Mapping vegetation species is critical to facilitate related quantitative assessment, and mapping invasive plants is important to enhance monitoring and management activities. Integrating high-resolution multispectral remote-sensing (RS) images and lidar (light detection and ranging) point clouds can provide robust features for vegetation mapping. However, using multiple sources of high-resolution RS data for vegetation mapping on a large spatial scale can be both computationally and sampling intensive. Here, we designed a two-step classification workflow to potentially decrease computational cost and sampling effort and to increase classification accuracy by integrating multispectral and lidar data in order to derive spectral, textural, and structural features for mapping target vegetation species. We used this workflow to classify kudzu, an aggressive invasive vine, in the entire Knox County (1362 km²) of Tennessee (U.S.). Object-based image analysis was conducted in the workflow. The first-step classification used 320 kudzu samples and extensive, coarsely labeled samples (based on national land cover) to generate an overprediction map of kudzu using random forest (RF). For the second step, 350 samples were randomly extracted from the overpredicted kudzu and labeled manually for the final prediction using RF and support vector machine (SVM). Computationally intensive features were only used for the second-step classification. SVM had constantly better accuracy than RF, and the producer’s accuracy, user’s accuracy, and Kappa for the SVM model on kudzu were 0.94, 0.96, and 0.90, respectively. SVM predicted 1010 kudzu patches covering 1.29 km² in Knox County. We found the sample size of kudzu used for algorithm training impacted the accuracy and number of kudzu predicted. The proposed workflow could also improve sampling efficiency and specificity. Our workflow had much higher accuracy than the traditional method conducted in this research, and could be easily implemented to map kudzu in other regions as well as map other vegetation species.

Invasive species experience biotic and abiotic conditions that may (or may not) resemble their native environment. We explored the methodology of determining climatic niches and compared the native and post-invasion niches of four invasive forest pests to determine if these species experienced shifts or changes in their new climatic niches. We used environmental principle components analysis (PCA-env) method to quantify climatic niche shifts, expansions, and temporal changes. Furthermore, we assessed the effect of variable selection in the delineation and comparison of niche space. We found that variable selection influenced the delineation and overlap of each niche, whereas the subset of climatic variables selected from the first two PCA-env axes explained more variance in environmental conditions than the complete set of climatic variables for all four species. Most focal species showed climatic niche shifts in their invasive range and had not yet fully occupied the available niche within the invaded range. Our species varied the proportion of niche overlap between the native and invasive ranges. By comparing native and invasive niches, we can help predict a species’ potential range expansion and invasion potential. Our results can guide monitoring and help inform management of these and other invasive species.

By the end of 2017, kudzu bug was reported in 652 counties in the United States since it was first observed in Georgia in 2009. Modeling its invasion dynamics is valuable to guide management through early detection and prevention of further invasion. Herein, we initially estimated the spread rate of kudzu bug with county-level invasion records and then determined important spatial factors affecting its spread during years 2010-2016. As kudzu bug infests a large heterogeneous area and shows asymmetric spread, we first utilized spatially constrained clustering (SCC), an unsupervised machine learning method, to divide the infested area into eight spatially contiguous and environmentally homogenous neighborhoods. We then used distance regression and boundary displacement methods to estimate the spread rates in all neighborhoods. Finally, we applied multiple regression to determine spatial factors influencing the spread of kudzu bug. The average spread rate reached 76 km/yr by boundary displacement method; however, the rate varied largely among eight neighborhoods (45-144 km/yr). In the southern region of the infested area, host plant density and wind speed were positively associated with the spread rate, whereas mean annual temperature, precipitation in the fall, and elevation had inverse relationships. In the northern region, January minimum temperature, wind speed, and human population density showed positive relationships. This study increases the knowledge on the spread dynamics of kudzu bug. Our research highlights the utility of SCC to determine natural clustering in a large heterogeneous region for better modeling of local spread patterns and determining important factors affecting the invasions.

Biological invasions are an ongoing threat for sustainability of ecosystems, and estimating the spread of invasive species is critical for making management decisions. Geopolitical unit-level data (GULD) are often used to estimate invasions due to their wide availability, and researchers had evaluated the abilities of multiple methods to estimate invasion with GULD. However, earlier studies were case based and only addressed limited information on the spread, thus making it inadequate to determine which method to choose to estimate invasions with GULD under various spread scenarios. Here, we conducted a simulation study to (1) evaluate performances of eight methods on estimating expansion patterns, spread rates, and spread dynamics of invasive species with GULD; (2) assess the impact of size and homogeneity of size of geopolitical unit on the estimations by studied methods; (3) evaluate the similarities of all studied methods. Additionally, we presented a concave hull boundary displacement method (Ctd_BD) and an area-based regression method (SqrtNA_R) to estimate spread with GULD. Three regions with varying sizes of counties in the United States (U.S.) were selected to conduct the simulations, and three spread scenarios and three expansion patterns were simulated. AIC, and R² and root mean square error (RMSE) were used to evaluate the accuracy of methods on estimating expansion pattern, and overall spread rate and spread dynamics, respectively. Correlation coefficient and RMSE were used to assess the similarity of eight methods. We found Ctd_BD and area-based regression methods consistently estimated the right expansion patterns. Boundary displacement and area-based regression methods estimated highly correlated spread rates and dynamics. Distance-based regression methods provided a high accuracy on estimating overall spread rate without long-distance jump dispersal but performed poorly on estimating the spread dynamics. We recommend boundary displacement method, especially Ctd_BD, for estimating spread with GULD, whereas for spread without clear infestation boundaries, distance-based regression can be used to estimate overall spread rate and area-based regression can be used to estimate spread dynamics.