Encroachment of shrubs into adjacent grasslands has become an increasingly reported phenomenon across the world. In many cases, such encroachment is either pulled forward by high population growth at the low-density encroachment front or pushed forward by higher-density areas behind the front. However, at sites such as Sevilleta National Wildlife Refuge in central New Mexico, little is known about whether encroachment is pushed or pulled, and the dynamics of encroachment are not well-understood. Here, long-term encroachment of creosotebush (Larrea tridentata), a native perennial shrub, stands in stark contrast with the stagnation in encroachment observed in recent decades.
In order to better understand creosotebush encroachment dynamics at the Sevilleta National Wildlife Refuge, we quantify this encroachment using a spatially structured population model where a wave of individuals travels at a speed governed by both dispersal and density-dependence. Density dependence was studied along a gradient of creosotebush densities, using observational and experimental approaches such as transplant experiments and annual surveys; dispersal was separately modelled using seed terminal velocity estimates in a wind-based dispersal framework. Demographic and dispersal data were then incorporated into the spatial population model, which estimates the speed of encroachment and determines whether it is pushed or pulled.
Results/Conclusions
Demographic analyses show that annual growth exhibited negative size- and density-dependence, while reproduction exhibited positive size-dependence but negative density-dependence. Probability of flowering and survival both exhibited positive size-dependence, with very weak density-dependence. Population growth rates generally increased with decreasing density, suggesting that encroachment was pulled by individuals at the low-density wave front.
Simulated dispersal kernels show that dispersal by wind was extremely limited in this species: the vast majority of seeds travelled less than 1 m from the parent plant and the most probable dispersal distance was only approximately 6 cm. Long-distance dispersal events by wind were exceedingly rare.
The spatial population model predicted that the wave moves at speeds less than 2 cm per year, which was consistent with observations of stagnant encroachment over recent decades. However, these results fail to explain long-term creosotebush encroachment at the study site, suggesting that this process may occur in pulses when recruitment, seedling survival, or dispersal significantly exceed typical rates. Overall, our work demonstrates that individuals at low densities are likely the biggest contributors to creosotebush encroachment at this site, and that this encroachment is likely a process that occurs in large but infrequent bursts rather than at a steady pace.