An empirical assessment of a Linaria dalmatica invasion in Yellowstone National Park

Background/Question/Methods

There are few invasions of non-indigenous plant species (NIS) where a new species initial location is documented then its locations recorded in following years. We discovered archived maps and information on the distribution of Linaria dalmatica (nonindigenous plant species) in Yellowstone National Park in the early stages of invasion (1946-1970). The maps were digitized and georectified. The region was also sampled for L. dalmatica occurrence from 2001 to 2007.  Our research questions were: 1) using the colonization and extinction rates from 1968 to 1970 how well could we predict the spatial and temporal pattern of invasion from the origin in 1946 to the observed in 1968, 1970 and 2007? 2) What were the primary driving variables for L. dalmatica colonization and extinction from 1968 to 1970? 3) What management strategy in 1968 and 1970 might have successfully arrested the invasion? To address these questions, competing occupancy models were constructed from the 1968 and 1970 data. Probabilities of patch colonization and extinction (changes in patch-specific occupancy status from 1968 to 1970) were estimated using generalized linear models that consider various combinations of potential driving variables. Simulations of the invasion were conducted using the colonization and extinction rates.

Results/Conclusions

The best supported models of occupancy dynamics indicated that the probability of new colonization of L. dalmatica populations was associated with habitat suitability, HS (+)¹ based on digital elevation map properties (slope, aspect, elevation, annual solar radiation, distance to stream), mean distance to nearest source patches (-), mean HS associated with the source patches (+), and distance to road (-). Probability of patch extinction was associated with HS (-), distance to road (+) and distance to trails (-).When the top model was used with the observed field covariates to simulate the invasion from 1946 to 1970, metapopulation number was over-predicted for 1968 and 1970, which suggests that the invasion might not have occurred at a constant rate (i.e. might have been episodic). These results suggest that the lag phase may involve source strength and dispersal limitations at the beginning of the invasion of this species. Simulation of different early-invasion management strategies applied when the L. dalmatica was first recognized to be problematic indicated that killing populations on the outer edge of the invasion might have been most effective at slowing the invasion.

¹The sign in the parentheses indicated the sign on the regression fitted parameter estimate.