*2.6. Statistical Analysis*

Edge erosion was calculated for two periods from 1994 to 2011 and 2011 to 2017. The salt marsh edge erosion was calculated based on starting edge length and the total area lost for each edge erosion object. Welch ANOVAs were used to compare rates of change from 1994 to 2011 and 2011 to 2017 for both pannes and edge erosion. Least-squares means was used to compare the two edge erosion rates and to see if edge erosion had increased following the hurricane breach and how this varied between barrier island, bayside islands, and the mainland. A two-way ANOVA was used to compare panne yearly change rates before and after the breach and between barrier island, bayside islands, and the mainland. Annual salt marsh change rates were compared with a Kruskal–Wallis rank sum test, a nonparametric statistical test, followed by a Wilcoxon rank sum test. Linear regression models were used to test the relationship between edge erosion and distance from the breach for the period 2011 to 2017 for 2000 m in both directions. A beta regression ANOVA-like table was produced testing the relationship between the percent change from 2015 to 2017 of interior areas with and without mosquito ditch hydrological connections. All statistical analysis was done within the R 3.5.1 statistical environment [63].
