Landscape scale spread of alien invasive plants: what are the key drivers and how can human behaviour affect it?

Invasive plants disrupt ecosystems from local to landscape scales, and there is a pressing need to predict spread rates accurately in order to inform management. Models of plant spread can aid management by identifying key population and dispersal processes to target, as well as areas at risk of invasion. This thesis aims to determine which vital rates drive spread under a variety of conditions for different types of species and how much biological detail is required to accurately model spread. This thesis also asks how control of invasive plants at local scales translates to landscape scale control when managers cannot access the entire invasive population. In chapter 2 the main drivers of spread, and thus potential targets for management, were identified using a spatially explicit simulation model tested on different life history categories in different spread and landscape scenarios. From our results we deduced four simple management guidelines: i) manage dispersal if possible as mean dispersal distance was important for spread under all conditions tested, ii) short bursts of rapid spread or more usual year on year spread can have different drivers, therefore managers need to decide what type of spread they want to slow, iii) efforts to manage spread will have variable outcomes due to interactions between, and non-linear responses to, key drivers of spread and iv) the most useful demographic rates to target depend on dispersal ability, life history and how spread is measured. Fecundity was found to be important for driving spread only when reduced to low levels. For longer lived species management should target survival or age of first seed production, especially when dispersal ability is limited. Chapter 3 describes a field study on the reproductive ecology of a population of Pinus nigra, an invasive pine species, at a site in New Zealand. We describe how cone production varied among trees using a negative binomial or mixed gamma-exponential distribution. Both of these distributions were right skewed and trees maintained fecundity hierarchies over time; suggesting that some trees in the population have much higher lifetime reproduction than others. We found that trees dropped significantly more seeds when conditions were dry and windy, potentially increasing the proportion of seeds that disperse mid to long distances. We also found taller neighbours reduced cone production much more than shorter neighbours; indicating shade was the primary mechanism by which neighbours reduced cone production. In chapter 4 the effect that individual level variation in fecundity has on spread rates was tested with an individual based simulation, parameterised with demographic data from chapter 3 and dispersal data from the same site. We found that including heterogeneity in fecundity lowered predicted spread rates, but the differences were small. However, the overestimation of spread rates caused by assuming homogeneous fecundity accumulated over time, leading to large differences in the extent of the invasion after 250 time steps. Thus, early in an invasion simpler models may be adequate to capture average spread rates, but over longer time spans heterogeneity in fecundity should be included for accurate prediction of extent and invasion impact. Given that current models of plant spread assuming homogeneous fecundity are commonly used for sensitivity analysis, an important finding was that sensitivities were similar under heterogeneous and homogeneous fecundity. In many situations control of damaging alien invasive plants is undertaken by multiple independent decision makers, each managing only a small part of the invader’s population. Using a spatially explicit agent based simulation we determined how individual manager behaviour, in concert with weed ecology, determined the prevalence of two invasive grass species in Australia, Nassella trichotoma and Eragrostis curvula (chapter 5). We expect damaging weeds with effective control strategies to be less prevalent because the majority of land managers will choose to control them. However, if there is long distance dispersal a small minority (ca. 10%) of land mangers reluctant to control can impact the whole landscape. The way decision makers react to the benefit of management can have a large effect on the extent of a weed. If all agents believed control was a good idea and acted on it, they tended to act synchronously, reducing the pool of infested agents available to spread the weed. The general findings of this thesis are as follows: i) characteristics of the dispersal kernel are the most influential factors driving spread rates, but demographic rates such as fecundity, establishment probability, age of first seed production and the frequency of high seed production years can all substantially influence spread rates, ii) individual level variation in fecundity did not greatly reduce spread rates and simpler models appear adequate to describe spread in many situations, iii) when there are multiple managers the translation of local control to landscape scale reductions in spread have the potential to be greatly influenced by human behaviour, especially if long distance dispersal is possible