Implications of Climate-Driven Fallowing for Ecological Connectivity of Species At Risk

Climate change and agricultural intensication are modifying the conguration of natural lands within agricultural landscapes, further impacting species’ ability to move freely between remaining natural areas. These working landscapes have inherently high opportunity costs, making the establishment of additional permanent reserves for species movement unlikely. Here we explore the potential for opportunistic and dynamic conservation reserves, in the form of temporary fallowed croplands, to increase connectivity in competing land use regions. We evaluate the potential for fallowed lands to facilitate habitat connectivity for at-risk species in the San Joaquin Valley (SJV), an intensive agricultural landscape in California. We perform landscape connectivity analyses to examine how historic drought-induced fallowing from 2011 to 2017 in the SJV region impacted connectivity within Kern County for the endangered, endemic San Joaquin kit fox (Vulpes macrotis mutica). We found that an increase in temporary fallowing from 2011 to 2015/2017 in Kern County likely increased habitat connectivity fox. This nding was represented by reductions in average Cost-Weighted Distances (CWD), Effective Resistances, and CWD-to-Least Cost Path Ratios between core habitat areas, indicating that cumulative costs incurred kit foxes travelling between primary habitats decreased. examine how changes to fallowed areas in Kern county from 2011 to 2017 inuence connectivity between kit fox habitat areas. Beyond illustrating the connectivity benets of fallowing for kit fox in this region, we discuss potential applications of these methods for future studies and collaborations around strategic fallowing and temporary connectivity corridors. show to 2015/2017 in likely landscape the San Joaquin kit fox. These illustrate the potential for co-benets to be derived amidst signicant land use changes associated with drought conditions and the impending implementation of SGMA. Though the opportunity costs of fallowing to farmers under SGMA will likely be high, those costs have the potential to be partially offset by tapping into the conservation potential of dynamic reserves comprised of permanent and temporary conservation corridors made available by such actions. Given the ubiquity and inuence of productive landscapes on human and natural systems and the increasing preponderance of uncultivated land therein, strategic and coordinated fallowing paired with dynamic and opportunistic conservation may be key to biodiversity conservation in agricultural landscapes.


Introduction
Global environmental change is putting increasing pressure on agricultural production and on the natural resources that support and are affected by production, from groundwater to biodiversity (Norris, 2008;Tilman, 1999). Where agriculture overlaps areas of high biodiversity there are di cult trade-offs between economic bene ts, resources needed to support human population growth, and ecological conservation (Dudley & Alexander, 2017;Fischer et al., 2017;Larsen et al., 2020;Shackelford et al., 2015). Given agriculture covers about 40% of ice-free land, understanding how to balance these inherent trade-offs is critical for the conservation of global biodiversity (Foley et al., 2005;Ramankutty et al., 2008).
Agricultural land conversion and intensi cation do not just affect habitat availability and quality, but also the spatial matrix of different land uses in the landscape (Bennett et al., 2006;Tscharntke et al., 2005). This spatial con guration, in turn, in uences the movement of species and associated gene ow and migration that may be critical to species persistence (Doherty & Driscoll, 2018;Fahrig, 2002 Historically, conservation initiatives have focused on separating nature from areas of human economic activity and resource use through the acquisition and protection of "untouched" wildland or land with restoration potential (Folke, 2006). More recently, the focus has been on the creation of permanent protected areas connected by xed natural corridors through the landscape (Folke, 2006). However, both of these static conservation approaches are not always practical in (semi) urban or high value agricultural zones where setting aside land for conservation has high opportunity costs. Rather, more opportunistic and dynamic reserves may offer a feasible mechanism for provisioning habitat and enhancing connectivity in regions unlikely to be protected in perpetuity due to competing land-uses California, US, presents a valuable opportunity to understand the potential conservation bene ts of dynamic reserves in the form of temporary or persistent fallowed croplands. California is considered a biodiversity hotspot with ~ 1500 plant and ~ 60 vertebrate endemic species across its 13 distinct level III ecoregions, and is also home to some of the most valuable agricultural lands in the country, which underpin the economic and social fabric of many inland regions (Gri th et  Management., 2014). As the majority of groundwater in California is withdrawn for agricultural uses, forecasts suggest that more than 300,000 ha of agricultural land in the San Joaquin Valley might need to be retired to achieve basin sustainability by the SGMA 2040 deadline (Bryant et al., 2020;Hanak et al., 2017Hanak et al., , 2018Hanak et al., , 2019. This potential policy-driven fallowing could provide signi cant environmental and conservation co-bene ts (Queiroz et al., 2014). In particular, fallowed lands, which we de ne here as land that does not produce a crop within a calendar year, could play a signi cant role as dynamic reserves in the San Joaquin Valley. The San Joaquin Valley is the agricultural powerhouse of California that is home to many endemic and endangered species greatly impacted by agricultural and urban development Here we consider the potential of fallowed lands to facilitate habitat connectivity in an intensive agricultural landscape. We focus on the highly productive and heavily ground-water reliant Kern County in the San Joaquin Valley. We select the endangered and endemic San Joaquin kit fox (Vulpes macrotis mutica) as our case study species, since it is considered an umbrella species for regional fauna, is disturbance sensitive, has well-studied historic distribution and ecology, and has designated high suitability core area in Western Kern (Cypher et  fox in this region, we discuss potential applications of these methods for future studies and collaborations around strategic fallowing and temporary connectivity corridors.

