Hughes and Styles (1989) drew attention to the fact that, at that time, the process of species introduction was characterized by (i) little knowledge of the real needs of those whom introduced trees were intended to benefit, (ii) uncontrolled distribution of germplasm around the world and (iii) a low level of concern regarding potential problems of weediness (invasiveness). Whilst tremendous progress has been made in relation to the first two points, many countries are still not addressing the issue of weediness, despite pressure from international conventions such as the CBD. Although there is now a greater awareness among stakeholders of the issue of invasiveness, the general lack of quantitative information on the benefits versus the risks and impacts of invasiveness has prevented many countries from taking action. The lack of appropriate tools for assessing and managing the risks associated with plant introductions (or for managing existing invasive tree species) are additional constraining factors. Furthermore, in many cases, there has been little study of the biology of the trees and woody shrubs used in agroforestry prior to introduction. With respect to biosecurity (the policy and regulatory framework for managing all risks associated with food, agriculture, fisheries and forestry), the forestry sector is not as advanced as some of the others (FAO 2002). However, attempts have been made to address this issue and some progress has been made over the last decade.
Some of the measures taken to address the real or potential negative effects of introduced trees are discussed below.
Preventative methods have focused on the development of management models and techniques for risk assessment. In general, trees and woody shrubs have formed just one part of a larger consideration of risk assessment for all alien plants. This is a relatively new and dynamic area of research, but some countries have already implemented assessment and management schemes in combination with their national quarantine services. Two situations are considered here: risk assessment and management of plants not yet introduced and risk assessment and management of alien plants already present. The attention of most countries has been focused on the former.
Hughes and Styles (1989), Cronk and Fuller (1995), Hughes (1994, 1995), Lonsdale (1994) and Panetta (1993) were among the first to call for and suggest improved methods for the introduction of exotic species of trees and woody shrubs. Hughes (1994, 1995) suggested that protocols for introductions should be based on an assessment of benefits and risks prior to seed distribution and should include some assessment of site adaptation and acceptability versus the risk of becoming invasive. Hughes (1995) also suggested some protocols (Box 1) for introductions in forestry based on IUCN (1987) guidelines on the translocation of living organisms.
In accordance with the principles suggested by the IUCN (1987), Hughes (1995) stressed the point that native species should always be assessed before exotics are considered for any new tree-planting programme. In agroforestry, there is now a range of criteria that are used to select species (e.g. yield, stability, microsite matching, product quality, compatibility with livestock, etc.) and there is thus a need to employ a diversity of planting material. This runs counter to the “multipurpose” tree concept and provides the opportunity of using many native species in agroforesty. It also has the advantage of reducing the risks associated with relying on a few exotic species in any one area.
The use of native species in commercial forestry seems to be a more complex and contentious issue. With pines, for example, the most productive species are known to be invasive (Richardson and Higgins 1998) and forestry companies often wield considerable economic and political influence locally (Richardson 1998), so that economic considerations may outweigh ecological concerns. Nonetheless, there are good examples of cases in which native species have been employed, or where less aggressive exotic species have been used. Part of the problem seems to be that in many countries, few native species have been adequately studied, especially in relation to their growth potential. For those countries with only a few native species suitable for forestry, the use of exotics is likely to continue to be the most viable option for their forestry industry.
The assessment protocols detailed in Box 1 still need to address certain issues. For example, there is frequently a time lag between the introduction of a species and it becoming invasive; for trees this can be 50 or more years. Thus longer-term factors that may lead to a species becoming invasive need to be incorporated into assessment protocols. There is also the problem referred to earlier that introduced species may hybridize unpredictably with native species and produce highly aggressive crosses. This risk also needs to be assessed.
As mentioned earlier, many biological characteristics are poor indicators of a species’ potential to become invasive. In view of this and the other factors mentioned above, some aspects of risk assessment may be problematic, at least for some tree species. This could be counterbalanced by placing more emphasis on assessing the benefits of a species introduction (Hughes 1995).
