Biol Invasions (2018) 20:1865–1880 https://doi.org/10.1007/s10530-018-1667-6 ORIGINAL PAPER Invasion ecology of wild pigs (Sus scrofa) in Florida, USA: the role of humans in the expansion and colonization of an invasive wild ungulate . . . Felipe A. Hernandez Brandon M. Parker Cortney L. Pylant . . . . Timothy J. Smyser Antoinette J. Piaggio Stacey L. Lance Michael P. Milleson James D. Austin Samantha M. Wisely Received: 27 April 2017 / Accepted: 13 January 2018 / Published online: 20 January 2018 The Author(s) 2018. This article is an open access publication Abstract Wild pigs (Sus scrofa) are the most widely sampled at 39 locations between 2014 and 2016. Our distributed invasive wild ungulate in the United States, data revealed the existence of genetically distinct yet the factors that inﬂuence wild pig dispersal and subpopulations (F = 0.1170, p \ 0.05). We found ST colonization at the regional level are poorly under- evidence of both ﬁne-scale subdivision among the stood. Our objective was to use a population genetic sampling locations, as well as evidence of long term approach to describe patterns of dispersal and colo- genetic isolation. Several locations exhibited signiﬁ- nization among populations to gain a greater under- cant admixture (interbreeding) suggesting frequent standing of the invasion process contributing to the mixing of individuals among locations; up to 14% of expansion of this species. We used 52 microsatellite animals were immigrants from other populations. This loci to produce individual genotypes for 482 swine pattern of admixture suggested successive rounds of human-assisted translocation and subsequent expan- sion across Florida. We also found evidence of Electronic supplementary material The online version of genetically distinct populations that were isolated this article (https://doi.org/10.1007/s10530-018-1667-6) con- tains supplementary material, which is available to authorized from nearby populations, suggesting recent users. F. A. Herna´ndez J. D. Austin S. M. Wisely (&) S. L. Lance School of Natural Resources and Environment, University Savannah River Ecology Laboratory, University of of Florida, 103 Black Hall, PO Box 116455, Gainesville, Georgia, PO Drawer E, Aiken, SC 29802, USA FL 32611, USA e-mail: wisely@uﬂ.edu M. P. Milleson United States Department of Agriculture, Animal and F. A. Herna´ndez B. M. Parker C. L. Pylant Plant Health Inspection Service, Wildlife Services, 2820 J. D. Austin S. M. Wisely East University Avenue, Gainesville, FL 32641, USA Department of Wildlife Ecology and Conservation, University of Florida, 110 Newins-Ziegler Hall, PO Box 110430, Gainesville, FL 32611, USA T. J. Smyser A. J. Piaggio United States Department of Agriculture, Animal and Plant Health Inspection Service, Wildlife Services, National Wildlife Research Center, 4101 LaPorte Avenue, Fort Collins, CO 80521, USA 123 1866 F. A. Herna´ndez et al. introduction by humans. In addition, proximity to wild the most economically costly livestock diseases in the pig holding facilities was associated with higher United States (Ca´rdenas-Canales et al. 2011). migration rates and admixture, likely due to the According to the Species Survival Commission of escape or release of animals. Taken together, these the World Conservation Union (IUCN), wild pigs (Sus results suggest that human-assisted movement plays a scrofa) are among the most ecologically destructive major role in the ecology and rapid population growth invasive species in the world (Lowe et al. 2000). of wild pigs in Florida. Multiple factors have contributed to the establishment of wild pig populations including deliberate releases Keywords Invasion ecology Sus scrofa Florida for hunting, the escape of individuals raised as Human-assisted movement Interbreeding livestock as a consequence of free-range practices, Immigration and the deliberate dumping of unwanted pets (e.g. Vietnamese pot-bellied pigs) (Mayer and Brisbin 2008; Caudell et al. 2013; Bevins et al. 2014). Since their ﬁrst introduction to the continental USA in the Introduction sixteenth-century by European explorers (Wood and Barret 1979), the species’ distribution and abundance Biological invasions are one of the most important have expanded dramatically. Although long-estab- factors contributing to the loss of biodiversity, degra- lished in the USA in regions of California, Texas and dation of ecosystems, and decline in ecosystem the Southeast, recent and rapid range expansion has services (Chapin et al. 1997; Sala et al. 2000; Pysek led to the establishment of wild pig populations in as and Richardson 2010). Understanding the pathways of many as 44 states (Hutton et al. 2006; Barrios-Garcia species introductions and range expansions informs and Ballari 2012; Bevins et al. 2014). The rapid wildlife and land management and can help mitigate expansion of wild pigs has been attributed to both or prevent further invasions (Hulme et al. 2008). Many intrinsic properties of the species (i.e. ability to adapt different processes contribute to the human-assisted to a variety of habitat types, omnivorous foraging introduction of exotic animals (Hulme et al. 2008; behavior, and high reproductive rates) and extrinsic Carpio et al. 2016), which include the unintentional causes (i.e. illegal transportation and release, frequent escape of managed animals (e.g. zoo mammals, escapes from farms and hunting preserves, the Cassey and Hogg 2015), the intentional/accidental propensity to thrive in human-altered landscapes, release of alien animals from managed environments and a lack of native predators) (Seward et al. 2004; (such as animals from fur farms (e.g. American mink Bevins et al. 2014). Regionally, wild pig abundance in (Neovison vison), Kidd et al. 2009), or unwanted pets Florida is second only to Texas with an estimated (e.g. domestic cats (Felis catus), Dickman 2009), and 500,000 to one million individuals in the state the intentional release of game species (e.g. roe deer (Giuliano 2010; FDACS 2016). (Capreolus capreolus), Randi 2005). The ﬁrst introduction of domestic swine in Florida There has been a long history of introductions of is believed to have occurred in the early 1500s when game species for the creation of hunting opportunities Spanish conquistadors arrived at Charlotte Harbor in (Yiming et al. 2006; Genovesi et al. 2012), but many Lee County, southwest Florida (Mayer and Brisbin of these species have proven to be damaging to the 2009). Through the early 1900s, European colonists function and health of native ecosystems. For exam- raised domestic swine in unfenced, semi-wild condi- ple, non-native browsers such as feral goats (Capra tions, with animals often becoming feral and expand- hircus), barbary sheep (Ammotragus lervia) and red ing across the broad central savannah and coastal areas deer (Cervus elaphus) negatively impact native plant of the state (Mayer and Brisbin 2008). Speciﬁcally, it communities, reduce vegetation densities and cause is believed that descendants of free-ranging domestic high levels of soil erosion (Wardle et al. 2001; swine maintained by homesteaders in the Kissimmee Acevedo et al. 2007). Other exotic game introductions, River Valley became a substantial component of the such as that of nilgai antelope (Boselaphus trago- wild pig populations established in Florida by the camelus) in Texas, have facilitated the spread of cattle 1980s (Mayer and Brisbin 2008). fever ticks, which transmit bovine babesiosis—one of 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1867 Currently, wild pig hunting is permitted in Florida. markers are a widely used molecular tool to infer From the 1940s through the 1970s, domestic pigs were population connectivity and dispersal among sampling allowed to range freely. Periodic introductions of pure locations, thus allowing a greater understanding of the Eurasian wild boar throughout Florida hybridized with location-speciﬁc ecology of this species (Vernesi et al. domestic and semi-feral swine to establish non-native 2003; Hampton et al. 2004; Nikolov et al. 2009; wild pig populations throughout the state (Mayer and