Overview
We examined how the fallowing of cropland in intensive agricultural areas as a result of widespread drought impacts species mobility throughout the landscape and what role it might play in increasing overall connectivity between critical species habitat. Our case study centered on the endangered San Joaquin kit fox within Kern County, the highest crop producing county in California (California Department of Food and Agriculture, 2019). Kern experienced increases in fallowed elds beginning in 2012 from widespread drought, which peaked in 2015. To examine species connectivity, we de ned the core habitat areas of the San Joaquin kit fox and created resistance layers for the species across the landscape derived from land cover as well as barriers to movement such as roads and rivers. We then ran spatial connectivity analyses to measure how the spatial distribution of fallowing affected landscape connectivity through its effect on landscape resistance. We statistically compared connectivity between the years of analysis, with 2011 as the base year of our analysis, 2015 as the year of the maximum extent and intensity of the drought in California, and 2017 as the nal year of analysis. All processing and analyses were performed in R Statistical Software v3.5.3 (R Core Team, 2019) and ArcGIS 10.7.1.

Kit Fox Core Areas
We de ned critical habitat of the kit fox using the California Wildlife Habitat Relationships (CWHR) Predicted Habitat Suitability raster (30m) (California Department of Fish and Wildlife California Interagency Wildlife Task Group, 2016), which depicts the mean habitat suitability score (0-1) for the kit fox based on the average value of expert de ned reproduction, cover, and feeding scores for the species in the habitat type. We created polygons of the habitat area by aggregating the raster at 270 meters around the median, and then de ned core habitat areas as > 5 ha polygons-to remove isolated fragments created by vectorizing the habitat data-with the highest possible suitability score (0.92) (SI Methods).

Resistance Layers
To analyze how species mobility through the landscape changed annually from 2011 to 2017, we created resistance layers for each of the years derived from land cover, slope, and barrier (roads and rivers) datasets. Resistance surfaces re ect the "energetic cost, di culty, or mortality risk" . The annual FAM datasets were used to de ne fallowed areas for these analyses, de ned as land that did not produce a crop within a calendar year. Non-irrigated lands exhibiting volunteer crop growth or evidence of a non-irrigated winter cover crop in January-February or November-December were considered fallow. After harmonizing the extent and spatial resolution of the three land use datasets, we combined them into one consolidated land use map for Kern county. As we wanted to evaluate how the change in fallow area, de ned as a binary variable based on the annual FAM data, in uenced the change in species connectivity across the years, we combined the FAM input in each year from 2011 to 2017 with the 2016 DWR dataset and the 2016 NLCD were then aggregated to 270-meter resolution for input into spatial connectivity analysis software.

Spatial Circuit and Connectivity Analyses
To examine how changes in fallowed areas impacted species movement through the landscape and overall landscape connectivity, we used Circuitscape and least-cost path approaches (

Statistical Analyses
In addition to comparing the annual change in species mobility through the landscape visually, we performed statistical comparisons of our spatial connectivity analyses to examine if there were signi cant changes in species connectivity as fallow area increased from 2011 (start of drought) to 2015 (peak) to 2017 (drought subsides). We focused on measures of cost-weighted distance and effective resistance for our least cost paths. Cost-weighted distance (CWD) for least-cost paths is the amount of resistance accumulated by an individual when moving optimally between core areas, and is in units of Given that we had panel data (2011,2015,2017) with a paired design (from and to core pairs are constant), we used a within-estimator model (also known as xed effects in causal inference terminology) predicting CWD, LCP Length, CWD to LCP Length Ratio, and ER as a function of year dummy variables absorbing core-to-core identi ers and clustering the standard errors at the same level. The coe cient in this model is equivalent to a paired-t test, though sample restrictions differ, and is more exible with respect to the treatment of the standard errors (Table S2).

