Fig 1.
Map showing the location of the study area, Pamlico Sound, within the Albemarle-Pamlico Estuarine System (APES); stars denote oceanic inlets (OI = Oregon Inlet, HA = Hatteras Inlet, OC = Ocracoke Inlet, DI = Drum Inlet, BI = Bardens Inlet).
Table 1.
Threshold layers and associated weights utilized to compute suitability in the three HSI scenarios.
Weights were applied to each layer, and the assigned weight corresponds to the relative importance of each threshold layer to siting oyster restoration efforts in a given HSI scenario. The assigned weights of all threshold layers sum to 100%.
Fig 2.
Habitat suitability based on: (A-C) aggregated threshold layers, (D) aggregated exclusion layers, (E-G) aggregated exclusion and threshold layers combined for the three model scenarios: ‘Water Filtration,’ ‘Water Filtration & Metapopulation Persistence,’ and ‘Metapopulation Persistence’.
Suitability in panels A-C and E-G is continuous, while suitability in panel D is binary. Suitability increases from low (red) to high (green) HSI. Panels C, D, and G adapted from Puckett et al. [18].
Fig 3.
Relationships between: (A) mean chlorophyll a concentration and coefficient of variation, and (B) mean flow velocity (cm/s) and percent frequency flow velocity exceeds 15 cm/s.
A description of the analysis methods used to identify these relationships can be found in ‘Methods, Water Filtration Ecosystem Services Layer Development’.
Fig 4.
Relationship between actual values of: (A) mean chlorophyll a concentration, (B) coefficient of variation of chlorophyll a concentration, (C) mean water flow velocity (cm/s), and (D) minimum observed dissolved oxygen concentrations, and their associated suitability values.
A description of the analysis methods used to develop these functions can be found in ‘Methods, Water Filtration Ecosystem Services Layer Development’.
Fig 5.
Suitability layer for: (A) mean chlorophyll a concentration, (B) coefficient of variation of chlorophyll a concentration, (C) mean water flow velocity, and (D) minimum observed dissolved oxygen concentration.
Suitability for oyster restoration increases from low (red) to high (green) HSI.
Fig 6.
Optimal locations for restoration (i.e., top 1% highest HSI scores) identified in each HSI scenario as derived from the final HSI (i.e., aggregated threshold combined with aggregated exclusion layers).
Locations that were identified within the top 1% for multiple HSI scenarios are indicated in blue (2 models) and gold (3 models). For example, ‘Top 1% for 2 Models’ may include areas identified within the top 1% for the ‘Water Filtration’ and the ‘Water Filtration & Metapopulation Persistence’ HSI scenarios, whereas ‘Top 1% for 3 Models’ includes areas identified within the top 1% for all three HSI scenarios.
Fig 7.
Results of model sensitivity analysis conducted for each of the three HSI scenarios where we: 1) removed the four threshold layers with the highest weight individually, 2) re-weighted the remaining layers proportionally based on the weight of the removed layer, 3) re-ran the model, and 4) calculated on a cell-by-cell basis the percent change in model output with removal of each layer.
Error bars represent standard error of the mean (n = 5,987) of the percent change in model output with removal of a layer. Weights associated with the threshold layers in each HSI scenario can be found in Table 1.