Space invaders: modelling the distribution, impacts and control of alien organisms W. Richard and J. Dean Trends in Ecology and Evolution 1998, 13:256-258 Percy FitzPatrick Institute of African Ornithology, University of Cape Town, Rondebosch 7701, South Africa

Invasive alien plants and other organisms have become a worldwide problem, threatening ecosystem functioning, biodiversity, the integrity of species, water availability and the attractiveness of natural areas. The management of aliens is an expensive business in developed countries. Economically important plants, such as Pinus spp. , have become invasive in some ecosystems. In developing countries, aid agencies recommend the planting of the very species that conservation agencies spend time and energy trying to control. Gardeners and plant nurseries import and export seeds of invasive plants without concern about the consequences. Causes of conflicting behaviour include inadequate policies, inability to predict which plants will be invasive and differences in perceptions.

The first SCOPE (Scientific Committee on Problems of the Environment) programme (1984–1986) sought to provide answers to some fundamental questions: (1) what factors determine whether a species is an invader or not, (2) what site properties determine whether an ecosystem will be relatively prone to invasion, and (3) how should management systems be developed using the knowledge gained from answering these questions? This programme stimulated a large amount of new research on the ecology of invasions, resulting in several important books and numerous papers. Even so, however helpful the insights gained from this programme may have been, they have not really fed into meaningful control programmes. At a global scale, the problem with invasive alien plants and animals has increased phenomenally over the past two decades. There is still a serious lack of essential technical tools to deal with the problem.

With these facts in mind, the Global Invasives Strategy Project, under the auspices of a new SCOPE programme, aims to (1) draw together the best management approaches for pest prevention and control and make these readily accesible to all nations, and (2) lay the groundwork for new tools in science, information management, education and policy that must be developed through collaborative international action. Project leaders in 11 topic areas will work with international teams to complete these tasks over the two-year (1998–2000) project timetable. The results of this work will be disseminated via published reports, international meetings, and, especially, through a network of information exchange and training to be developed as part of this project. The goal is to implement an effective strategy, not simply to record ideas on paper. Marcel Rejmánek (University of California, Davis, USA) and Dave Richardson (University of Cape Town, South Africa) were appointed as co-leaders of the section of the project dealing with the ecology of invasives and will be convening several international fora over the next year or two to consider the important question of exactly what we need to know about the ecology of invasives in order to have a meaningful input into control operations that really will make a difference at a global scale.

As part of this initiative, a workshop organized by Dave Richardson at the Council for Scientific and Industrial Research (C.S.I.R.) in Stellenbosch, South Africa, in March 1998 considered whether the currently available techniques of modelling are an effective way of making our knowledge of the ecology of invasives useful 1. The C.S.I.R., through its Division of Water, Environment and Forestry Technology (Environmentek), is providing scientific input into one of the largest alien plant control programmes in the world—the South African Working for Water Programme, designed not only to control aliens, but also to restore water production in catchment areas and to provide much needed employment and training. Modelling is being used in the programme to estimate costs, evaluate results of control methods, and to identify key factors in alien plant invasions.

Speakers at the workshop included researchers who had modelled the rate and extent of plant and animal invasions, who had estimated costs and benefits of controlling aliens and who had developed expert systems for screening methods for potential invasive plants. Marcel Rejmánek first discussed the population dynamics of different invasive alien species—populations of some translocated species rapidly build up and then decline but remain present (and potentially invasive) in ecosystems. Populations of other species increase slowly and then build up to large numbers. Translocated species that have wide latitudinal and altitudinal ranges in their native environments are more likely to be invasive in new environments. An exception is Pinus radiata, which has a small distributional range in its native California but has become invasive in mountainous areas of southwestern South Africa. Rates of invasions in pines are related, in part, to size of seed crop, interval between successive seed crops, mean age of first reproduction and dispersal distance2. Small seeds, short generation times and short intervals between large seed crops all increase invasive ability in this group. Vertebrate-dispersed large seeded species follow different rules to those derived for pines.

The genome size of plants is inversely related to invasive ability, determining factors such as growth rate and age of first reproduction. Species with small genomes tend to show r-selected traits, which promote the ability to exploit disturbances. Invasive species may have a `general purpose' genotype3 that enables species to reproduce successfully over a wide range of habitats. Genetic polymorphism plus phenotypic plasticity leads to population fitness homeostasis in some non-native species, with the resulting ability to edge out native species. Phylogenetic distance from native species in any ecosystem may also be important—species more distant from native species may be successful invaders of the native species habitat, either because they are incompatible with local predators and pathogens or because they use resources only marginally used by native plants. Invasive potential of plants is also related to dispersal mechanisms, and many of the most invasive species of plants are vertebrate dispersed. Vertebrate dispersal, particularly of fleshy fruits, allows alien species to penetrate natural habitats, from where they may be difficult to dislodge.