Scandura et al. 2011). Previous population genetic Brisbin 2008; W. Frankenberger pers. comm.). In studies of wild pigs, largely conducted in Europe and addition, legal translocations were conducted to Oceania, have identiﬁed individual membership to restock state-controlled wildlife management areas. particular populations and levels of population admix- For example, from 1950 through the 1970s, approx- ture (i.e. interbreeding among isolated populations imately 3000 wild pigs were collected from various which produces offspring with a mixture of alleles state parks and other ecologically sensitive areas and from different ancestral populations) (Vernesi et al. relocated to wildlife management areas in Palm 2003; Hampton et al. 2004; Spencer and Hampton Beach, Glades and Collier/Monroe counties in south 2005; Nikolov et al. 2009; Scandura et al. 2011; Lopez Florida to establish or augment locally hunted popu- et al. 2014). Although these data will help inform lations (Belden and Frankenberger 1977; Mayer and population management and control efforts, little is Brisbin 2008). known about wild pig dispersal and expansion The Florida Department of Agriculture and Con- throughout North America. sumer Services (FDACS) authorizes registered dealers The goal of this study was to use population genetic to capture wild pigs on federal, state, municipal or techniques to describe movement patterns of wild pigs private lands, and transport them to transitory holding and to identify the potential factors that may inﬂuence facilities, prior to being sold for meat or released at their dispersal across Florida. We hypothesized that private game preserves for hunting (Gioeli and wild pigs would exhibit genetic population structure Huffman 2012; FDACS 2016). During the course of consistent with both historic and contemporary pat- this study, approximately 400 transitory holding terns of human-assisted introductions. Speciﬁcally, in facilities were registered by FDACS in Florida. the Kissimmee Valley region, where populations have Despite current state regulations, animals can escape been long established, we expected to ﬁnd signiﬁcant from holding facilities, or alternatively, they can be levels of both interbreeding and immigration among illegally transported by recreational hunters and wild pig populations, consistent with a long history of landowners over large distances and introduced to natural and human-assisted movement in the valley hunting areas without documentation of the move- and surrounding regions. If recent human-assisted ment. The willingness of people to translocate wild introductions from outside the Kissimmee Valley were pigs has facilitated range expansion of this species in occurring, we would expect to ﬁnd pockets of Florida and other states in the southern USA (Seward genetically distinct populations with limited genetic et al. 2004; Mayer and Brisbin 2009; Bevins et al. exchange with other nearby populations. Finally, 2014; W. Frankenberger pers. comm.). Illegal intro- because both escapes from holding facilities and ductions represent a growing concern because of the intentional release at wildlife management areas have impacts that wild pigs have on biodiversity, agricul- been identiﬁed as a source of introductions in the ture production, and animal and human health (Crooks southeastern USA (Seward et al. 2004; Mayer and Brisbin 2009; Bevins et al. 2014; W. Frankenberger 2002; Hone 2002; Bankovich et al. 2016). Population genetic analysis can provide informa- pers. comm.), we hypothesized that populations near tion regarding patterns of connectivity and interbreed- animal holding facilities and at wildlife management ing among populations and can be useful for areas would support higher frequencies of interbred differentiating natural patterns of animal dispersal wild pigs and genetic immigrants than other sites from human-assisted translocations. Microsatellite around Florida. December 2016, Gainesville, Florida (U.S.). 2 3 December 2016, Gainesville, Florida (U.S.). December 2016, Gainesville, Florida (U.S.). 123 1868 F. A. Herna´ndez et al. Materials and methods wildlife management areas, military bases, and private properties. We recorded demographic data for each Sample collection of wild pig tissue animal, which included sex, age, and sampling location. Speciﬁcally, we used body size, reproductive From January 2014 to March 2016, we collected tissue traits, and tooth eruption patterns (Matschke 1967)to samples from 482 wild pigs at 39 sites across the state classify animals as adults (C 1 yr), subadults (2 mo– of Florida, USA (Fig. 1). We sampled animals oppor- 1 yr), or juveniles (\ 2 mo). From 431 animals, we tunistically as part of a national wild pig disease collected whole blood (0.5 ml) by cardiac puncture or monitoring effort led by the United States Department orbital draw and stored the sample immediately in of Agriculture, Animal Plant and Health Inspection, 1 ml mammalian lysis buffer (Qiagen, Valencia, CA, Wildlife Services, National Wildlife Disease Program USA). We stored blood samples on ice packs and then (NWDP). We acquired genetic samples from wild pigs refrigerated at 4 C. From 51 animals, we collected that were trapped and euthanized during animal hair, which was stored in paper envelopes in the ﬁeld. control efforts conducted throughout the study period Both whole blood and hair samples were transported by state or federal agencies. Additionally, we collected to the University of Florida and stored at - 80 C until samples at check-stations from animals that were DNA could be extracted. This study was approved by legally harvested by hunters on federal and state Fig. 1 Sample size of wild pigs (Sus scrofa) collected per site through the state of Florida (U.S.) (2014–2016) 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1869 University of Florida’s Institutional Animal Care and for 20–30 cycles with a ﬁnal elongation at 72 C for Use Committee. 40 min. We analyzed PCR products by capillary electrophoresis on an ABI 3130xl Genetic Analyzer DNA isolation and microsatellite genotyping (Applied Biosystems, Foster City, CA, USA) and scored fragments using GeneMarker version 2.6.2 We extracted DNA from blood using the Qiagen (SoftGenetics, State College, PA, USA) at the Univer- DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA, sity of Florida. USA) and from hair using the QIAamp DNA Micro Kit (Qiagen, Valencia, CA, USA). For both proce- Validation of genotypes and calculation dures, we followed the manufacturer’s protocol, with of genotyping error slight modiﬁcations to increase DNA yields including vigorously mixing blood samples prior to extraction, We attempted to re-amplify loci that were initially increasing the amount of starting material (i.e. 200 ll unsuccessful; however, if subsequent efforts failed for blood and 1–21 collected hair follicles), using (i.e. second and third attempts), these genotypes were 20 ll 1M DTT to increase hair tissue digestion, and a categorized as missing data for the sake of analysis. To longer incubation period prior to ﬁnal DNA elution assess genotyping error and allelic dropout, 52 blood (i.e. up to 15 min with shaking). We quantiﬁed the samples (i.e. approximately 12% of the dataset) were chosen at random and re-genotyped. We then com- concentrations of recovered nucleic acids using the Epoch Microplate Spectrophotometer running the pared the 52 duplicated genotypes to the originals, and Gen5 software, version 2.09 (BioTek Instruments, any discrepancies were reconciled by conducting a Inc., Winooski, VT, USA). We stored isolated DNA at third genotyping run. Six markers (S0215, Susc18, - 20 C. S0005, CGA, SW1680, SW13) exhibited C 5% geno- Sixty-one microsatellite markers were initially type error and were removed from the ﬁnal dataset. selected for multilocus genotyping, 42 of which were Additionally, we eliminated two loci that exhibited previously described (Ellegren et al. 1993; Robic et al. high ampliﬁcation failure ([ 20%) and one monomor- 1994; Alexander et al. 1996; Rohrer et al. 1996) and 19 phic locus from the ﬁnal dataset (SW1816, S0090, novel