## Introduction

The Anthropocene is the current geological age on earth, characterized by the dominant influence of humans on the environment, including impacts on native ecosystems and spread of invasive species globally1. Indeed, invasive species have been identified as one of the greatest threats to global biodiversity2. Human activities can facilitate the spread of invasive species, directly through intentional and unintentional movements of plants and animals, and indirectly through habitat fragmentation and change associated with agriculture, urban expansion, and other anthropogenic land-use changes, climate change, and over-harvest of native species3,4,5. Efforts to mitigate the spread of invasive species have been limited by the lack of timely and accurate maps of occurrences and spatial expansion, especially over very large areas (>1 million km2). Processes required to complete comprehensive national scale mapping of a species are limited by project budgets. New approaches are needed that are cost-effective and repeatable, especially for species expected to expand rapidly over large areas and that are associated with greatest impact.

Wild pigs (Sus scrofa), also referred to as feral swine, wild hogs, or feral hogs6, currently have the largest global range of any non-domesticated terrestrial mammal on earth7. Native to Eurasia and part of North Africa, wild pigs have expanded their range, primarily through human introductions combined with natural dispersal, across all continents except Antarctica7,8. The broad geographic extent of their native range, coupled with the generalist nature of the species has allowed wild pigs to easily adapt and survive in new environments that span a broad range of climate, habitat, and resources9,10. The widespread success of wild pigs is explained by their extremely high fecundity11, early sexual maturity12, plastic diet7, long lifespans13, and highly adaptive nature9.

Wild pigs are an invasive species in North America and are descendants of Eurasian wild boar (S. scrofa scrofa), domestic pigs (S. scrofa domesticus), and hybrids of the two14,15. Long-established populations have existed in southern parts of the U.S. for hundreds of years, with areas of high population densities in Texas, Florida, and California. In the continental United States, there has been a well-documented expansion in the distribution and abundance of wild pigs in recent decades, from 17 to 38 states during the last 30 years16. Wild pigs were first introduced to Canada during a federal and provincial agriculture diversification initiative in the 1980’s and 1990’s to diversify livestock species and supplement producer incomes17,18. Escapes and intentional releases from domestic wild boar farms have led to the feral populations that are established on the Canadian Prairies18,19. Brook and van Beest18 provided a coarse-scale distribution of wild pigs in Saskatchewan at the Rural Municipality level. However, prior to this current study there has not been a comprehensive national scale map of the species range in Canada.

Wild pigs are considered to be the most damaging invasive species in the U.S.20, posing numerous ecological and socio-economic threats within their introduced range. Referred to as ecological train wrecks, wild pigs alter ecosystem processes, vegetation successional stages, nutrient cycles, and cause erosion, sedimentation, and eutrophication to riparian areas and water bodies7,16,20,21,22,23. The generalist nature and plastic diet of wild pigs allows them to utilize and compete for a wide variety of resources, as well as predate small mammals, amphibians, invertebrates, and ground nests16. The significant disturbance of habitat, resources, and ecosystem processes has direct and indirect effects on native wildlife and has the ability to decrease biodiversity and cause extirpations and extinctions20,24,25. Species extirpations and population declines as a result of wild pig presence have been documented in the United States, Galapagos Islands, and Australia16,26,27,28. In some areas within wild pig’s introduced range, populations have expanded to the point where eradication is no longer feasible29. Negative impacts associated with wild pigs have been well-documented across Europe, Australia, and the United States; however, these negative impacts have not been characterized in Canada.

A key challenge in managing rapidly expanding invasive species such as wild pigs at national and continental scales is having up-to-date information on their spatial distribution. Mapping the locations of invasive species is central to guiding effective management and is essential to determine if control efforts are effective at controlling and limiting, or even reducing, their spatial expansion30. However, identifying cost-effective methods to accurately and repeatedly map a species at a national scale represents a significant time and financial commitment. Conventional ecological monitoring used for large mammals such as aerial surveys, trail cameras, and mark-recapture can be effective at relatively small scales (<100,000km2), but become time and cost prohibitive at much larger scales. The use of local and traditional knowledge accumulated by people living and working on the land through personal observations and shared knowledge has been used for documenting species occurrences31,32. Such data can be collected using personal interviews, mail surveys, internet surveys, telephone surveys, and open source mapping33,34. Similarly, citizen science engages large numbers of lay people in collecting species occurrence data, such as the annual Breeding Bird Count and the Christmas Bird Count across the United States and Canada35, though this also requires considerable logistics coordination and has rarely been used for large mammals. Efforts to incorporate local knowledge in data collection using rigorous social science methods have rarely been applied over large areas (>1 million km2). Much work remains to evaluate the efficacy of these methods and determine the benefits and limitations of the different approaches in the face of immediate needs for detailed information on invasive species in general and wild pigs in Canada specifically. As such, the objectives of this study were to: (1) identify the past and current spatial distribution of wild pigs across Canada, and (2) evaluate the benefits and limitations of four different data collection methods and four different evaluation/validation approaches for national-scale, repeatable mapping of wild pig spatial distribution.

