MODELING SEAGRASS LANDSCAPE PATTERN AND ASSOCIATED ECOLOGICAL ATTRIBUTES

Abstract

A predictive model for seagrass bed coverage (presence/absence at 1-m resolution) and ecological attributes of the bed, such as biomass and shoot density, would be a valuable management tool. But forming such a predictive model is complicated by a number of factors that strongly influence seagrass bed structure and our interpretation of its ecological function. The factors include the effects of waves and water depth (hydrodynamic setting) and the spatial and temporal scales of the sampling technique itself. In this study, we examined the coherence of predictions of seagrass cover and ecological attributes of temperate, mixed-species seagrass derived from two common sampling techniques, (video) line transect (commonly used by biologists) and grid-sampled surveys (often used in remote sensing). Mapping resolution was held constant at 1 m, and the two techniques applied across seagrass beds of varying coverage that reflected the effect of a hydrodynamic gradient ranging from patchy, high-energy beds to continuous cover, low-energy beds. We found that the prediction of seagrass coverage as a function of hydrodynamic setting can be improved not only by increasing the spatial extent of sampling at a fixed resolution (1 m), but also by ensuring that data for both dependent (e.g., percent cover) and independent (e.g., wave exposure) variables are averaged over similar scales (spatial extent and resolution). Large-scale features of the landscape, such as patches several meters in width, appeared to be best quantified by sampling over a large spatial extent, as with the video transects. Therefore, contiguous sampling over a broad spatial extent, as opposed to our numerous, somewhat smaller sampling (grid-sampled, 50 × 50 m areas) is the more appropriate strategy for predicting the probability of seagrass bed cover. Conversely, we found that ecological attributes of the seagrass bed (biomass, shoot density, and sediment composition) were best characterized by sampling over a shorter spatial extent (i.e., <50 m), indicating that very localized conditions may have influenced patterns of seagrass community attributes. Generalizing information about seagrass bed ecological attributes obtained from high-resolution samples (fine scale) taken over a broad spatial extent (coarse or landscape scale), as may occur with resource surveys and impact assessments, has the potential to be highly misleading, especially in patchy environments. The influence of sampling scale and survey method on the prediction of coverage and ecological attributes of seagrass beds reveals the need to carefully choose sampling designs to evaluate seagrass distribution and their associated ecological characteristics in the Beaufort, North Carolina (USA) area, and perhaps in other like habitats.