Reclaiming urban vacant land to manage stormwater and support insect habitat
Data files
Jan 08, 2024 version files 1.07 MB
-
compiledRG.csv
50.29 KB
-
diversityRG.csv
29.73 KB
-
nativeBeeRG.csv
27.36 KB
-
nativeBeeRG.xlsx
772.96 KB
-
predatorRG.xlsx
177.20 KB
-
README.md
7.84 KB
Abstract
Urban green spaces can provide important wildlife habitat and ecosystem services. In legacy cities, built structures are demolished as populations dwindle, resulting in vacant land. Vacant land constitutes an opportunity to establish green infrastructure that provides multiple ecosystem services. Our objective was to determine whether establishing green infrastructure on vacant land to manage stormwater could provide insect habitat in the legacy city of Cleveland, Ohio, U.S.A. Two green infrastructure treatments were implemented on vacant land in the historic Slavic Village neighborhood in 2014 and 2015: rain gardens (lower cost) and bioswales (higher cost). We hypothesized that rain gardens and bioswales would support more abundant, species rich insect communities compared to unaltered vacant lots. Wild bees (Hymenoptera: Aculeata) and lady beetles (Coleoptera: Coccinellidae), two insect groups of conservation concern, were sampled during the summer (June–August) from 2014 to 2016 using pan traps and yellow sticky card traps. Local vegetation and temporal variables were measured. Generalized linear mixed effects models evaluated whether insect biodiversity varied with treatment, habitat variables, site, and time. We collected 3,004 bees from pan traps and 5,438 lady beetles from yellow sticky card traps during this study. Bee biodiversity was similar among treatments. In 2014, alien Coccinellidae abundance was higher in vacant lots compared to rain gardens. In 2015 and 2016, alien Coccinellidae were marginally more abundant in rain gardens compared to vacant lots and bioswales, while native Coccinellidae abundance was significantly higher in vacant lots. In the short term, establishing green infrastructure on vacant land can improve stormwater management without compromising the quality of vacant land as insect habitat.
https://doi.org/10.5061/dryad.dv41ns255
The dataset contents include count data, local vegetation variables, temporal variables, and site information. These data were used to create generalized linear models in R, using a backwards stepwise selection approach.
Description of the data and file structure
The Excel spreadsheets are databases containing information about insect taxa we collected as well as the local vegetation variables, while the CSV files include information about insect abundance and diversity. The CSV files were brought into R for analysis. The data is structured according to time of sampling, site, and quadrat. You will find in some sites, we only sampled insects and vegetation from 4 quadrats, whereas in others, we sampled from up to 6 quadrats. This was to account for variation in size, as well as variation in cover. For example, in the rain gardens, we placed additional sampling equipment in the mulched depression. Note that at the beginning of our analysis, we considered cover variables which were collected but ultimately decided to exclude them, as they did not fit our models well and were not strong predictors of insect abundance. Columns where variables are listed as NA indicate that either the information was not collected (e.g., cover or vegetation variables) or that sampling equipment was damaged and sampling effort was compromised (e.g., bee abundance, richness, diversity).
Datasets include
compiledRG.csv | Information about sampling location and time (columns A-F), local vegetation (G-J), bee abundance, richness, and number of traps (K-O), percent cover in each quadrat (P-T, totaling to 100 per observation), number of sticky cards (U), and number of bees and lady beetles collected per sticky card (V-AE). This dataset was used to run GLMMs of bee abundance, bee richness, and lady beetle abundance. | ‘_e’ indicates alien abundance where as ‘_n’ indicates native abundance. ‘_y’ indicates a response variable was measured with sticky cards, while ‘_b’ indicates a response variable was collected using bee bowls. ‘lasio’ refers to the bee genus Lasioglossum, and ‘halictidae’ refers to the bee family Halictidae. P14 (column X) references the abundance Propylea quatuordecimpunctata. ‘cocc’ (AB-AD) indicates the subfamily Coccinellinae, while ‘scym’ indicates the subfamily Scymninae. All lady beetles were collected using sticky cards. |
---|---|---|
diversityRG.csv | Includes information about sampling location and time (columns A-G), number of bee bowls (H), number of bees collected (I), Shannon-Weiner diversity (J), and local vegetation variables (K-M). This dataset was used to run GLMMs of bee diversity. | |
