Home away from home — objective mapping of high-risk source areas for plant introductions

INTRODUCTION

Efforts are underway in most parts of the world to manage biological invasions. These range from local-scale efforts to eradicate particular invasive species, or to mitigate their harmful effects, to systematic management programs at the scale of landscapes or regions, through to national and global measures aimed at identifying important vectors and pathways of invasion, raising awareness of the problem, and providing legal instruments (Mooney et al., 2005). Because any given region has alien species at every stage of the introduction–invasion continuum (Richardson et al., 2000b), all these types of intervention are required. Emerging national policies and legal instruments are grappling with the challenge of managing and integrating all these types of invention.

Clearly, the most cost-effective way of reducing future problems with invasive species is to prevent the introduction of species that have a high risk of becoming invasive. The essential role of prevention is stipulated in the Convention on Biological Diversity (http://www.biodiv.org/) and the Global Strategy of the Global Invasive Species Programme (McNeely et al., 2001). Preventative measures are also required post-introduction. The invasive potential of recently introduced alien species needs to be reassessed regularly, since many alien species undergo a clear ‘lag phase’, sometimes for decades following introduction, before the species shows any signs of becoming invasive (Crooks & Soulé, 1999). Many alien species already present in a region and that currently show no signs of being invasive will invade in the future.

Identifying future invaders is hugely challenging. There has been good progress in the search for robust generalizations in plant invasion ecology (Rejmánek et al., 2005). Several traits are clearly associated with invasiveness (Pyšek & Richardson, 2007). Generally, however, predictive power is fairly limited and is practically applicable only for closely related taxa. The most reliable and practical predictor of plant invasiveness is still whether a species is invasive in other parts of the world, especially in areas with similar environmental conditions (Westbrooks, 1981; Rejmánek, 2000; Union of Concerned Scientists, 2001). Consequently, environmentally similar parts of the world (e.g. the five regions with mediterranean-type climate) have reasonably similar invasive alien floras; the degree of similarity is directly related to the degree of environmental similarity, but is modified by different cultural histories, soil, biogeography, and topography (Kruger et al., 1989). This shows that broadscale habitat compatibility, e.g. as defined by climatic conditions, is an important first-cut determinant of invasibility, and thus has value in preventing invasions.

Homoclime analysis has a long history of application in ecology. Duncan et al. (2001) found that climatic suitability significantly predicts introduction success and the subsequent geographical range size of introduced birds. Similar results have emerged for introduced plants (Panetta & Mitchell, 1991; Mack, 1996; Weber, 2001; Welk et al., 2002; Robertson et al., 2004; Thuiller et al., 2005). Climatic matching is definitely not the final answer, as biotic factors and other aspects play an important role (Stohlgren & Schnase, 2006; Thuiller et al., 2006b), but climate matching is, in many cases, the most important single factor. It clearly has considerable potential for use in invasion ecology (Peterson & Vieglais, 2001). With the recent increase in computing power, substantial improvements have been made on the techniques and approaches (called niche-based models) used to predict suitable habitats for species, biomes, or plant functional types in the context of climate change or invasion (for a review, see Guisan & Thuiller, 2005). These methods are now reasonably accurate and allow for a quick first extrapolation of habitat suitability.

Several currently used screening procedures include an assessment of whether a given species is ‘invasive elsewhere’ or ‘naturalized where grown’(Tucker & Richardson, 1995; Reichard & Hamilton, 1997; Pheloung et al., 1999; Daehler & Carino, 2000; Tye et al., 2000; Robertson et al., 2003; Daehler et al., 2004; Meyer & Lavergne, 2004; Nel et al., 2004; Mgidi et al., 2007). A problem is that such assessments seldom define such ‘other’ areas, where naturalization or invasiveness should be taken as suggesting a high risk of invasiveness in the region in question. It is clearly best to look for evidence of invasiveness of given species in parts of the world with similar climates and soil conditions. On the other hand, invasiveness in regions with very different climatic and other environmental conditions may not necessarily translate to a high risk of invasiveness in the region of interest. Applying the general rule of thumb of ‘invasive elsewhere = high risk of invading here’ uncritically is counterproductive and is likely to result in the rejection of many species (including potentially beneficial species) that really have a low risk of becoming invasive. Such ‘false rejections’ reduce the confidence of managers and the public in screening systems and generally undermine efforts to inform policy through the application of scientific methods. Furthermore, increasingly strong emphasis is placed on the need for ‘scientific validity’ in any imposition of trade restrictions (Penman, 1998). Can we fine-tune the climate-matching modules of screening systems to improve the accuracy of assessments, but without making the systems more cumbersome?

