Satellite-Based Change Recognition and Tracking

Defoliation in NC's Roanoke River Bottomland Forests


The Forest Tent Caterpillar (Malacosoma disstria) is a widespread native defoliator of deciduous forests in the Eastern US. While host trees differ regionally, the insect prefers sweet gum (Liquidambar styraciflua), tupelo gum, black gum (Nyssa spp.) and species of oak (Quercus spp.) in the bottomland forests of North Carolina's Coastal Plain.

The larvae hatch in early spring as the tree buds swell, then larvae defoliate emerging and growing leaves in April and early May. After a few weeks, caterpillars form pupae, then adults emerge in late May.

This early spring timing explains why this caterpillar often has a limited impact on its host trees. By pupating by early May, there is plenty of time for trees to leaf out again so that they can function for the remainder of the growing season. It is thought that the largest impact to host trees is that defoliation reduces the duration of the growing season and tree growth. Nonetheless, host tree mortality and dieback can occur in the South, particularly during multi-year outbreaks (Baker 1972: 332).

Multi-year runs of landscape scale defoliation are a phenomenon for many forest defoliating insects. As shown by the accompanying slides, in the Roanoke River bottomland hardwoods, this current outbreak has been notable since 2013, so 2016 is year four. As the insect population expands, natural enemies grow, and these include egg parasites and insect predators.

ForWarn's systematic monitoring reveals something remarkable about the landscape population dynamics of this defoliator. Like other species', the hardest hit areas often shift from year to year. This pattern may result from local increases in enemies that eventually contribute to the outbreak's collapse at a broader scale. In the Roanoke bottomland forests, the last time defoliation was widespread was in 2000, and Landsat images and survey reports suggest that an outbreak much like that of recent years may have lasted from 1995 to 2000.

High flood waters can complicate remote sensing based assessments of defoliation in this environment. Erratic water releases from an upstream dam are common and floods are relatively long-lasting compared to natural events (Pearsall and others 2005). By relying on the maximum observed NDVI over a 24-day window, ForWarn's change product avoids short term flood signals compared to single-date imagery, such as Landsat, yet long lasting floodwaters may still occasionally influence ForWarn's NDVI. That said, the detectability of standing water in the understory is likely greatest prior to greenup (i.e., winter and early spring) and where insects or wind storms have defoliated the canopy in late spring or summer. That is, by late spring, understory water may only be discernible where defoliators have unmasked it. Thus, by mid to late May, ForWarn's change products are likely to mostly show areas of defoliation, with some increase in change intensity resulting from floodwaters, when present.

Within a broader context, it has been suggested that outbreaks may have increased due to this altered flow regime that is associated with flood control by the US Army Corps of Engineers and hydroelectric generation (US FWS 2005: 3, 42). These managed flows are less extreme, of longer duration, and occur more often during the growing season than what occurred naturally (Hochman 2004). The hypothesized mechanism is that growing season flooding may reduce populations of ground-dependent parasitic wasps so that they are less able to regulate forest tent caterpillar population levels (Pearsall and others. 2005).

Long-term monitoring is an important part of addressing forest impacts, and only by accurately mapping annual defoliation can multi-year severity be addressed based on both the seasonal magnitude and the multi-year duration of those events. Because trees are long-lived perennials, it is often the cumulative effects of multi-year stress that influences how stands evolve.


Baker, WL 1972. Eastern Forest Insects. USDA Forest Service Misc. Pub. 1175.

Hochman, ER. 2004. Lower Roanoke River hydroperiods: altered hydrology and implications for forest health and species response. MS Thesis, Univ. North Carolina, Chapel Hill.

Pearsall, SH., BJ McCrodden, and PA Townsend. 2005. Adaptive management of flows in the Lower Roanoke River, North Carolina, USA. Environmental Management. 35 (4) 353-367.

US Fish and Wildlife Service. 2005. Roanoke River National Wildlife Refuge Comprehensive Conservation Plan.