Agricultural margins could enhance landscape connectivity for pollinating insects across the Central Valley of California, U.S.A

Abstract

One of the defining features of the Anthropocene is eroding ecosystem services, decreases in biodiversity, and overall reductions in the abundance of once-common organisms, including many insects that play innumerable roles in natural communities and agricultural systems that support human society. It is now clear that the preservation of insects cannot rely solely on the legal protection of natural areas far removed from the densest areas of human habitation. Instead, a critical challenge moving forward is to intelligently manage areas that include intensively farmed landscapes, such as the Central Valley of California. Here we attempt to meet this challenge with a tool for modeling landscape connectivity for insects (with pollinators in particular in mind) that builds on available information including lethality of pesticides and expert opinion on insect movement. Despite the massive fragmentation of the Central Valley, we find that connectivity is possible, especially utilizing the restoration or improvement of agricultural margins, which (in their summed area) exceed natural areas. Our modeling approach is flexible and can be used to address a wide range of questions regarding both changes in land cover as well as changes in pesticide application rates. Finally, we highlight key steps that could be taken moving forward and the great many knowledge gaps that could be addressed in the field to improve future iterations of our modeling approach.

Fig 1. Workflow used to generate the least-cost paths.

The five sources of input data include pesticide application reporting (State of California Department of Pesticide Regulation Pesticide Use Reporting database), toxicology databases (US Environmental Protection Agency ECOTOX database and the US National Institute of Health PubChem database), land cover maps (LandIQ crop map, NOAA Coastal Change Analysis Program, and US Geological Survey National Land Cover Map), estimates of landscape resistance for urban areas derived from Jha and Kremen (2013), and source and destination points created along the margins of the study area. Relative resistance of each crop/land cover type is shown in Table 1. Tabular data are depicted with the table icon whereas mapped data are shown using map icons. Italicized text indicates steps taken to process data. Non-italicized text is used to show input data.

Fig 2. Major land cover types in the Central Valley study area.

All 26 crop types have been combined for the purpose of map legibility. The agricultural margins appear as tiny dots scattered throughout the Central Valley (left panel), but appear much more clearly as linear features in close-up map (right panel). There is a black line around the total study area (Central Valley and adjoining lands; see text for details) as well as the predominantly agricultural inner study area, which is also shown with the gray line in the right panel. The darker green areas are conservation lands that are treated as natural in our analyses.

Fig 3. Resistance to movement maps used to parameterize the Central Valley pollinator connectivity models.

Maps depict the resistance surfaces used for the a) restored margins scenario, b) current margins scenario, and c) no margins scenario. Colors range from low resistance (blue) to high resistance (red). A resistance value of 1 is equivalent to isolation-by-distance, while a value of 100 indicates that a grid cell is 100 times more difficult to traverse than the lowest-resistance cell.

Fig 4. Least-cost paths across the Central Valley and the close-up study area.

Line colors depict all possible combinations of resistance levels (low, moderate, high, low+moderate, etc.) for the a) restored margins scenario, b) current margins scenario, and c) no margins scenario across the Central Valley. The same line colors are used to depict the same combinations of resistance for the d) restored margins scenario, e) current margins scenario, and f) no margins scenario for the close-up study area. An interactive version of all scenarios can be found at: https://arcg.is/1eGvDv.

Fig 5. Summary statistics describing least-cost paths across the Central Valley and the close-up study area.

a) Total path length of least-cost paths connecting the western and eastern sides of the Central Valley including the outer area, b) unique path length in the inner area for the Central Valley, c) average distance to the nearest path in the inner area for the Central Valley, d) average number of paths in the inner area for the Central Valley, e) total path length of least-cost paths connecting the western and eastern sides of the close-up study area including the outer area, f) unique path length in the inner area for the close up study area, g) average distance to the nearest path in the inner area for the close up study area, and h) average number of paths in the inner area for the close-up study area. Note the differences in the scale of the y-axis among graphs.

Fig 6. Land cover classes traversed by least-cost paths for each of the three agricultural margin scenarios crossed with the three resistance scenarios.

a) Land cover classes for the Central Valley study area and b) the close-up study area. All of the crops are condensed into a single category, and the three urban categories have been condensed into a single category.

Fig 7. Close-up image showing cumulative costs incurred as a function of distance moving outward from three source points (yellow dots) for A) the current conditions with moderate resistance and B) the scenario in which all agricultural margins are converted to natural habitat with moderate resistance for the other land cover classes.

Red colors indicate a low cost of movement (i.e. high probability of connectivity with the source points) and blue colors indicate a movement cost approaching 15,000 cost-distance units (number of cells traversed x cost of moving over the cell).