Identification of high-risk contact areas between feral pigs and outdoor-raised pig operations in California: Implications for disease transmission in the wildlife-livestock interface.

The US is currently experiencing a return to raising domestic pigs outdoors, due to consumer demand for sustainably-raised animal products. A challenge in raising pigs outdoors is the possibility of these animals interacting with feral pigs and an associated risk of pathogen transmission. California has one of the largest and widest geographic distributions of feral pigs. Locations at greatest risk for increased contact between both swine populations are those regions that contain feral pig suitable habitat located near outdoor-raised domestic pigs. The main aim of this study entailed identifying potential high-risk areas of disease transmission between these two swine populations. Aims were achieved by predicting suitable feral pig habitat using Maximum Entropy (MaxEnt); mapping the spatial distribution of outdoor-raised pig operations (OPO); and identifying high-risk regions where there is overlap between feral pig suitable habitat and OPO. A MaxEnt prediction map with estimates of the relative probability of suitable feral pig habitat was built, using hunting tags as presence-only points. Predictor layers were included in variable selection steps for model building. Five variables were identified as important in predicting suitable feral pig habitat in the final model, including the annual maximum green vegetation fraction, elevation, the minimum temperature of the coldest month, precipitation of the wettest month and the coefficient of variation for seasonal precipitation. For the risk map, the final MaxEnt model was overlapped with the location of OPOs to categorize areas at greatest risk for contact between feral swine and domestic pigs raised outdoors and subsequent potential disease transmission. Since raising pigs outdoors is a remerging trend, feral pig numbers are increasing nationwide, and both groups are reservoirs for various pathogens, the contact between these two swine populations has important implications for disease transmission in the wildlife-livestock interface.

Introduction

Although a majority of commercial swine production in the United States (US) occurs indoors with high levels of biosecurity, the US is currently experiencing a return to raising domestic pigs outdoors [1, 2]. Before the 1950s, most swine operations in the US were small-scale family farms and either a hybrid of indoor/outdoor or solely outdoor-based [1, 3]. Beginning in the 1960s, commercial swine production began transitioning to indoor systems, based on goals to increase efficiency and reduce swine disease transmission (e.g., brucellosis) as well as a public health mandate to decrease human trichinosis cases [4-7]. However, consumer demand for sustainable or pasture-raised animal products within the past few decades has revived traditional methods of raising swine outdoors or on pasture (i.e., outdoor-raised pigs, pasture-based) [1, 2, 8, 9]. While primarily considered a niche production method in the US, outdoor-raised pig operations (OPO) (e.g., commercial pork producers, backyard operations) are broadly distributed throughout California.

A challenge in raising pigs outdoors is the possibility of these animals interacting with wildlife disease reservoirs, such as feral pigs, and an associated risk of zoonotic and/or swine pathogen transmission [10-16]. Both domestic and feral pigs share the same genus and species (Sus scrofa) and can be reservoirs for zoonotic pathogens (e.g., swine influenza virus, Shiga toxin-producing Escherichia coli) [17-23]. Also, swine diseases eradicated in conventional indoor-raised herds (e.g., pseudorabies, brucellosis) have been documented in feral swine in California and contact between feral pigs and outdoor-raised swine herds is a risk factor for the reintroduction of these diseases to domestic herds in the US [6, 7, 15, 19, 24–36]. For example, a 2016 human case of brucellosis in New York state was traced to a feral pig intrusion event on a pasture-raised pig farm. Brucella suis was then transmitted to domestic pigs raised outdoors in 13 other states through animal sales [25, 27]. Feral pigs could also play a significant role in the transmission and maintenance of transboundary animal diseases (TAD) introduced to North America [11, 16, 28, 29]. For instance, African swine fever (ASF) is actively spreading in eastern Europe, with wild boars transmitting this devastating disease between and within countries [30]. Similarly, wild boars abet the transmission of ASF in South Korea, spreading the virus to outdoor-raised swine [31, 32]. And most recently, ASF was identified in domestic swine in the Dominican Republic, which is the closest to the US that ASF has spread in this century [33].

During the past few decades, feral pig populations have greatly increased in the US from 17 to 41 states [34-36]. California has one of the largest and widest geographic distributions of feral pigs and this invasive species has the broadest habitat range of any large mammal except humans, which is in part due to their ability to adapt to a diverse range of ecological habitats and their opportunistic omnivore diet [36-41]. Feral pig population distribution and abundance is dynamic yet has not been documented at fine spatial units less than 1km [42]. Additionally, previous presence maps reported feral pigs for an entire county, even if there had only been a single occurrence recorded countywide [38, 43, 44].

Hypothetically, an area is at higher risk of disease transmission if it is more likely to experience interactions between feral pig and domestic pigs raised outdoors, as these outdoor-based pigs can serve as a conduit for disease spread from wildlife to humans. Locations at greatest risk for increased contact between both swine populations are those regions that contain feral pig suitable habitat located near OPO, especially those OPO with relatively low levels of biosecurity [24, 26, 36, 45]. Contact between feral pigs and outdoor-raised pigs in California has been documented, as feral pigs are attracted to agricultural regions for food, water, and mates [10, 12, 29, 46–49]. There is enormous value in identifying agricultural regions with a higher probability of feral pig contact, because these areas could benefit from targeted cost-effective disease surveillance and risk-mitigation strategies to prevent disease transmission.

Predicting suitable habitat for feral pigs (i.e., likelihood of feral pig presence) in combination with spatially characterizing the distribution of OPO can provide an important tool to ascertain possible high-risk areas of contact at the feral-domestic pig interface and identify future disease spillover areas [48, 50, 51]. Species distribution modeling (SDM) methods have been widely used in ecological studies and are becoming popular for use in epidemiological investigations of disease transmission between wildlife and livestock [49, 52–54]. Maximum Entropy (MaxEnt), which is one type of SDM, allows usage of presence-only data for the species of interest (i.e., feral pigs) [55]. In combination with biologically-appropriate covariate factors, MaxEnt is able to spatially predict the probability of suitable habitat for a species for a chosen spatial unit (i.e., pixel) [56].

These two parallel trends of expanding feral pig populations and a resurgence of raising domestic swine outside has important implications for disease transmission, which could negatively impact both public health and California’s agricultural industry. To the best of our knowledge, there are no maps characterizing where suitable feral pig habitat overlaps with domestic pigs raised outdoors at the farm-level in California. The overall objective of this study entailed spatially identifying potential high-risk areas of disease transmission between these two swine populations. This objective was achieved by a three-step process: 1) predicting suitable feral pig habitat in California using MaxEnt; 2) mapping the spatial distribution of OPO in California; and 3) identifying high-risk regions where there is spatial overlap between feral pig suitable habitat and OPO, as potential disease transmission areas.

Materials and methods

This study used secondary data, does not included field work, nor animal primary data. The survey instrument and protocols were reviewed and determined to be exempted by the Institutional Review Board (IRB) of the University of California-Davis (No. 1180798–1).

