TY - JOUR
AU - Daugherty, Matthew P
AB - Abstract The spread and impact of invasive species in exotic ranges can be mitigated by increased understanding of pest invasion dynamics. Here, we used geospatial analyses and habitat suitability modeling to characterize the invasion of an important vineyard pest, vine mealybug (Planococcus ficus Signoret, Hemiptera: Pseudococcidae), using nearly 15,000 trapping records from throughout Napa County, California, between 2012 and 2017. Spatial autocorrelation among P. ficus detections was strongest at distances of ~250 m and detectable at regional scales (up to 40 km), estimates of the rate and directionality of spread were highly idiosyncratic, and P. ficus detection hotspots were spatiotemporally dynamic. Generalized linear model, boosted regression tree, and random forest modeling methods performed well in predicting habitat suitability for P. ficus. The most important predictors of P. ficus occurrence were a positive effect of precipitation in the driest month, and negative effects of elevation and distance to nearest winery. Our results indicate that 250-m quarantine and treatment zones around P. ficus detections are likely sufficient to encompass most local establishment and spread, and that implementing localized regulatory procedures may limit inadvertent P. ficus spread via anthropogenic pathways. Finally, surveys of P. ficus presence at >300 vineyard sites validated that habitat suitability estimates were significantly and positively associated with P. ficus frequency of occurrence. Our findings indicate that habitat suitability predictions may offer a robust tool for identifying areas in the study region at risk to future P. ficus invasion and prioritizing locations for early detection and preventative management efforts. invasion dynamics, invasion risk, invasion hotspot, integrated pest management Biological invasions are increasingly frequent events that can result in a multitude of ecological and economic impacts (Simberloff 2013, Liebhold et al. 2017). For example, it is estimated that an average of 10 arthropod crop pests are introduced into the United States each year (Work et al. 2005). Invasive species can subsequently threaten food security, substantially reduce agricultural production and revenues, affect trade of agricultural commodities, and incur tremendous management costs (Pimentel et al. 2005, Paini et al. 2016, Simmons et al. 2018). These threats posed by invasive agricultural pests underscore the need for effective management strategies that mitigate pest spread and impacts. It is often difficult to anticipate which pest invasions will ‘fade-out’ and which will persist long term (Pulliam 2000, Soberon and Peterson 2005, Jiménez-Valverde et al. 2011). This is due, in part, to poor understanding of the pest’s spatiotemporal dynamics during its initial invasion phases. In instances where a pest spreads and persists throughout an exotic range, understanding of its spatiotemporal invasion dynamics and the factors that govern these patterns must be quickly gained and integrated to develop pest management efforts—including prevention, detection, and eradication components—that can mitigate invader spread and impacts (Leung et al. 2012, Baker and Bode 2016, Baker 2017). Geospatial analyses and niche-based/habitat suitability models are useful for these purposes (Worner and Gevrey 2006, Margosian et al. 2009, De Meyer et al. 2010, Svobodovà et al. 2013, Hahn et al. 2017). To this end, results of these analytical approaches can inform multiple aspects of invasive species management efforts, including, early detection strategies, prioritization of high-risk areas, minimizing management expenditures and ecological effects, and guiding eradication efforts (Rout et al. 2014, Berec et al. 2015, Vicente et al. 2016, Baker 2017, Deleon et al. 2017). The vine mealybug, Planococcus ficus (Signoret) (Hemiptera: Pseudococcidae), is an invasive agricultural pest whose management strategies may benefit from increased understanding of its invasion dynamics. Although of Paleartic origin, P. ficus has spread to and established in multiple countries and regions including South Africa, India, the Caribbean, and South America (Walton and Pringle 2004, Daane et al. 2012). In California, P. ficus was first reported in Coachella Valley vineyards (Gill 1994) and soon spread throughout most of the state’s grape-growing regions (Godfrey et al. 2002; Daane et al. 2004a, b). The U.S. range of P. ficus continues to expand—a point of considerable concern because of the direct and indirect damage the mealybug can cause to host plants. P. ficus actively feeds on and debilitates grapevines and, in the process, excretes honeydew that contaminates fruit, making it unsuitable for sale or wine-making (Daane et al. 2006). In addition to direct feeding damage, P. ficus can transmit plant pathogens, including Grapevine leafroll-associated virus-3 (Daane et al. 2012, Almeida et al. 2013), that further reduce grapevine health and productivity. The ability of P. ficus to complete 3–10 generations per year under California climatic conditions can exacerbate both its within-vineyard damage and spread. First instars, the most mobile life stage, begin to disperse in May (Walton et al. 2006) and their spread can be facilitated by wind or anthropogenic means. This is especially concerning because first-instar mealybugs are the most efficient life stage at transmitting grapevine leafroll viruses (Tsai et al. 2008). As the growing season progresses, P. ficus generations overlap, further indicating the potential for P. ficus to develop large populations that exacerbate within-vineyard impacts and spread (Walton et al. 2006). Management of P. ficus is challenging and costly because, if left unchecked, its populations can destroy entire crops and kill their host vines (Daane et al. 2006, Daane et al. 2012). Management tactics include biological control, mating disruption, and chemical control but are often less effective than desired (Daane et al. 2008). This is especially the case in coastal wine grape-growing regions where climatic conditions are favorable for P. ficus and Argentine ants (Linepithema humile), whose activity disrupts biological control (Daane et al. 2007). Once established, P. ficus populations are difficult to eradicate and must be aggressively and continually managed to limit damage and contain spread. A better understanding of the factors governing P. ficus invasion dynamics may inform early detection and management efforts in grape-growing regions dealing with or at risk of severe impacts from P. ficus populations. Here, we address these issues using a large-scale geodatabase of P. ficus trapping records throughout Napa County, CA. Trapping records were first interrogated with a range of geospatial tools to describe spatiotemporal patterns in P. ficus distribution throughout the study region. Next, we developed habitat suitability models to quantify how key climatic, environmental, and anthropogenic variables underpin P. ficus invasion dynamics. Finally, we used a combination of field surveys and grower interviews of P. ficus presence to validate predictions from the suitability modeling, to gauge its predictive value as a risk assessment tool. Materials and Methods Description of the Study Region Napa County, which is located in northern California, encompasses ~2000 km2 and is noted for its highly variable topography. The county’s variable topography generates pronounced elevation and moistures gradients, especially in its northernmost regions (Underwood-Russel et al. 2001). These elevation and moisture gradients are buffered by the county’s predominantly Mediterranean climate and subsequently promote highly diverse vegetative communities (Thorne et al. 2004). Viticulture accounts for nearly $938 million of the county’s annual revenue (Napa County Department of Agriculture and Weights and Measures 2019). Planococcus ficus Monitoring and Trapping Program In early 2012, the continued spread of P. ficus throughout California