South Africa is, for several reasons, a good region to explore whether a more systematic approach to climate matching in the screening of plant introductions is appropriate. The country already has a major problem with invasive species (Macdonald et al., 1986), and there is ongoing interest in introducing additional species for use in various enterprises (Richardson et al., 2003). Ecologically, South Africa has been called ‘a world in one country’ (Cowling et al., 1997) with the concentration of several major biome types in a relatively small area. New legislation (the Environmental Management: Biodiversity Act 100 of 2004) provides the legal framework for far-reaching new measures to prevent the introduction of additional invasive species to South Africa. These call for the development of screening techniques that draw on the best available science to ensure the most accurate assessment of risk. As discussed above, there is clear evidence that accurate matching of climate is a useful first cut in identifying high-risk introductions. This paper (1) provides an objective delimitation of regions in other parts of the world with ‘similar’ climatic conditions to those that exist in South Africa; (2) evaluates whether the current set of major invaders would have been flagged as ‘high risk’; and (3) suggests a protocol for the effective implementation of objective climate matching in future screening systems.

METHODS

Validation

To determine whether the current set of invasive alien plants in South African biomes show close climate matching between their current (adventive range) in South Africa and their native range or adventive range elsewhere, we compiled representative lists of prominent invasive species for five of the biomes (quantitative data were unavailable for the desert biome). Our list was taken from data compiled by Henderson (2006). Using the full data set from the South African Plant Invader Atlas (details in Richardson et al., 2005), she ranked species according to ‘prominence’ which is defined as: (total species records of species × in biome y)/(sum of the records of all species in biome y) × 100. We consider this to be the best available objective lists of occurrence of species within the geographical domains of each of the biomes. We arbitrarily chose to include only species with a prominence value > 1 since most species with lower values have localized distributions. This gives us an uneven number of species for the different biomes, but we suggest results in an overall list that closely match the likelihood of encountering different species in the field. Although all the species listed in SAPIA are invasive in natural or seminatural vegetation in part of their adventive range in South Africa, a proportion of records is also from more disturbed sites.

For each species in the list thus derived, we plotted the native distribution (derived from numerous published sources and online herbaria, notably: the Atlas of Florida Vascular Plants —http://www.plantatlas.usf.edu/ Australia's Virtual Herbarium http://www.cpbr.gov.au/avh/ the Burke Museum of Natural History and Culture http://www.washington.edu/burkemuseum/collections/herbarium/index.php; CalFlora http://www.calflora.org/ and Flora Europaea http://rbg-web2.rbge.org.uk/FE/fe.html) as accurately as possible onto the generated maps showing regions with analogous climates to the South African biomes (as shown in Fig. 1). We also consulted numerous sources to map regions of the world where the taxon is known to be invasive. This enabled us to determine the degree of climate matching between the South African biome where the taxon is ‘prominent’ (see above) and the natural and adventive range(s) of the taxon. A few points need to be made regarding this assessment. First, accurate climate matching could not be done for several taxa. For some prominent invasive taxa in South Africa, the level of taxonomic resolution is inadequate to enable us to map with any confidence the natural distribution. There are two main issues here. First, in several groups, species are difficult to identify in the field, and taxa were mapped in SAPIA as species groups (e.g. taxa in the genera Eucalyptus and Rubus). Second, some invaders in South Africa are hybrids with no original natural range, e.g. Lantana camara (an artificial hybrid taxon that has been subject to horticultural improvement for centuries), Populus taxa, and Prosopis taxa. Several other taxa have been used and moved around the world so much and for so long by humans, that the precise natural range is unknown (e.g. Melia azedarach, Salix babylonica). We also did not do climate matches for aquatic species such as Azolla filiculoides and Eichhornia crassipes, as the distribution of such species is known to be poorly controlled by broad-scale climate. Despite the above-mentioned caveats, we argue that this analysis provides a useful way of assessing the role of climate matching in determining invasive success.

