A MIXED METHODS APPROACH TO RESOLVING CONSERVATION AND LIVELIHOOD CONFLICTS IN RURAL COMMUNITIES IN THE AMERICAN WEST

Project Methods
Activity 1will consist of a collaborative mapping exercise implemented by Maher with GYE landowners. Collaborative mapping, often referred to as "Participatory GIS," is characterized by the crowd-sourcing of geographic information, and prioritizes community engagement and social learning, especially in rural areas (Brown and Kyttä 2014). Maher will identify properties representing key demographic and geographic traits through chain-referral sampling (Cohen and Arieli 2011) from NGO, agency, and PAG contacts and invite them to participate in a 3-part exercise. Part 1 will consist of aspatial mapping exercise, where landowners will mark-up satellite imagery of their properties to create detailed maps of fences, roads, water sources, crop and grazing rotations, wildlife presence, and other features involved in the experienced conflict. Part 2 will be atemporal mapping exercise, in which landowners create a timeline of conflict events and the costs of preventative measures. Part 3 will consist of ascenario analysisportion, in which Maher will propose voluntary programs to landowners and request feedback on their advantages and barriers to adoption. Maher will digitize the results from Part 1 to enable descriptive statistical analysis following Wilkinson et al. (2021) and use Parts 1-3 to write up case studies and design flow charts that match properties with conservation tools.Response to reviewer comment on Activity 1:The panel requested additional details about survey/mapping protocol and recruitment process. Landowners will be recruited for the mapping program based on the extent to which they meet two criteria: (1) The parcel they own is ecologically valuable to migratory wildlife (e.g., elk, mule deer), and (2) The parcel is at risk of either development (subdivision) or a high degree of conflict with wildlife (based on crop cover, stocking densities). The research and mentorship team was worked closely with with landowners in the Cody area over the last years and maintains open lines of communications with a large percentage of properties that meet these criteria. Landowners will receive a digital or paper flyer summarizing the program protocol and goals and will be able to opt in to participating. Landowners will be scoredon a spectrum of "amenity" to"working" and from"small" to"large", and of those that opt in, 20 total will be selected such that the following quotas are met: 5 small-parcel "amenity" landowners; 5 large-parcel "amenity:landowners; 5 small-scale "working" landowners; 5 large-scale "working" landowners, thus ensuring properties that span the range of conditions present in the region.Activity 2will build on work by NIFA-supported researchers at UC Berkeley, Arizona State, and the Univ. of Wyoming by integrating experiential knowledge from landowners into econometric models. This analysis will focus on two main areas of inquiry: (i) The extent to which increases in irrigated land over the last decade provide resource subsidies that cause elk to spend more time in private land, and (ii) The role that changes in fine-scaled hunting access play in creating conflict by altering elk use of habitat on private properties. Maher will use data generated by the collaborative mapping process and interviews performed prior to this proposal (i) ground-truth previous work by the broader NIFA team and (ii) perform additional econometric analyses that explicitly incorporate community-generated data into model parameterization and structuring. The econometric approach will utilize a series of well-established quasi-experimental regression techniques to leverage naturally-arising arrangements of landscape features and traits to isolate causal relationships (Greenstone and Gayer 2009). For the first area of inquiry, Maher will use difference in difference regression, which evaluates causal relationships by comparing panel data from different sites (Butsic et al. 2017). For the second, Maher will use spatial regression discontinuity analysis, which exploits differences in treatments across borders (e.g., between private and public land) to specific causal relationships.Response to reviewer comment on Activity 2:The panel requested more details on the proposed data analysis and methodology and suggested considering the use of quantitative ecological data capturing wildlife movement. To address the latter, the Middleton Lab has access to 30+ years of satellite telemetry data from mule deer and elk in the Cody area. PD Maher has already cleaned and begun a time series analysis on the elk data with the ultimate goal of using metrics of wildlife movement(step-length, occupancy in different land ownership types, migration timing, site fidelity) as the response variable for the proposed econometricmodels.In response to a request for more details on the econometric analysis making up Activity 2, Maher has updated her proposed methods to include the following 2part-modelling program:1. Seasonal Integrated Step Selection Model for Elk: ISSFs predict relative probability of use of different habitat typesfor wildlife based on movement patterns observable from satellite telemerty data. PD Maher will evaluate elk preferences for different environmental and socioeconomic covariates (forage quality, predation risk, land use type, presence of agricultural subsidies, hunting pressure) during key conflict periods identified by landowners (instead of summer/winter seasons, which is what ISSFs are usually confined to). Covariates will be constructed using already existing data layers (USDA crop scape, parcel boundary maps, AI-detected center pivots and stackyards, hunting access boundaries from WGFD). The model will be trained using elk collar data owned by the Middleton Lab.2. Two-part Causal Inference Model: Based on the results from the ISSF model described above(which can indicatecorrelations but lacks causal identification power), Maher will select key environmental and socioeconomic variables that predict wildlife overlap with private properties and isolate them using a two-part causal inference methodology. First, the model will use Difference in Difference regression tocompare elk use of areas before and after a key change in the target variable (e.g., before and after a center pivot with alfalfa was put in, before and after hunting access was reduced). Second, the model will use spatial regression discontinuitycompare elk use of areas within and outside of areas with target policies (e.g., across different types of hunting access).Potential Pitfalls: First, landowners might be too busy or unwilling to participate in collaborating mapping, though there is already interest in participation. Participation may be biased towards landowners with favorable attitudes towards conservation. Second, data on hunting access might be difficult to assemble at the scale needed to capture property-level heterogeneity. Last, existing data sets distinguish between public and private land, but can struggle to capture fine-scaled attributes of land cover. Response to reviewer comment on Potential Pitfalls:The reviewers requested a plan for detailing with the potential pitfalls identified above, which are listed here as they correspond to each item: (1) While it is impossible to completely get rid of participation bias without coercing participation, the long-term building of trust and open communication between researchers and landowners in this area has already successfully decreased suspicion of outside researchers and broadened landowner participation in our research program. (2) Hunting and land use data at the scaleneeded for analysis can be collected over a the geographic extent of the project with moderate effort. Maher will initiate a pilot program in which a fine-scaled hunting access and land use data set is assembled for a small area through conversations with game wardens, outfitters, and landowners to assess if it is worth collecting for the entire region.