Maximum Entropy model

MaxEnt is an established SDM method that produces an output prediction map containing estimates of the relative probability of suitable habitat areas for the species of interest (i.e., feral pigs) within each pixel, using presence-only points and predictors (i.e., covariate spatial layers) [32, 47, 51, 55, 57–61]. For feral pig presence data, we obtained feral pig hunting tags from 2012–19 that were cleaned and recorded with GPS coordinates by the California Department of Fish and Wildlife (CDFW). Hunters in California are voluntarily asked to report feral pig harvest locations by submitting hunting tags to CDFW. Using hunting records for presence-points of feral pigs or wild boars has been used in previous studies [49, 62]. CDFW 2012–19 feral pig hunting tags totaled 5,148 after removing duplicates. Due to the large amount of data points, hunting tags were also manually filtered (i.e., subsampled) by year as a way to decrease the abundance of points before running models to reduce sampling bias and increase model stability, as suggested by previous analyses of MaxEnt (Fig 1) [49, 61, 63–65].

Fig 1

California feral pig hunting tags from 2017.

Each point represents a GPS location of a feral pig hunting tag, after removing duplicate locations.

Predictor spatial layers available online, including biotic (e.g., land cover, vegetation) and abiotic (e.g., temperature, precipitation, elevation), were included in variable selection steps, and were chosen based on known feral pig behaviors, habitat, and food preferences. Table 1 displays a subset of the more than 30 predictors initially standardized and analyzed for inclusion in model building steps for MaxEnt (Table 1) [37, 47, 66–70]. For example, AVGMODIS was the annual maximum green vegetation fraction (MGVF) combined with 12 years of normalized difference vegetation index data (NDVI) and relates to food and shrub cover for feral pigs [71-73]. NDVI measures vegetation gathered by the Moderate Resolution Imaging Spectroradiometer (MODIS) as part of NASA’s satellite systems. Other variables included elevation, as feral pigs may prefer specific altitudes, and nineteen environmental variables from the WorldClim set of 30 year trend climatic factors [74]. Examples of environmental variables used from the WorldClim site included the minimum temperature of the coldest month (BIO6), precipitation of the wettest month (BIO13) and the coefficient of variation for seasonal precipitation (BIO15) [75]. Layers were reprojected to the Albers Equal-Area coordinate reference system for California (“California Albers” (meters)) and masked for the entire state of California, using QGIS 3.6 [76]. Rasters were all converted to the same resolution of 270m x 270m, which used a reasonable amount of computer computation time, while maintaining fine-scale for suitable habitat modeling at the farm-level. Predictors were assessed for correlation using Spearman’s rank and a cut-off threshold of 0.80, a threshold used in previous studies [54]. Two correlated variables were not included at the same time, during variable selection steps.

Table 1

Predictor layers assessed during variable selection for Maximum Entropy model building.

Name	Short Description	Year	Original Resolution	Source
AVGMODIS *	Annual maximum green vegetation fraction: 12 years of normalized difference vegetation index data	2001–2012	250 m	modis.gsfc.nasa.gov/data/
Cropland Data Layer	USDA National Agricultural Statistics Service Cropland Data	2017	30 m	https://nassgeodata.gmu.edu/CropScape/
Elevation *	Altitude	NA	30 arc seconds	www.worldclim.org/
FVEG	Statewide vegetation with WHR types, size, and density.	2015	30 m	https://frap.fire.ca.gov/mapping/gis-data/
GAP	USGS GAP analysis project: land cover	2011	30 m	https://www.usgs.gov/core-science-systems/science-analytics-and-synthesis/gap/science/
Global Human Influence Index	Nine global data layers: human population pressure, human land use and infrastructure, and human access	1995–2004	1 km	https://sedac.ciesin.columbia.edu/data/set/wildareas-v2-human-influence-index-geographic/maps
NLCD	National Land Cover Database	2016	30 m	https://www.mrlc.gov/data
Open Water	Multiple integrated global remote sensing-derived land-cover products and prevalence of 12 land-cover classes	2005–2006	1 km	https://www.earthenv.org/landcover
PRISM	Seven climatic variables for the US annual and monthly precipitation and temperature	2016	270 m	http://www.prism.oregonstate.edu/
Streams	Distance to water	2003	NA- Shapefile	https://catalog.data.gov/dataset/cdfg-100k-streams-2003
USDA zones	Hardiness zones based on mean extreme annual minimum temperatures	2012	NA- Shapefile	https://planthardiness.ars.usda.gov/
WorldClim *	19 bioclimatic variables: 30-year averages 1970–2000	1970–2000	30 arc seconds	www.worldclim.org/bioclim

*Indicates variables included in the final MaxEnt model.

MaxEnt models were built in R Statistical Software version 0.98.110253 © [77]. The following R packages were used to run MaxEnt: dismo, sp, and raster [58, 78–81]. MaxEnt settings were chosen based on previously published literature and included using 25 random test points, 15 replicates, 5000 maximum iterations and the 10-percentile training for the threshold rule [52, 56, 58, 82]. A regularization multiplier of one through five was assessed to avoid overfitting and the default one was determined to be the optimal setting for the final model [82]. Logistic values for output were used as well as cross validation, which separates presence points into 80% training and 20% testing data (i.e., model validation), using k-fold sub-sampling to fit a model [52, 56, 83]. The relative contribution of each variable in a MaxEnt model was assessed comparing both percent contribution and permutation of importance, averaged over the number of iterations run and ascertained by jackknife tests [52, 55, 83]. Predictors for the final model were assessed using a backward variable selection approach: variables remained at each step if their percent contribution or permutation importance were approximately 10% or more [32, 52]. The response curves generated within MaxEnt showed the predicted probability of suitable feral pig habitat for each individual variable, changing per each level of the predictor [66, 69, 84].

MaxEnt model performance was assessed using the area under the curve (AUC) of the receiver operator characteristic (ROC), averaged over the number of chosen replicate runs [52, 54, 85]. AUC reflects a model’s prediction ability, on a scale of 0 to 1.00, with 0.50 representing random chance. In general, the following guidelines are used to assess AUC: 0.90 and above indicates an excellent model, a good model ranges from 0.80–0.90, a fair model runs between 0.70–0.80 and poor or failed model is any value under 0.70 [47, 56, 86]. While AUC is a standard diagnostic method to evaluate MaxEnt models, some authors suggest calibrating the AUC (i.e., AUCc), which removes spatial sorting bias (ssb) (i.e., spatial autocorrelation) by using point-wise distance sampling [58, 78, 87]. A ssb close to 1 indicates no spatial sorting bias, whereas a ssb close to 0 suggests a large spatial bias, and the need to use AUCc [87]. The final model was chosen based on the highest AUCc, relative to other models.

Risk map and OPO

A risk map that categorized areas at greatest risk for contact between these two swine groups was built by overlapping California OPO locations with the final MaxEnt feral pig suitable habitat raster. Between 2014–2019, OPO in California were compiled through various sources (e.g., agricultural festivals, local farmers markets, University of California Cooperative Extension (UCCE) advisors, web-based searches (search terms: “pasture-raised pork”, “pastured pigs”). GPS coordinates for all OPO were identified using Google Earth Pro v7.3.3. [88] Additionally as part of our objective to gather locations of OPOs in California, an online survey that contained an interactive map component was built with Survey 123 v3.6 [89]. The survey contained 29 questions that consisted mainly of multiple choice questions, with a few open ended questions about the number of animals raised (e.g., how many sows or boars raised on average each year). The survey included questions regarding biosecurity practices, swine health and feral pig presence. This online survey was announced electronically (e.g., media, e-newsletters) to swine related groups and organizations or conducted in-person at events, such as agricultural fairs. The survey instrument and protocols were reviewed and exempted by the Institutional Review Board (IRB) of the University of California-Davis (No. 1180798–1) (S1 Appendix).

