Biological invasion, the spread of non-native organisms, is occurring rapidly worldwide, and many desert areas currently show a dramatic increase in the arrival and spread of non-native invasive plant species. Among the detrimental effects are alterations in fire regimes and direct negative impacts on native plant species performance. Prior to invasion, native annual plants were mostly restricted to nutrient-rich areas under desert shrubs and infrequently occurred within the open areas between shrubs. This project examined the hypothesis that some of the now dominant and problematic non-native invasive annuals are able to spread into the areas between the shrubs by employing population strategies that sharply contrast with those of native species. This greatly increases the fuel load in the matrix, which has historically produced a natural firebreak between shrubs. This project’s particular aims were to: (1) gain an understanding of the landscape-scale population dynamics of fire promoting and fire retarding plant species; (2) test the hypothesis that once fire becomes important, naturally formed islands of fertility will break down and a negative feedback will enhance fire even further; (3) apply the results to aid management practices that will help restore the original environmental pattern of islands of fertility in a low-nutrient matrix and therefore prevent future wildfires; and (4) understand the effects of non-native invasive plant species on fire regimes.

This project explored an apparently novel population dynamics strategy of non-native invasive plant species in southwestern United States deserts, which is not used by native species. Field observational and controlled experimental studies in two contrasting desert sites (Mojave and Sonoran Deserts) were set up to parameterize detailed, landscape-scale, spatially explicit population models. The major components of the approach were: experimental studies to obtain demographic data under different environmental conditions for several target species, both non-native invasive and native; characterization of spatial patterns of fertility using data from the experimental studies; development of landscape-scale, spatially explicit simulation models of the spread of non-native invasive species (NIS) in matrix habitat, based on parameters obtained in the experimental studies; and simulation studies of fire spread and efficacy of different management strategies under varying climatic regimes, based on a ground-truthed version of the simulation models.

In the initial years of the project, permanent research sites were established in creosote bush communities in the Sonoran Desert (Barry M. Goldwater Range) and the Mojave Desert (Fort Irwin) and spatially explicit density data was gathered for shrubs and herbaceous plants. The team explored how the spatial pattern of shrubs and native and non-native herbaceous plants can lead to desert wildfires. In the following years, further investigations occurred through experimental studies as to whether these initial conclusions held and whether the underlying mechanisms for this could be elucidated. These experiments examined the combined effects of fire, disturbance, and precipitation amounts on the demographics of native and non-native invasive annual plants. Factorial experiments were conducted to determine the effects of fire, rainfall change, seed limitation, and disturbance on the populations of native and non-native desert annuals. Methods included burning individual shrubs, installing rain-out shelters and irrigating plots to mimic changing rainfall amount, experimentally disturbing soils, and adding seeds of already present NIS to the experimental sites. Spatial patterns of fertility and soil moisture availability were characterized using data from the experimental studies. Basic landscape-scale, spatially explicit simulation models of the spread of NIS in matrix habitat have been developed based on initial parameters obtained in the experimental studies.

Project data on annual and perennial plant densities and their spatial distributions suggest that different processes have the potential to promote fire in the two contrasting desert sites. In the Mojave the rise of NIS also occupying the areas between shrubs indeed has the potential to promote fire. In the Sonoran Desert native species also occupy the areas between shrubs and potentially provide enough fuel to carry wildfires. In addition, higher shrub densities and lesser shrub segregation in the Sonoran Desert might be the key factor for promoting wildfires, even in the absence of NIS. Population parameter studies in the Mojave and Sonoran Deserts are consistent with the hypothesis that one of the populations of the primary NIS in the study sites (Schismus arabicus) indeed uses a strategy that has elements of source-sink dynamics in the Mojave Desert, but not in the Sonoran Desert. In the Mojave Schismus maintains high densities in the area between shrubs that seem to be supported by higher seed production under shrub canopies.

Responses of annual plant populations and communities to the treatments show that native species and NIS in both deserts react differently. As expected, annual biomass increased in both deserts with increasing rainfall; however, fire increased biomass only in the Sonoran and had little effect in the Mojave. Disturbance had strong increasing effects on biomass in both deserts. In the Mojave Desert disturbance and decreasing rainfall favored NIS; whereas, burns did not cause a relative increase of NIS. In contrast, annual NIS in the Sonoran Desert did not become more abundant with fire, most likely due to a strong reaction of native species; rather, they increased with disturbance and drought. This suggests that the invasion processes differ between the deserts and that the impacts of fire are regionally quite different.

This project explored the dynamics of fire spread in a simple, fairly abstract version of FireGrid, the simulation model developed for this project. The results are consistent with percolation theory from landscape ecology and suggest that catastrophic spread of fire will only occur if 60% or more of an area has enough fuel to burn. However, the potential for fire spread will be altered dramatically depending upon the flammability state of the creosote shrubs, which can help connect the landscape even when adequate fuel loads between shrubs are below 60% coverage. The latter will occur when creosote is in a high state of flammability. Under conditions when creosote is less prone to catch fire, it can actually act to slow fire spread under otherwise appropriate conditions. Another factor that can play a role in fire spread is the rate at which fire moves through the annual litter layer between shrubs. This can be lowered when the hydraulic status of the site is elevated or if plants are not evenly distributed at short distances. Under these circumstances fire only spreads under conditions of greater fuel loads than predicted by traditional percolation theory. There is also a large stochastic element to the process. Even if a fire can easily spread through the landscape, given the current fuel loads and potential of spread, there is still a great deal of variability in the degree of spread that occurs from a localized fire source depending on the local distribution of fuels. This makes prediction of fire spread in any one location less precise when considering individual events.

The results derived from this project’s experimental and simulation modeling approaches facilitate a better understanding of the association between annual plants and desert shrubs with respect to key interactions and the development of spatial patterns that may influence fire risk. It also provides insights into the different role the exotic species Schismus arabicus plays in fire spread within the Mojave and Sonoran Desert sites. This understanding is a first step in characterizing the interaction of fire and soil disturbance in changing the likelihood of future fire occurrences through direct influence on the creosote shrub plant community.