Adult butterflies were sampled using visual observation transect protocols described in Van Swaay and others (2015) and identified using Belth (2013). While avoiding mowed hiking trails, a total of 61 transects were sampled over all the habitats (table 1; Albers and others, 2021), with old fields having only one sampling event owing to logistical time constraints. Each transect either started or ended at a randomly generated point location on the site, was a minimum of 100 meters (m) long, and was sampled between 0800 and 1700 hours (h) on sunny, non-windy days between June 21 and August 12, 2016. One person walked the transect with a mean speed of 0.423±0.106 meter per second (mean ± standard deviation, number of observations [n] = 59) and documented all observed butterflies within 5 m of the transect and as much as 10 m in front of the observer. Mean transect length was 247.49±135.79 m, and the mean transect area was 2,574.92±1357.90 square meters (m²) (n=61). A geographic positioning system (GPS; Trimble Juno 3B, Survey Supplies Ltd, Liverpool, United Kingdom) was used to record the transect start location and time, transect route, and transect end location and times. While sampling the transect, the GPS was also used to record the time and location of each butterfly encounter, visually identified taxa, and number of individuals in each taxon.

Live butterflies were identified to species level using visual observations of the individuals, except for groups of species similar in color, shape, and flight behavior, which prevented positive species-level identification. We grouped the following species: silvery checkerspot (Chlosyne nycteis) and pearl crescent (Phyciodes tharos) into the Chlosyne/Phyciodes spp. group; all small sulphurs (Coliadinae) into the Colias spp. group; all skippers (Hesperiidae) into the Hesperiidae group; all Satyrinae subfamily and two other species in the Nymphalidae family, hackberry emperor (Asterocampa celtis), and northern pearly-eye (Lethe anthedon anthedon) into the Satyrinae/Apaturinae spp. group; all question mark (Polygonia interrogationis), eastern comma (P. comma), and gray comma (P. progne) into the Polygonia spp. group; all Theclinae subfamily (hairstreaks) and Polyommatinae subfamily (blues) into the Theclinae/Polyommatinae spp. group; and all painted lady (Vanessa cardui) and American lady (V. virginiensis) into the Vanessa spp. group.

Passive and active sampling techniques were used for invertebrate collections on study sites (table 2). Passive sampling methods are designed to catch a representative sample of all invertebrate functional groups in the sampled habitats (Leather, 2005), whereas active methods yield data only on invertebrates that are present within the sample area at the time of sampling.

For passive invertebrate sampling, we deployed 18 PEMITs (6 at Bluffton Native Habitat Waterway, 6 at Deetz Nature Preserve, 4 at Bell, and 2 at Holden) that consisted of a sea, land, and air Malaise trap (SLAM; BioQuip Products, Inc., Compton, California; 1.1 × 1.1 × 1.1 m), with additional pitfall traps as described below. The SLAM is fabricated with a white roof (108 × 32 mesh per square inch) and black Polyester no-see-um fabric (96 × 26 mesh per square inch; fig. 2). Each of the two perpendicular interception areas are 165 × 110 centimeters (cm), and a 500-milliliter (ml) collection bottle is attached at the peak of the trap. For this study, the large insect excluder was not used. The collection bottle contained 250 ml of soapy water (unscented, undyed dish soap) at a ratio sufficient to remove surface tension and increase invertebrate collection success. Plastic collection trays containing additional soapy water were positioned under the two crossed walls of the trap to collect falling invertebrates that encountered the fabric and as pit fall traps for ground crawling invertebrates. We dug two perpendicular trenches and inserted the trays into the ground so that the lip of the tray was even with ground surface and had good soil-tray contact. Individual plastic trays were 29.2 × 19.1 cm, contained soapy water that was at maximum 5.7 cm deep, and had a 250-micrometer (μm) mesh overflow screen in case of precipitation. The SLAM was aligned over the 14 trays that covered an area of 0.781 m² below the trap (fig. 2). After sampling, all trap water was filtered with a 250-μm mesh to collect invertebrates. Invertebrates from the top collector and bottom trays were combined into one sample for the sampling day. All specimens were preserved in 70-percent ethanol.

Traps were randomly placed within each of only two habitats (mature and restored) on studied sites owing to logistical time constraints (table 2). The traps were deployed between 0800 and 1200 h, and invertebrates were collected approximately every 24 h for 2 days. Traps were deployed three times at each site from June 21 to August 12, 2016, at approximately 2-week intervals. Mean trap first day deployment was 23.1±0.50 h (n=18), and mean trap second day deployment was 23.5±0.62 h (n=18).

Active sampling for terrestrial invertebrates employed seven sweep net transects in only the restored habitat owing to logistical time constraints (table 2). We used a net that was 0.38 m wide, 0.41 m high, and 0.64 m deep, with a 0.91-m handle and 24 count mesh (MidLakes Corporation, Knoxville, Tennessee). Sweep transects were 1.82 m wide, at approximately 1.6 sweeps per second or 80 sweeps per 50-m transect. Each sweep transect started from a random point and proceeded in a random direction. Direction was randomized by the sampler looking at the second hand on a clock. The sampler walked about 50 m while sweeping all vegetation and collecting all invertebrates within approximately 0.91 m of the transect between 1200 and 1700 h from July 17 to August 11, 2016. Mean transect length was 44.1±8.86 m, and the mean transect area was 81.8±16.20 m² (n=7). Invertebrates were euthanized with gaseous acetone while still in the net and then were removed and preserved in 70-percent ethanol.

Biaggini and others (2007) suggest that identification of individual invertebrates to order is adequate to determine differences in invertebrate communities owing to land use changes. Consequently, all terrestrial invertebrate specimens were identified to order or the lowest taxonomic level possible following keys in Borror and White (1998), Bouchet and others (2005), Chu (1949), Cupul-Magaña and Flores-Guerrero (2016), Ramel (2017), Sierwald (2007), Stehr (1987, 1991), Treadwell (1996), and Triplehorn and others (2005). Identifications were made in the invertebrate taxonomy laboratory at the U.S. Geological Survey Columbia Environmental Research Center in Columbia, Missouri.