grape-growing regions spurred efforts to delimit and monitor the scope of P. ficus establishment in Napa County. The California Department of Food and Agriculture (CDFA) State Wide Grid System was used for these monitoring and detection efforts. The geographical extent of the trapping effort and number of traps deployed varied between years; trapping was conducted county-wide in some years (2012, 2016–2017; Fig. 1), whereas the Carneros region, in southwest Napa County, was excluded from trapping in others (2013–2015). Each year at the beginning of May, one sex pheromone-baited Delta trap was deployed at the approximate center of every CDFA State Wide Grid System subgrid cell (~0.1 km2) selected for trapping and checked every 2 wk through November. These pheromone-baited traps are effective at capturing adult male P. ficus at distances up to 100 m from infestation sources, and resulting captures are positively associated with local infestation densities (Millar et al. 2002, Walton et al. 2004). The number of adult male P. ficus caught on each trap was recorded and traps were replaced as needed. In total, 14,996 traps were deployed from 2012 to 2017 (Table 1) at 4,526 unique locations throughout Napa County. Fig. 1. Open in new tabDownload slide Map of the study region showing the location of traps deployed to monitor for Planococcus ficus in 2017. Dark grey points denote trap detections versus black points where P. ficus was not detected. Fig. 1. Open in new tabDownload slide Map of the study region showing the location of traps deployed to monitor for Planococcus ficus in 2017. Dark grey points denote trap detections versus black points where P. ficus was not detected. Table 1. 2012–2017 trapping and detection records of Planococcus ficus occurrence in Napa County, CA Year . No. of traps deployed . No. of traps recording captures . No. of male captured . 2012 4,021 577 49,327 2013 3,437 327 16,488 2014 3,419 296 43,444 2015 3,473 841 26,577 2016 4,000 1,415 49,785 2017 4,116 1,602 55,723 Year . No. of traps deployed . No. of traps recording captures . No. of male captured . 2012 4,021 577 49,327 2013 3,437 327 16,488 2014 3,419 296 43,444 2015 3,473 841 26,577 2016 4,000 1,415 49,785 2017 4,116 1,602 55,723 Open in new tab Table 1. 2012–2017 trapping and detection records of Planococcus ficus occurrence in Napa County, CA Year . No. of traps deployed . No. of traps recording captures . No. of male captured . 2012 4,021 577 49,327 2013 3,437 327 16,488 2014 3,419 296 43,444 2015 3,473 841 26,577 2016 4,000 1,415 49,785 2017 4,116 1,602 55,723 Year . No. of traps deployed . No. of traps recording captures . No. of male captured . 2012 4,021 577 49,327 2013 3,437 327 16,488 2014 3,419 296 43,444 2015 3,473 841 26,577 2016 4,000 1,415 49,785 2017 4,116 1,602 55,723 Open in new tab All yearly trapping data were checked for duplicate georeferenced records; duplicates were removed prior to analysis. Throughout all analyses, we first considered whether a given trap recorded any P. ficus captures (i.e., trap detection or P. ficus presence–absence) rather than number of P. ficus caught per trap (Table 1). However, the number of adult male P. ficus caught per trap was considered when estimating yearly rates of P. ficus spread. The R statistical language, version 3.4.3 (R Core Development 2017), was used for all geospatial analyses and habitat suitability modeling. Geospatial Analyses of P. ficus Invasion Dynamics We employed geospatial analyses to quantify spatiotemporal patterns in yearly P. ficus detections as a first step toward understanding P. ficus invasion dynamics. More specifically, we quantified the magnitude of clustering, and thus the degree of spatial autocorrelation (SAC), among P. ficus detections in each study year. We also identified hotspots of P. ficus detections in each study year. The degree of clustering among P. ficus detections was quantified to characterize local P. ficus movement distances (Thomas et al. 2017) and identify the need to control for SAC-induced inflation when estimating predictor variable significance and predicting organism distributions (Veloz 2009, Merow et al. 2014). We quantified the degree of clustering and magnitude of SAC among yearly P. ficus detections at a regional scale because the standardized spacing of the traps deployed within the CDFA State Wide Grid System limited our ability to detect and quantify these patterns at fine (i.e., less than ~300 m) spatial scales. Yearly P. ficus detections were first transformed into point-pattern processes (R package ‘spatstat’; Baddeley and Turner 2005) and then analyzed via pair correlation functions bootstrapped over 999 simulations to quantify the magnitude and spatial scale of clustering among detections, as well as develop 95% CIs around these estimates. Next, we used the 2012–2017 P. ficus trapping records to capture the extent of spatial heterogeneity and identify hotspots of P. ficus detections. First, we superimposed a grid of 1-km2 cells over Napa County and summed the number of traps recording P. ficus detections in each year within each cell. We then calculated a local Getis-Ord Gi statistic (Z-score; Getis and Ord 1992) for each 1-km2 grid cell (R package ‘spdep’; Bivand et al. 2005) to identify cells where the number of P. ficus detections were greater or lesser than expected by chance. The Z-scores associated with each grid cell were then transformed into probabilities; grid cells with a statistically significant (P < 0.05) Z-score were classified as P. ficus detection hotspots. Estimating P. ficus Spread We also estimated the rate and direction of P. ficus spread using square-root area regression, distance-based regression, and boundary displacement methods (Tobin et al. 2015). Three abundance thresholds (presence-only, 10 and 100 adult P. ficus captured) were imposed to evaluate rates of P. ficus spread relative to trap catch abundances. Both the distance-based regression and boundary displacement methods involve calculating trap distance from a reference point that can either be arbitrarily chosen or represent the initial location of invader introduction or detection (Tobin et al. 2015); we arbitrarily chose a reference point in Napa County and used this location for both methods. The square-root area and distance-based methods of estimating spread regress the square root of the total area occupied and distances of all detections from a reference point, respectively, in each year as a function of time. Spread is consequently represented by the slope of the fitted regression line. For square-root area and distance-based analyses in each year, we considered the area and distance from our chosen reference point, respectively, of all CDFA State Wide Grid System subgrid cells that recorded P. ficus detections. The boundary displacement method considers the distances between pairs of consecutive invasion boundaries relative to a reference point. Invasion boundaries were delimited by generating a convex hull around the traps recording P. ficus detections in each year and for each abundance threshold. Transects originating from the reference point at 0.5° intervals were then generated and the distance from the reference point to where each transect intersected each yearly invasion boundary was calculated. The lengths of each transect were then subtracted between invasion boundaries in consecutive years to calculate between-year boundary displacement distances and directionality in P. ficus spread relative to the imposed abundance thresholds. Habitat Suitability and Ensemble Modeling Here, we provide an overview of our habitat suitability modeling methodology following the ODMAP (Overview, Data, Model, Assessment, and Prediction) protocol for species distribution models (Zurell et al. 2020). Specific methodological details for all ODMAP sections are presented as supplementary materials (Supp Table 1 [online only]). Habitat suitability models were developed to characterize key variables underpinning P. ficus invasion dynamics and identify regions at risk of its future spread in Napa County, California. All habitat suitability models were generated at a 1-km2 resolution. All models were restricted to and projected within the extent of the political boundaries of Napa County, California (spatial extent [Lon/Lat]: −122.6464, −122.0614, 38.15489, 38.86424 [xmin, xmax, ymin, ymax]; Fig. 1). Habitat suitability models were trained and validated using P. ficus trapping (i.e., presence–absence) data that were compiled from 2012 to 2017. All trapping data utilized for habitat suitability modeling were prepared as described for analyses characterizing P. ficus spread. We hypothesized that climatic, environmental, and anthropogenic factors influence habitat suitability for P. ficus and its invasion dynamics. We subsequently assumed that predictor variables included in habitat suitability models were relevant ecological drivers of P. ficus distributions in the study region. We assumed that P. ficus detectability via pheromone-baited traps did not change across habitat gradients and that all P. ficus monitoring and trapping data were collected via sampling efforts that were adequate and free of observational biases. All trap detections of P. ficus were assumed to represent established infestations because pheromone-baited trap captures are significantly and positively associated with local P. ficus densities (Walton et al. 2004). All predictor variables included in habitat suitability models were assumed to be measured without error, and all predictor–P. ficus relationships were fit under current conditions. These predictor–P. ficus relationships and subsequent habitat suitability models also were not projected beyond the spatial extent of the P. ficus trapping and predictor variables. Habitat suitability models were developed using generalized linear model (GLM), boosted regression tree (BRT; Friedman 2001), random forest algorithm (Breiman 2001), and ensemble modeling methods. Boosted regression trees and the random forest are machine-learning methods noted for their flexibility in fitting ecological relationships and predictive accuracy in modeling species distributions (Elith and Leathwick 2009). Seven climatic, environmental, and anthropogenic variables were selected based on their limited multicollinearity and included in all habitat suitability models. Due to the presence of SAC among our P. ficus trapping records, a total of 200 spatial eigenvectors were generated (package ‘spmoran’; Murakami 2017) and considered for inclusion in our habitat suitability models to reduce the signature of SAC in fitted model residuals and model misspecification errors, and thus ultimately increase model accuracy (Diniz-Filho and Bini 2005, Tiefelsdorf and Griffith 2007, Thayn and Simanis 2013). Spatial eigenvectors are orthogonal and uncorrelated by nature, suggesting that stepwise selection methods guided by information criteria may select a robust subset of eigenvectors balancing model complexity, information loss, and the minimization of SAC effects (Getis and Griffith 2002, Tiefelsdorf and Griffith 2007, Murakami 2017, Murakami and Griffith 2017). Spatial eigenvectors were subsequently selected for inclusion in our models using a two-step approach. First, a GLM was fit with our selected climatic, environmental, and anthropogenic predictors. Forward stepwise selection, guided by Akaike’s Information Criterion, was then used to identify spatial eigenvectors to include in all GLM, boosted regression tree, and random forest habitat suitability models. Spatial eigenvectors selected for inclusion in our habitat suitability models were then interpolated into rasters via inverse-distance weighting and used as predictor variables in the suitability modeling process. Habitat suitability models generated via generalized linear modeling methods utilized P. ficus presence–absence as the dependent variable and were fit with logistic link, binomial error, and no first-order interactions among included predictor variables. Models generated by BRT methods included default interaction depth (seven) and learning rate (0.001) settings, and a total of 5,000 trees were fit. Random forest models were generated by growing a default number of trees (500). Settings for all habitat suitability modeling methods were chosen to accurately predict the potential distribution of P. ficus in the study region and develop fitted responses detailing predictor-P. ficus occurrence relationships. Training and testing data for each modeling method were generated via an 80/20 split of the compiled P. ficus trapping data. Ten replicate predictions were generated for each modeling method, weighted by their true skill statistic (TSS; Allouche et al. 2006) scores, and combined to generate method-specific ensemble models. All replicate models generated by each modeling method were weighted by their TSS scores and combined to generate a grand ensemble model. The predictive performance of all habitat suitability models was assessed using the area under the receiver operating curve (AUC) and TSS metrics. An evaluation strip method (Elith et al. 2005) was used to extract univariate response curves from our method-specific and grand ensemble predictions. The R statistical language, version 3.4.3 (R Core Development 2017), and package ‘biomod2’ (Thuiller et al. 2016) were used to generate all habitat suitability models, evaluate model performance and predictor contribution, and extract method-specific fitted response curves detailing predictor–P. ficus occurrence relationships. All R code used to generate these habitat suitability models is available upon request. Model Validation via Field Surveys of P. ficus Presence In the final part of the study, we assessed independently how well the suitability modeling captured in-field distribution of P. ficus. To do this, we first determined the invasion status by P. ficus of vineyards spread throughout the study region via a combination of grower interviews, in-field inspection, or compilation of reports of newly invaded sites (Supp Fig. 1 [online only]). For interviews, cooperating growers or vineyard managers were asked whether they had observed P. ficus-infested vines at the site and, if so, what year it was first observed. We used pesticide reports and applications of P. ficus effective products to gauge when the block was first invaded in the few instances for which the year of establishment was not known. Grower interviews yielded information on >300 vineyards (Supp Fig. 1A [online only]), 279 of which had all information needed for inclusion in the analysis. In-field monitoring and inspection of vines was conducted at 24 additional vineyards (Supp Fig. 1B [online only]). At each site, 250 vines were inspected among multiple rows spread throughout up to ~5 ac of the block. Vines were selected haphazardly and visually inspected for 30 s by peeling bark and checking leaf veins for live P. ficus of any stage. If live individuals were found, the search was ended and the site was considered invaded. The third source of site data was more serendipitous, from growers reporting suspected newly invaded vineyards, which were later confirmed by one of the researchers (17 sites total). Next, at each of the vineyards we extracted the habitat suitability value from the grand ensemble model ( Habitat Suitability and Ensemble Modeling section) closest to the center of the block. In addition, to capture effects of dispersal from surrounding areas with an established P. ficus population, we calculated for each year between 2014 and 2019 the distance from the edge of the vineyard to the center of the nearest trapping grid cell containing a P. ficus trap detection in the previous year. Because trapping records were not available for 2018, the lagged distance for 2019 was calculated based on trap detections from 2 yr prior. Among the 320 vineyard sites in the dataset, estimated habitat suitability values ranged from 0.02 to 0.99, and lagged distance to prior trap detection from 0 m (i.e., detection on a trap within or directly adjacent to the vineyard) to >5 km. We analyzed the frequency of P. ficus presence in the vineyard using a generalized linear model with habitat suitability and lagged distances [ln(x + 1) transformed] as continuous variates, and binomial error (Crawley 2012). Results Yearly P. ficus detections were nonrandomly distributed and the strength of spatial autocorrelation varied among study years (Supp Fig. 2 [online only]). In each study year, the strongest clustering among P. ficus detections occurred at approximately the spatial resolution at which traps were deployed (~250 m). The strongest and weakest degree of clustering among P. ficus detections occurred in 2013 and 2016, respectively. Autocorrelation among yearly P. ficus detections was detectable at regional scales up to 40 km (Supp Fig. 2 [online only]). We also observed substantial, between-year spatial heterogeneity in the location of P. ficus hotspots (Fig. 2). Early P. ficus hotspots were located in the southern portion of the study region, surrounding the city of Napa, with smaller clusters near St. Helena and Rutherford, CA. Over time, hotspots became increasingly common up valley (i.e., northwest), ultimately resulting in large, contiguous hotspot clusters throughout much of the southern and western portions of the county (Fig. 2). Fig. 2. Open in new tabDownload slide Local Getis-Ord hotspots of 2012–2017 Planococcus ficus detection within 1-km2 grid cells throughout the study region. The Carneros region was not trapped in 2013–2015 and is shaded gray in these years. Fig. 2. Open in new tabDownload slide Local Getis-Ord hotspots of 2012–2017 Planococcus ficus detection within 1-km2 grid cells throughout the study region. The Carneros region was not trapped in 2013–2015 and is shaded gray in these years. Estimates of yearly P. ficus spread varied substantially among methods and based on the abundance thresholds imposed on the calculations (Table 2; Supp Figs. 3–7 [online only]). The distance-based regression and boundary displacement methods produced the most conservative and liberal, respectively, estimates of spread. Estimates from these two methods were less variable than the square-root area method. Implementing abundance thresholds (i.e., 10 and 100 P. ficus captured per trap) generally resulted in lower estimates of P. ficus spread but greater variability. Boundary displacement estimates of P. ficus spread were idiosyncratic in both magnitude and direction (Supp Figs. 5–7 [online only]). For example, when a presence-only threshold was imposed, P. ficus spread in 2012–2013 occurred primarily toward the southwest at a maximum magnitude of slightly >4 km (Supp Fig. 5 [online only]). However, in 2013–2014, P. ficus spread toward the northeast with a maximum magnitude of ~13 km (Supp Fig. 5 [online only]). Table 2. Distance-based, square-root area-based, and boundary displacement estimates of Planococcus ficus yearly spread Threshold . Distance . Square-root area . Boundary displacement . Mean Error Mean Error Mean Error Presence-only 365.9 31.0 779.8 261.2 832.6 41.0 10 310.1 39.9 572.8 230.1 848.1 41.0 100 299.5 87.8 94.9 257.3 890.3 62.6 Threshold . Distance . Square-root area . Boundary displacement . Mean Error Mean Error Mean Error Presence-only 365.9 31.0 779.8 261.2 832.6 41.0 10 310.1 39.9 572.8 230.1 848.1 41.0 100 299.5 87.8 94.9 257.3 890.3 62.6 Open in new tab Table 2. Distance-based, square-root area-based, and boundary displacement estimates of Planococcus ficus yearly spread Threshold . Distance . Square-root area . Boundary displacement . Mean Error Mean Error Mean Error Presence-only 365.9 31.0 779.8 261.2 832.6 41.0 10 310.1 39.9 572.8 230.1 848.1 41.0 100 299.5 87.8 94.9 257.3 890.3 62.6 Threshold . Distance . Square-root area . Boundary displacement . Mean Error Mean Error Mean Error Presence-only 365.9 31.0 779.8 261.2 832.6 41.0 10 310.1 39.9 572.8 230.1 848.1 41.0 100 299.5 87.8 94.9 257.3 890.3 62.6 Open in new tab All the methods employed performed well (all AUC/ROC values > 0.8 and TSS values > 0.6) in predicting habitat suitability for P. ficus in the study region. The grand ensemble method performed the best of all employed methods (Table 3; AUC/ROC = 0.953, TSS = 0.753). The relative importance of the seven climatic, landscape, and anthropogenic variables varied among the modeling and ensemble methods employed (Fig. 3). The probability of P. ficus occurrence was most strongly associated with the amount of precipitation in the driest month, elevation, and distance to nearest winery. More specifically, the probability of P. ficus occurrence was negatively associated with precipitation in the driest month and distance to nearest winery, but increased slightly with increasing elevation (Fig. 4). The probability of P. ficus occurrence was least affected by distance to nearest road across all modeling methods. Table 3. Performance of GLM, random forest algorithm (RF), BRT, and a grand ensemble method of predicting habitat suitability for Planococcus ficus in Napa County, CA Metric . GLM . RF . BRT . Grand . TSS 0.645 0.664 0.698 0.753 AUC/ROC 0.913 0.921 0.936 0.953 Metric . GLM . RF . BRT . Grand . TSS 0.645 0.664 0.698 0.753 AUC/ROC 0.913 0.921 0.936 0.953 Open in new tab Table 3. Performance of GLM, random forest algorithm (RF), BRT, and a grand ensemble method of predicting habitat suitability for Planococcus ficus in Napa County, CA Metric . GLM . RF . BRT . Grand . TSS 0.645 0.664 0.698 0.753 AUC/ROC 0.913 0.921 0.936 0.953 Metric . GLM . RF . BRT . Grand . TSS 0.645 0.664 0.698 0.753 AUC/ROC 0.913 0.921 0.936 0.953 Open in new tab Fig. 3. Open in new tabDownload slide The mean (±SE) relative importance of select landscape, climatic, and anthropogenic predictors for each method of generating habitat suitability for Planococcus ficus. Note that the scale of the y-axes differs among panels. Fig. 3. Open in new tabDownload slide The mean (±SE) relative importance of select landscape, climatic, and anthropogenic predictors for each method of generating habitat suitability for Planococcus ficus. Note that the scale of the y-axes differs among panels. Fig. 4. Open in new tabDownload slide Fitted responses of select landscape, climatic, and anthropogenic variables from the grand ensemble prediction of habitat suitability for Planococcus ficus. Solid black lines represent the mean fitted response of TSS-weighted, mean grand ensemble prediction. Light-gray dashed lines represent 95% CIs around model predictions. Fig. 4. Open in new tabDownload slide Fitted responses of select landscape, climatic, and anthropogenic variables from the grand ensemble prediction of habitat suitability for Planococcus ficus. Solid black lines represent the mean fitted response of TSS-weighted, mean grand ensemble prediction. Light-gray dashed lines represent 95% CIs around model predictions. The grand ensemble prediction showed substantial heterogeneity in habitat suitability over the study region. The areas surrounding the cities of Napa and St. Helena, as well as the central-eastern portion of Napa County, were predicted to be of highest suitability for P. ficus (Fig. 5). Method-specific predictions of habitat suitability were largely congruent (Supp Fig. 9 [online only]). Regions of Napa County predicted to be poorly suitable for P. ficus, such as the northeastern portion of the county, coincide with regions where viticulture is largely absent (Fig. 5; Supp Fig. 9 [online only]). Fig. 5. Open in new tabDownload slide Habitat suitability for Planococcus ficus as predicted by the TSS-weighted grand ensemble prediction. Fig. 5. Open in new tabDownload slide Habitat suitability for Planococcus ficus as predicted by the TSS-weighted grand ensemble prediction. Of the 320 sites for which the invasion status of P. ficus was determined from grower interviews, field inspections, or other reporting, 24% (76) had P. ficus present on grapevines. The analysis of invasion frequency showed a significant effect of habitat suitability (χ 2 = 63.902, df = 1, P < 0.0001), a nonsignificant effect of distance to the nearest trap detection the year prior (χ 2 = 1.898, df = 1, P = 0.1683), and a significant interaction between habitat suitability and lagged distance (χ 2 = 7.186, df = 1, P = 0.0073). The interactive effect appears to stem from invasion rates that were relatively moderately dependent on habitat suitability at vineyards nearby prior detections while being strongly dependent on habitat suitability at vineyards further away (Supp Fig. 10 [online only]). Overall, habitat suitability had a strong positive effect on invasion rate, with none of the sites showing P. ficus presence at the lowest values up to nearly 60% of sites being invaded at the highest values (Fig. 6A). Conversely, overall, lagged distance showed a nonsignificant negative trend, from approximately half of sites directly adjacent to prior trap detections having P. ficus to approximately 10% of sites furthest away (Fig. 6B). Fig. 6. Open in new tabDownload slide Main effects of (A) estimated habitat suitability and (B) distance to trap detections 1 yr prior (on an untransformed scale) on the proportion of vineyards where Planococcus ficus was present based on field surveys. Dashed line denotes the fit for a significant variable in the model. Fig. 6. Open in new tabDownload slide Main effects of (A) estimated habitat suitability and (B) distance to trap detections 1 yr prior (on an untransformed scale) on the proportion of vineyards where Planococcus ficus was present based on field surveys. Dashed line denotes the fit for a significant variable in the model. Discussion Understanding patterns of invader distribution can aid in developing detection, management, and eradication strategies (Thuiller et al. 2005, Jiménez-Valverde et al. 2011, Vicente et al. 2016). We found substantial heterogeneity in the distribution of P. ficus detections over the study region, with statistical hotspots representing those locations where P. ficus is well-established and likely to act as source populations from which individuals disperse to nearby uninvaded locations. Isolated hotspots of P. ficus detections became increasingly contiguous over the course of this study in a way that is consistent with a simultaneous increase in the pace and geographical extent of the invasion. In such cases, localized surveillance and management efforts targeting early, isolated hotspots may be cost-effective and crucial efforts for limiting subsequent spread along the regional invasion front (Baker 2017, Epanchin-Niell 2017). Yet, identifying those areas at highest risk of invasion requires deeper understanding of the factors underlying spatial heterogeneity in P. ficus invasion dynamics. Here, we investigated some of the processes that may explain P. ficus invasion dynamics and discuss their implications for future surveillance and management efforts. The magnitude and ‘shape’ of invader dispersal dynamics (i.e., invasion kernel) can strongly influence the patterns, pace, and outcomes of biological invasions (Wilson et al. 2009, Lindström et al. 2011). Our analysis of P. ficus trapping records found spatial autocorrelation peaked at distances of ~250 m, which was consistent with both the more conservative estimates of yearly spread and field survey results indicating invasion rates declined at vineyard sites over the first few hundred meters from previous trap detections of P. ficus. Collectively, these results support prior studies demonstrating that many mealybug species generally disperse short distances or settle on their natal host plants (Franco et al. 2009). Furthermore, these results indicate quarantine, treatment, and control measures applied at the scale of 10s of acres may be appropriate for containing local P. ficus spread and logistically achievable. However, some caution is warranted regarding this conclusion as P. ficus records were collected on a regular grid that limited our ability to estimate spatial autocorrelation at finer spatial scales and therefore may have biased inferences regarding the characteristic scale of P. ficus dispersal and appropriate management scales. The multitude of pathways by which invasive species are introduced into and disperse across an invaded range can generate highly variable patterns and rates of spread (Melbourne and Hastings 2009, Ochocki and Miller 2017, Sullivan et al. 2017). For example, anthropogenic pathways concerning live plant trade and movement have been implicated in facilitating the spread and introduction of other phytophagous arthropod invaders (Buchan and Padilla 1999, Lockwood et al. 2005, Hulme et al. 2008, Liebhold et al. 2012, Meurisse et al. 2018). An understanding of the primary invasion pathways is important for limiting future spread, preferably via local efforts that can reduce the potential for regional and landscape-level invasion impacts (Wilson et al. 2009, Epanchin-Niell 2017). The movement of plant material or farm equipment likely introduced P. ficus and facilitated its spread among California grape-growing regions, including Napa County (Haviland et al. 2005). While spatial autocorrelation of P. ficus trap detections in Napa County peaked at relatively short distances, significant spatial autocorrelation was detected at distances up to ~40 km, which is more consistent with occasional long-distance, human-assisted movement events than natural dispersal processes (Thomas et al. 2017). This result, in addition to the most liberal estimates of yearly P. ficus spread, suggest local control efforts around P. ficus detections may need to be implemented at larger, less achievable spatial scales (i.e., 100s of acres). Indeed, these longer potential dispersal distances and the idiosyncratic nature of yearly spread indicate that other anthropogenic dispersal mechanisms likely played an important role in regional P. ficus invasion dynamics. However, additional research is needed to clarify the specific mechanism (e.g., plant material, farm equipment, harvested fruit) and the phase of in the invasion over which that mechanism is most important (i.e., introduction versus postestablishment, within-region spread). We used habitat suitability modeling to understand how climatic, landscape, and anthropogenic factors contributed to the observed heterogeneity in P. ficus distributions in the study region. The amount of precipitation in the driest month, elevation, and trap distance to nearest winery were identified as the most important predictors of P. ficus occurrence. The relative importance and relationship between distance to winery and P. ficus presence, as well as estimates that the probability of P. ficus occurrence at winery-adjacent sites was twice that of more distant sites, further highlights the potential role of anthropogenic factors in contributing to P. ficus spread and is consistent with other invasive species spread dynamics (Buchan and Padilla 1999, Lockwood et al. 2005, Wilson et al, 2009, Thomas et al. 2017). Yet, our analyses found modest effects of distance to major roadways on P. ficus occurrence, a factor that similar studies have found to be important during the early phases of other arthropod invasions (Thomas et al. 2017). With respect to environmental conditions, temperature is generally considered to be the primary determinant of mealybug distribution (Daane et al. 2012). However, our modeling only indirectly captures temperature effects via other climatic and environmental variables (e.g., precipitation and elevation) due to extensive multicollinearity between candidate temperature predictors and those predictors ultimately included in the habitat suitability models. For example, increasing elevation corresponds with increased exposure to direct sunlight. Increased exposure to sunlight can raise environmental and grapevine temperatures, which can, in turn, increase egg and larval mortality rates for some grape pests (e.g., European grapevine moth, Lobesia botrana; Kiaeian Moosavi et al. 2017). Increased temperatures (up to 40°C) can decrease P. ficus developmental and generation times (Daane et al. 2012) and may account for the observed pattern of greater P. ficus occurrence at higher elevations. In addition, the observed importance of precipitation in the driest month may signify that grapevine health (Jackson and Lombard 1993) is an important factor underlying P. ficus distribution. Specifically, greater precipitation may correspond with an increased potential for grape fungal infections (Jackson and Lombard 1993) that reduce vine quality as a host for P. ficus. Such a precipitation–vine quality–P. ficus relationship could explain the observed negative relationship between precipitation in driest month and P. ficus occurrence. Correlative approaches that rely on occurrence records, such as species distribution or habitat suitability modeling, are often used to explain ongoing invasions or project invasion risk for locations ranging from a local habitat up to a global scale (Lockwood et al. 2005, Broennimann and Guisan 2008, Beaumont et al. 2009, Jimenez-Valverde et al. 2011, Brummer et al. 2013). Such risk mapping has proven useful in guiding surveillance programs and optimizing management (Peterson and Robins 2003, Gormley et al. 2011) but has also been criticized based on some of the assumptions made or inherent limitations in the approach (Araújo and Peterson 2012). One of the commonly cited limitations of these approaches in invasion biology is the challenge of independently validating model predictions (Barbett-Massin et al. 2018). Our evaluations of model performance indicated that the grand ensemble model provides a robust description of P. ficus occurrence in the region. This conclusion is bolstered further by field surveys that assessed empirically the explanatory power of the suitability modeling, the results of which confirmed important effects of model estimates of habitat suitability on P. ficus establishment. In other words, the habitat suitability model may prove useful in a risk assessment framework (Jimenez-Valverde et al. 2011). Such information can help optimize responses to the P. ficus invasion in the region (e.g., Gormley et al. 2011, Brummer et al. 2013), including the development of efficient and cost-effective surveillance programs (Epanchin-Niell et al. 2012, Berec et al. 2015, Epanchin-Niell 2017) and identification of locations most favorable for P. ficus (e.g., Iacarella et al. 2015) that should be targeted aggressively with existing control measures (Haviland et al. 2005, Daane et al. 2012) to limit further spread. Acknowledgments We thank the Napa Valley Winegrape Pest & Disease Control District for access to the geodatabase of Planococcus ficus records required by this project, A. Napolitano and the Napa County Agricultural Commissioner’s Office for sharing the trapping records and ancillary data, S. MacDonald for GIS support needed for field site selection and mapping, and those grape growers and vineyard managers that contributed to this research. This project was funded by a grant (17-0330-000-SA) from the Pierce’s Disease Control Program to MPD and MLC. References Cited Allouche , O. , A. Tsoar, and R. Kadmon. 2006 . Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS) . J Appl Ecol . 43 : 1223 – 1232 . Google Scholar Crossref Search ADS WorldCat Almeida , R. P. , K. M. Daane, V. A. Bell, G. K. Blaisdell, M. L. Cooper, E. Herrbach, and G. Pietersen. 2013 . Ecology and management of grapevine leafroll disease . Front. Microbiol . 4 : 94 . Google Scholar Crossref Search ADS PubMed WorldCat Araújo , M. B. , and A. T. Peterson. 2012 . Uses and misuses of bioclimatic envelope modeling . Ecology . 93 : 1527 – 1539 . Google Scholar Crossref Search ADS PubMed WorldCat Baddeley , A. , and R. Turner. 2005 . Spatstat: an R package for analyzing spatial point patterns . J. Stat. Soft . 12 : 1 – 42 . Google Scholar Crossref Search ADS WorldCat Baker , C. M . 2017 . Target the source: optimal spatiotemporal resource allocation for invasive species control . Con. Lett. , 10 : 41 – 48 . Google Scholar Crossref Search ADS WorldCat Baker , C. M. , and M. Bode. 2016 . Placing invasive species management in a spatiotemporal context . Ecol. Appl . 26 : 712 – 725 . Google Scholar Crossref Search ADS PubMed WorldCat Barbet-Massin , M. , Q. Rome, C. Villemant, and F. Courchamp. 2018 . Can species distribution models really predict the expansion of invasive species? PLoS One . 13 : e0193085 . Google Scholar Crossref Search ADS PubMed WorldCat Beaumont , L. J. , R. V. Gallagher, W. Thuiller, P. O. Downey, M. R. Leishman, and L. Hughes. 2009 . Different climatic envelopes among invasive populations may lead to underestimations of current and future biological invasions . Div. Distrib . 15 : 409 – 420 . Google Scholar Crossref Search ADS WorldCat Berec , L. , J. M. Kean, R. Epanchin-Niell, A. M. Liebhold, and R. G. Haight. 2015 . Designing efficient surveys: spatial arrangement of sample points for detection of invasive species . Biol. Invasions . 17 : 445 – 459 . Google Scholar Crossref Search ADS WorldCat Bivand , R. , A. Bernat, M. Carvalho, Y. Chun, C. Dormann, S. Dray, R. Halbersma, N. Lewin-Koh, J. Ma, and G. Millo. 2005 . The spdep package. Comprehensive R Archive Network, 05–83 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Breiman , L . 2001 . Random forests . Mach. Learn . 45 : 5 – 32 . Google Scholar Crossref Search ADS WorldCat Broennimann , O. , and A. Guisan. 2008 . Predicting current and future biological invasions: both native and invaded ranges matter . Biol. Lett . 4 : 585 – 589 . Google Scholar Crossref Search ADS PubMed WorldCat Brummer , T. J. , B. D. Maxwell, M. D. Higgs, and L. J. Rew. 2013 . Implementing and interpreting local -scale invasive species distribution models . Div. Distrib . 19 : 919 – 932 . Google Scholar Crossref Search ADS WorldCat Buchan , L. A. J. and D. K. Padilla. 1999 . Estimating the probability of long-distance overland dispersal of invading aquatic species . Ecol. Appl . 9 : 254 – 265 . Google Scholar Crossref Search ADS WorldCat Crawley , M. J . 2012 . The R book . John Wiley & Sons , Southern Gate, Chichester, West Sussex, United Kingdom . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Daane , K. M. , E. A. Weber, and W. J. Bentley. 2004a . Vine mealybug –formidable pest spreading through California vineyards . Pract. Wine. Vine . May/June: (www.practicalwinery.com) Google Scholar OpenURL Placeholder Text WorldCat Daane , K. M. , R. Malakar-Kuenen, and V. M. Walton. 2004b . Temperature development of Anagyrus pseudococci (Hymenoptera: Encyrtidae) as a parasitoid of the vine mealybug, Planococcus ficus (Homoptera: Pseudococcidae) . Biol. Control . 31 : 123 – 132 . Google Scholar Crossref Search ADS WorldCat Daane , K. M. , W. J. Bentley, J. G. Millar, V. M. Walton, M. L. Cooper, P. Biscay, and G. Y. Yokota. 2006 . Integrated management of mealybugs in California vineyards . pp. 235 – 252 . In P. G. Adsule, S. Indu, I. S. Sawant, and S. D. Shikhamany (eds.), Proceedings of the International Symposium on Grape Production and Processing , International Society for Horticultural Science , Madison, WI. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Daane , K. M. , K. R. Sime, J. Fallon, and M. L. Cooper. 2007 . Impacts of Argentine ants on mealybugs and their natural enemies in California’s coastal vineyards . Ecol. Entomol . 32 : 583 – 596 . Google Scholar Crossref Search ADS WorldCat Daane , K. M. , M. L. Cooper, S. V. Triapitsyn, V. M. Walton, G. Y. Yokota, D. R. Haviland, W. J. Bentley, K. Godfrey, and L. R. Wunderlich. 2008 . Vineyard managers and researchers seek sustainable solutions for mealybugs, a changing pest complex . Cal. Agric . 62 : 167 – 176 . Google Scholar Crossref Search ADS WorldCat Daane , K. M. , R. P. Almeida, V. A. Bell, J. T. Walker, M. Botton, M. Fallahzadeh, M. Mani, J. L. Miano, R. Sforza, V. M. Walton, et al. 2012 . Biology and management of mealybugs in vineyards, pp. 271 – 308 . In N. J. Bostanian, R. Isaacs, C. Vincent (eds.), Arthropod management in vineyards, Springer Science and Business Media , Dordrecht, the Netherlands . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC De Meyer , M. , M. P. Robertson, M. W. Mansell, S. Ekesi, K. Tsuruta, W. Mwaiko, J. F. Vayssières, and A. T. Peterson. 2010 . Ecological niche and potential geographic distribution of the invasive fruit fly Bactrocera invadens (Diptera, Tephritidae) . Bull. Entomol. Res . 100 : 35 – 48 . Google Scholar Crossref Search ADS PubMed WorldCat Deleon , L. , M. J. Brewer, I. L. Esquivel, and J. Halcomb. 2017 . Use of a geographic information system to produce pest monitoring maps for south Texas cotton and sorghum land managers . Crop Prot . 101 : 50 – 57 . Google Scholar Crossref Search ADS WorldCat Diniz-Filho , J. , and L. M. Bini. 2005 . Modelling geographical patterns in species richness using eigenvector-based spatial filters . Glob. Ecol. Biogeogr . 14 : 177 – 185 . Google Scholar Crossref Search ADS WorldCat Elith , J. , and J. R. Leathwick. 2009 . Species distribution models: ecological explanation and prediction across space and time . Ann. Rev. Ecol. Evol. Syst . 40 : 677 – 697 . Google Scholar Crossref Search ADS WorldCat Elith , J. , S. Ferrier, F. Huettmann, and J. Leathwick. 2005 . The evaluation strip: a new and robust method for plotting predicted responses from species distribution models . Ecol. Model . 186 : 280 – 289 . Google Scholar Crossref Search ADS WorldCat Epanchin-Niell , R. S . 2017 . Economics of invasive species policy and management . Biol. Invasions . 19 : 3333 – 3354 . Google Scholar Crossref Search ADS WorldCat Epanchin-Niell , R. S. , and J. E. Wilen. 2012 . Optimal spatial control of biological invasions . J. Environ. Econ. Manage . 63 : 260 – 270 . Google Scholar Crossref Search ADS WorldCat Franco , J. C. , A. Zada, and Z. Mendel. 2009 . Novel approaches for the management of mealybug pests . pp. 233 – 278 . In I. Ishaaya, A. R. Horowitz, (eds.), Biorational control of arthropod pests , Springer , Dordrecht, the Netherlands . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Friedman , J. H . 2001 . Greedy function approximation: a gradient boosting machine . Ann. Stat . 29 : 1189 – 1232 . Google Scholar Crossref Search ADS WorldCat Getis , A. , and D. A. Griffith. 2002 . Comparative spatial filtering in regression analysis . Geogr. Anal . 34 : 130 – 140 . Google Scholar Crossref Search ADS WorldCat Getis , A. , and J. K. Ord. 1992 . The analysis of spatial association by use of distance statistics . Geogr. Anal . 24 : 189 – 206 . Google Scholar Crossref Search ADS WorldCat Gill , R . 1994 . Vine mealybug . California plant pest and disease report, January-June . California Department of Food and Agriculture , Sacramento, CA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Godfrey , K. E. , K. M. Daane, W. J. Bentley, R. J. Gill, and R. Malakar-Kuenen. 2002 . Mealybugs in California vineyards . UCANR Publ. 21612, Oakland, CA Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Gormley , A. M. , D. M. Forsyth, P. Griffioen, M. Lindeman, D. S. Ramsey, M. P. Scroggie, and L. Woodford. 2011 . Using presence-only and presence-absence data to estimate the current and potential distributions of established invasive species . J. Appl. Ecol . 48 : 25 – 34 . Google Scholar Crossref Search ADS PubMed WorldCat Hahn , N. G. , C. Rodriguez-Saona, and G. C. Hamilton. 2017 . Characterizing the spatial distribution of brown marmorated stink bug, Halyomorpha halys Stål (Hemiptera: Pentatomidae), populations in peach orchards . PLoS One . 12 : e0170889 . Google Scholar Crossref Search ADS PubMed WorldCat Haviland , D. R. , W. J. Bentley, and K. M. Daane. 2005 . Hot water treatments to control Planococcus ficus (Hemiptera: Pseudococcidae) in grape nursery stock . J. Econ. Entomol . 98 : 1109 – 15 . Google Scholar Crossref Search ADS PubMed WorldCat Hulme , P. E. , S. Bacher, M. Kenis, S. Klotz, I. Kühn, D. Minchin, W. Nentwig, S. Olenin, V. Panov, J. Pergl, et al. 2008 . Grasping at the routes of biological invasions: a framework for integrating pathways into policy . J. Appl. Ecol . 45 : 403 – 414 . Google Scholar Crossref Search ADS WorldCat Iacarella , J. C. , J. T. Dick, M. E. Alexander, and A. Ricciardi. 2015 . Ecological impacts of invasive alien species along temperature gradients: testing the role of environmental matching . Ecol. Appl . 25 : 706 – 716 . Google Scholar Crossref Search ADS PubMed WorldCat Jackson , D. I. , and P. B. Lombard. 1993 . Environmental and management practices affecting grape composition and wine quality—a review . Am. J. Enol. Vitic . 44 : 409 – 430 . Google Scholar OpenURL Placeholder Text WorldCat Jimenez-Valverde , A. , A. Peterson, J. Soberon, J. M. Overton, P. Aragon, and J. M. Lobo. 2011 . Use of niche models in invasive species risk assessments . Biol. Invasions . 13 : 2785 – 2797 . Google Scholar Crossref Search ADS WorldCat Kiaeian Moosavi , F. , E. Cargnus, F. Pavan, and P. Zandigiacomo. 2017 . Mortality of eggs and newly hatched larvae of Lobesia botrana (Lepidoptera: Tortricidae) exposed to high temperatures in the laboratory . Environ. Entomol . 46 : 700 – 707 . Google Scholar Crossref Search ADS PubMed WorldCat Leung , B. , N. Roura-Pascual, S. Bacher, J. Heikkilä, L. Brotons, M. A. Burgman, K. Dehnen Schmutz, E. Essl, P. E. Hulme, et al. 2012 . Teasing apart alien-species risk assessments: a framework for best practices . Ecol. Lett . 15 : 1475 – 1493 . Google Scholar Crossref Search ADS PubMed WorldCat Liebhold , A. M. , E. G. Brockerhoff, L. J. Garrett, J. L. Parke, and K. O. Britton. 2012 . Live plant imports: the major pathway for forest insect and pathogen invasions of the US . Front. Ecol. Environ . 10 : 135 – 143 . Google Scholar Crossref Search ADS WorldCat Liebhold , A. M. , E. G. Brockerhoff, and M. A. Nuñez. 2017 . Biological invasions in forest ecosystems: a global problem requiring international and multidisciplinary integration . Biol. Invasions . 19 : 3073 – 3077 . Google Scholar Crossref Search ADS WorldCat Lindström , T. , N. Håkansson, and U. Wennergren. 2011 . The shape of the spatial kernel and its implications for biological invasions in patchy environments . Proc. Biol. Sci . 278 : 1564 – 1571 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Lockwood , J. L. , P. Cassey, and T. Blackburn. 2005 . The role of propagule pressure in explaining species invasions . Trends Ecol. Evol . 20 : 223 – 228 . Google Scholar Crossref Search ADS PubMed WorldCat Margosian , M. L. , K. A. Garrett, J. S. Hutchinson, and K. A. With. 2009 . Connectivity of the American agricultural landscape: assessing the national risk of crop pest and disease spread . BioScience . 59 : 141 – 151 . Google Scholar Crossref Search ADS WorldCat Melbourne , B. A. , and A. Hastings. 2009 . Highly variable spread rates in replicated biological invasions: fundamental limits to predictability . Science . 325 : 1536 – 1539 . Google Scholar Crossref Search ADS PubMed WorldCat Merow , C. , M. J. Smith, T. C. Edwards, A. Guisan, S. M. McMahon, S. Normand, W. Thuiller, R. O. Wüest, N. E. Zimmermann, and J. Elith. 2014 . What do we gain from simplicity versus complexity in species distribution models? Ecography . 37 : 1267 – 1281 . Google Scholar Crossref Search ADS WorldCat Meurisse , N. , D. Rassati, B. P. Hurley, E. G. Brockerhoff, and R. A. Haack. 2018 . Common pathways by which non-native forest insects move internationally and domestically . J. Pest Sci . 92 : 1 – 15 . Google Scholar OpenURL Placeholder Text WorldCat Millar , J. G. , K. M. Daane, J. S. McElfresh, J. A. Moreira, R. Malakar-Kuenen, M. Guillén, and W. J. Bentley. 2002 . Development and optimization of methods for using sex pheromone for monitoring the mealybug Planococcus ficus (Homoptera: Pseudococcidae) in California vineyards . J. Econ. Entomol . 95 : 706 – 714 . Google Scholar Crossref Search ADS PubMed WorldCat Murakami , D . 2017 . spmoran: an R package for Moran’s eigenvector-based spatial regression analysis . arXiv preprint arXiv:1703.04467. Murakami , D. , and D. A. Griffith. 2017 . Eigenvector spatial filtering for large data sets: fixed and random effects approaches . Geogr. Anal . 51 : 23 – 49 . Google Scholar Crossref Search ADS WorldCat Napa County Department of Agriculture and Weights and Measures. 2019 . 2019 Agricultural Crop Report . https://www.countyofnapa.org/DocumentCenter/View/17671/2019-Napa-County-Agricultural-Crop-Report-PDF?bidId=. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ochocki , B. M. , and T. E. Miller. 2017 . Rapid evolution of dispersal ability makes biological invasions faster and more variable . Nat. Commun . 8 : 14315 . Google Scholar Crossref Search ADS PubMed WorldCat Paini , D. R. , A. W. Sheppard, D. C. Cook, P. J. De Barro, S. P. Worner, and M. B. Thomas. 2016 . Global threat to agriculture from invasive species . Proc. Natl. Acad. Sci. U. S. A . 113 : 7575 – 7579 . Google Scholar Crossref Search ADS PubMed WorldCat Peterson , A. T. , and C. R. Robins. 2003 . Using ecological-niche modeling to predict barred owl invasions with implications for spotted owl conservation . Conserv. Biol . 17 : 1161 – 1165 . Google Scholar Crossref Search ADS WorldCat Pimentel , D. , R. Zuniga, and D. Morrison. 2005 . Update on the environmental and economic costs associated with alien-invasive species in the United States . Ecol. Econ . 52 : 273 – 288 . Google Scholar Crossref Search ADS WorldCat Pulliam , H. R . 2000 . On the relationship between niche and distribution . Ecol. Lett . 3 : 349 – 361 . Google Scholar Crossref Search ADS WorldCat R Development Core Team. 2017 . R: a language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria , ISBM 3900051-07-0, http://www.R-project.org. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rout , T. M. , R. Kirkwood, D. R. Sutherland, S. Murphy, and M. A. McCarthy. 2014 . When to declare successful eradication of an invasive predator? An. Conserv . 17 : 125 – 132 . Google Scholar Crossref Search ADS WorldCat Simberloff , D. , J. L. Martin, P. Genovesi, V. Maris, D. A. Wardle, J. Aronson, F. Courchamp, B. Galil, E. García-Berthou, M. Pascal, et al. 2013 . Impacts of biological invasions: what’s what and the way forward . Trends Ecol. Evol . 28 : 58 – 66 . Google Scholar Crossref Search ADS PubMed WorldCat Simmons , G. S. , L. Varela, M. Daugherty, M. Cooper, D. Lance, V. Mastro, et al. 2018 . Area wide eradication of the European grapevine moth, Lobesia botrana in California, USA . In Proceedings of the Third FAO-IAEA International Conference on Area-wide Management of Insect Pests: Integrating the Sterile Insect and Related Nuclear and Other Techniques. Vienna, Austria . http://www-naweb.iaea.org/nafa/ipc/Greg-Simmons.pdf. Soberon , J. , and A. T. Peterson. 2005 . Interpretation of models of fundamental ecological niches and species’ distributional areas . Biodivers. Inf . 2 : 1 – 10 . Google Scholar OpenURL Placeholder Text WorldCat Sullivan , L. L. , B. Li, T. E. X. Miller, M. G. Neubert, and A. K. Shaw. 2017 . Density dependence in demography and dispersal generates fluctuating invasion speeds . Proc. Natl. Acad. Sci. U. S. A . 114 : 5053 – 5058 . Google Scholar Crossref Search ADS PubMed WorldCat Svobodová , E. , M. Trnka, M. Dubrovský, D. Semerádová, J. Eitzinger, Z. Žalud, and P. Štěpánek. 2013 . Pest occurrence model in current climate–validation study for European domain . Acta Univ. Agric. Silvicult. Mendel. Brunensis . 61 : 205 – 214 . Google Scholar Crossref Search ADS WorldCat Thayn , J. B. , and J. M. Simanis. 2013 . Accounting for spatial autocorrelation in linear regression models using spatial filtering with eigenvectors . Ann. Am. Assoc. Geogr . 103 : 47 – 66 . Google Scholar Crossref Search ADS WorldCat Thomas , S. M. , G. S. Simmons, and M. P. Daugherty. 2017 . Spatiotemporal distribution of an invasive insect in an urban landscape: introduction, establishment and impact . Land. Ecol . 32 : 2041 – 2057 . Google Scholar Crossref Search ADS WorldCat Thorne , J. H. , J. A. Kennedy, J. F. Quinn, M. McCoy, T. Keeler-Wolf, and J. Menke. 2004 . A vegetation map of Napa County using the manual of California vegetation classification and its comparison to other digital vegetation maps . Madroño . 51: 343 – 363 . Google Scholar OpenURL Placeholder Text WorldCat Thuiller , W. , D. M. Richardson, P. Pyšek, G. F. Midgley, G. O. Hughes, and M. Rouget. 2005 . Niche-based modelling as a tool for predicting the risk of alien plant invasions at a global scale . Global. Change Biol . 11 : 2234 – 2250 . Google Scholar Crossref Search ADS WorldCat Thuiller , W. , D. Georges, R. Engler, and F. Breiner. 2016 . biomod2: ensemble platform for species distribution modeling. R package version 3.3-7 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tiefelsdorf , M. , and D. A. Griffith. 2007 . Semiparametric filtering of spatial autocorrelation: the eigenvector approach . Environ. Plan . 39 : 1193 – 221 . Google Scholar Crossref Search ADS WorldCat Tobin , P. C. , A. M. Liebhold, E. A. Roberts, and L. M. Blackburn. 2015 . Estimating spread rates of non-native species: the gypsy moth as a case study, pp. 131 – 144 . In R. C. Venette (ed.), Pest risk modelling and mapping for invasive alien species . CABI International and USDA , Wallingford, United Kingdom . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Tsai , C. W. , J. Chau, L. Fernandez, D. Bosco, K. M. Daane, and R. P. Almeida. 2008 . Transmission of grapevine leafroll-associated virus 3 by the vine mealybug (Planococcus ficus) . Phytopathology . 98 : 1093 – 1098 . Google Scholar Crossref Search ADS PubMed WorldCat Underwood-Russell , E. K. , A. D. Hollander, and K. Willett. 2001 . Napa County biodiversity mapping report . Information Center for the Environment, University of California , Davis, CA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Veloz , S. D . 2009 . Spatially autocorrelated sampling falsely inflates measures of accuracy for presence -only niche models . J. Biogeogr . 36 : 2290 – 2299 . Google Scholar Crossref Search ADS WorldCat Vicente , J. R. , A. S. Vaz, R. F. Fernandes, J. P. Honrado, D. Alagador, M. B. Araujo, C. Guerra, J. M. Alonso, C. Kueffer, C. Kueffer, et al. 2016 . Cost effective monitoring of biological invasions under global change: a model-based framework . J. Appl. Ecol . 53 : 1317 – 1329 . Google Scholar Crossref Search ADS WorldCat Walton , V. M. , and K. L. Pringle. 2004 . Vine mealybug, Planococcus ficus (Signoret)(Hemiptera: Pseudococcidae), a key pest in South African vineyards. A review . S. Afr. J. Enol. Vitic . 25 : 54 – 73 . Google Scholar OpenURL Placeholder Text WorldCat Walton , V. M. , K. M. Daane, W. J. Bentley, J. G. Millar, T. E. Larsen, and R. Malakar-Kuenen. 2006 . Pheromone-based mating disruption of Planococcus ficus (Hemiptera: Pseudococcidae) in California vineyards . J. Econ. Entomol . 99 : 1280 – 1290 . Google Scholar Crossref Search ADS PubMed WorldCat Wilson , J. R. U. , E. E. Dormontt, P. J. Prentis, A. J. Lowe, and D. M. Richardson. 2009 . Something in the way you move: dispersal pathways affect invasion success . Trends Ecol. Evol . 24 : 136 – 144 . Google Scholar Crossref Search ADS PubMed WorldCat Work , T. T. , D. G. McCullough, J. F. Cavey, and R. Komsa. 2005 . Arrival rate of nonindigenous insect species into the United States through foreign trade . Biol. Invasion . 7 : 323 – 332 . Google Scholar Crossref Search ADS WorldCat Worner , S. P. , and M. Gevrey. 2006 . Modelling global insect pest species assemblages to determine risk of invasion . J. Appl. Ecol . 43 : 858 – 867 . Google Scholar Crossref Search ADS WorldCat Zurell , D. , J. Franklin, C. König, P. J. Bouchet, C. F. Dormann, J. Elith, G. Fandos, X. Feng, G. Guillera-Arroita, A. Guisan, et al. 2020 . A standard protocol for reporting species distribution models . Ecography . 43 : 1 – 17 . Google Scholar Crossref Search ADS WorldCat © The Author(s) 2020. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Quantifying Planococcus ficus (Hemiptera: Pseudococcidae) Invasion in Northern California Vineyards to Inform Management Strategy
JO - Environmental Entomology
DO - 10.1093/ee/nvaa141
DA - 2021-02-17
UR - https://www.deepdyve.com/lp/oxford-university-press/quantifying-planococcus-ficus-hemiptera-pseudococcidae-invasion-in-B9vN33vBDc
SP - 138
EP - 148
VL - 50
IS - 1
DP - DeepDyve
ER -