Skip to main content
. 2017 May 4;12(5):e0175018. doi: 10.1371/journal.pone.0175018

Table 3. Results for the regression model of coral bleaching as a function of different explanatory variables.

Variables Coefficients Average Marginal Effect
Observed DHW 0.201*** 3.31***
(0.014) (0.16)
Maximum PAR Anomaly 0.0268*** -0.0274
(0.0054) (0.059)
Distance from Shore -0.0363*** -0.72***
(0.0137) (0.27)
Population 0.0311*** 0.618***
(0.0059) (0.12)
Wind 0.0132 0.263
(0.0085) (0.17)
Depth 0.0734*** 0.375***
(0.011) (0.085)
Depth×Observed DHW -0.00336**
(0.0010)
Depth×Maximum PAR Anomaly -0.00246***
(0.00049)
Bermuda 0.114 -19.8***
(0.137) (1.7)
Bahamian 1.05*** 2.8**
(0.245) (1.4)
Eastern Caribbean 0.0185 0.86
(0.156) (0.64)
Greater Antilles 0.325* 1.6**
(0.188) (0.77)
Southern Caribbean 1.16*** -3.8**
(0.148) (1.9)
Southwestern Caribbean 0.605*** -0.69
(0.148) (0.63)
Floridian 0.476** -2.2**
(0.176) (0.89)
Constant -2.421***
(0.198)
Log-likelihood -1299
Cross-validated MSE 607.0
Cross-validated RMSE 24.6
Cross-validated R2 0.29
Number of Observations 2,945

Estimated coefficients and average marginal effects using a fractional logit model, where coral bleaching is explained as a function of environmental stressors. Robust standard errors are shown in parentheses (*** p<0.01, ** p<0.05, * p<0.1).