Least Cost Paths
As expected, fallowed land within Kern increased from 2011 (~ 45 kha) to 2015 (~ 93 kha) and 2017 (~ 67 kha) (Fig. 2a-c). The total number of least cost paths links-the pathway between core areas with the lowest cost to an individual-uctuated slightly between the years as pathways computed through core areas were dropped: there were 384 least cost path (LCP) links in 2011, 386 in 2015, and 391 in 2017 (Fig. 2d).
The increase in fallowed parcels from 2011 to 2015 altered current ow in the movement corridors, or the importance of the areas for preserving species probable movement, by opening up some alternative, independent routes to movement (Fig. 3) (Table 1). Conversely, the Least Cost Path Length (km), or the distance an individual travels following the least cost path from one core area to another, showed the opposite trend. Thus, while the paths became slightly longer in 2015 and 2017 relative to 2011 in terms of LCP Length, they were less costly and there was a lower resistance distance to individuals. However, while these summaries help illustrate overall trends in mobility across the landscape, averaging across all pathways does not account for the paired nature of pathways.

Within-estimator Models
For our within-estimator models, which evaluate differences in core-to-core connectivity metrics, we see similar patterns between years ( Table 2). The average CWD to LCP Length Ratio, which measures movement costs accumulated per km traveled on the LCP, was signi cantly (p < 0.05) lower in 2015 relative to 2011 (-14.25 ± 3.26 (+/-SE)). The CWD to LCP ratio was also signi cantly reduced in 2017 relative to 2011, but to a lesser degree (-9.  (Table  S2). Overall, the trends highlight that the decrease in cost weighted and effective distances was greatest for 2015-the height of the drought with the greatest peak in fallowed lands.  Groundwater overdraft in the San Joaquin Valley has resulted in the largest groundwater de cit of any region in California (Hanak et al., 2018). In order to meet SGMA guidelines by 2040 it has been suggested that upwards of 300 kha of agricultural land may need to be fallowed, with a yet to be determined mix of rotational and permanent fallowing (Bryant et al., 2020;Hanak et al., 2018). It is anticipated that many farmers will have options to temporarily fallow some elds, bringing them into production as short-term climate conditions yield su cient surface water supply. Strategic collaboration as to which elds different farmers choose to temporarily fallow could provide important dynamic habitat connectivity throughout the San Joaquin Valley. Furthermore, temporary corridors paired with increased protection and restoration of quality habitat, via permanently fallowed land, could create dynamic reserve networks that help achieve both short-and long-term conservation objectives (Bengtsson et al., 2003;D'Aloia et al., 2019). Incentives provided to farmers who temporarily or permanently fallow particular elds based on objectives of these dynamic reserves could offset opportunity costs they would incur by not selecting elds with the lowest economic value. Funding mechanisms that seek to capitalize on the biodiversity conservation potential of fallow land under SGMA while decreasing costs to farmers have grown in recent years. Broadly speaking, agri-environmental schemes for biodiversity conservation have been used successfully across the US and globally, including direct payments and subsidies to farmers who adopt environmentally bene cial practices and fallowing rotations ( To derive the requisite bene ts from tax payer funded agri-environmental schemes necessitates a thorough understanding of the socio-ecological system. Our study adds to understanding of how temporary reserves may function to improve connectivity in working landscapes, yet has several limitations of note. First, our case study focuses on one species in one agriculturally-dominated county. Second, we lack data on species movement to validate our connectivity models. Further examination of strategic fallowing and its potential co-bene ts at a larger spatial scale and inclusion of groundvalidation of species movement is needed to understand cost-effectiveness of opportunistic fallowing. Strategic corridors of fallowed lands could be functional for kit foxes, who have been historically documented living near and foraging in fallow and less intensive agriculture elds. Indeed, the species recovery plan suggests farmland areas periodically set aside for more than 2-3 years would be useful for maintaining connectivity corridors in our study region (Williams & Fitton, 1997). However, this schema may not be generalizable to other species or regions. More research is needed to understand how connectivity bene ts and co-bene ts scale with fallowing duration.
Strategic, yet temporary conservation actions have the potential to reduce the con ict between biodiversity preservation and agricultural production in agricultural landscapes. Here we show that an increase in temporary fallowing from 2011 to 2015/2017 in western Kern County likely increased landscape connectivity for the San Joaquin kit fox. These results illustrate the potential for co-bene ts to be derived amidst signi cant land use changes associated with drought conditions and the impending implementation of SGMA. Though the opportunity costs of fallowing to farmers under SGMA will likely be high, those costs have the potential to be partially offset by tapping into the conservation potential of dynamic reserves comprised of permanent and temporary conservation corridors made available by such actions. Given the ubiquity and in uence of productive landscapes on human and natural systems and the increasing preponderance of uncultivated land therein, strategic and coordinated fallowing paired with dynamic and opportunistic conservation may be key to biodiversity conservation in agricultural landscapes.

Declarations Funding
No funding was received for conducting this study.

Con icts of interest
We declare no con icts of interest.

Ethics approval
No ethics approvals were required for conducting this study.

Consent to participate
Not applicable.

Consent for publication
Not applicable.
Availability of data and material FAM data will be available upon request from reviewers, and will be posted to Github. All other data is publicly downloadable from the sources cited, and processed in the manner described in the Methods and SI Methods.

Code availability
Code is available upon request from reviewers.