To build the risk map for California, the final MaxEnt model predicting suitable habitat for feral pigs was overlapped with the location of OPO to categorize areas at greatest risk for disease transmission, due to contact between these two swine populations, and characterize risk at the farm-level. The underlying assumption presumed that direct or indirect contact between feral pigs and domestic pigs raised outdoors is a risk for disease transmission [90]. The probability of suitable habitat for feral pigs was extracted from the final MaxEnt model for each OPO location, using the Sample Raster Value tool in QGIS and added to the OPO shapefile. Then the Kernel Density tool in QGIS was used to make the risk map, matching the 270m x 270m resolution of the MaxEnt model and using the MaxEnt model probabilities as weights. Additionally, we used a radius of 5 km at each OPO location, which was an extrapolated average estimate from US based studies that measured home range of feral pigs, understanding that home ranges vary depending on age and sex of animal, as well as resource availability [47, 68, 91, 92]. The Kernel Density map was overlaid with the final MaxEnt model.

Results

MaxEnt model

The final MaxEnt model was chosen based on the highest AUCc of 89.7, relative to other models. Probability values for suitable habitat were divided into five equal interval categories: minimal (< 0.01); low (0.01–0.22); moderate (0.23–0.43); high (0.44–0.65); and extremely high (0.66–0.87), with 0.87 being the highest predicted probability in the final MaxEnt model (Fig 2). Areas with the highest likelihood of suitable feral pig habitat in California (i.e., orange, and red categories) included the north coast from Mendocino County all the way south along the coast to Santa Barbara County, and counties that border these coastal counties (e.g., Lake, Napa, Contra Coast, Santa Clara, and San Benito). Additionally, suitable habitat areas included the foothills of the Sierra mountains, from Shasta County south to Tulare County. Least likely suitable habitat included the Central Valley and eastern counties of California, from the most northern county of Modoc all the way to Imperial County in the south.

Fig 2

Final MaxEnt model predicting suitable feral pig habitat in California.

Color-coded categories represent the probability of suitable feral pig habit on a scale of almost zero (<0.01) to extremely high (0.66–0.87), based on equal intervals.

Five variables were identified as significant in predicting suitable feral pig habitat. The five significant variables were the annual maximum green vegetation fraction (AVGMODIS), the minimum temperature of the coldest month (BIO6), precipitation of the wettest month (BIO13) and the coefficient of variation for seasonal precipitation (BIO15) and elevation. All five variables provided approximately 10% or more percent contribution and permutation importance to the final model (S1 Table). The jackknife test results provided more information regarding the importance of each variable in the final model (S1 Fig). For example, BIO15 was the variable with the highest gain when used alone and elevation had the most information that was not available in the other variables. The response curves for the significant five variables indicated the predicted suitability range of each variable for feral pigs (i.e., the x-axis values above 0.50 on the y-axis) (Fig 3). For instance, feral pigs are predicted to prefer vegetative cover (i.e., AVGMODIS) of at least 60% or more.

Fig 3

MaxEnt response curves for the five significant variables used in the final MaxEnt model.

The response curves generated by MaxEnt show the predicted probability of suitable feral pig habitat for each individual variable per each level of the predictor. Significant layers included the minimum temperature of the coldest month (BIO6), the annual maximum green vegetation fraction (AVGMODIS), the precipitation of the wettest month (BIO13), the variation of annual precipitation (BIO15) and elevation.

A total of 305 OPOs were identified between 2014–2019, from 79.30% (46/58) of California’s 58 counties (i.e., no OPO data for 12 counties). The most OPO were identified in the following counties: Sonoma (n = 48), Mendocino (n = 19), Nevada (n = 16) and Yolo (n = 12). From the online survey, 39 OPO locations were gathered from 44 respondents and included in the final total. All survey respondents raised domestic swine outdoors and 25.00% (11/44) had seen feral pig presence within 3.22 km or less of their domestic swine raised outdoors, with 15.91% (7/44) observing feral pigs within 152.4 m of their pig herd. Domestic pig herd size ranged from 1–350 animals, with a mean of 24 and median of six. Hectares (ha) dedicated to raising pigs ranged from 0.026 ha to 12.14 ha with a mean of 1.89 ha and median of two 0.81 ha, with nine not answering.

The risk map reflects areas at greatest risk for contact between feral swine and domestic pigs raised outdoor and subsequent potential disease transmission (Fig 4). Risk levels start at green for low-risk areas and range up to orange and red for the highest risk areas. Areas with the most risk for contact between these two swine populations are denoted in orange or red, with sharper colors representing denser clustering of OPO. The counties with the highest likelihood of suitable feral pig habitat and densest clustering of OPO included: Sonoma, Marin, Napa, Yolo, Nevada, Mendocino, and Lake counties. Areas at lowest risk include the full eastern edge of California, which includes the Cascadian and Sierra Nevada Mountain ranges as well as deserts in the south. Table 2 categorizes the distribution of OPO at each level of probable suitable feral pig habitat using the final MaxEnt model levels. The results indicated that 49.18% of the 305 OPO were located near extremely high or highly suitable feral pig habitat.

Fig 4

Risk map demonstrating areas in California at greatest risk for contact between feral pigs and outdoor-raised domestic pigs within a 5km radius from each farm, using the Kernel Density tool in QGIS.

Colors are based on the probability of suitable feral pig habitat from the final MaxEnt model at each OPO, with sharper colors representing denser clustering of OPO.

Table 2

Percentage of 305 OPO identified in each MaxEnt suitable feral pig habitat level.

The final MaxEnt model contains a probability scale of 0.00 to 0.87 and was divided into equal intervals.

Levels	%OPO (ct/305)
Minimal (< 0.01)	0.98% (3/305)
Low (0.01–0.22)	19.67% (60/305)
Moderate (0.23–0.43)	30.16% (92/305)
High (0.44–0.65)	25.90% (79/305)
Extremely high (0.65 +)	23.28% (71/305)

Discussion

In this study, we investigated the predicted distribution of feral pigs in California and their spatial overlap with domestic pigs raised outdoors, to determine areas for surveillance in the case of an emerging or reemerging disease outbreak. The MaxEnt model results indicated heterogenous feral pig suitable habitat in each California county, instead of a homogenous distribution, as suggested by past maps. Additionally, this study overlapped predicted suitable feral pig habitat and OPOs to create a risk map for potential disease transmission at the feral pig-domestic pig interface. Although previous studies discussed the possibility of feral pig populations spreading disease to outdoor-raised pigs at the county-level, this is the first study to predict risk at the farm-level in California.