• The taxonomic identification of the proposed introduction needs to be confirmed.
Only if these first four conditions are met should further assessment be undertaken.
• Reports of weediness from other areas should be assessed. If the proposed introduction is a reported weed, in most cases this is grounds for rejection, unless overriding benefits can be demonstrated that outweigh likely costs. A full Environmental Impact Assessment is justified in such cases.
(iv) Likelihood of interspecific hybridization with closely related native or other introduced species, and risk of contamination of native gene pools through introgression or evolution of new and potentially aggressive polyploid species.
• Introductions should be made initially in small, closely monitored field trials under quarantine conditions. Monitoring should include assessment of seed production and dispersal and natural regeneration into surrounding areas. Collection of seed from trials by station workers or visitors should be prevented by harvesting all seed before it ripens. When assessment is complete, trials should be completely destroyed, including any soil seed bank that has developed.
• The case for proposed introduction, including data on benefits and risks as outlined above, should then be put before the relevant national and/or state quarantine authorities for final decision. Costs should be borne by the intending importer.
Source: Hughes (1995) (modified from IUCN (1987) and amalgamating points from Panetta (1993), Hughes (1994) and Lonsdale (1994)).
Despite these drawbacks, information such as whether the plant is invasive elsewhere can be used as the basis of a risk assessment for proposed introductions (Panetta 1993). Practical risk assessments based on this and other information (mostly biological characters – see Section 5) are now in use in Australia, New Zealand and the USA. A few other initiatives are either under discussion or development (e.g. in Hawaii) but are not yet operational; these are discussed by Groves et al. (2001). Other, more ecologically based risk assessments, such as habitat models (e.g. Zalba et al. 2000) are not considered here since they are currently only proposals in the scientific literature.
One method of risk assessment in current use is that based on some form of numerical scoring system. Perhaps the most advanced of these is that developed by the Australian Quarantine and Inspection Service (AQIS). The AQIS model, known as the Weed Risk Assessment (WRA) system, forms the second tier of the Australian protocols for plant introductions. The three tiers (Walton 2001)are:
1. Identification of the species with reference to current lists of prohibited and permitted species and determination of its Australian distribution.
2. If the species is not listed and is not established in Australia, a pre-entry assessment procedure is applied to determine the risk of the species becoming a weed in Australia (the WRA system); possible recommendations are “accept”, “reject” and “further evaluate”. Rejected or accepted species are added to the prohibited or permitted lists.
3. If a recommendation for acceptance or rejection cannot be obtained from the second tier, and the importer wishes to proceed, then the species in question is subjected to post-entry evaluation either in the field or in glasshouse trials in order to examine its weed potential more directly (and/or to verify potential uses) so that, eventually, the species can be placed on a prohibited or permitted list.
The WRA system was developed by P.C. Pheloung (see Pheloung 2001). Plants are scored using 49 criteria, including life history characteristics and evidence of weediness elsewhere; the latter in particular is heavily weighted in the analysis. The system was initially calibrated using 370 known plant introductions to Australia. Hughes (1998) reported a “test” of the system using the legume Leucaena leucocephala subsp. leucocephala. This species fell within the “evaluate further” category – an appropriate result for such a species, which has many benefits but which is also known to be highly invasive. The WRA model is practical and was formally adopted by AQIS in 1997; it is currently being used to evaluate 600 plants of potential value to Australia, including some new tree species for forestry (Walton 2001). It has also been modified for use in New Zealand (Williams et al. 2001).
Tucker and Richardson (1995) suggested a model for risk assessment that was developed with a specific purpose in mind, namely the screening of plants for potential invasiveness in the South African fynbos biome. In the USA, the Animal and Plant Health Inspection Service (APHIS) has adopted another risk assessment model based on the FAO’s International Standards for Phytosanitary Measures Part 1 – Guidelines for Pest Risk Analysis. The FAO guidelines identify three stages in pest risk analysis (FAO 1996):
1. Identifying species that may qualify as quarantine pests, and/or pathways that may allow the introduction or spread of such pests.
