Methods for quantifying changes to Lesser Prairie-Chicken lek connectivity and habitat

Lesser prairie-chickens (Tympanuchus pallidicinctus) are an endangered bird of the Southern Great Plains (USA). They use spatially explicit locations called leks for reproductive activities; these leks manifest as a habitat network that females traverse to select mates. As the prairie-chicken population has declined in recent decades, leks have been abandoned, resulting in diminished connectivity that may further exacerbate declines. In this thesis, I examined how graph-theoretic multiscale connectivity was indicative of spatial patterns of lek persistence and decline through time. Additionally, because the connectivity of the prairie-chicken habitat network is thought to be negatively affected by woody plant encroachment, I developed a new workflow for using remotely sensed imagery to identify honey mesquite (Neltuma [=Prosopis] glandulosa) on the landscape; this work is published in the International Journal of Applied Earth Observation and Geoinformation (volume 136, article 104362, 2025). The prairie-chicken lek network represents a prime testing ground for examining how analyses of connectivity may help inform why and where population declines occur, which may be useful for predictive and preventative management. I compared how patterns of decline and abandonment may be associated with lek structural connectivity. I used data on locations and years of recorded activity of leks in the southern population of lesser prairie-chickens. Using the Python package igraph, I generated networks for a series of biologically relevant threshold distances representing a range of dispersal distances. Kleinberg’s hub and betweenness centrality metrics were calculated for each lek at each distance. I used centrality metrics to score each lek based on its contribution to the network, then I weighted the number of years in which lekking behaviors were observed. I also used these scores to build models that can predict whether a lek is active (occupied by lesser-prairie chickens) or not based on centrality score collected at multiple scales. I found that the number of years in which a lek was recorded as being active was positively associated with a lek’s betweenness centrality score and, to a lesser extent, hub score. These associations varied across scales, with leks having high persistence of activity being more central in smaller-scale clusters (maximum nesting distance), and leks known to have failed being less central at broader scales (average and maximum dispersal distances). I also found that centrality score could be used to predict whether a lek was active or not with 80% accuracy. The results of this analysis provide insight into how network connectivity may influence the decline or continued persistence of species reliant on spatially explicit features such as leks. The results of this work may also provide a basis for additional study of environmental factors, such as woody plant encroachment, that may constrain landscape connectivity for species such as lesser prairie-chickens. Curtailing encroachment is dependent on effectively identifying where problematic species occur. However, traditional classification methods struggle to distinguish spectrally similar species. New remote-sensing techniques that incorporate environmental variables (e.g, edaphic, climatic, and topographic characteristics) into classification can refine predictions and help identify important factors associated with species occurrence. I developed a workflow to improve classification of honey mesquite in the Southern Great Plains, examining 70 environmental variables to determine which were most associated with mesquite presence. I used Google Earth Engine to run X-means clustering on high-resolution aerial imagery from 50 replicate 78-km2 areas in New Mexico and Texas. I then refined the classification using XGBoost to generate accuracy assessment points for each area to confirm locations of mesquite clusters. This method improved classification accuracy from 36% to 83%. I then performed an ex-situ ground-truthed validation study that achieved 74% accuracy. Inclusion of environmental data increased the accuracy of mesquite classification and allowed me to estimate the influence of each variable in determining whether a given point was classified as mesquite. Shallow, alkaline soils with low water-storage capacity, high electrical conductance, and low cation exchange capacity were associated with mesquite presence; these areas tended to be associated with flat, low-elevation drainages in regions that experience wide annual temperature ranges. These methods provide an easily reproducible and scalable way to assist with image classification of rangeland shrubs from remotely sensed imagery, which may prove useful in managing the further encroachment of problematic species like honey mesquite.