While Rejmánek concentrated on identifying the attributes that define successful invaders, Steve Higgins (Institute for Plant Conservation, University of Cape Town, South Africa) attempted to predict the rates and patterns of invasive plant spread. Two types of model—reaction-diffusion analytical and simulated cellular automata—may be appropriate for modelling spread rates. The ability of these two modelling paradigms to predict the spread of Pinus spp. invading the macchia type vegetation (fynbos) of south-western South Africa was contrasted. The ability of the simulation model to explicitly simulate the event-driven nature of fynbos invasions meant that its predictions were strongly influenced by interactions between plant attributes and fire return intervals. This, in effect, meant that under some conditions the simulation model predicted faster rates of invasion than the reaction-diffusion model.

Neither of these models simulated rare long-distance dispersal. Long-distance dispersal events are disproportionately important biologically, and they allow alien plants to establish colonies within the matrix of natural vegetation, where new centres of dispersal develop. Higgins showed how the stratified nature of invasions could be modelled by considering dispersal as a mixture of statistical distributions that describe local and long-distance dispersal. The predictions of the simulation models were compared with a series of aerial photographs from 1938 to the early 1990s. This showed that the spread of Pinus pinaster closely matched model predictions of the rate and pattern of invasion of this species at a local scale. Important differences between the two modelling paradigms are that reaction-diffusion models seem to work for large areas (at a regional and continental scale: 10²–10⁶km²) and where the species net productive rate is fairly well understood. But when the details (such as processes) or dispersal patterns in smaller areas become important considerations, then simulation models become essential. Spatial scale is thus a critical factor.

In exploring the tactical applications of models, Higgins first outlined the basics—which species, with what probability, are likely to invade which environments, and how far will they go? Incorporating a gradient of disturbance, from natural disturbance regimes to highly modified disturbance regimes, in models illustrated the classic result: disturbance promotes invasions. However, Higgins showed why this generalization fails in many cases; interactions between disturbance level, plant attributes and environment greatly influences the results of the model. Modelling the plant-environment interactions may be of heuristic value—allowing understanding of the invasion process. Finally, two scenarios of modelling control methods were explored—either clearing sparse-outlier stands first and then dense stands, or vice versa. Lending support to Moody and Mack's4 theoretical model, this biological model suggests that if dense stands are cleared first, plants can invade the cleared areas faster than they can be cleared. However, once invasion rates exceed a certain threshold, clearing tactics become irrelevant.

Animal invasions differ from plant invasions only in that animals can abandon patches of poor quality habitat. Rob Hengeveld (Institute for Nature Studies, Wageningen, The Netherlands) talked about the potential and limitations of modelling invasions by animals, and once again emphasized the importance of rare long-distance dispersal events. Optimality of habitat can greatly influence movements in animals, and a study on the movement of muskrats (Ondatra zibethicus) across Germany5 showed that the invasion of this alien species was at variable rates, because of climatic variability and the patchiness of habitat. Because this species allocates more energy to reproduction in wet years (and in optimal habitats) and more energy to movement in dry years, rates of invasion are determined by climate and the dispersion of optimal habitat. Invasive terrestrial animals may thus move through suboptimal habitats faster than optimal habitats, so that predictions generated using homogeneous habitats may be quite unrealistic. For other species, however, observed and expected invasion rates based only on rates of reproduction and dispersal (and assuming homogeneous habitat) are reasonably well-correlated.

Water is scarce in South Africa, and modelling the effect of alien plant removal from catchment areas and rivers is one of the priority projects. Arthur Chapman (Environmentek) presented some startling statistics of the extent of alien invasions in catchment areas in southwestern South Africa (some 97 000 ha are invaded in one catchment area alone) and the impact of alien trees on run-off water (33.3 million m³ of water is lost annually because of alien trees). Biological control of alien trees has been effective for some species but is sometimes slow. However, there is a pressing need to clear alien trees to stop them spreading. Anticipated costs of removing alien trees from catchment areas are very great and resources need to be efficiently allocated to get the maximum return on the money. Against this background, Environmentek are working on modelling the rate of spread of alien trees, the impact on water yields, the cost of clearing and the need for follow-up operations to remove seedlings. The model, as yet at an early stage, is complex and incorporates the difficulties of physical access into mountain catchment areas (where clearing teams must be flown in by helicopter) and the cost to local (human) communities of clearing alien trees (e.g. loss of an important source of firewood).