2. Assessing pest risk (determining which species are quarantine pests, based on the likelihood of their entry, establishment, spread and economic importance).
3. Managing pest risk (developing, evaluating, comparing and selecting options for dealing with the risk).
APHIS adapted stages one and two to produce a qualitative assessment. In conducting the evaluation, APHIS makes considerable use of available databases and other published sources of information. After assessment, plants are classified in terms such as “high” or “low” risk.
There is an urgent need for these risk assessment models to be evaluated further for use in forestry and agroforestry. If validated, they could then be built into decision support systems that include socio-economic and other factors as well as biological risk. However, suitable ways still need to be found of assessing plants for which there is little or no information concerning their invasive tendencies elsewhere.
Some research groups have been working on a different approach to reducing the potential invasiveness of economically important trees and woody shrubs. Much of this work has been devoted to the production of either sterile hybrids or varieties with low levels of seed production. For example, in South Africa, work has started on the production of seedless clones of Pinus elliottii, P. patula and P. radiata by irradiating seed. Germination trials of such treated seed are now under way (Richardson 1998).
Research on Leucaena has been focused on producing sterile or near-sterile hybrids (Hughes 1998). The first such hybrid was a triploid produced from L. leucoephala × L. pulverulenta. This arose spontaneously in the tea- and coffee-growing areas of Indonesia where L. leucocephala is grown as a shade tree. The hybrid was propagated artificially by grafting and sold commercially but proved to be very susceptible to damage by the Leucaena psyllid (Heterosylla cubana). Widespread planting has therefore stopped except in areas where there is no pressure from the pysllid. Another triploid hybrid has been produced in Hawaii by crossing L. leucocephala with L. esculenta. This hybrid is very resistant to psyllid attack and has good properties as a tree for reforestation. However, the problem remains of how to propagate such hybrids in large numbers.
The New Zealand Department of Conservation has implemented a model for plants that are already present in the country but have not yet become invasive (Tye 2001). This model includes assessment of the following:
• types of community potentially at risk;
• potential effects on the ecosystem;
• biological success rating;
• additional information (e.g. fire risk, competitive ability and resistance to management).
Other information is also taken into account (e.g. the year of naturalization and the bioclimatic zones in which the species might be a pest in other countries).
More generally, several authors (Hughes 1995; Richardson 1998; Reichard 2000) have emphasized the need for monitoring schemes to be set up once a plant species is introduced, both at the site of planting trials and in the surrounding area (Hughes 1995). At present, however, there are no practical guidelines on monitoring forestry species and this is an area where further work is urgently needed.
Diverse control methods have been used for forestry trees that have already become invasive, although implementation of these measures on a global scale has been patchy. Early attempts to eradicate woody legumes in grassland were largely unsuccessful, although similar campaigns in other ecosystems were more promising. Taking into consideration both these experiences and changing perspectives on the risks posed by invasive species compared to their potential benefits, more emphasis is now being placed on more appropriate types of control. These are discussed in more detail below. Methods for the control of invasive plants in general have been reviewed by Cronk and Fuller (1995).
Various attempts have been made to eradicate species from a number of genera, mostly those used in agroforestry, including Acacia, Albizia, Cedrela, Leucaena, Maesopsis, Melia, Parkinsonia, Prosopis, Psidium, Sesbania, Swietenia and Toona (T. Simons, personal communication, 2002). Some of the most intensive efforts have been directed against Prosopis and Leucaena. In the case of the former, attempts have been made to eradicate species in Argentina, Australia, Pakistan, South Africa, Sudan and the USA (Pasiecznik et al. 2001). In the USA, control of the invasive native P. glandulosa was initially based on hand clearance and was followed by mechanical site clearance with tractors. However, these methods proved too labour intensive, and current control approaches rely heavily on chemical methods. At present, the most effective herbicides are clopyralid, dicamba, picloram and triclopyr, either alone or in combination (Pasiecznik et al. 2001). The most effective approach is a combination of mechanical and chemical control techniques. Fire was traditionally used as a control tool in American grasslands, but is no longer used to any great extent. Overall, experience has shown that Prosopis can be controlled but not eradicated.
