Hauling field gear up Gothic Mountain…
Apply climatic variable selection
Applied nonparametric Spearman rank correlation to explore relationships among climatic variables. Excluded 13 climatic variables that were highly correlated over a pre-selected threshold of ±0.6, retaining six variables: mean annual temperature, temperature seasonality, maximum temperature of warmest month, minimum temperature of coldest month, annual precipitation, and precipitation seasonality.
Quantities: Reduced from 19 to 6 climatic variables using ±0.6 correlation thresholdDuration: Not specifiedConditions: Pre-selected correlation threshold of ±0.6
Equipment: R software
Fit generalized additive mixed-effects models
Fitted GAMMs for family richness with region as a random intercept and climatic variables as fixed effects. Used Poisson distribution and log link function. Applied backward elimination procedure, retaining only significant GAMMs (p < 0.05) and variables that made a significant contribution to explained deviance.
Quantities: 12 GAMMs fitted (6 climatic variables × 2 wetland types), with 21 levels for temporary and 17 levels for permanent wetlands as random interceptsDuration: Not specifiedConditions: Separate models for temporary (n=459) and permanent (n=301) wetlands
Equipment: R software, mgcv package
Calculate beta diversity metrics
Calculated Local Contribution to Beta Diversity (LCBD) index using the R package 'adespatial' and total pairwise beta diversity using the R package 'betapart' following Baselga (2010). LCBD assesses degree of uniqueness in assemblage composition, with values ranging from 0 to 1.
Quantities: LCBD values calculated for all 769 sites, pairwise beta diversity for all site combinationsDuration: Not specifiedConditions: Using Hellinger-transformed presence-absence data
Equipment: R package 'adespatial', R package 'betapart'
Fit generalized dissimilarity models
Used GDMs for modelling beta diversity patterns along climatic gradients. Fitted separate GDMs for temporary wetlands (n=459) and permanent wetlands (n=301), based on total beta diversity. Used backward elimination procedure with three I-splines per predictor, retaining only significant GDMs (p < 0.05).
Quantities: 2 separate GDM models fitted (temporary vs permanent wetlands), with 3 I-splines per predictorDuration: Not specifiedConditions: Default of three I-splines per predictor, backward elimination with p < 0.05 significance threshold
Equipment: R software, gdm package