+6oC) and two light regimes (closed forest canopy (low light) and gap conditions (high light)).
Our hierarchical modeling techniques allow us to draw inference at level of individual plants.
Whereas traditional design-based inference could be limited to 3 replicates, we can fully
assimilate both treatment and covariate information, accommodate the correlations among
individual responses of trees in competition, and allow for plot-level effects (LaDeau and Clark
2006, Mohan et al. 2007b, Clark 2007).: