Open Collections

Description

<b>Abstract</b><br/>

Ecological restoration is a leading approach to mitigating biodiversity decline. While restoration often leads to an immediate increase in abundance or diversity, it is rarely clear whether it supports longer-term biodiversity gains at the landscape scale. To examine the impacts of urban restoration on pollinator biodiversity, we conducted a three-year natural experiment in 18 parks across a large metropolitan area. We applied an occupancy model to our survey data to determine how restoration, woody plant density, and pollinator specialization impacted interannual pollinator metacommunity dynamics. Restoration drove a rapid increase in pollinator species occurrence that was maintained through a positive balance between colonization and persistence, resulting in pollinator species richness gains that are retained. We conclude that urban restoration can effectively conserve pollinator biodiversity by influencing the processes that underlie long-term population stability. Our results highlight the need to study the long-term effects of restoration in different landscape contexts.</p>; <b>Methods</b><br />

The dataset includes: (1) Pollinator detection data. Pollinator detection data was collected by visiting urban park sites and sweep netting for insects for 20 minutes on 6 visit occasions per year for three years (2021, 2022, and 2023). (2) Flower resource data from herbaceous restoration areas or, in control sites, managed herbaceous turfgrass. These data were collected by placing a transect of twenty 1m squared quadrats through the restored or turfgrass area and counting the number of flowers of each plant species within the quadrats. These data were collected on each of the 6 visits per year for three years (2021, 2022, and 2023). (3) Flower resource data from existing woody plants within each park. These data were collected by walking complete transects through the park space and counting (or estimating) every flower on all woody trees and shrubs. These data were collected on each of the 6 visits per year for three years (2021, 2022, and 2023). (4) Land cover data. These data summarize the landscape cover surrounding each site. The original landcover data were sourced from publicly available Metro Vancouver GIS resources, cited in the text and code. In addition, these data include the latitude and longitude of the urban park sites as well as the categorical classification of whether the sites were restored or not. (5) A subset of pollinator interaction data from a publicly available published dataset. We subsetted the original data to the interactions occurring within our local area. The full, publicly available published dataset is cited in the text and code. (6) Appendix S2 - pollinator species information. These data summarize the interaction specialization estimated by our interaction network as well as species-specific phenology as estimated by our model. (7) Pollination data. These data describe the seed set of paired pollen-supplemented and naturally-pollinated <em>Clarkia amoena</em> plants. These data were obtained by placing plants at 11 of the 18 field sites during flowering. The ripened fruiting capsules were then harvested and dissected, and seeds were counted. </p>

The dataset here also includes novel code written to analyze these data. In summary, we transformed the pollinator detection data into an array of binary species/site/year/visit detections (0=not detected or 1= detected). We wrote a custom dynamic occupancy model in the programming language Stan, which estimated the interannual transitions in species occurrences while accounting for imperfect detection (false negatives) of species occurrence. Separately, we wrote a custom logistic regression model, also in Stan, to estimate the effect of park site restoration on the probability that a flower (of the annual plant species <em>Clarkia amoena</em>) is pollen limited. The code can also be accessed and run from a github repository: <a href="https://github.com/jensculrich/urban_pollinator_occupancy_model">https://github.com/jensculrich/urban_pollinator_occupancy_model</a>. </p>