markers that were designed and contributed by Susc11). Ultimately, we considered 52 loci in the ﬁnal us (Online Resource 1). We screened markers and dataset. arranged loci into multiplexes using the program Considering that locus ampliﬁcation and genotyp- Multiplex Manager, version 1.2 (Holleley and Geerts ing error rates potentially may be affected by using 2009) based on their primer annealing temperatures different tissue types (blood and hair) that yield and the likelihood of primer-product hybridization. different quality and quantity of DNA (e.g. noninva- Ultimately, 52 markers were either polymorphic or sive samples, such as hair, have been shown to have were successfully ampliﬁed to produce fragment higher allelic dropout because of lower quantity and peaks with a clear topology (see next subsection). quality of DNA recovered relative to other tissue We performed multiplex PCRs in 15 ll reaction types, Bonin et al. 2004), we conducted an indepen- volumes using the Qiagen Type-it Microsatellite dent validation study from parallel kidney and hair PCR Kit (Qiagen, Valencia, CA, USA) as follows: samples collected from an additional 34 wild pigs. 1X master mix, 0.2 lM 10X primer mix (see Online Speciﬁcally, kidney samples were collected from fresh Resource 2 for optimized primer concentrations), 0.5X carcasses, placed into a cooler in the ﬁeld, and then Q-solution additive, 3.5 ll sterile water, and 25 ng shipped on ice packs overnight to the NWDP for template DNA. We used touchdown PCR protocols to processing. We stored both kidney and hair samples at reduce the occurrence of non-speciﬁc ampliﬁcation - 20 C until DNA could be extracted. For each with the following protocol: initial denaturation at kidney sample, we extracted DNA independently in 95 C for 15 min, followed by cycling at 95 C for triplicate using a Qiagen DNeasy Blood and Tissue Kit 30 s, annealing for 30 s with a 0.5 C decrease with (Qiagen, Valencia, CA, USA), following the manu- each subsequent cycle to reach optimum annealing facturer’s recommended protocol. Similarly, we temperature, and elongation at 72 C for 30 s (see extracted DNA from hair follicles in triplicate using Online Resource 2 for speciﬁc starting temperatures) QIAamp DNA Micro Kit (Qiagen, Valencia, CA, 123 1870 F. A. Herna´ndez et al. USA) with 15 follicles used for each independent Monte Carlo permutations, and linkage disequilibrium extraction. We modiﬁed Qiagen’s recommended (LD) (i.e. signiﬁcant correlation of alleles at different extraction protocol by disrupting follicle samples loci) using the R package pegas version 0.9 (Paradis immediately prior to incubation by vibrating samples 2010). We adjusted signiﬁcance levels for multiple with a TissueLyser LT (Qiagen, Valencia, CA, USA) tests of HWE and LD using sequential Bonferroni at 30 Hz for 6 min with sterile stainless steel 5 mm corrections (Holm 1979; Rice 1989)in R. bead as recommended by Smith et al. (2011). We quantiﬁed the quality and quantity of DNA extracted Estimating population genetic structure from both kidney and hair samples with a Nanodrop (Thermo Fisher Scientiﬁc, Waltham, MA, USA) and We characterized wild pig population genetic struc- diluted extractions to 10 ng/ll for PCR ampliﬁcation. ture to infer historical and contemporary patterns of Extraction replicates with an elution concentra- animal introduction and dispersal throughout Florida. tion \ 10 ng/ll were re-extracted. Each replicate of We calculated F-statistics to examine the hierarchical kidney and hair DNA was ampliﬁed and genotyped partitioning of inbreeding within sampling locations using the same multiplex PCRs and fragment analysis (F ), relative to the inbreeding that can be explained IS conditions described above. We compared the geno- by drift among different sampling locations (F ), and ST types derived from triplicate DNA extractions from the individual inbreeding relative to the total popula- kidney (n = 3 9 34) and hair (n = 3 9 34) samples tion (F ) (Wright 1951, 1965). The statistical signif- IT to validated multilocus genotypes and quantify geno- icance of F-statistics was tested using 999 typing error between putatively high quality (kidney) permutations using the G-statistic Monte Carlo test and putatively low quality (hair) DNA sources. implemented in the R package hierfstat 0.04–26 Genotypes were assigned with GeneMapper 4.0 (Goudet 2005). We calculated pairwise F values ST (Applied Biosystems) and we analyzed genotypes (Weir and Cockerham 1984) among all sampling with software package ConGenR (Lonsinger and locations, and their statistical signiﬁcance determined Waits 2015) in R version 3.3.1 (R Core Team 2016) by 999 permutations, using GenAlEx version 6.5 to identify allelic dropout and false alleles among the 6 (Peakall and Smouse 2012). (3 kidney, 3 hair) replicates. Before quantifying migration into and out of populations, we evaluated the level of genetic clus- Estimating genetic diversity tering (i.e., the assortment of genotypes into distinct genetic clusters) using two different Bayesian cluster- We calculated descriptive statistics of basic measures ing methods implemented in STRUCTURE version of genetic diversity to assess sampling bias, population 2.3.4 (Pritchard et al. 2000; Falush et al. 2003) and structure, and the robustness of molecular marker data BAPS version 6.0 (Corander and Marttinen 2006; of wild pigs across sampling locations. For all genetic Corander et al. 2008). Both methods assign individuals analyses, we only considered genotypic data from to clusters (K) by minimizing deviations from HWE sampling sites with C 5 individuals (n = 454 ani- and linkage disequilibrium. STRUCTURE derives a mals). To describe locus polymorphism, we calculated posterior probability for each K examined across the number of alleles (N ), observed heterozygosity multiple Markov Chain Monte Carlo (MCMC) repli- (H ) and expected heterozygosity (H ) using GenAlex cates (Pritchard et al. 2000; Falush et al. 2003). BAPS o e version 6.5 (Peakall and Smouse 2012). We calculated employs a greedy stochastic optimization method to the mean allelic richness per sampling location (A ,El search for the most probably number of clusters Mousadik and Petit 1996), corrected for the smallest (Corander et al. 2008). sample size, using the R package PopGenReport To assess genetic clustering of individuals, we version 2.2.2 (Gruber and Adamack 2015). We tested the likelihood of K = 1–25 clusters using 20 evaluated deviation from Hardy–Weinberg equilib- replications at each K using the program STRUC- rium (HWE) (i.e. derived from the comparison TURE. Because of the long history of human-assisted between observed and expected heterozygosity) at introductions and high natural dispersal capabilities of each locus across the entire dataset and per each wild pigs, we assumed an admixture ancestry model sampling location using an exact test based on 100,000 and correlated allele frequencies without including 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1871 sampling location for each individual. We established continuous dispersal. When isolation by distance is not 100,000 iterations for the burn-in period (i.e. simula- observed, it is inferred that factors other than proxim- tion run previous to data collection to minimize the ity, such as human-assisted movement, shape the effect of the starting conﬁguration) and 100,000 population movement patterns. Using data from iterations post burn-in (i.e. simulations run after the individuals from all sampling locations, we ran a burn-in to obtain parameter estimates). We compared Mantel test with 10,000 permutations implemented in likelihood values across replicates for each value of the R package ade4 version 1.7-4 (Dray and Dufour K and calculated DK, a statistic based on the rate of 2007). Geographical distances among sampling loca- change in log-likelihood of the data (Evanno et al. tions were calculated from either the geometric 2005), with STRUCTURE HARVESTER version centroid of hunting check-stations or the mean center 0.6.94 (Earl and vonHoldt 2012) to estimate the point of trap clusters (depending on collection optimal value for K. DK has been shown to identify method). only the uppermost hierarchical level of genetic structure (Evanno et al. 2005). Further, the utility of Estimating migration rates and admixed DK to accurately identify the genetic structure is individuals limited by unequal sample sizes, as is the case here (Puechmaille 2016). Thus, we also used the suite of We assigned individuals as recent immigrants into a metrics developed by Puechmaille (2016) (i.e. cor- population or as nonimmigrants. These individual rected PP, MedMeaK, MaxMeaK, MedMedK and assessments were derived from population migration MaxMedK) to infer the number of genetic clusters rates within each sampling location using BAYESASS present within our dataset. version 3.0 (Wilson and Rannala 2003). We ran To assess the robustness of our genetic clusters, we 100,000,000 MCMC iterations following a conducted a Bayesian mixture-clustering analysis 10,000,000 burn-in period, and we used a sampling among individuals without considering sampling interval of 500 steps. We tested multiple delta values location (i.e. the inclusion of sampling locations did for the mixing parameters of migration rates, allele not generate different clustering results in preliminary frequencies and inbreeding values. Delta values set to analyses) using the program BAPS. Initially, we ran 1 resulted in optimal acceptance rates for changes to the program with 5 replications of K = 1–25 and each mixing parameter (between 20 and 60%). We subsequently, we conducted 20 replications on the conducted multiple runs initialized with different best-visited K values with highest likelihood random seeds and compared the posterior mean (K = 15–17). parameter estimates for convergence. We calculated To better visualize our clustering approaches to 95% credible intervals (CI) for pairwise migration rate population assignment, we used a Discriminant Anal- estimates between sampling locations, considering ysis of Principal Components (DAPC, Jombart et al. credible intervals that did not include zero to be 2010) using the R package adegenet version 2.0.1 statistically signiﬁcant. Finally, we compared the (Jombart and Ahmed 2011). DAPC is a multivariate signiﬁcant rates of recent (ﬁrst and second-generation) approach that does not make any assumption about descendants of migrants with their corresponding HWE or linkage equilibrium, maximizing the among- genetic cluster assignment to evaluate congruence population variation and minimizing the variation among all the conducted analyses. within predeﬁned groups (Jombart et al. 2010). The To determine whether interbreeding among indi- optimal value of K was determined based on Bayesian viduals from isolated clusters was occurring, we Information Criterion (BIC) scores. estimated individual genetic admixture using BAPS. We tested the relationship between genetic differ- This measure assessed whether an individual had the entiation (F /1 - F ) and geographic (Euclidean) signature of one or more distinct genetic clusters. ST ST distance (km) to assess patterns of isolation-by- Once the most likely K value was determined, distance. This hypothesis posits that the regular admixture inference was conducted using 100 simu- increase in genetic differentiation among individuals lations from posterior allele frequencies. We com- (or populations) is positively correlated with geo- pared the mean posterior proportion of each graphic distance due to geographically limited but individual’s ancestry (admixture coefﬁcient) relating 123 1872 F. A. Herna´ndez et al. Results to each estimated K (clusters with C 5 individuals). The admixture coefﬁcient was bound between 0 and 1; Genetic diversity and tissue validation animals with a coefﬁcient closer to 1 had a less admixed ancestry than coefﬁcient estimates closer to The 52 microsatellite loci in the ﬁnal dataset produced 0. Statistical signiﬁcance was set at a = 0.05 to determine whether individuals had evidence of admix- a genotyping error rate (both allelic dropout and false alleles) of 0.7% (21 genotyping discrepancies/2704 ture. The p value reﬂected the proportion of simulated individuals (n = 200) from the cluster to which one scored genotypes) across the 52 replicate genotypes speciﬁc individual was originally assigned that had from blood samples. From the comparison of geno- admixture coefﬁcient B to that speciﬁc individual. types from paired triplicate hair follicle and kidney samples, we obtained a genotyping error of 1.7% (196 genotyping discrepancies/11,628 scored loci) indicat- Predictors of wild pig admixture and migration rates ing that we were able to generate robust genotypes from hair samples. Thus, for subsequent analyses, we We tested whether human-related land use practices treated genotypes generated from hair follicles as equivalent to genotypes generated from blood. by hunters and trappers were related to the probability that an individual was (1) a genetic mixture of two Allelic diversity across loci ranged from 6 (locus SW174) to 42 (locus SW856) alleles per locus (Online genetically distinct populations (admixture) or was a recent immigrant (as inferred from the population Resource 2). Within sampling locations, we detected migration rate). Admixed offspring or recent immi- an average of 5.1 alleles per locus with a mean allelic grants would be present if individuals from one richness of 3.35 (when resampling individuals within genetically distinct population migrated (either natu- site to an n = 5) (Table 1). H and H values across o e loci ranged from 0.163 (locus Susc34) to 0.866 (locus rally or with human assistance) to another area and mated with individuals from a different and geneti- SW1067), and from 0.352 (locus S0227) to 0.942 (locus SW856), respectively. Although we found cally distinct population. Land use factors were categorical (i.e. public signiﬁcant differences between H and H , the differ- o e ence was small (Online Resource 2). Mean H and H hunting = 1 and no public hunting = 0) and contin- o e uous (i.e. geographic distance to the nearest wild pig were 0.626 and 0.616, respectively, across all the holding facility) variables. Individual traits were age locations (Table 1). Of the 52 loci, we detected (i.e. juvenile, sub-adult, or adult) and sex. We used signiﬁcant deviations from Hardy–Weinberg equilib- generalized linear regression models within an Akaike rium (HWE) at 46 loci (Bonferroni adjusted p\ 0.05) information criterion (AIC) framework for model across the entire dataset. Forty-three of 46 markers comparison (Burnham and Anderson 2002) to test for with HWE deviations exhibited a deﬁcit of heterozy- gotes (H \ H ) (Online Resource 2). Evidence of relationships between predictor (i.e. human-related o e land use and individual age and sex) and response (i.e. linkage disequilibrium (LD) between genotyped loci was demonstrated for 63% (830/1326) of pairwise loci probabilities of admixture and migration) variables. We used the R package MuMIn version 1.15.6 (Barton comparisons (Bonferroni adjusted p\ 0.05). The relatively high HWE deviations and LD likely due to 2015) to ﬁt the global multinomial model and all additive subsets, and to calculate model-averaged the genetic population structure we observed across regression coefﬁcients, 95% conﬁdence intervals and sampling locations (see next subsection). We detected cumulative AIC weight of evidence of each predictor signiﬁcant deviations from HWE at 13 out of the 1508 variable (Burnham and Anderson 2002). Prior to their tests conducted (Bonferroni adjusted p\ 0.05) across inclusion of predictor variables in the models, we all sampling locations. A maximum of three deviations tested all predictor variables for collinearity using from HWE were detected per location, and one marker (locus Susc2) exhibited HWE deviation in six out of Pearson correlation coefﬁcients. We conducted statis- tical tests in R using a = 0.05 for determination of 29 sampling locations (Table 1). statistical signiﬁcance. 