We measured recruitment of fishes and crabs to artificial and natural seagrass beds in Back Sound, North Carolina from June to August 2018.  We sampled 26 artificial seagrass landscapes defined by 18m x 13m landscape extents (234 m2). Within each landscape, we varied habitat amount and fragmentation independently in a crossed design. We created a custom algorithm to generate random landscapes using the randomHabitat function in the secr package in R (Efford 2018, R Core Team 2018). Landscapes were generated along two orthogonal axes of habitat cover (10-60%) and fragmentation using a random modified clusters method (percolation probability = 0.10-0.59 which determines patch number; higher percolation coefficients correspond to less patchiness)(Saura and Martinez-Millan 2000). Landscapes were constrained to fall within 2% of the area input parameter, while holding patch number similar within fragmentation levels (1, 2-3, 4-5, 6-7, or 8-10 patches for each level). Using this approach, we constructed landscapes in which seagrass area and number of patches were uncorrelated (R = -0.02), allowing us to independently assess the effects of seagrass area and habitat configuration on fish communities across experimental study sites. Artificial seagrass units (ASUs) were used to create artificial seagrass landscapes.  ASUs were made by tying green plastic ribbon to polyethylene mesh cut into 1.2m x 0.86m rectangles, with ~450, 15 cm tall shoots (with 2 blades each) attached each unit. ASUs were deployed by affixing ASUs in designated configurations to bare sediment using metal lawn staples. These artificial landscapes were deployed between May 23 to May 31, 2018.  We compared the artificial seagrass landscapes to natural landscapes in the same area.  We sampled 18 haphazardly selected natural seagrass patches that were at least 100 m away from the artificial seagrass landscapes.

We used Standardized Monitoring Units for Reef Fishes (SMURFs, Ammann 2004) to measure recruitment of fishes and invertebrates into each landscape.  SMURFs used in this study were 60 x 15 x 20 cm in size with an outer shell made from 5 cm x 7.6 cm polyethylene mesh.  Each SMURF was filled with crumpled 2.5 x 2.5 cm polyethylene mesh and plastic mesh produce bags (https://www.amazon.com/dp/B01LWL1YFT/ref=cm_sw_r_cp_api_i_J9TzCbZZ5GKR6). For deployment, SMURFs were deployed attached to the bottom at one end, but floated vertically in the water column.  SMURFs were deployed for ~24hrs, including overnight, and collected before noon on the following day (pick-up times are included in the data).  Upon retrieval, the SMURF was enclosed in a custom made BINKE net (Anderson and Carr 1998) made out of 1.5 mm square nylon netting.  The SMURF was brought to the boat within the enclosed BINKE net and placed into a large plastic tub.  It was shaken vigorously while being rinsed with buckets of seawater and then the rinse water in the tub was visually inspected for organisms which we collected, placed in a resealable plastic bag, and quickly froze in an ice and seawater slurry. This process was repeated until three successive rinses yielded no new organisms.  The water in the tub was passed through a 1 mm mesh sieve to ensure no organisms remained in the tub. All organisms were returned to the lab for processing. Each landscape was sampled on 9 different dates between June and August, representing the major recruitment period for most fishes in this system.