## Results

Wild pig occurrence data were collected from four independent data collection methods (stakeholder snowball sampling, expert interviews, bounty data, and rural telephone survey) and were used directly to develop maps of wild pig distribution across Canada using all datasets combined for three time periods, 1990–2000, 2001–2010, and 2011–2017 (Fig. 1). An additional four methods (GPS collars, citizen science photos, research trail camera networks, and media search) were used to evaluate benefits and limitations of methods and validate observations. A total of 1,489 occupied watersheds were identified, out of 37,578 watersheds in the study area (3.9% occupied by wild pigs), based on pooled results of all data collection methods across all years. The area of watersheds occupied in Canada has increased exponentially from 1990 to 2017 (Fig. 2). The large majority of the spatial expansion (92%) occurred in the three prairie provinces of Alberta, Saskatchewan, and Manitoba (Fig. 3). Indeed 58% of the national spread of wild pigs occurred within Saskatchewan. Wild pigs are also established in localized populations in British Columbia, Ontario, and Quebec. Of the ten provinces in Canada, only the four eastern provinces in Atlantic Canada (Newfoundland and Labrador, New Brunswick, Prince Edward Island, and Nova Scotia; 5% of Canada combined) have no confirmed sightings of wild pigs.

Based on analysis at the Level 9 watershed, the cumulative range of wild pigs in Canada is 777,783 km2. The average annual cumulative increase in wild pig range from the period 1990–2017 was 40,936 km2. The greatest increase in range expansion has occurred in the current time period (2011–2017), with an average annual cumulative increase of 88,094 km2.

As expected, the number of detected watersheds in Canada occupied by wild pigs was different among sampling methods (Table 1). There was an overall high consistency in the spatial distribution of occupied watersheds for each method. However, specific correspondence of individual occupied watersheds between data collection methods was low overall (<50%) for all methods (Table 2). Correspondence was strongest between the occupied watersheds determined using expert interviews and stakeholder snowball sampling methods (21% correspondence). The number of occupied watersheds was significantly and positively associated with the level of correspondence among all possible pairs of data collection methods (R2 = 0.90, df = 5, p < 0.001).

Overall mean response rate for the expert interviews was 51% (S.E. = 5.9). The mean number of completed districts within provinces was 55% (S.E. = 7.7). Total number of individual participants in the stakeholder snowball sampling method and rural telephone survey was 275 and 3,000 respectively. 272 detections were received from the bounty data. Response from all methods combined provided 95.5% coverage of the study area using the Canadian census sub-division as stratification units (Fig. 4).

### GPS collars

We captured 21 female and 17 male wild pigs by net-gun fired from a helicopter (n = 33) and through trapping using corral-style traps (n = 5) from 2015 to 2017 in southeastern and east-central Saskatchewan. After capture, each wild pig was physically immobilized and fitted with a Global Positioning System (GPS) tracking collar (Telonics, Mesa Arizona, USA). All collars were programmed to record a location every three hours and transmit the data via Iridium satellite link.

### Citizen science photos

Citizen science photos consisted of images from trail cameras, digital cameras, and cellular phones. Photos were provided by stakeholders and experts as supplementary information associated with wild pig locations when available.

### Trail camera network

We used images from a previous study (O’Brien et al., unpublished data) in our study design as an evaluation/validation method. Over a two-year period from 2011 to 2013 a network of 17 research trail cameras were deployed to capture wild pigs over an area of 275 km2 in east-central Saskatchewan (O’Brien et al., unpublished data).

### Media search

Media searches were conducted between 2014 and 2017. Searches were conducted in Google, media outlets, hunting forums, and social media and included key words such as “Eurasian wild boar”, “wild boar”, “feral boar”, “feral wild boar”, “feral pig”, and “wild pig”.

### Response coverage

Coverage by each method was determined using pre-defined provincial and national units that provided stratified and comparable coverage across the entire study area. Experts were stratified by provincial wildlife management zones, which are designated areas they manage wildlife and work within. The snowball stakeholder and bounty methods used provincial municipal administrative units and the telephone survey used national Forward Sortation Areas. All methods combined used the Canadian census sub-division units, as this stratification unit was consistent across the entire study area and provided a conservative unit size at the national scale with an average area of 1,195 km285.

### Wild pig occurrence mapping

We followed the MaNIS Georeferencing protocol (2001) and converted all wild pig locations into UTM coordinates using Google Maps (2018) and the Legal Land Description Converter (2017). Estimates of spatial accuracy of the data were based on information from the provider and using map scale. Two Level 9 (North America and Arctic) watershed shapefiles were downloaded from HydroSHEDS (2017) following the Pfafstetter coding system86. The 37,578 Level 9 watershed sub-basins in the study area have a mean area of 267 km287. Watershed units were chosen to model wild pig distribution as they are ecologically stratified units and allow for consistency across the landscape based on abiotic and biotic factors10,32,88. Level 9 watershed sub-basins are large enough to include the home range of at least one individual or sounder, which average 3.6 km2 for sows and 4.91 km2 for boar89,90,91,92, while encompassing considerable landscape heterogeneity10, The use of watershed units also maintains confidentiality of exact locations to ensure private land is not easily accessible to hunters, which was a condition of many landowners for providing information. Each wild pig occurrence was buffered by a radius of 10 km, a conservative estimate of annual home range movements and an area larger than the expected spatial error from any of the data collection methods. Watersheds that intersected a wild pig location were selected to create a watershed occurrence layer illustrating wild pig presence by watershed units. Two ten-year interval maps were created from 1990–2000 and 2001–2010. A current wild pig location map was created for the years 2011–2017.