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1873 Table 1 Summary of genetic diversity of 454 wild pigs across related individuals (within family units called soun- the 29 sampling locations; number of individuals per sampling ders) being sampled within a site. This pattern also location (N), average number of alleles per locus (N ), mean likely inﬂuenced the ﬁnding of a signiﬁcant LD and allelic richness per sampling location (A ), observed (H ) and R o heterozygote deﬁcit. expected (H ) heterozygosities, and deviation from the Hardy– Weinberg equilibrium (H-W Bonferroni corrected p value) All pairwise F values estimated between sam- ST given as number of signiﬁcant values per location pling locations were signiﬁcantly different from zero (p\ 0.01), which indicated genetic differentiation Location N N A H H H-W p value A R o e among sampling locations. F values ranged from ST 1 31 7.058 3.779 0.676 0.678 0.020 (between locations 14 and 22) to 0.256 (between 2 6 3.442 2.921 0.598 0.563 locations 3 and 10). Fourteen of 29 sampling sites 3 21 3.654 2.701 0.521 0.523 showed moderate level of genetic differentiation (all 4 10 2.808 2.448 0.539 0.498 F values [ 0.05) compared to the rest of sampling ST 5 6 4.058 3.185 0.599 0.585 sites (Online Resource 3). A Mantel test did not reveal 6 21 6.462 3.722 0.636 0.671 2 a signiﬁcant correlation between genetic and geo- 7 6 4.750 3.608 0.708 0.639 graphic distances across the sampling locations 8 12 5.500 3.554 0.543 0.659 1 (r = 0.081, p [ 0.05) (Online Resource 4), suggesting 9 31 5.692 3.015 0.575 0.589 1 that the patterns of genetic differentiation could not be 10 7 2.269 2.102 0.473 0.391 explained by Euclidian distance. 11 7 4.808 3.554 0.647 0.641 STRUCTURE analyses revealed a peak in the mean 12 10 4.712 3.350 0.679 0.615 posterior probabilities L(K)at K = 21 accompanied 13 17 4.269 3.051 0.643 0.599 by the lowest variance (L(K) =- 70,901.035 ± 14 53 8.558 3.961 0.654 0.697 3 204.443) among replicates (Online Resource 5). We 15 7 4.192 3.207 0.632 0.591 detected the highest DK peak at K = 2(DK = 9.9906), and a second highest DK peak at K = 21 16 20 7.269 4.012 0.689 0.711 17 9 4.654 3.335 0.645 0.608 (DK = 2.1179) (Online Resource 5). Evaluation of the STRUCTURE results with the statistics introduced by 18 17 5.173 3.381 0.634 0.630 Puechmaille (2016) provided support for a range of 19 10 4.904 3.445 0.637 0.645 K values varying by metric (corrected PP = 20; 20 13 4.692 3.191 0.631 0.594 MedMeaK = 21, MaxMeaK = 23, MedMedK = 22 21 6 3.865 3.090 0.608 0.577 and MaxMedK = 23). We chose to interpret K = 21 22 45 8.788 4.076 0.707 0.721 3 given that was included within the distribution of 23 5 4.731 3.737 0.668 0.644 Puechmaille’s statistics and coincided with the 24 20 5.788 3.521 0.675 0.654 L(K) peak with the lowest variance according to 25 18 6.192 3.716 0.658 0.669 STRUCTURE. Thus, the mean membership coefﬁ- 26 5 3.692 3.098 0.536 0.579 cient (Q) of each sampling location to the inferred 27 15 6.308 3.820 0.654 0.679 2 clusters was divided into 21 distinct groups, consid- 28 13 4.808 3.284 0.666 0.614 ering locations with Q C 0.5 in any inferred cluster. 29 13 4.731 3.191 0.621 0.594 1 Mean membership coefﬁcient per location ranged Mean 5.097 3.347 0.626 0.616 from 0.546 to 0.976 for individuals across the 21 inferred clusters (Online Resource 6). One distinct cluster included all the animals from location 3 in northwest Florida. We found other clusters where the Genetic population structure majority of individuals comprising each cluster were from one sampling location, such as location 1 in the The overall F-statistics resulted in signiﬁcant values for F = 0.0281, F = 0.1419, and F = 0.1170 northwest; locations 2 and 4 in the northcentral; IS IT ST locations 6, 8, 9, 10, 13, 19 and 21 in the northeast; (G-statistic = 38,470, p = 0.001). The small but signiﬁcant F value indicated low levels of inbreeding locations 15, 18, 20, 24, 25, 26 and 29 in the IS southwest; and location 28 in the south. Two other within sampling locations, which was likely driven by clusters were composed of individuals from two 123 1874 F. A. Herna´ndez et al. sampling locations in each cluster, such as locations 5 assigned to a speciﬁc cluster (Online Resource 6, (northeast) and 17 (southwest), and locations 12 Fig. 2a). (northeast) and 16 (southwest), respectively (Fig. 3a). Mixture-clustering analyses in BAPS resulted in a The rest of the six locations (7, 11, 14, 22, 23 and 27) probability of [ 0.999 (log (ml) of optimal partition: had Q \ 0.5 in any inferred cluster and were not - 80159.3593) of there being K = 16 genetic clusters of wild pigs in the study area (Fig. 2b). Other cluster Fig. 2 Geographic location and number of genetic clusters of individuals to one of 16 genetic clusters (left) and mixture (K) inferred by three statistical methods across the 29 sampling clustering output (right), and c DAPC: sampling sites colored locations of wild pigs in Florida. a STRUCTURE (corrected by according to the predominant assignment of individuals to one Puechmaille’s statistics): sampling sites colored according to of 5 genetic clusters (above) and projection of clusters in the predominant assignment of individuals to one of 21 genetic discriminant space using the ﬁrst two principal components clusters (left) and Bayesian clustering output (right) (sites not (proportion of variance conserved by PCA principal compo- assigned to a speciﬁc cluster are colored in white), b BAPS: nents = 0.932) (below) sampling sites colored according to the predominant assignment 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1875 partitions, such as K = 15 and K = 14, had lower log the descendant of recent migrants. Except for two (ml) values (- 80188.3407 and - 80224.8726, locations, all the sites with signiﬁcant migration into respectively). BAPS analyses were able to detect a or out of location 22 were assigned to the major similar ﬁne-scale population structuring as STRUC- genetic cluster inferred by the BAPS analysis. Six out TURE. BAPS identiﬁed 13 clusters where the majority of 16 locations with signiﬁcant migration into or out of of individuals comprising each cluster were from one location 22 corresponded to the locations that were not sampling location (i.e. locations 1, 2, 3, 4, 6, 9, 10, 13, assigned to any speciﬁc genetic cluster by STRUC- 18, 20, 21, 28 and 29). One major cluster was TURE, after calculation of Puechmaille’s statistics. comprised of animals from the rest of the 16 locations Mean posterior proportion of each individual’s (Online Resource 7). Two clusters only included two ancestry showed that 6.2% (28/450) of wild pigs had individuals in each cluster (from locations 26 and 27, signiﬁcant evidence (p \ 0.05) of genetic admixture respectively), and they were not included in the (i.e. mixture of alleles from different ancestral popu- admixture analysis. lations due to interbreeding events) across 14 out of 16 DAPC analyses suggested an ‘optimal’ value of inferred clusters, and 75% (21/28) of admixed indi- K = 5 (i.e. lower BIC value at K = 5); but the viduals were assigned to the major cluster. Individual relatively ﬂat pattern of the elbow in the curve wild pigs from other genetic clusters were not (representing the relationship between BIC and num- signiﬁcantly admixed, and thus, the majority of their genome was related to one particular ancestral ber of clusters) suggested that values of K ranging from 3 to 10 may also represent ‘optimal’ number of population. clusters summarizing the observed genetic structure of wild pigs. Considering a K = 5 as the most probable Predictors of wild pig admixture and migration number of clusters, animals sampled at locations 3 (K5), 9 and 10 (K2), and 13 (K3) seemed to be A total of 390 transitory holding facilities were genetically distinct from individuals sampled at all the identiﬁed and located throughout the state of Florida. rest of 25 locations (divided in both K1 and K4) Proximity to the nearest wild pig holding facility (Online Resource 8, Fig. 2c). (range: 2–40 km) was the only variable that signiﬁ- Deviations from genetic equilibrium were likely a cantly predicted both the probability of admixture and product of biological processes