As sea level rises, the landward extent of high tide begins to inundate terrestrial uplands. Most often this upland is forest which responds to the increase in inundation frequency and salinization first with a loss of seedlings and saplings in the understory, limiting regeneration of the forest. Ultimately, the mature trees succumb as well, leading to the replacement of the forest ecosystem with marsh. The standing dead trees surrounded by marsh are known as “ghost forests” and feature prominently in media portrayal of sea-level rise impacts on the Atlantic coast of the United States. This process of forest conversion to marsh is especially well studied in the mid-Atlantic region where extensive areas of low-lying land, coupled with high rates of relative sea-level rise (RSLR) have led to the conversion of 236 km2 of forest to marsh since 1984 (Chen & Kirwan, 2023). While we think about sea-level rise in very small units, just millimeters of rise per year, the realized impact of even small amounts of change can be staggering. Lateral rates of forest loss in the mid-Atlantic have been measured to reach up to 40 m yr-1 in areas of extremely low slope on the Eastern Shore on Maryland (Chen & Kirwan, 2023).
Driving north on Route 13 through the Eastern Shores of Virginia and Maryland, and up into Delaware, you pass by hundreds of square kilometers of soy, corn, and wheat fields which have replaced forests in the low-lying coastal land vulnerable to sea-level rise. Through conversations with landowners across the region, we knew that farmland was already being impacted by saltwater intrusion. But there had never been a regional assessment of how rapidly we were losing valuable agricultural land to sea level-rise. So we set out to quantify where farmland had already been lost and at what rate over almost four decades of Landsat imagery. To do so we adapted remote sensing methods previously developed to detect forest conversion to marsh (Chen & Kirwan, 2023) to detect agricultural land that had converted to marsh in the region based on training samples from our collective field-based knowledge and input from collaborators.
Our results showed that over 100 km2 (24,982 acres) of agricultural land converted to marsh from 1984 to 2022 around the Chesapeake and Delaware Bays. Unsurprisingly, the hotspots of farmland lost were largely driven by low slope, just like the forestland. Geometrically, a little sea level rise over flat ground results in a large area of inundation. But we were interested in understanding the role people play in sea level-driven land loss in these rural land uses. To do so, we calculated vertical retreat rate, or how fast the boundary between marsh and agriculture moved upslope over the 38-year study period. This method removes slope from the equation and instead offers a way to directly compare the rates of land use change with the rate of sea level-rise. If sea level-rise was the only driver of vertical retreat, then the rate of sea level-rise would equal the rate of vertical retreat of the marsh-upland boundary. Farmland retreated upslope 3.22 mm yr-1, which was faster than forestland (2.71 mm yr-1), but at a rate slower than regional RSLR (4.13 - 6.74 mm yr-1 between 1984 and 2022; Ezer 2023). We were initially surprised by this finding as we had expected the farmland retreat rate to be much faster than that of forestland. Our initial hypothesis had been that the replacement of mature trees, which can take a decade or more to die from saltwater intrusion (Chen & Kirwan, 2023), with annual crops would hasten the land use change. But these results suggested that something, or in this case, someone, was successfully resisting the landward encroachment of marsh.
Defending land from inundation is nothing new in the mid-Atlantic region. Landowners have been building small earthen berms around their properties since the mid-1800s. Most of these features are small, less than a meter in height and width, and were likely dug by hand. Colloquially, landowners on the Eastern Shore know about these berms. Many have berms on their property that were inherited from previous landowners. If you know what to look for, you can see them in aerial imagery across the entire region. But you won’t find them in any federal or state databases. Could these berms be the reason the farmland was not converting as fast as expected? To test this, we hand digitized over 100 km of berms in Somerset County, Maryland, where extensive agricultural land conversion had also been documented. We then examined the rates of farmland conversion for fields that had barriers to inundation and those that did not. We found that farmland without structural barriers retreated 3.45 mm yr-1, while farmland containing barriers retreated 2.14 mm yr-1. The rate of forest retreat in Somerset County was 2.31 mm yr-1 which suggests that the barriers are slowing the vertical rate of retreat below that of the “natural” ecosystem.
This work has important implications for models forecasting sea level-driven land use change which currently either lump agricultural land in with developed land use classes that are assumed to not convert, or with forestland where the assumption is that the conversion process is the same. We demonstrate that there is widespread farmland conversion to marsh in the mid-Atlantic region. However, there is a scale-dependency to this process such that when we zoom in to the level of individual farm fields, we find that landowners are able to temporarily delay the loss of their land to sea-level rise. Farmers are currently experimenting with other methods such as applying gypsum to neutralize the salts ions in the soil or planting more salt-tolerant crops, which could continue to influence the rate of agricultural land loss far into the future. The physical, biological, and chemical methods of resistance culminate in a process of coastal land use change that is fundamentally different for agricultural land than other land use types and suggests a need to reframe how we view to the conversion of economically valuable, private property to marsh in rural areas.