Christo Marais (Working for Water Programme, Department of Water Affairs and Forestry, South Africa) presented a model that predicted the costs and benefits (in terms of water yield) of management impacts on alien plant invasions in inland and coastal mountain catchment areas of the Western Cape in South Africa. Marais adopted a state-and-transition approach to model invasions as influenced by fires. This model was linked to extensive surveys of the distribution and density of alien plants in the catchments of this region. By simulating the clearing of catchments, the benefits of clearing in terms of increased water yield could be predicted. Clearing increased water yield, and in higher-rainfall catchments the value of the increase in water yield was greater than the cost of clearing. However, in the drier inland catchments, the value of the increases in water yield may not cover the cost of clearing. The conclusion is that cost-benefit analysis may be a useful way of demonstrating the social benefits of alien tree removal, provided the perception of the benefit is wider than water yield.

An expert-systems approach, with rules based on our understanding of invasion processes, was used by Kim Tucker (Environmentek) to screen potential woody invaders in the fynbos biome. The system starts by comparing broad-scale environmental conditions of the species' native habitat with fynbos, and if not ruled out by these criteria, assesses a range of characteristics, including basic life history traits (including dispersal syndromes), population characteristics, regeneration biology, habitat preferences and adaptations to fire—particularly fynbos fire regimes. The system is precautionary, and acts as a filter excluding `innocent' taxa—if in doubt, taxa stay in. Species are classified as potentially high- or low-risk invader species, and the system can be used to help plan control operations. The advantages of using an expert system approach is that it is based on the biology of the species, it is useful when data are thin and the decision process is transparent. There are limitations with the system, and it really can only be used for tightly defined and reasonably well-understood ecosystems. It does not work even for other ecosystems (e.g. riparian ecosystems) in the same general environment, and a general screening system will probably not be all that effective because of the wide range of ecosystems and taxa involved.

The last of the formal presentations was by Craig Walton [Australian Quarantine and Inspection Service (AQIS), Department of Primary Industries and Energy, Canberra, Australia] who outlined policies of AQIS and the implementation of a new system to assess the weed potential of new plants imported into Australia. All species not yet found or introduced to Australia are regarded as potential pests—that is, as invaders unless proved otherwise. Strict border controls include a list of permitted plants (this means that the only plants allowed in without question are those that are already in Australia but are not declared pests), a questionnaire for all proposed new imports not on the list, and all potential new imports assessed with a proforma Weed Risk Assessment (WRA) with, for certain plants, a post-entry quarantine period. The WRA is based on test-case assessments by a suite of experts and is only indirectly based on species characteristics. All serious weeds are rejected by the WRA, but about 60% of non-weeds are accepted.

The workshop closed with a discussion facilitated by Brian van Wilgen (Environmentek) aimed at elucidating some of the issues that had been raised at the meeting and how people can use these ideas to model and control alien invasions. The question arises of whether the types of models and the approaches discussed during the meeting are optimizing our ability to manage invasions, and if not, how can these be improved. Points that were raised in discussion outlined some of the practical difficulties in the control of aliens, and the advantages of using models to extend present knowledge to gain a predictive understanding of alien invasions. Real problems in developing a national strategy to control alien vegetation includes the tension between pragmatists and academics, and the biased views of different biologists—conservationists and agriculturalists may have very different ideas on what constitutes an invasive alien plant. In a South African context, very little is spent on research compared with relatively vast amounts on control of alien vegetation. There are still many problems with models—databases in South Africa are still inadequate, biocontrol and integrated management of aliens is seldom (if ever?) incorporated into models, funding for developing models is scarce, and modellers are a relatively small subset of all people who are concerned about aliens. Nevertheless, modelling alien invasions and modelling management options for aliens represents a significant advance on `spraying and praying'. At least we now have some understanding of how and why plants and other organisms can be invasive, and what environmental factors promote their success. Invasive alien organisms are not going to go away, but with modelling of rates of invasions and of costs and benefits of control programmes, plus a few decades of fighting in the trenches, they might be slowed down a little.



References

[1] Higgins S.I. and Richardson D.M. (1996) A review of models of alien plant spread.
Ecol. Model., 87:249-265. [ScienceDirect] [Cited by]

[2] Rejmánek M. and Richardson D.M. (1996) What attributes make some plant species more invasive?
Ecology, 77:1655-1661. [Cited by]

[3] Baker, H.G. and Stebbins, G.L. (1965) The Genetics of Colonizing Species, Academic Press

[4] Moody M.E. and Mack R.N. (1988) Controlling the spread of plant invasions: the importance of nascent foci.
J. Appl. Ecol., 25:1009-1021. [Cited by]

[5] Schroepfer R. and Engsfeld C. (1983) Die Ausbreitung des Bisams (Ondatra zibethicus Linné, 1766, Rodentia, Arvicolidae) in der Bundesrepublik Deutschland.
Z. Angew Zool., 70:13-37.