RESULTS

Figure 1(a–f) shows regions of the world with similar climatic conditions to each of South Africa's biomes. Table 1 shows the top 20 countries — those with the largest combined total area with climatic conditions similar to those in South Africa. Eight countries have larger areas of ‘SA-like’ climate than the total area of South Africa. Australia has more than four times more SA-like climate than South Africa, with samples of all of the main climate categories (biomes). The USA, Argentina, and China also have large areas of analogous climate. Table 2 shows the top 20 countries ranked in descending order of area of climatic conditions analogous to each of South Africa's seven biomes. The position of South Africa in the different biome listings shows the markedly different representation of different major climatic types for each. For example, South Africa has only a tiny fraction of the total world area that experiences climatic conditions that define the region's desert biome, but a larger proportion of the total world area of ‘Nama-karoo’ climate.

Figure 2 shows the level of climate matching between prominent invasive plant species in five of South Africa's biomes and their native and adventive ranges elsewhere. For those species for which matches could be made (see Methods), there was generally a high degree of climate matching between their ranges in South Africa and their natural and/or their adventive ranges.

Results of the climate matching for known invasive species in other parts of the world (and, in some cases, the native range of the species) are summarized in Appendix 1. The analysis shows that many highly invasive species in other parts of the world occur (either as natives or invasive aliens, or both) in regions with very similar climatic conditions to those that occur in South Africa. For the sample of species/localities selected to illustrate the utility of the approach, the closest matches (similarity = 0.8) were for taxa occurring in conditions similar to South Africa's grassland biome. Species thus identified are known to possess traits that enable them to invade and must be considered likely to invade if introduced to South Africa. The taxa include some already present in South Africa, but not (yet) widely invasive, such as Gleditsia triacanthos and Ulex europaeus (both noted as ‘emerging invaders’ in Nel et al., 2004), Robinia pseudoacacia (a ‘major invader’ in Nel et al., 2004; but currently still largely confined to areas near plantings), Tradescantia fluminensis (currently reported as invasive at only one locality; Alston & Richardson, 2006), and others known to be present but with no records of invasiveness.

DISCUSSION

This paper has provided an objective delimitation of regions of the world with similar climatic conditions to those that occur in South Africa. We showed that the most prominent invasive alien plants in South African biomes have native ranges, or are known to be invasive, in parts of the world with climates matching those that occur in South Africa. Although most currently invasive species in South Africa are most prominent in biomes with climates that match their native or adventive ranges elsewhere, some species are invasive outside the closest matched climatic conditions. Such species either occur in fragments of other biomes within the borders of a particular biome (e.g. fynbos islands within the succulent karoo biome), occur in riparian zones that enable them to extend beyond climatic conditions in which they occur naturally, or are widely planted by humans. Recent studies have shown that the extent and distribution of the main invasive plant species are influenced by interactions between environmental conditions (notably climate) and human usage factors (e.g. Thuiller et al., 2006b). The fact that some species are recorded as invasive in biomes that are poorly matched in terms of climate to the native range and/or adventive ranges of the species should not be seen as lack of support for the dominant role of climate matching. Such poor matches are especially evident for the succulent karoo biome, where species such as Acacia cyclops, Acacia mearnsii, Hakea sericea, and Sesbania punicea are poorly matched. This is attributable to the fact that the biomes were defined for the purpose of formulating this table as geographical entities. In the case of the succulent karoo, many enclaves of fynbos vegetation occur within the boundaries of the biome. Also, for the succulent karoo and other biomes, species that occur frequently in riparian zones or that are abundant in disturbed sites show less clear patterns of climate matching between the invaded biome and the native or adventive ranges. Examples are A. mearnsii in the succulent karoo and savanna biomes and Acacia dealbata and Acacia saligna in the savanna biome. Species with the less close matches are all taxa that have been widely planted, often beyond climatically optimum conditions and/or those that proliferate in disturbed sites. All the species listed in Fig. 2 (excluded those for which climate matching could not be done; see above) have natural or adventive ranges with similar climatic conditions to those that occur in one of South Africa's biomes.