More effective attempts at eradication have been made in South Africa against a range of exotic trees that invade watersheds. A detailed cost–benefit analysis showed that clearance of the trees was justified because the cost of clearance was much less than the value of benefits from the improved water supply (van Wilgen et al. 1996). This led to the Government of South Africa funding a large and successful invasive tree programme that engaged large numbers of unemployed and unskilled people. This has become known as the “Working for Water” programme.
On a smaller scale, Casuarina equisetifolia (Australian pine) and other plants have been removed from a nature reserve in Florida by volunteers (Randall et al. 1997). There has also been a successful project in Mauritius to eradicate guava and other invasive plants from plots in natural forest areas by hand weeding. These plots are also fenced to keep out invasive mammals such as deer (Cronk and Fuller 1995).
Biological control of weeds through the importation of natural enemies from the area of origin has a long history and a good success rate (McFadyen 1998). Some attempts have been made against invasive forestry trees, mostly in Australia, Southeast Asia and South Africa. A long-running programme of biological control of Mimosa pigra in Australia and Southeast Asia has been undertaken by CSIRO (Commonwealth Scientific and Industrial Research Organization) along with other Australian partners and the government of Thailand (Lonsdale et al. 1995; Napompeth 1982). Several insect and fungal natural enemies from the neotropics have been assessed and released in Australia and Thailand; the results are still being evaluated but it is predicted that a larger complex of natural enemies will be needed to exert full control. Similarly, efforts are being made in South Africa to control Acacia saligna through the use of a gall-forming rust fungus, Uromycladium tepperianum, which kills saplings and some older trees and reduces seed production in others (Morris 1995). Again, however, further efforts will be needed if complete control is to be achieved.
Researchers in South Africa have also pioneered a novel approach to the biological control of several economically important woody legumes by using introduced bruchid seed predators (Neser 1996; Zimmerman 1991). Seed predator communities can be quite diverse in the native ranges of some of these trees (e.g. see Hughes (1998) for the species that feed on Leucaena in Central America). The aim of this approach is to “control” the invasiveness of the trees by seed predation. Host-specific bruchids have now been successfully established for the control of Acacia melanoxylon and various Prosopis species, and the same approach is being used for an ongoing programme to control black wattle (Acacia mearnsii) in South Africa (Adair 2002) and Prosopis juliflora in Ascension Island (S.V. Fowler, personal communication, 1998). Another bruchid species has also been introduced into South Africa for the control of Leucaena. This species was also accidentally released in Australia, where its impact is currently being assessed (Hughes 1998).
There are relatively few case studies of fully integrated control. An important exception is the successful management of Hakea sericea in South African rangelands by combining mechanical, chemical and biological control methods (Kluge et al. 1986).
In South Africa, management methods have been developed for some invasive forestry trees that are based on the biological control of seed production and/or eradication of the plant in watershed catchment areas. This approach was developed on the basis of economic and environmental models that were intended to facilitate the resolution of conflicts over the benefits and negative impacts of invasive forestry trees (Geldenhuys 1986; Higgins et al. 1997). This example demonstrates the value of these assessments and models. More recently, new legislation in South Africa requires land users who plant crop species, including forestry trees, to be responsible for the control of water and seed pollution; the species have to be planted within designated areas and any invasions outside these areas have to be controlled (Klein 2002).
In some grassland areas (such as the southern USA and northern Africa), research has been conducted on the ways in which invasive woody legumes can complement pasture grass. Woody legumes can be a valuable resource in times of drought and, in some circumstances, can even improve grass production. Thus, some foresters have called for “bush management” rather than control (Hughes and Styles 1989; Pasiecznik et al. 2001).