Since the exact location of most feral pig populations is unknown, predictive methods for species distribution (e.g., MaxEnt) are important to understand where feral pigs could potentially interface with domestic swine raised outside, either currently or in the future. Our final prediction model provided a more informative picture of suitable habitat for feral pigs than previous studies, which only showed single presence points or reported feral pigs at the county-level, even if only one feral pig was identified in that county [36, 38, 43, 44]. For instance, although previous county-level maps stated that all California counties except for Imperial County harbored feral pigs, our model shows almost no suitable habitat in an additional five counties: Modoc, Mono, Alpine, Lassen and Inyo. This result may reflect that few feral pigs have been seen in those counties. Additionally, our results were based on a fine spatial scale and indicated heterogenous suitable habitat in counties, not a uniform distribution, which is compatible with the fact that feral pigs need shrub cover and food to survive, which would not be found in cities or deserts [41]. Earlier feral pig mapping studies by the Southeastern Cooperative Wildlife Disease Study and National Feral Swine Program (NFSP) focused on county-level occurrence in the US [43, 44, 93]. A 2015 United States Department of Agriculture (USDA) study overlapped NFSP county-based feral pig locations with data from the 2012 National Animal Health Monitoring System (NAHMS) report of small-enterprise swine operations, specifically whether these survey respondents had seen feral pigs on their premises or within the same county, to ascertain the level of agreement between the two datasets [36]. They identified five counties in California that were in agreement with our model findings for suitable feral pig habitat: Mendocino, Tehama, Nevada, El Dorado, and San Luis Obispo, and two counties that differed: Ventura and Los Angeles counties. Although these county-based maps are important to demonstrate the trend of increasing feral pig populations nationwide, stakeholders and feral pig disease surveillance agencies could benefit from targeting outreach and mitigation strategies to specific regions within a county using our maps.

The results of our final model indicated five variables that were useful in predicting suitable feral pig areas in California, including three WorldClim layers: the minimum temperature of the coldest month (BIO6), precipitation of the wettest month (BIO13) and the coefficient of variation for seasonal precipitation (BIO15). Other studies also used WorldClim factors to predict the distribution of wild boar or feral pigs. These bioclimatic variables have been widely used in environmental studies and are now becoming popular for use in epidemiological investigations [75]. These climate variables are 30 year averages and “capture broader biological trends better than the temperature or the amount of precipitation for a given day due to the inherent variability associated with weather.” [75]. Bosch et al (2014) built a MaxEnt model for wild boar in Spain and their model also contained BIO6 and BIO15 as did regional models built by Pittiglio et al (2018) with BIO13 being significant as well [70, 86]. BIO6 is the minimum temperature of the coldest month and is interpreted as being a useful variable when deciding if the species of interest is affected by extreme cold events throughout a year [75]. Hill et al (2014) used MaxEnt to predict the distribution of Trichinella spp. and Toxoplasma gondii in feral pigs in the US and also identified BIO6 and elevation as significant predictor variables, along with land cover and other WorldClim factors [26]. The response curve for BIO6 in our model peaks at the predicted ideal range for feral pigs, with both ends indicating extreme cold temperatures that may be avoided by feral pigs. A 2015 study by McClure et al (2015) indicated that suitable feral pig habitat may be limited by cold temperatures, precipitation, and water availability, which reflects our findings [47]. BIO13 is defined as precipitation of the wettest month and is useful if extreme rainfall patterns influence the range of feral pigs [75]. BIO15 measures the variation in annual precipitation totals per month (i.e., seasonality of precipitation) and reflects the variability of rainfall that may affect a species [75]. According to the Jackknife graph, the variable with the highest gain when used alone was BIO15, and therefore had the most important information for predicting suitable feral pig habitat. Snow et al (2017) used Bayesian methods to predict the expansion of feral pigs in the US, but also detected that temperature and precipitation levels were significant predictors [38]. The final model gain is decreased the most if elevation is ignored and therefore it has significant information that is not available from the other variables in predicting feral pig suitability. Elevation was also significant in the MaxEnt models built by Hill et al (2014) [26]. These results combined with the response curve possibly reflect feral pigs preference for lower altitudes in the US. AVGMODIS, a measure of the annual maximum green vegetation fraction on a scale of 0 to 100, was also an important predictor of suitability, which reflects feral pigs’ need for available food and vegetative cover [72]. Garza et al (2018) identified NDVI, which AVGMODIS is based upon, and precipitation as important variables in predicting home ranges of feral pigs or wild boar worldwide, using generalized linear models [66].

The significant layers identified in our study to predict feral pig suitability are not unique, and this may be due to the fact that feral pigs are highly adaptable and opportunistic omnivores [38]. Lobe et al (2008) stated that MaxEnt AUC values will be lower for generalist species that are widely distributed [94]. However, the AUCc of our final model was 89.7, which indicates a good model. Another factor to consider when analyzing the distribution of feral pigs, is the effect of anthropogenic movement. Tabak et al (2017) analyzed anthropogenic factors that might affect the expansion of feral pigs, using hierarchical Bayesian models [13]. They determined that feral swine movement was affected by similar factors included in our study, for instance, the number of domestic pig farms, the amount of public land and hunter pressure [13]. Although they did not specifically identify domestic swine raised outdoors, they found that presence of a domestic pig farm did predict movement of feral pigs into California counties, since domestic pigs are known to escape and can readily adapt [13]. Additionally, using a model fitted with 2017 hunting tags (n = 1,745) vs. all 5,148 provided the best prediction model, based on the AUCc. MaxEnt is an important method to predict the distribution of rare species, and an upper maximum range for the number of species occurrence points has not been previously determined. However, our result fits with a study conducted by Chen et al (2012) to determine the sample size for the outcome variable in building MaxEnt models [63]. They reported that standard deviation decreased and MaxEnt models became more stable using species occurrence points of 1,000–1,200 [63-65]. Possibly the sample size of the outcome variable that reaches asymptote is dependent on the geographic extent and characteristics of the species of interest.

Regarding feral pig presence on farms, the most recent NAHMS survey asked participating swine small-enterprise producers in the US (i.e., those raising 100 pigs or less) about presence of feral swine in their county, but did not separate farms based on whether they raised domestic swine indoors or outdoors. However, a 2015 USDA report regarding overlap of feral and domestic pigs in the US used this NAHMS dataset and reported that of the 320 participating US counties, 74% of these counties had small-enterprise swine producers who allowed their pigs some level of outdoor access [36]. The NAMHS results indicated that 52.9% of small-enterprise swine producers in the West/South region, which included California, reported feral pigs in the same county, with 16.2% of those having feral pig presence on their operation, similar to our survey results that showed 15.91% of respondents had seen feral pigs within 500 feet of their pig herd [95]. Another study that measured the co-occurrence of feral pigs and agriculture to understand the risk of disease transmission, but that did not separate outdoor versus indoor herds, reported that on average 47.7% of all types of farms had feral pigs in the same counties, including California, showing a significant increase in the decade from 2002–2012, which aligns with the fact that feral pigs numbers are increasing nationwide [96, 97]. Additionally, our risk map identified eastern counties as having the lowest risk. However, we did not identify OPO in many of these counties, therefore we cannot say there is no or low risk in those regions. The results of these aforementioned survey-based studies indicated that more than 45% of farms have feral pig presence within the same county in areas with a large populations of feral pigs, which matches the results from our risk map that showed almost half of the identified OPO in California had suitable feral pig habitat nearby [36, 95, 96]. Targeting outreach and surveillance to highly connected farms may be warranted, even if that farm is not near highly suitable feral pig habitat, given their effect on other farms. Additionally, OPO with the highest risk of disease spillover may not be connected to other OPOs. Nevertheless, these findings together indicate the need for targeted outreach and mitigation strategies for those farms at highest-risk for feral pig contact, due to the potential for disease transmission between these two swine groups.

