Developing a Detection Method for New Invaders at the Landscape Scale

By Lisa C. Jones and Timothy Prather, University of Idaho. Presented at the Western Society of Weed Science Annual Meeting, 2018.

Abstract: The ability to predict plant invasions and detect them early in the process are important considerations for invasive plant management. While agencies and landowners typically take the approach of on-the-ground searches and some may utilize habitat suitability models, these tools may not facilitate detection of incipient infestations when the species is unknown. We set out to develop a method to identify where to look for a new invader to assist managers in focusing search efforts to areas more prone to invasion. We used habitat suitability models (also referred to as species-specific susceptibility models) of seven plant species to investigate whether creating weed “hotspots” of overlapping models was an effective tool to infer areas more invaded within the boundaries of a 4,200-ha ranch in southern Idaho. We tested this by sampling vegetation cover by species, in five, 0.125 m2 quadrats placed along each of 24 transects located in areas modeled to be suitable habitat for either zero, two, four, or six weed species located in the northeast section of the ranch. Since it is well-documented that roads and trails provide corridors for dispersal, we located transects either near (within 60 m) or far (more than 60 m) from unimproved roads. We hypothesized that non-native species richness and/or cover would be higher in hotspots where a greater number of suitability models overlapped closer to roads. Of the 46 unique species in our quadrats, five species (11%) were non-native, of which Japanese brome (Bromus japonicus) and downy brome (Bromus tectorum) were the most abundant. Among non-native species, there was no significant difference in richness or foliar cover between hotspots or proximity to roads. Among native species, richness and foliar cover were not significantly different between hotspots, but they were curiously greater in transects closer to roads. To further aid the development of a detection method for new invaders, we examined indicator species that are positively or negatively associated with Japanese and downy brome. Notably, when downy brome cover was high, two perennial native forbs were in greater abundance, and when downy brome was not present, Sandberg’s bluegrass (Poa secunda) cover was high. There were no positive indicator species for Japanese brome, though there were 11 native species negatively associated with it. Overall, our initial foray to develop a detection method using existing weed habitat suitability models was not successful in identifying areas at greater risk of invasion as evidenced by current diversity and cover of non-native species. However, we recognize the limits of our small sample size and narrow extent of the area surveyed (15% of the ranch). Identifying sites at high risk to invasion when the life history traits and environmental niche of the invader is unknown is a complex challenge, but one that has the potential to help land managers prioritize areas for invasive plant monitoring. Future tests will investigate if there are specific modeled weed species combinations that are suggestive of areas generally susceptible to invasion; for example, more non-native species were along transects located where leafy spurge (Euphorbia esula) habitat was predicted. Further, indicator species may be used to reveal which models are better candidates for estimating invasibility.