and not null alleles. individual migration patterns among all top models. We reran STRUCTURE without the four loci that had The best-ranked AIC model (Log-lik =- 148.26, the highest deviations from HWE (loci Susc2, Susc15, AIC = 302.51) predicting wild pig admixture only Susc34, and Susc20), and found no effect on the included distance to nearest holding facility as a clustering results. Thus, we left the loci in all of our predictor variable. This model presented the highest analyses. AIC weight of evidence (w = 0.32) and the cumula- tive AIC weight of evidence of the predictor variable Migration and ancestry analysis was 0.72 across the four best-ranked candidate models (Online Resource 9). Probability of wild pig admixture Analysis of gene ﬂow patterns revealed low and was higher in wild pigs collected in sites near holding statistically insigniﬁcant migration rates for the facilities (b = 0.0048, 95% CI 0.0011, 0.0085, majority of sampling locations (i.e. mean migration p = 0.0105), i.e. the closer the proximity to a holding rates for which the 95% CI included the zero). facility the lower the ancestry coefﬁcient and the However, we found statistically signiﬁcant migration higher the individual admixture. rates between one particular ‘core’ site (location 22 in The best-ranked AIC model (Log-lik =- 269, the southwest) and 16 other sampling sites throughout AIC = 546) predicting wild pig migration included Florida (ranging from 3 to 14% immigrants between distance to nearest holding facility and sex as predictor sites, Fig. 3). For other locations that had signiﬁcant variables. This model presented the highest AIC migration rates with the core location, 84.2% (16/19) weight of evidence (w = 0.19) and the cumulative of these animals exhibited a probability [ 0.9 to be AIC weights of evidence for both predictor variables either ﬁrst or second-generation migrants which were 1 (distance to nearest holding facility) and 0.56 suggested that these animals were recent migrants or (sex), respectively, across the eight best-ranked 123 1876 F. A. Herna´ndez et al. Fig. 3 Signiﬁcant migration rates between one particular ‘core’ site (location 22, red circle) in the Kissimmee Valley and 16 other surrounding locations (green circles). Entire lines denote 3–6.7% immigrants between sites, and dashed lines denote 7.4–14% immigrants between sites candidate models (Online Resource 9). Probability of the global multinomial model (all Pearson correlation an individual being a ﬁrst/second-generation migrant coefﬁcients with p [ 0.05). signiﬁcantly increased with the proximity of the sampling site to animal holding facilities (b =- 0.0106, 95% CI - 0.0157, - 0.0059, Discussion p \ 0.001), but sex was not signiﬁcantly related to individual migration (b =- 0.0684, 95% CI The genetic patterns of wild pigs observed in Florida - 0.1521, 0.0174, p = 0.113). support the hypothesis that ongoing human-assisted Although candidate models including the rest of movement is a source of their introduction and predictor variables exhibited DAIC \ 2 and similar dispersal throughout the state. We found evidence of weight of evidence compared to the best-ranked AIC multiple unique genetic groupings, and patterns of models, neither public hunting nor individual covari- both admixture and isolation that are not easily ates of age or sex were signiﬁcantly related to explained by natural dispersal. The lack of isolation admixture or migration patterns across all the candi- by distance signal suggests that patterns of dispersal date generalized linear regression models (i.e. all b are driven by processes other than geographic prox- coefﬁcients with p [ 0.05). No correlation was imity as would be expected under a stepping-stone detected between the predictor variables included in model of gene ﬂow. We suggest that human-assisted 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1877 movement at least partially explains the observed opportunities (Vernesi et al. 2003; Spencer and pattern, aligning with a previous population genetic Hampton 2005; Scandura et al. 2011; Lopez et al. study of wild pigs in California (Tabak et al. 2017). 2014). We demonstrated that locations proximal to wild Human-assisted movement is a likely explanation pig holding facilities were associated with a higher for the signal of admixed, recent immigrants detected probability of (1) individuals with a mix of genetic between a particular ‘core’ site in southwest Florida signatures from two or more genetically distinct and the other 16 sites that were mainly in one cluster, populations, and (2) ﬁrst or second-generation immi- yet distributed across the state. Several anecdotal grant individuals. Human-assisted movement also reports suggest that trappers have successively intro- explains the high migration rates in populations near duced up to several thousand wild pigs per year (from the holding facilities. Our results suggest that holding 2000 through 2008) into a private hunting club on the facilities may act as foci for genetic exchange within northern border of the ‘core’ site (both sites were landscapes through multiple potential routes, such as located on the border of Polk and Highlands Counties). escapes from the facility, escapes during animal These animals could have been trapped at multiple transport, escapes during transfer from dealers to unidentiﬁed preserves and parks across northeast and holding facility, and/or deliberate releases. These southwest Florida (W. Frankenberger pers. comm. ), arguments support previous research that speculates creating stocks from different genetic sources in the that the inﬂuence of animal escapes from farms and hunting club. Intensive and prolonged hunting pres- hunting preserves (Bratton 1975), and illegal transport sures may expand the movement ranges of wild pigs and release (Waithman et al. 1999; Zivin et al. 2000)is from the hunting club to the ‘core’ site (Choquenot responsible for increasing the range expansion and et al. 1996; Mayer and Brisbin 2009), likely resulting population densities of wild pigs across other states in in the admixture and production of F1/F2 individuals U.S. from the mating between animals from the ‘core’ site Three genetic clusters associated with unique and other source populations (up to 84% of admixed locations were consistently inferred by each clustering individuals were ﬁrst/second immigrants from another method. Recent wild pig introductions from multiple population). Ultimately the emerging picture of wild genetic sources may explain the existence of these pigs in the Kissimmee Valley region of Florida is a three distinct genetic clusters that were genetically long and continuous history of movement, both natural distinct from wild pigs found elsewhere in Florida. and human-assisted within the valley. One cluster was composed of animals sampled on an The small but signiﬁcant inbreeding (as measured island located in Franklin County (northwest Florida) by F ) we detected across all populations can be IS where purebred domestic Brown Russian and Poland explained by an interaction between wild pigs breed- China swine were introduced in the early 1940 s to ing strategy/social structure, and our collection restock hunted wild pigs on the island. Since the last scheme. Aggregation of related individuals in family introduction, the insular population was assumed to be groups, high levels of female philopatry, and a few disconnected from other wild pigs inhabiting the polygynous males siring the next generations, con- mainland (Mayer and Brisbin 2008), and these data tribute to increase the genetic similarities among suggest that, indeed, no migration to the island from individuals of the same group (Gabor et al. 1999; the mainland has occurred. The other two clusters Hampton et al. 2004; Kaminski et al. 2005; Poteaux were composed of animals sampled at locations from et al. 2009). Considering our collection scheme of wild Lake and Orange counties, respectively (northeast pigs, where multiple individuals were often harvested Florida). No ofﬁcial records exist about wild pig or trapped simultaneously in sampling locations, we translocations into these particular sites, but the likely genotyped related individuals belonging to the genetic uniqueness compared to wild pigs from other same family group, increasing the estimation of surrounding locations suggests that animals at these inbreeding with subpopulation (F ). The large num- IS sites were recently introduced. This introduction ber of loci with small deviations in HWE is likely due would likely be associated with unreported/illegal transport and release to increase local hunting December 2016, Gainesville, Florida (U.S.). 