The results of our risk map indicated that 49.18% of the 305 OPO identified in California are located near highly suitable feral pig habitat, indicating that spillover of an emerging or transboundary disease is a possibility, given the correct drivers. Possible drivers of disease spillover in the US between feral pigs and domestic swine raised outdoor include the density of animals (both feral and domestic swine), shared natural areas and increasing contact between these two growing swine populations [98, 99]. Although spatial overlap of these two swine populations does not necessarily demonstrate direct contact, direct or indirect contact between a pathogen host and susceptible individuals is one factor that facilitates disease transmission [90]. A Spanish study by Kukielka et al (2013) used camera traps to measure interactions between wild boar and domestic swine, to understand the transmission of Mycobacterium tuberculosis [100]. Interactions between wild boar and domestic swine were mainly recorded as indirect contact at water sites during wet seasons. Additionally, domestic pigs followed wild boar most often, instead of vice versa, and the authors suggested the spread of tuberculosis would occur mainly from exposure of domestic pigs to wild boar, through indirect contact. A study by Yang et al (2021), used GPS collars to quantify direct and indirect contact rates between feral pigs and cattle in Florida, a state with a high abundance of feral pigs like California [90]. While they found that direct contact was infrequent, indirect contact at water troughs or mineral blocks was significant, indicating that contact between feral pigs and livestock can contribute to pathogen transmission, assuming other disease transmission requirements are also met [90]. These aforementioned studies demonstrate the continued need to study transmission dynamics between feral pigs and domestic swine through direct and indirect interactions, as contact between feral and OPOs is possible. Other components of disease transmission that affect spillover include the density of the pathogen host (i.e., feral pigs), the number of domestic swine raised outdoors on each farm and the connections between OPO [90]. Since the exact number and location of feral pigs is currently unknown in California, using camera traps to estimate abundance might be one way to improve this current study. Additionally, we were unable to gather the number of acres and swine raised on each OPO, as this information is not readily available in the US. Regarding connections between OPO, unless a farm has a closed herd, one might assume that connections between OPO exist if the owners share tools or sell animals between farms. However, these dynamics were not measured in this study. A network analysis of connections between OPO would add more dimensionality to the current risk map.

Our risk map reflected the heterogeneity of feral pig habitat in each region and identified high-risk contact areas between farms and feral pigs in California. Studies that identified high-risk areas in California between feral pigs and domestic swine raised outdoors are sparse. A 2015 USDA report extracted outdoor operations with NFSP feral swine populations and did not identify any hot spots of overlap in California as seen in our results [36]. However, they did not report the number of OPO per state or county and most likely our state-focused study identified more OPO than their survey-based national study. A study by Miller et al (2017) also assessed possible disease transmission between feral pigs and farm at the county-level [96]. They reported that domestic swine, either raised indoors or outside, have been increasing in counties that also had feral pig presence. The lack of maps identifying areas at high-risk for disease transmission between these two swine populations indicates a need for further research.

A limitation of this study involves using hunting tags as a proxy for presence of feral pigs to predict suitable habitat. Hunting tags are voluntarily submitted to CDFW by hunters and estimated to account for only 30% of all hunted pigs and most likely biased toward easy to access areas. Also, only half of the land in California is public land and accessible to hunters, therefore feral pigs hunted on private land are not included in our data sets. However, Rutten et al (2019) used similar hunting bags and MaxEnt to successfully predict the distribution of wild boar in Belgium [49]. And Alexander et al (2016) also used hunting records to predict wild boar habitat in Europe [62]. Additionally, MaxEnt assists in overcoming these challenges by identifying similar habitats in all parts of California and predicting suitable areas.

Both the MaxEnt model and risk map are limited because they are static maps that use fixed layers as their foundation; consequently, they do not incorporate dynamic events over various years (e.g., wildfires, landscape changes, weather fluctuations). Also, feral pigs may migrate seasonally due to shifting weather, resource availability, hunting pressure or wildfire and future research could focus on species distribution modeling that includes dynamic real time variables or remote sensing data; however, seasonal or dynamic spatial data are not available yet for most spatial predictors in California [47, 49, 66, 91, 101–103]. However, our approach is valuable as a first step in identifying multiple high-risk areas for future research, where additional data could be collected. Furthermore, future research could add feral pig disease data collected statewide to evaluate if high-risk areas for feral-domestic pig contact equates to those areas with higher prevalence of diseases [50].

There are some challenges and limitations to the risk map generated in this study. For instance, farms and ranches in California, including backyard and commercial operations, are not required to register with state agricultural agencies, therefore, the total number, distribution, and size of OPO remains unknown and are underrepresented in this study. A majority of the identified OPO in this study were commercial pork producers with an online presence or ones that attend conferences, farmers markets and fairs. If more OPO locations could be identified, than a more comprehensive map of high-risk areas could be generated. Additionally, because we are based at the University of California, Davis in Yolo County, there is selection bias in the OPO identified as our agricultural networks are within the UCCE network. Overrepresented counties reflected either sampling bias or clustering of these niche operations or both. Nevertheless, the number of OPO included in this study (n = 305) and the fact that more than 40% of these operations were in highly suitable areas for feral pig contact is relevant as an initial approximation of a likely much larger risk of disease transmission at the feral-domestic swine interface in California. In the future, adding disease cases to this risk map would add additional epidemiological information regarding possible pathogen transmission.

Conclusion

This study evaluated the feral-domestic pig interface of two parallel trends: expanding feral pig populations and an increase in outdoor-raised pig operations in California, as related to the risk for future disease transmission. Since both swine populations are reservoirs for various pathogens, the contact between these two swine groups has important implications for disease transmission in the wildlife-livestock interface. This study provides a foundation to design targeted, cost-effective disease surveillance and risk mitigation programs in regions at highest risk for wild-domestic pig contact and can serve as a template for similar efforts nationwide. Moreover, the results of this study provide a framework to create an outreach extension program and inform all stakeholders (e.g., farmers, government agencies) that may be called upon to respond to future zoonotic or TAD outbreaks, such as ASF. The results of this study, despite limitations, can provide important information to stakeholders and organizations that handle swine diseases or public health problems originating from any swine group in California.

Survey for outdoor-raised pig owners in California.

(PDF)

Click here for additional data file.

The analysis of variable contribution table provided estimates of the relative contribution of each variable to the final MaxEnt model.

(TIF)

Click here for additional data file.

Jackknife results for final MaxEnt model.

The Jackknife graph indicated importance of key variables: BIO6 was the minimum temperature of the coldest month, AVGMODIS was the annual maximum green vegetation fraction, BIO13 was the precipitation of the wettest month, BIO15 was the variation of annual precipitation and elevation.

(TIF)

Click here for additional data file.

(ZIP)

Click here for additional data file.

