File S1 - Winter Climate Limits Subantarctic Low Forest Growth and Establishment

Contains the following files: Figure S1. Changes in the mean annual temperature (°C), mean annual precipitation (mm), and growth (mean detrended ring width across all plot) with year (1941–2000). Dashed line indicates the best fit line between year and either mean temperature, mean precipitation, or mean growth. Figure S2. Relationship between mean annual temperature (°C) between 1941 and 1989 at Campbell Island and a) Christchurch and b) Dunedin, the two closest mainland New Zealand climate stations with climate records extending back prior to 1940. Figure S3. Seasonal lag effects on growth when winter temperature range and winter precipitation are lagged by a) one and b) two years. The linear regression model was fit within a Bayesian framework specifying a lag in either one (climate: 1941–1998, growth: 1942:1999) or two years (climate: 1941–1997, growth: 1943:1999). Note that climate records were not available prior to 1941. The mean (filled dots) and 95% credible intervals (solid lines) for the parameter estimates describe the effect of each explanatory variable on plot-level annual growth rates. Categorical variables are relative to a reference class (good soil drainage, eutrophic soil type, D. scoparium species) as described in the Methods section. Credible intervals crossing the zero line (dashed) are not significant. Figure S4. Seasonal lag effects on establishment (occurred or did not) when maximum winter temperature is lagged by a) one and b) two years. The logistic regression model was fit within a Bayesian framework specifying a lag in either one (climate: 1941–1992, establishment: 1942:1993) or two years (climate: 1941–1991, establishment: 1943:1993). Note that climate records were not available prior to 1941. The mean (filled dots) and 95% credible intervals (solid lines) for the parameter estimates describe the effect of each explanatory variable on plot-level age-class structure. Categorical variables are relative to a reference class (good soil drainage, eutrophic soil type, D. scoparium species) as described in the Methods section. Credible intervals crossing the zero line (dashed) are not significant. Table S1. Summary statistics for chronologies. (DOC)