123 1878 F. A. Herna´ndez et al. References to the large number of admixed and migrant individ- uals in our samples. When individuals are the product Acevedo P, Cassinello J, Hortal J, Gorta´zar C (2007) Invasive of parents from genetically distinct populations this exotic aoudad (Ammotragus lervia) as a major threat to produces linkage disequilibrium, which by deﬁnition native Iberian ibex (Capra pyrenaica): a habitat suitability produces deviations in HWE. model approach. Divers Distrib 13:587–597 As a whole, our study contributes novel insights Alexander LJ, Troyer DL, Rohrer GA, Smith TPL, Schook LB, Beattie CW (1996) Physical assignments of 68 porcine regarding the role of human-assisted movement in the cosmid and lambda clones containing polymorphic maintenance and spread of wild pigs in Florida and the microsatellites. Mamm Genome 7:368–372 inﬂuence of holding facilities as foci of translocation Bankovich B, Boughton E, Boughton R, Avery ML, Wisely SM activities. We identiﬁed areas where long-term and (2016) Plant community shifts caused by feral swine rooting devalue South Florida rangelands. Agric Ecosyst ongoing wild pig introductions have taken place, Environ 220:45–54 reﬂected in high interbreeding due to wild pig Barrios-Garcia MN, Ballari SA (2012) Impact of wild boar (Sus dispersion between different locations through the scrofa) in its introduced and native range: a review. Biol Invasions 14:2283–2300 Kissimmee Valley region and surrounding regions. Barton K (2015) R package ‘MuMIn’: multi-model inference We have also identiﬁed isolated genetic groupings (version 1.15.6). http://CRAN.R-project.org/package= with limited genetic exchange with other nearby MuMIn. Accessed 21 Apr 2017 populations, which is suggestive of recent transloca- Belden RC, Frankenberger WG (1977) Management of feral hogs in Florida—past, present, and future. In: Wood GW tions. Finally, we have shown that transition holding (ed) Research and management of wild hog populations. facilities for wild pigs are not secure and likely result Clemson University, Georgetown, pp 5–10 in escapes or intentional releases into surrounding Bevins SN, Pedersen K, Lutman MW, Gidlewski T, Deliberto areas. These human activities have shaped the demo- TJ (2014) Consequences associated with the recent range graphic structure of wild pigs at the regional level. Our expansion of nonnative feral swine. Bioscience 64:291–299 ﬁndings inform both legislative and regulatory man- Bonin A, Bellemain E, Bronken Eidesen P, Pompanon F, agement focused on this invasive wild ungulate in Brochmann C, Taberlet P (2004) How to track and assess Florida and other southeastern states in the U.S. by genotyping errors in population genetics studies. Mol Ecol highlighting the role of transportation and escapes 13:3261–3273 Bratton SP (1975) The effect of the European wild boar, Sus from holding facilities in maintaining and expanding scrofa, on gray beech forest in the Great Smoky Mountains. invasive wild pigs in Florida. Ecology 56:1356–1366 Burnham KP, Anderson DR (2002) Model selection and mul- Acknowledgements We thank USDA/APHIS/WS ﬁeld timodel inference: a practical information-theoretic personnel and several hunting check station operators for approach. Springer, New York graciously collecting samples on our behalf. We thank JC Cardenas-Canales EM, Ortega-Santos JA, Campbell TA, Gar- Grifﬁn, R. Boughton and M. Legare for repeatedly assisting us cı´a-Va´zquez Z, Cantu´-Covarrubias A, Figueroa-Milla´n JV, with our sampling efforts in the ﬁeld; M. Lopez, M. Anderson, P. DeYoung RW, Hewitt DG, Bryant FC (2011) Nilgai Royston, B. Pace-Aldana, D. Watkins, and B. Camposano for antelope in northern Mexico as a possible carrier for cattle quickly getting us the necessary sampling permits. We thank M. fever ticks and Babesia bovis and Babesia bigemina. Tabak, H. Ernest and R. Beasley for providing essential J Wildl Dis 47:777–779 technical support and molecular data to work with wild pig Carpio AJ, Guerrero-Casado J, Barasona JA, Tortosa FS, microsatellite loci. This study was partially funded by U.S. Vicente J, Hillstrom L, Delibes-Mateos M (2016) Hunting Department of Energy under Award Number DE–FC09– as a source of alien species: a European review. Biol 07SR22506 to the University of Georgia Research Invasions 19:1–15 Foundation, and by funding to SMW from USDA NIFA Cassey P, Hogg CJ (2015) Escaping captivity: the biological McIntire-Stennis Cooperative Forestry Research Program, invasion risk from vertebrate species in zoos. Biol Conserv Award No. FLA-WEC-005166. FAH was supported by the 181:18–26 Comisio´n Nacional de Ciencia y Tecnologı´a de Chile Caudell JN, McCann BE, Newman RA, Simmons RB, Backs (CONICYT-Chile). SE, Schmit BS, Sweitzer RA (2013) Identiﬁcation of putative origins of introduced pigs in Indiana using Open Access This article is distributed under the terms of the microsatellite markers and oral history. Proc Wildl Dam- Creative Commons Attribution 4.0 International License (http:// age Manag Conf 15:39–41 creativecommons.org/licenses/by/4.0/), which permits unre- Chapin FS, Walker BH, Hobbs RJ, Hooper DU, Lawton JH, Sala stricted use, distribution, and reproduction in any medium, OE, Tilman D (1997) Biotic control over the functioning of provided you give appropriate credit to the original ecosystems. Science 277:500–504 author(s) and the source, provide a link to the Creative Com- mons license, and indicate if changes were made. 123 Invasion ecology of wild pigs (Sus scrofa) in Florida, USA 1879 Choquenot D, McIlroy J, Korn T (1996) Managing vertebrate Hampton JO, Spencer PBS, Alpers D, Twigg L, Woolnough A, pests: feral pigs. Australian Government Publishing Ser- Doust J, Higgs T, Pluske J (2004) Applying molecular vice, Bureau of Resource Sciences, Canberra ecology to wildlife management: population structure and Corander J, Marttinen P (2006) Bayesian identiﬁcation of dynamics of feral pigs in south-western Australia. J Appl admixture events using multilocus molecular markers. Mol Ecol 41:735–743 Ecol 15:2833–2843 Holleley CE, Geerts PG (2009) Multiplex manager 1.0: a cross- Corander J, Marttinen P, Sire´n J, Tang J (2008) Enhanced platform computer program that plans and optimizes Bayesian modelling in BAPS software for learning genetic multiplex PCR. Biotechniques 46:511–517 structures of populations. BMC Bioinform 9:539 Holm S (1979) A simple sequentially rejective multiple test Crooks JA (2002) Characterizing ecosystem-level conse- procedure. Scand J Stat 6:65–70 quences of biological invasions: the role of ecosystem Hone J (2002) Feral pigs in Namadgi National Park, Australia: engineers. Oikos 97:153 dynamics, impacts and management. Biol Conserv Dickman CR (2009) House cats as predators in the Australian 105:231–242 environment: impacts and management. Hum Wildl Hulme PE, Bacher S, Kenis M, Klotz S, Kuhn I, Minchin D, Interact 3:41–48 Nentwig W, Olenin S, Panov V, Pergl J, Pysek P, Roques Dray S, Dufour AB (2007) The ade4 package: implementing the A, Sol D, Solarz W, Vila M (2008) Grasping at the routes of duality diagram for ecologists. J Stat Softw 22:1–20 biological invasions: a framework for integrating pathways Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a into policy. J Appl Ecol 45:403–414 website and program for visualizing STRUCTURE output Hutton T, DeLiberto T, Owen S, Morrison B (2006) Disease and implementing the Evanno method. Conserv Genet risks associated with increasing feral swine numbers and Resour 4:359–361 distribution in the United States. Michigan Bovine El