18 Jan 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

W. David Walter, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match.

When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

3. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

4. We note that Figures 1 and 2 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

a. You may seek permission from the original copyright holder of Figures 1 and 2 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

Additional Editor Comments:

Line 84: Change "outdoor-raised domestic pig premises" to "OPO" unless there its a difference between OPOs and a "premise" which does not seem the case.

Line 125: Change "Publicly available predictor spatial layers online" to "Predictor spatial layers were publicly available online." The authors often and excessively use adjective modifiers throughout their writing and should avoid if possible. Using 3-5 adjectives to describe your subject gets confusing to interpret exact meaning of your topic.

Line 149: Change "1" to "one" or delete altogether? Not clear if you are stating that the default was determined to be the optimal setting so is 1 even needed?

Line 262: Change "outdoor-raised pig operations" to "OPOs."

Line 264-265: While I believe this is appropriate language for a Cover Letter, I suggest the authors delete this here because you say nearly the same thing in the subsequent sentence. Also, remove "to our knowledge" in the Lines 265-267.

Lines 268-269: Change "species distribution predictive methods" to "predictive methods for species distribution."

Line 270: remove "MaxEnt" because it is not necessary.

Line 275: Again remove "MaxEnt" throughout Discussion. You state the models run in your Methods so no reason to refer to MaxEnt each time you reference your models or results in the Discussion.

Line 278: Change "the final MaxEnt model was" to "our results were"

Line 284: NAHMS? Spell out.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Comments

--------

In this study, Patterson et al. use hunter-harvested pigs and landscape

variables to develop an occupancy map of feral swine in California. They then

use farmer surveys to identify outdoor pig operations and overlay their

occupancy and OPO map to predict areas of high contact risk between feral swine

and domestic swine. The study provides one of the first high resolution maps

of potential contact zones between domestic and feral pigs in California.

Overall, I found the results and conclusions sound and consistent with previous

analyses on feral swine resource selection and occupancy.

My main comment is that I think more discussion should be provided on

the "implications for disease transmission" portion of this manuscript. The

bulk of the discussion (6 paragraphs) focuses on the MaxEnt results while the

discussion of transmission implications are given in 2 paragraphs. The MaxEnt

results are largely consistent with previous analyses on feral swine and, given

the paper's title, I think there should be more balance between predictors of

pig relative occupancy and transmission implications. For example, beyond

actual disease data, what are some additional components of transmission that

the proposed measure of contact risk is lacking? At least three come to mind:

i) this measure is not accounting for the size of OPOs (though it looks like

this was measured in the surveys) ii) the measure is not accounting for the

density of feral pigs in an area and iii) the measure is not accounting for the

connectivity of the OPOs (mentioned in my minor comments below). All of these

factors could significantly alter how contact risk translates to transmission

risk and spread among farms. Balancing the discussion with a more in-depth

look at the transmission implications of these results would strengthen the

paper.

Minor Comments

--------------

Line 77: What about Lewis et al. "Historical, current, and potential population

size estimates of invasive wild pigs (Sus scrofa) in the United States" who map

feral swine density at the 1 km scale?

Line 130: Change "NDIV" to "NDVI"

Line 138: This is a bit vague. What does "maintaining adequate detail for

suitable habitat modeling" mean? For example, could you clarify why a 1 km x 1

km scale would have been inadequate for the goals of this study?

Table 1: Please provide the spatial resolution of the layers (where applicable)

Line 159: Change "was" to "were"

Figure 1: It would be useful if the author's could overlay a map of the observed

hunter harvest points.

Line 228: Change "OPO" to "OPOs"

Results: I could not figure out how to access the Supplementary Material so I

was unable to review the response curves and S1 Table, S1 Fig, and S2 Fig.

The authors may have been limited by space, but if possible it would be

useful to see the response curves in the main text.

Figure 2: Colors on this figure are hard to see. Also, I don't understand the

color bar label. Why is Lowest Risk greater than (>) Moderate Risk greater

than (>) Highest Risk?

Line 265: Pepin et al. 2021 (Prev. Vet. Medicine) also does this, but at the

county scale and crossing the wildlife-livestock-human interface.

Line 276-277: But don't you have the observed point so you know whether or not

feral swine were observed in these counties? The "may indicate" is confusing.

Perhaps "reflects"?

Line 335-336: Could you remind the reader what criteria you are using to make

this assessment. Also, if I am interpreting this sentence correctly, it would

be helpful to rephrase similar to the following: "Additionally, using a model

fitted with 2017 hunting tags (n=1,745) vs. all 5,148 points for 2012-19

provided the best out-of-sample predictions".

Line 361-362, 415-417: It would be pertinent to mention that farms with the

highest spillover risk might not be the most connected to other farms and might

be less important for among-farm spread than suggested by the maps given here.

Mitigation might be most effective if highly connected farms are targeted even

if they have lower feral swine suitability.

Line 373-374: I don't understand how the "therefore" clause follows in this

sentence. I would recommend re-wording this sentence to make it more clear.

Reviewer #2: Overall this manuscript looks great! I think it is a nice body of work that is translational and has true One Health implications. I have provided additional comments in the manuscript but I do think this manuscript is worthy of acceptance.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PONE-D-21-36324 (1)_VB.pdf

Click here for additional data file.

23 May 2022

PONE-D-21-36324

Identification of high-risk contact areas between feral pigs and outdoor-raised pig operations in California: implications for disease transmission in the wildlife-livestock interface

PLOS ONE

Editor Comments:

I agree with the comments by Reviewer 1 but I believe that Major Revision is more appropriate. I don't believe the level of detail of the dataset and spatial layers is adequate. In their revision I would appreciate the authors including details for their methods outlined below:

AR1: The authors really appreciate the comments and suggestions of the reviewers. The authors have done their best to address all of the points addressed by the editor and reviewers. We updated the tables and figures and added two more figures, including the feral pig hunting tags. The Journal requirements were followed and updated. Additionally, text regarding disease transmission and anthropogenic movement was added to the discussion section.

1. As mentioned by Reviewer 1, the level of detail of how each spatial layer was presented into MaxEnt is not adequate. I needed to see Figure 1 before realizing that this was throughout the state of California and not a portion of it. Table 1 could include resolution of each layer and how each layer was presented. The level of detail in Description column is not nearly enough. I would suggest moving this to the Methods text and add more details. I don't believe anyone could replicate your models with this level of detail for your spatial layers.

AR2: Thank you for raising this concern. All layers' initial resolutions and projections are available at each website source presented in Table 1. For the purpose of this study, we reprojected all layers to the Albers Equal-Area coordinate reference system for California (“California Albers” (meters)) using QGIS 3.6. If a layer is not specific to California, we cropped it to fit specifically to the extent of California and added a line explaining that layers were masked for entire state of California: “and masked for the entire state of California”. Lines 142-143.

Maxent requires that all layers have the same size (pixel size) which we fixed to 270 x 270 m. This has been clarified in the text and Table 1 to facilitate replication. Additionally, we have included the standardized five layers that were significant in the final model in the Supplementary section. More details about this modeling approach can be found in Hijmans RJ, Phillips S, And JL, Elith A. dismo: Species Distribution Modeling. R package version 1.3-3. 2019. Additionally, we added more text to the Materials and Methods section to clarify descriptions of the main layers used in the final MaxEnt model. See Table 1 and Lines 130-136; 142-147.

2. Was AVGMODIS an average of 12 years of NDVI.

AR3: Yes, AVGMODIS is an average of 12 years of NDVI as is written here: “For example, AVGMODIS was the annual maximum green vegetation fraction (MGVF) combined with 12 years of normalized difference vegetation index data (NDVI) and relates to food and shrub cover for feral pigs.” We added this text to clarify about this layer: NDVI measures vegetation gathered by the Moderate Resolution Imaging Spectroradiometer (MODIS) as part of NASA’s satellite systems. Lines 134-136.

3. If AVGMODIS is NDVI is it correlated to NDVI? Were any correlations included in your data sources prior to using them in MaxEnt?

AR4: Thank you for the suggestion. We did run correlations and I added in this text to clarify: “Predictors were assessed for correlation using Spearman’s rank and a cut-off threshold of 0.80, a threshold used in previous studies (LaHue et al, 2016). Two correlated variables were not included at the same time, during variable selection steps.” Lines 145-147.

4. What year was NLCD used as a spatial layer? NLCD is released every 5 years so which version should be placed in the methods.

AR5: NCLD layer used was from 2016. We added a ‘Years’ column to Table 1.

5. As requested by Reviewer 1, feral pig tag locations across the state needs to be included if the authors are presenting a risk map statewide.

AR6: We have included a map containing feral pig hunting tags from 2017. MaxEnt uses a variety of layers to predict suitable habit, along with presence points, so overlaying hunting tag points is contrary to the purpose of using MaxEnt. Hunting tags represent a subset of feral pig locations, because legal hunting is only conducted on public lands, which encompasses approximately 50% of California.

6. Was the OPO survey done as part of this study? The response rate of these surveys is important to report. Do the authors believe their survey adequately reflects OPOs (13 percent response rate) in California based on their results? It seems the survey questions should be a supplement or a separate manuscript? If the authors have coordinates for OPOs but only a 13% response rate, is the survey of any value in this manuscript?

AR7: Yes, the survey was conducted to gather the locations of OPO and investigate feral pig presence nearby these type of operations. We added this text to clarify: ‘Additionally as part of our objective to gather locations of OPOs in California…’ Line 186.

The survey only had 44 respondents, because of the recruitment strategy (snow-ball approach) and there is not a census of those operations, therefore, we are not able to estimate a response rate. See this text within the manuscript: “This online survey was announced electronically (e.g., media, e-newsletters) to swine related groups and organizations or conducted in-person at events, such as agricultural fairs.” Lines 191-193.

Regarding whether we “believe their survey adequately reflects OPOs (13 percent response rate) in California, farms are not required to register with any government agency, therefore no one really knows the number of OPOs nationwide, although the USDA has been editing their national surveys over time to include operations that raise swine outdoors. Please see this text that was in the limitations section “For instance, farms and ranches in California, including backyard and commercial operations, are not required to register with state agricultural agencies, therefore, the total number, distribution, and size of OPO remains unknown and are underrepresented in this study. A majority of the identified OPO in this study were commercial pork producers with an online presence or ones that attend conferences, farmers markets and fairs.” Lines 460-465.

The authors believe it is important to include this survey because 1) it demonstrates how challenging it is to identify OPOs and 2) although a small number of famers participated, it provides demographic details regarding this population that is not reflected in mapping locations only.

Journal Requirements:

1. Manuscript Meets PLOS One’s style requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

2. Financial Disclosure

We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match

3. Data Availability:

In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found.

4. We note that Figures 1 and 2 in your submission contain map images which may be copyrighted.

AR8: The manuscript and files were revised and modified to meet the style requirements. Financial disclosure and funding information were corrected.

Funding: This study was supported by the Agriculture and Food Research Initiative grant no. 2019-67011-29609 from the USDA National Institute of Food and Agriculture (LP) and a University of California, Davis Academic Federation grant (LP and AFAP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data statement was modified to: OPO data cannot be shared publicly, as they contain GPS coordinates and participants provided locations based on confidentiality. Feral pig hunting tags are provided in the text as Fig1. Predictor layers used to build a MaxEnt model are publicly available online, but rasters included in the final MaxEnt model are included in KML format as supplementary files.

Figures: Figure 1 (Hunting tags), Figure 2 (Maximum Entropy model that predicts suitable feral pig habitat) and Figure 4 (Risk map) were all created by the first author (i.e., originals) and are not copied nor subjected to copyrights.

Additional Editor Comments:

Line 84: Change "outdoor-raised domestic pig premises" to "OPO" unless there its a difference between OPOs and a "premise" which does not seem the case.

AR9: Suggestion accepted. Line 84.

Line 125: Change "Publicly available predictor spatial layers online" to "Predictor spatial layers were publicly available online." The authors often and excessively use adjective modifiers throughout their writing and should avoid if possible. Using 3-5 adjectives to describe your subject gets confusing to interpret exact meaning of your topic.

AR10: Suggestion accepted. Line 128.

Line 149: Change "1" to "one" or delete altogether? Not clear if you are stating that the default was determined to be the optimal setting so is 1 even needed?

AR11: Correction was made. Line 157-168.

Line 262: Change "outdoor-raised pig operations" to "OPOs."

AR12: Suggestion accepted. Line 263 and throughout the manuscript (see yellow highlights).

Line 264-265: While I believe this is appropriate language for a Cover Letter, I suggest the authors delete this here because you say nearly the same thing in the subsequent sentence. Also, remove "to our knowledge" in the Lines 265-267.

AR13: Thank you for the suggestion. The first paragraph of discussion was modified. Lines 281-289

Lines 268-269: Change "species distribution predictive methods" to "predictive methods for species distribution."

AR14: Suggestion accepted. Lines 290-291.

Line 270: remove "MaxEnt" because it is not necessary.

AR15: Suggestion accepted. Lines 292-293.

Line 275: Again remove "MaxEnt" throughout Discussion. You state the models run in your Methods so no reason to refer to MaxEnt each time you reference your models or results in the Discussion.

AR16: Thank you for the suggestion, we removed “MaxEnt” throughout the discussion.

Line 278: Change "the final MaxEnt model was" to "our results were".

AR17: Suggestion accepted. Lines 299

Line 284: NAHMS? Spell out.

AR18: Suggestion accepted. Lines 305-36.

Comments to the Author

Reviewer #1: Comments

In this study, Patterson et al. use hunter-harvested pigs and landscape variables to develop an occupancy map of feral swine in California. They then use farmer surveys to identify outdoor pig operations and overlay their occupancy and OPO map to predict areas of high contact risk between feral swine and domestic swine. The study provides one of the first high resolution maps of potential contact zones between domestic and feral pigs in California. Overall, I found the results and conclusions sound and consistent with previous

analyses on feral swine resource selection and occupancy.

My main comment is that I think more discussion should be provided on the "implications for disease transmission" portion of this manuscript. The bulk of the discussion (6 paragraphs) focuses on the MaxEnt results while the discussion of transmission implications are given in 2 paragraphs. The MaxEnt results are largely consistent with previous analyses on feral swine and, given the paper's title, I think there should be more balance between predictors of pig relative occupancy and transmission implications. For example, beyond actual disease data, what are some additional components of transmission that the proposed measure of contact risk is lacking? At least three come to mind: i) this measure is not accounting for the size of OPOs (though it looks like this was measured in the surveys) ii) the measure is not accounting for the density of feral pigs in an area and iii) the measure is not accounting for the connectivity of the OPOs (mentioned in my minor comments below). All of these factors could significantly alter how contact risk translates to transmission risk and spread among farms. Balancing the discussion with a more in-depth look at the transmission implications of these results would strengthen the paper.