nativeBeeRG.csv | Includes information about sampling location and time (columns A-E), native bee abundance (F), alien bee abundance (G), number of bee bowls (H), and local vegetation variables (I-K). This dataset was used to run GLMMs of native bee abundance and alien bee abundance. | |
nativeBeeRG.xslx | This file includes a database of bee species ID (compiledBeeID) which was used to derive metrics of bee abundance (abundance, abundanceFinal). | Under Species ID (B) in the sheet titled “abundance”, NA indicates that no bees were collected during that observation period. Hence, there is no species to identify. |
predatorRG.xslx | This file includes more detailed response variables of insect predators that were not included in the final publication. This includes information about sampling location and time (columbs A-F), local vegetation variables (G-H), abundance of lacewings (I), assassin bugs (J), rove beetles (K), pirate bugs (L), hoverflies (M), longlegged flies (N), and lady beetles (O). Lady beetles are further broken down into different lady beetle species (P-AF). | Propylea quatuordecimpunctata (P), Cycloneda munda (Q), Coleomegilla maculata (R), Scymnus spp. (S), Brachiacantha spp. (T), small black unidentified lady beetles (U), Hyperapsis undulata (V), Harmonia axyridis (W), Hippodamia parenthesis (X), Hippodamia variegata (Y), unknown lady beetle (Z), Psyllobora vigintimaculata (AA), Hippodamia convergens (AB), Epilachna varivestis (AC), Hyperapsis signata (AD), Coccinella septempunctata (AE), and unknown lady beetle (AF). |
Sharing/Access information
N/A
Code/Software
We used R Studio to upload and run these files. The R Script is included here. The code is greatly informed by Zuur et al. 2009.
Pan traps were set in 2014 (July 10, July 31), 2015 (June 9, July 6, Aug 5), and 2016 (June 9, July 26, Aug 8) to sample bees. Pan traps consisted of 96 ml plastic soufflé bowls (Solo®) painted fluorescent blue, yellow, or left white. One pan trap of each color was filled halfway with soapy water (1% blue Dawn dish soap/water solution) and deployed at ground level for 24 h in each quadrat used for vegetation sampling. At the time of collection, the number of viable pan traps per site was recorded to account for trap loss during each sampling. All bees from pan traps were washed, blown dry, pinned, and preserved with museum level archival labels. Bees collected using pan traps were identified using Discoverlife.org (Ascher and Pickering 2017), revisions by Jason Gibbs for Lasioglossum (Gibbs 2011; Gibbs et al. 2013), and Bumble Bees of North America (Williams et al. 2014). Voucher specimen identifications were verified by Sam Droege at the USGS Bee Inventory and Monitoring Lab in Laurel, Maryland and pinned bee specimens were deposited in the Museum of Biological Diversity in Columbus, Ohio. From the verified bee species data, total bee abundance, Halictidae abundance, Lasioglossum abundance, and native bee abundance were counted. Bee species richness and Shannon diversity were also calculated. Bycatch, including hoverflies (Syrphidae), longlegged flies (Dolichopodidae), minute pirate bugs (Anthocoridae: Orius), and rove beetles (Staphylinidae), were counted but not analyzed for the purposes of this study. To sample lady beetles (Coleoptera: Coccinellidae), Trece Pherocon ® Unbaited AM Yellow Sticky Traps (hereafter sticky traps) (23 cm W × 28 cm L with a 18 cm W × 23 cm L grid) were deployed for seven days at a time in 2014 (June 23, July 21), 2015 (June 10, July 6, Aug 3), and 2016 (June 9, July 14, Aug 2). One sticky trap was set per quadrat and secured to a step-in fence post above maximum vegetation height. At the time of collection, the number of viable sticky traps per site was recorded. Lady beetles were counted and identified to species using a stereomicroscope (Spring 2017). Bees were also counted and identified on these sticky traps, but the majority were only identifiable to genus due to the adhesive obscuring diagnostic features in the abundant Lasioglossum and Hylaeus specimens. This adhesive also made it challenging to remove bees collected from sticky traps without damaging them, and as a result, these specimens were not archived. From the lady beetle species data, total lady beetle abundance, lady beetle abundance by subfamily, lady beetle abundance by species, alien lady beetle abundance, and native lady beetle abundance were calculated. Some sampling equipment (pan traps, sticky traps, stakes) was lost due to theft or destruction by mowing equipment. Sticky traps on fallen stakes were not included in sticky trap analyses. Some traps had obvious contamination from animal fur or feathers but were still included in the analyses since a large portion of the sticky surface was still viable. Pan traps were also occasionally knocked over or washed out due to high rainfall during the sampling period, and these traps were not included in the analyses.