Tables

Table 1. Model averaging results (or best model, when no other models were within Δ AIC < 2, indicated with ’*’) for assessing importance of different microhabitat variables in explaining likelihood of recruitment (i.e. did seeds germinate or not), relative recruit counts (i.e. how many seedlings recruited relative to maximum recruit counts for the species), and 1- or 2-year seedling survival. ‘+’ indicates a positive parameter estimate and ‘” indicates a negative parameter estimate. Columns are colored by the expectation of those microhabitat variables being directly (blue) or indirectly (green) affected by climate change. Blank cells indicate that the parameter was not chosen in model selection. Because many of our soil loggers were compromised by animal disturbance, we only used those variables for species that showed no response at sites with compromised data: recruitment and recruit counts for M. nervosa, E. lanatum (summer maximum soil temperature, summer minimum soil moisture); seedling survival for R. ursinus (summer plant-height minimum temperature, summer minimum soil moisture). These variables are listed as ’NA’ for other species. Summer plant-height minimum temperature was never selected so it is not included in this table. Because of the link function in our binomial GLMs, the results can be interpreted as increasing or decreasing the log odds of recruitment or seedling survival, corresponding to lower or higher likelihoods, respectively. Abbreviations are as follows: Yr = year; Tr = transect; Canopy = canopy openness; F:B = fungus:bacteria; C:N = carbon:nitrogen; WHC = water holding capacity; Sum Soil Temp = summer maximum soil temperature; Sum Soil Moist = summer minimum soil moisture; Wint Soil Temp = winter minimum soil temperature; Snow = spring days with snow.