AR18: The authors appreciated the suggestion of the reviewer. We added a section in the discussion section regarding implications for disease transmission. Lines 398-428

Minor Comments

Line 77: What about Lewis et al. "Historical, current, and potential population size estimates of invasive wild pigs (Sus scrofa) in the United States" who map feral swine density at the 1 km scale?

AR19: The reviewer raised a very good point. Lewis et al, measured at a 1km scale, but they were trying to estimate the abundance of feral pigs per 1km x 1km cell size for the entire US, whereas we were identifying possible areas of contact between a feral pig and OPOs, which is a smaller scale level. (See lines 102-104 “To the best of our knowledge, there are no maps characterizing where suitable feral pig habitat overlaps with domestic pigs raised outdoors at the farm-level in California.”) Additionally, because they were running models for the entire nation, using a smaller pixel size (as we did in the current study) would have taken a lot of computer processing time. Because we were working only with one state, California, we could run models with smaller cell sizes, which provided a finer scale prediction per 270m x 270m area. We modified the introduction to: ‘Feral pig population distribution and abundance is dynamic yet has not been documented at fine spatial units less than 1km. Additionally, previous presence maps reported feral pigs for an entire county, even if there had only been a single occurrence recorded countywide’ Lines 76-79.

Line 130: Change "NDIV" to "NDVI"

AR20: Correction made. Line 134.

Line 138: This is a bit vague. What does "maintaining adequate detail for suitable habitat modeling" mean? For example, could you clarify why a 1 km x 1 km scale would have been inadequate for the goals of this study?

AR21: The OPO are generally small-scaled and since we were modeling the interface at the farm-level, we wanted to use a resolution that matched closer to a small farm size (270 meter vs 1 km). We changed the text to “Rasters were all converted to the same resolution of 270m x 270m, which used a reasonable amount of computer computation time, while maintaining fine-scale for suitable habitat modeling at the farm-level”. Lines 143-145.

Table 1: Please provide the spatial resolution of the layers (where applicable)

AR22: Suggestion accepted. We adjusted Table 1 to include years and original layer resolution.

Line 159: Change "was" to "were".

AR23: Correction made. Line 153

Figure 1: It would be useful if the author's could overlay a map of the observed hunter harvest points.

AR24: Thank you for the suggestion. We included a map of 2017 hunting tags. Figure 1.

Line 228: Change "OPO" to "OPOs" .

AR25: Corrected. Line 248.

Results: I could not figure out how to access the Supplementary Material so I was unable to review the response curves and S1 Table, S1 Fig, and S2 Fig. The authors may have been limited by space, but if possible it would be useful to see the response curves in the main text.

AR26: Thank you for the suggestion. We added the response curves to the main document. Lines 240-245. Figure 3.

Figure 2: Colors on this figure are hard to see. Also, I don't understand the color bar label. Why is Lowest Risk greater than (>) Moderate Risk greater than (>) Highest Risk?

AR26: Thank you for the suggestion. We re-formatted the figure and legend of the Risk map. The background of the risk map now matches the legend better and the legend was fixed to remove the “>”. Figure 4

Line 265: Pepin et al. 2021 (Prev. Vet. Medicine) also does this, but at the county scale and crossing the wildlife-livestock-human interface.

AR27: This is a good point raised by the reviewer. We updated the first paragraph of the discussion section. See Lines 281-289.

Line 276-277: But don't you have the observed point so you know whether or not feral swine were observed in these counties? The "may indicate" is confusing. Perhaps "reflects"?

AR28: This is a good point. We changed to “reflects” Line 258. Hunting tags contain bias, because hunting is only conducted on public lands and not in all counties. Additionally, not all hunters report GPS locations accurately. This limitation is described in Lines 440-444.

Line 335-336: Could you remind the reader what criteria you are using to make this assessment. Also, if I am interpreting this sentence correctly, it would be helpful to rephrase similar to the following: "Additionally, using a model fitted with 2017 hunting tags (n=1,745) vs. all 5,148 points for 2012-19 provided the best out-of-sample predictions".

AR29: Thank you for the suggestion. We rephrased this sentence and added a new sentence explaining the AUC assessment, in Materials Methods section. “In general, the following guidelines are used to assess models with AUC: 0.90 and above indicates an excellent model, a good model ranges from 0.80-0.90, a fair model runs between 0.70-0.80 and poor or failed model is any value under 0.70.” Lines 171-173 and 362-363.

Line 361-362, 415-417: It would be pertinent to mention that farms with the highest spillover risk might not be the most connected to other farms and might be less important for among-farm spread than suggested by the maps given here. Mitigation might be most effective if highly connected farms are targeted even if they have lower feral swine suitability.

AR30: This is a good point. We added a new sentence to the discussion: “Targeting outreach and surveillance to highly connected farms may be warranted, even if that farm is not near highly suitable feral pig habitat, given their effect on other farms. However, OPO with the highest risk of disease spillover may not be connected to other OPOs.” Lines 391-394.

Line 373-374: I don't understand how the "therefore" clause follows in this

sentence. I would recommend re-wording this sentence to make it more clear.

AR31: This sentence was removed.

Reviewer #2: Comments

Overall I think this manuscript is a really nice body of work. There is a lot of work done on the feral swine side, a lot of work done on the domestic swine side, and a lot of work done on the human health side but you don't often see comprehensive studies so I really appreciated that. One important consideration that I think needs to be included in the discussion is anthropogenic movement. Feral swine are routinely translocated via human-mediated movement which is very different than most other wildlife species. I think that is an important consideration when thinking about environmental conditions and warrants discussion. Great job!

AR32: Thank you for the positive feedback. We added a paragraph about anthropogenic movement in the discussion section. Lines 359-367.

Other minor changes:

Line 71: ‘as’ =‘has’. AR33: Changed. Line 72

Line 122: Capitalized ‘methods’ to Methods. AR34: Changed Line 112

Line 155: Add ‘were’. AR35: Changed. Line 167

Line 195: Change ‘sex’ for ‘gender’. AR35: Changed. Line 210

Submitted filename: Response to Editor R2.docx

Click here for additional data file.

13 Jun 2022

Identification of high-risk contact areas between feral pigs and outdoor-raised pig operations in California: implications for disease transmission in the wildlife-livestock interface

PONE-D-21-36324R1

Dear Dr. Pires,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

W. David Walter, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

I appreciate the authors addressing my comments as well as those of the reviewers. I believe the level of detail added to this draft by the authors provided sufficient information to your Methods, spatial layers, and MaxEnt models that was requested by reviewers.

Reviewers' comments:

20 Jun 2022

PONE-D-21-36324R1

Identification of high-risk contact areas between feral pigs and outdoor-raised pig operations in California: implications for disease transmission in the wildlife-livestock interface

Dear Dr. Pires:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. W. David Walter

Academic Editor

PLOS ONE