Mousadik A, Petit RJ (1996) High level of genetic differen- Tuberculosis Bibliography and Database tiation for allelic richness among populations argan tree Jombart T, Ahmed I (2011) adegenet 1.3-1: new tools for the [Argania spinosa (L.) Skeels] endemic to Morocco. Theor analysis of genome-wide SNP data. Bioinformatics Appl Genet 92:832–839 27:3070–3071 Ellegren H, Johansson M, Chowdhary BP, Marklund S, Ruyter Jombart T, Devillard S, Balloux F (2010) Discriminant analysis D, Marklund L, Brauner-Nielsen P, Edfors-Lilja I, Gus- of principal components: a new method for the analysis of tavsson I, Juneja RK, Andersson L (1993) Assignment of genetically structured populations. BMC Genet 11:1–15 20 microsatellite markers to the porcine linkage map. Kaminski G, Brandt S, Baubet E, Baudoin C (2005) Life-history Genomics 16:431–439 patterns in female wild boars (Sus scrofa L., 1758): Evanno G, Regnault S, Goudet J (2005) Detecting the number of mother–daughter postweaning associations. Can J Zool clusters of individuals using the software structure: a 83:474–480 simulation study. Mol Ecol 51:672–681 Kidd A, Bowman J, Lesbarre`res D, Schulte-Hostedde A (2009) Falush D, Stephens M, Pritchard JK (2003) Inference of popu- Hybridization between escaped domestic and wild Amer- lation structure using multilocus genotype data: linked loci ican mink (Neovison vison). Mol Ecol 18:1175–1186 and correlated allele frequencies. Genetics 164:1567–1587 Lonsinger RC, Waits LP (2015) ConGenR: rapid determination FDACS (Florida Department of Agricultural and Consumer of consensus genotypes and estimates of genotyping errors Services) (2016) Intrastate movement of feral swine. www. from replicated genetic samples. Conserv Genet Resour freshfromﬂorida.com/Divisions-Ofﬁces/Animal-Industry/ 7:841–843 Consumer-Resources/Consumer-Protection/Animal-Move Lopez J, Hurwood D, Dryden B, Fuller S (2014) Feral pig ment/Intrastate-Movement-of-Feral-Swine. Accessed 15 populations are structured at ﬁne spatial scales in tropical Dec 2016 Queensland, Australia. PLoS ONE 9:e91657 Gabor TM, Hellgren EC, Van Den Bussche RA, Silvy NJ (1999) Lowe S, Browne M, Boudjelas S, Poorter M (2000) 100 of the Demography, sociospatial behaviour and genetics of feral world’s worst invasive alien species: a selection from the pigs in a semi-arid environment. J Zool 247:311–322 global invasive species database. The Invasive Species Genovesi P, Carnevali L, Alonzi A, Scalera R (2012) Alien Specialist Group, Species Survival Commission, World mammals in Europe: updated numbers and trends, and Conservation Union IUCN, p 12 assessment of the effects on biodiversity. Integr Zool Matschke GH (1967) Aging European wild hogs by dentition. 7:247–253 J Wildl Manag 31:109–113 Gioeli KT, Huffman J (2012) Land managers’ feral hog man- Mayer JJ, Brisbin IL Jr. (2009) Wild pigs: biology, damage, agement practices inventory in Florida. Proc Fla State Hort control techniques and management. SRNL-RP-2009- Soc 125:2012 00869. Savannah River National Laboratory: Aiken, SC Giuliano W (2010) Wild hogs in Florida: ecology and man- Mayer JJ, Brisbin IL Jr (2008) Wild pigs in the United States: agement. IFAS# WEC277 University of Florida Institute of their history, comparative morphology, and current status. Food and Agricultural Sciences University of Georgia Press, Athens Goudet J (2005) Hierfstat, a package for R to compute and test Nikolov IS, Gum B, Markov G, Kuehn R (2009) Population variance components and F-statistics. Mol Ecol Notes genetic structure of wild boar Sus scrofa in Bulgaria as 5:184–186 revealed by microsatellite analysis. Acta Theriol Gruber B, Adamack AT (2015) Landgenreport: a new R func- 54:193–205 tion to simplify landscape genetic analysis using resistance surface layers. Mol Ecol Resour 15:1172–1178 123 1880 F. A. Herna´ndez et al. Paradis E (2010) pegas: an R package for population genetics Seward NW, VerCauteren KC, Witmer GW, Engeman RM with an integrated–modular approach. Bioinformatics (2004) Feral swine impacts on agriculture and the envi- 26:419–420 ronment. Sheep Goat Res J 19:34–40 Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysis in Smith B, Li N, Andersen AS, Slotved HC, Krogfelt KA (2011) Excel. Population genetic software for teaching and Optimising bacterial DNA extraction from faecal samples: research—an update. Bioinformatics 28:2537–2539 comparison of three methods. Open Microbiol J 5:14–17 Poteaux C, Baubet E, Kaminski G, Brandt S, Dobson FS, Spencer PBS, Hampton JO (2005) Illegal translocation and Baudoin C (2009) Socio-genetic structure and mating genetic structure of feral pigs in Western Australia. J Wildl system of a wild boar population. J Zool 278:116–125 Manag 69:377–384 Pritchard J, Stephens M, Donnelly P (2000) Inference of pop- Tabak MA, Piaggio AJ, Miller RS, Sweitzer RA, Ernest HB ulation structure using multilocus genotype data. Genetics (2017) Linking anthropogenic factors with the movement 155:945–959 of an invasive species. Ecosphere 8:e01844 Puechmaille SJ (2016) The program STRUCTURE does not Vernesi C, Crestanello B, Pecchioli E, Tartari D, Caramelli D, reliably recover the correct population structure when Hauffe H, Bertorelle G (2003) The genetic impact of sampling is uneven: subsampling and new estimators demographic decline and reintroduction in the wild boar alleviate the problem. Mol Ecol Resour 16:608–627 (Sus scrofa): a microsatellite analysis. Mol Ecol Pysek P, Richardson DM (2010) Invasive species, environ- 12:585–595 mental change and management, and health. Annu Rev Waithman JD, Sweitzer RA, Van Vuren D, Drew JD, Brinkhaus Environ Resour 35:25–55 AJ, Gardner IA, Boyce WM (1999) Range expansion, R Core Team (2016) R: a language and environment for sta- population sizes, and management of wild pigs in Cali- tistical computing. R foundation for statistical computing, fornia. J Wildl Manag 63:298–308 Vienna, Austria. ISBN 3-900051-07-0. http://www.R- Wardle DA, Barker GM, Yeates GW, Bonner KI, Ghani A project.org/ (2001) Introduced browsing mammals in New Zealand Randi E (2005) Management of wild ungulate populations in forests: aboveground and belowground consequences. Ecol Italy: captive-breeding, hybridization and genetic conse- Monogr 71:587–614 quences of translocations. Vet Res Commun 29:71–75 Weir BS, Cockerham CC (1984) Estimating F-statistics for the Rice WR (1989) Analyzing tables of statistical tests. Evolution analysis of population structure. Evolution 38:1358–1370 43:223–225 Wilson GA, Rannala B (2003) Bayesian inference of recent Robic A, Dalens M, Woloszyn N, Milan D, Riquet J, Gellin J migration rates using multilocus genotypes. Genetics (1994) Isolation of 28 new porcine microsatellites reveal- 163:1177–1191 ing polymorphism. Mamm Genome 5:580–583 Wood G, Barrett R (1979) Status of wild pigs in the United Rohrer GA, Alexander LJ, Hu Z, Smith TPL, Keele JW, Beattie States. Wildl Soc B 7:237–246 CW (1996) A comprehensive map of the porcine genome. Wright S (1951) The genetical structure of populations. Ann Genome Res 6:371–391 Eugenic 15:323–354 Sala OE, Chapin FS, Armesto JJ, Berlow E, Bloomﬁeld J, Dirzo Wright S (1965) The interpretation of population structure by R, Huber-Sanwald E, Huenneke LF, Jackson RB, Kinzig F-statistics with special regard to systems of mating. A, Leemans R, Lodge DM, Mooney HA, Oesterheld M, Evolution 19:395–420 Poff NL, Sykes MT, Walker BH, Walker M, Wall DH Yiming L, Zhengjun W, Duncan RP (2006) Why islands are (2000) Global biodiversity scenarios for the year 2100. easier to invade: human inﬂuences on bullfrog invasion in Science 287:1770–1774 the Zhoushan archipelago and neighbouring mainland Scandura M, Iacolina L, Cossu A, Apollonio M (2011) Effects China. Oecologia 148:129–136 of human perturbation on the genetic make-up of an island Zivin J, Hueth BM, Zilberman D (2000) Managing a multiple- population: the case of the Sardinian wild boar. Heredity use resource: the case of feral pig management in Cali- 106:1012–1020 fornia rangeland. J Environ Econ Manag 39:189–204
Biological Invasions – Springer Journals
Published: Jan 20, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera