Improving species distribution forecasts by measuring and communicating uncertainty: An invasive species case study.

Forecasting invasion risk under future climate conditions is critical for the effective management of invasive species, and species distribution models (SDMs) are key tools for doing so. However, SDM-based forecasts are uncertain, especially when correlative statistical models extrapolate to nonanalog environmental domains, such as future climate conditions. Different assumptions about the functional form of the temperature-suitability relationship can impact predicted habitat suitability under novel conditions. Hence, methods to understand the sources of uncertainty are critical when applying SDMs. Here, we use high-resolution predictions of lake water temperatures to project changes in habitat suitability under future climate conditions for an invasive macrophyte (Myriophyllym spicatum). Future suitability was predicted using five global circulation models and three statistical models that assumed different species-temperature functional responses. The suitability of lakes for M. spicatum was overall predicted to increase under future climate conditions, but the magnitude and direction of change in suitability varied greatly among lakes. Variability was most pronounced for lakes under nonanalog temperature conditions, indicating that predictions for these lakes remained highly uncertain. Integrating predictions from SDMs that differ in their species-environment response function, while explicitly quantifying uncertainty across analog and nonanalog domains, can provide a more robust and useful approach to forecasting invasive species distribution under climate change.

Quantifying Spillover of an Urban Invasive Vector of Plant Disease: Asian Citrus Psyllid (Diaphorina citri) in California Citrus.

Urban environments frequently play an important role in the initial stages of biological invasions, often serving as gateways for non-native species, which may propagate to nearby natural and agricultural ecosystems in the event of spillover. In California, citrus trees are a dominant ornamental and food plant in urban and peri-urban environments. We studied the invasion dynamics of the Asian citrus psyllid (Diaphorina citri), which became widespread in urban areas of southern California starting in 2008, to understand the factors driving its more recent invasion in commercial citrus groves. Using a multi-year monitoring database, we applied a suite of models to evaluate the rate at which groves accrued their first D. citri detection and the cumulative number of detections thereafter. Grove characteristics and landscape context proved to be important, with generally higher invasion rates and more cumulative detections in groves that were larger, had more edge, or had more perforated shapes, with greater urbanization intensity favoring more rapid invasion, but with inconsistent effects of distance to roads among models. Notably, distance to urban or other grove occurrences proved to be among the most important variables. During the early phase of D. citri invasion in the region, groves closer to urban occurrences were invaded more rapidly, whereas more recently, invasion rate depended primarily on proximity to grove occurrences. Yet, proximity to urban and grove occurrences contributed positively to cumulative D. citri detections, suggesting a continued influx from both sources. These results suggest that inherent features of agroecosystems and spatial coupling with urban ecosystems can be important, temporally dynamic, drivers of biological invasions. Further consideration of these issues may guide the development of strategic responses to D. citri's ongoing invasion.

Species distribution models for invasive Eurasian watermilfoil highlight the importance of data quality and limitations of discrimination accuracy metrics.

AIM: Availability of uniformly collected presence, absence, and abundance data remains a key challenge in species distribution modeling (SDM). For invasive species, abundance and impacts are highly variable across landscapes, and quality occurrence and abundance data are critical for predicting locations at high risk for invasion and impacts, respectively. We leverage a large aquatic vegetation dataset comprising point-level survey data that includes information on the invasive plant Myriophyllum spicatum (Eurasian watermilfoil) to: (a) develop SDMs to predict invasion and impact from environmental variables based on presence-absence, presence-only, and abundance data, and (b) compare evaluation metrics based on functional and discrimination accuracy for presence-absence and presence-only SDMs. LOCATION: Minnesota, USA. METHODS: Eurasian watermilfoil presence-absence and abundance information were gathered from 468 surveyed lakes, and 801 unsurveyed lakes were leveraged as pseudoabsences for presence-only models. A Random Forest algorithm was used to model the distribution and abundance of Eurasian watermilfoil as a function of lake-specific predictors, both with and without a spatial autocovariate. Occurrence-based SDMs were evaluated using conventional discrimination accuracy metrics and functional accuracy metrics assessing correlation between predicted suitability and observed abundance. RESULTS: Water temperature degree days and maximum lake depth were two leading predictors influencing both invasion risk and abundance, but they were relatively less important for predicting abundance than other water quality measures. Road density was a strong predictor of Eurasian watermilfoil invasion risk but not abundance. Model evaluations highlighted significant differences: Presence-absence models had high functional accuracy despite low discrimination accuracy, whereas presence-only models showed the opposite pattern. MAIN CONCLUSION: Complementing presence-absence data with abundance information offers a richer understanding of invasive Eurasian watermilfoil's ecological niche and enables evaluation of the model's functional accuracy. Conventional discrimination accuracy measures were misleading when models were developed using pseudoabsences. We thus caution against the overuse of presence-only models and suggest directing more effort toward systematic monitoring programs that yield high-quality data.

Climate and landscape conditions indirectly affect fish mercury levels by altering lake water chemistry and fish size.

Mercury pollution is a global environmental problem that threatens ecosystems, and negatively impacts human health and well-being. Mercury accumulation in fish within freshwater lakes is a complex process that appears to be driven by factors such as individual fish biology and water chemistry at the lake-scale, whereas, climate, and land-use/land-cover conditions within lake catchments can be influential at relatively larger scales. Nevertheless, unravelling the intricate network of pathways that govern how lake-scale and large-scale factors interact to affect mercury levels in fish remains an important scientific challenge. Using structural equation models (SEMs) and multiple long-term databases we identified direct and indirect effects of lake-scale and larger-scale factors on mercury levels in Walleye and Northern Pike - two species that are valued in inland fisheries. At the lake-level, the most parsimonious path models contained direct effects of fish weight, DOC, and pH, as well as an indirect effect of DOC on fish mercury levels via fish weight. Interestingly, lakeshed-, climate-, and full-path models that combine the effects of both lakeshed and climate revealed indirect effects of surrounding landscape conditions and latitude via DOC, pH, and fish weight but no direct effects on fish mercury levels. These results are generally consistent across species and lakes, except for some differences between stratified and non-stratified lakes. Our findings imply that understanding climate and land-use driven alterations of water chemistry and fish biology will be critical to predicting and mitigating fish mercury bioaccumulation in the future.

Spatiotemporal dynamics of the Southern California Asian citrus psyllid (Diaphorina citri) invasion.

Biological invasions are governed by spatial processes that tend to be distributed in non-random ways across landscapes. Characterizing the spatial and temporal heterogeneities of the introduction, establishment, and spread of non-native insect species is a key aspect of effectively managing their geographic expansion. The Asian citrus psyllid (Diaphorina citri), a vector of the bacterium associated with huanglongbing (HLB), poses a serious threat to commercial and residential citrus trees. In 2008, D. citri first began expanding northward from Mexico into parts of Southern California. Using georeferenced D. citri occurrence data from 2008-2014, we sought to better understand the extent of the geographic expansion of this invasive vector species. Our objectives were to: 1) describe the spatial and temporal distribution of D. citri in Southern California, 2) identify the locations of statistically significant D. citri hotspots, and 3) quantify the dynamics of anisotropic spread. We found clear evidence that the spatial and temporal distribution of D. citri in Southern California is non-random. Further, we identified the existence of statistically significant hotspots of D. citri occurrence and described the anisotropic dispersion across the Southern California landscape. For example, the dominant hotspot surrounding Los Angeles showed rapid and strongly asymmetric spread to the south and east. Our study demonstrates the feasibility of quantitative invasive insect risk assessment with the application of a spatial epidemiology framework.