S6_Scraped_code.R is the R code to run competing models to examine if scraping is an important predictor of recruitment in 2011 and 2012. Models included scraping and size of clearing as fixed effects and the effects of years, bays, sites nested within bays and all interactions as random effects. Average recruitment from 1997 to 2008 was included as a covariate. Models are hierarchical, that is models lacking a particular main effect would not include interactions involving that effect. Code uses lmer() in the lme4 package and then model.sel() in the MuMIn package to obtain corrected AIC, deltas and AIC weights.
Data file, S6_Recruit_2011-2012_R_ready.csv, is organized in nearly identical manner as S3_Recruit_1997-2008_R_ready.csv in Supplement 3.
