**4. Discussion**

The change analysis between the 1997 classification and the 2015 classification revealed salt marsh loss (Figure 8). High marsh area fell from 199.6 ha to 109.8 ha. Previous studies mapping salt marsh change from 1974 to 2005/2008 for the entirety of Long Island, NY found similar change, including a 35.5% reduction in the high marsh for a region from Fire Island inlet to Smith Point, and a decrease in Phragmites on the south shore [15]. The conversion of upland and Phragmites to low and high marsh categories suggests salt marsh migration in response to SLR. The utility of the comparison between 1997 and 2015 was limited due to the different classification schemes.

In general, pannes/pools demonstrated several periods of statistically significant expansion. Of the 475 pannes, 46% were present in 1994. Meaning there was a doubling of pools and pannes from 1994 to 2015. Two hundred and twelve of the 475 pannes were in areas classified as high marsh in 1997. These pannes accounted for 12.51 ha out of a total of 21.81 ha, i.e., the largest area of pannes occurred in the high marsh. These non-vegetated pannes/pools are essentially tidal mudflats which provide some essential ecosystems services. However, ecosystem service valuations sugges<sup>t</sup> salt marsh to be over five times more valuable than mudflats [64].

The expected evolution of an interior salt marsh pool is expansion until hydrological connectivity is established leading to drainage and possible vegetation regrowth [22]. In our analysis, pannes/pools connected to mosquito ditches in 2015 had a mean change rate of −3.52 m<sup>2</sup> y<sup>−</sup><sup>1</sup> compared to 30.87 m<sup>2</sup> y<sup>−</sup><sup>1</sup>

for non-hydrologically connected pannes/pools. This is encouraging for the possibility of vegetation regrowth. However, natural creeks are infrequent landscape features having remained relatively stable from 1930 to 2007 [61]. In contrast mosquito ditches are common across Fire Island, leading to hydrological connectivity with mosquito ditches being common. However, besides providing a hydrological connection, mosquito ditches likely drive drowning by altering marsh hydrology, and plugged mosquito ditches cause subsidence and loss of salt marsh function [65]. Additionally, the berms surrounding ditches can lead to poor drainage [66]. Highly variable accumulation of sediment in Fire Island's mosquito ditches has led to the infill of some ditches and little to no accumulation in others [61] (Figure 8). The varied rate of infilling could be influencing observed rates of panne/pool expansion. The landscape legacy of the mosquito ditches is a site-specific factor that is critical for understanding salt marsh change on Fire Island.

**Figure 8.** Land cover of the area surrounding the 2012 Fire Island breach for 1997 and 2015. Land cover change both due to the breach and overwash are evident in the 2015 classification. *Spartina alterniflora* classes are shown as a single class due to the 1997 class no differentiating between percent cover. Upland and dune vegetation classes are also shown as a single class.

Whether vegetation regrowth occurred within the pannes/pools is a critical question. Vegetation regrowth is limited by the growth range of *S. alterniflora* at the site. The lowest elevation of living *S. alterniflora* at the site was 25.7 cm below NAVD 1988 [67]. The minimum growing elevation of *S. alterniflora* for FIIS was determined using the tidal range of 45.5 cm and the methods of [10]. Finding a minimum growth elevation of 12.7 cm below NAVD 1988, which is a more conservative estimate than the observed minimum growth elevation. Six of the 475 pannes analyzed were below the vegetation range of *S. alterniflora* at the site, meaning vegetation could grow on nearly all of the observed pannes. However, only 30 of the 475 2015 pannes/pools were entirely vegetated in 2017. Complete vegetation regrowth was rare but did occur in the pannes and pools analyzed.

Significant increases in edge erosion were observed following the breach. These areas likely experienced changes in currents, wave energy, or LMSL. For example, the William Floyd estate site saw no significant di fference in edge erosion before and after the breach. This site is approximately 8 km away from the breach and approximately 5 km from the stabilized Moriches Inlet. In contrast, the area immediately surrounding the breach to the east experienced significant loss from increased edge erosion. An increase in edge erosion as you neared the breach was evident towards the east. However, there was no such trend to the west of the breach. The high variability of the bayside salt marsh erosion demonstrates the importance of geospatial monitoring to understand how these systems are changing spatially. As previous studies reported, Surface Elevation Table (SET)-derived accretion estimates at the site are below the rates of SLR [5]. FIIS' wilderness areas have little infrastructure limiting the migration of salt marsh. However, the islands width and interconnectedness of the barrier island systems means salt marsh migration alone will not maintain the barrier island.

The bayside of barrier islands have low energy and small tidal range (Watch Hill, NY on Fire Island's tidal range is 45.5 cm between MLLW and MHHW [43]), which can result in slower expansion of pools due to edge erosion [25]. The establishment of an inlet can cause an increase in tidal range; however, there were no statistical di fferences between panne/pool expansion rates between the examined time periods (1994–2011 and 2011–2017). Possibly due to the scarcity of pools in our analysis which would be expected to expand more rapidly with increased tidal range. The breach caused by Hurricane Sandy did not appear to accelerate or slow the interior salt marsh change. However, edge erosion significantly increased following the breach. Continued monitoring is necessary to determine if the observed trend continues.