The maps and tables presented here can be used to improve the objectivity and accuracy of screening systems. Any species known to be invasive in the areas shown in Fig. 1 and Tables 1 and 2 must be considered a high-risk introduction in South Africa. Also, lists of invasive species in countries such as Australia, the USA, Argentina, and China should be scrutinized when preparing ‘blacklists’ (lists of species that will not be permitted to be introduced). For some of these countries, detailed regional lists of invasive species are available (e.g. for states of the USA and for states and territories of Australia), making it possible to include only high-risk species that occur in those parts of these countries with the closest match to South African climates. For example, for the USA, species known to be invasive in the central states would not be included on the blacklist (unless the species were known to be invasive in riparian habitats or other ecosystems, such as agricultural lands, where environmental modification is likely to reduce the dependence of the species on prevailing climatic conditions). Many species already invasive in regions with similar climates to South Africa are already in this country, but may not have had enough time to start invading. The compilation of a list of such species, with details of parts of South Africa at the highest risk of invasion, would be very useful.

The maps and tables of course also provide an objective basis for the assessment of species that are not known to be invasive anywhere. Despite the value of asking ‘invasive elsewhere?’ in screening protocols, many potentially invasive species have yet to be moved outside their native ranges and therefore have no history of invasiveness. Some screening systems explicitly avoid assessments of whether species have ‘invaded elsewhere’ for this reason (e.g. Frappier & Eckert, 2003). Since various modules of most screening systems assess ‘invasiveness’ fairly generally (i.e. without detailed reference to particular localities), the ability to assess the climate match is especially important for potential introductions with no history of invasiveness. A close climatic match should not, however, automatically be taken as a reason for assigning high-risk status to a species (see also Krivánek & Pyšek, 2006).

There are certain limitations of the approach outlined in this paper. Being a correlative method, the approach does not consider directly the effects of biotic interactions that are known to be fundamentally important for the recruitment, establishment, and spread of introduced species. What our approach does is to delineate zones that are climatically suitable for a species from a given biome. The roles of biotic factors such as competition or facilitation in defining actual invasive potential need to be assessed using emerging methods in invasion ecology (see Rejmánek et al., 2005 for a review). Considerable advances have been made recently in this regard, for example regarding the importance of mutualisms (e.g. Richardson et al., 2000a; Callaway et al., 2004) or the resistance of the native community (Levine et al., 2004). An example from our analysis is Araucaria araucana. This species is invasive in Argentina and Great Britain, but depends on squirrels and/or birds for dispersal of its large seeds (Richardson & Rejmánek, 2004). Despite the close match between the climate of A. araucana’s adventive range in Argentina and South Africa's grassland biome (Appendix 1), the lack of suitable dispersers in South Africa will probably prevent it from becoming invasive. Improved understanding of such factors definitely improves our ability to predict whether a particular species will ‘fit in’ and potentially become invasive at a given locality (Rejmanek et al., 2005). Such assessments must be made separately for individual taxa, and cannot be incorporated in a large-scale analysis. We suggest that the approach presented in this paper is useful for mapping zones of high risk within a given region. Future work could use more dynamic models (e.g. succession models) that include landscape structure, vegetation type, competition, dispersal, and disturbance to assess the potentiality of the given species to spread across the landscape. Examples of work along these lines include attempts to model the intensity and frequency of disturbance (Grigulis et al., 2005) and the dispersal ability of species (Higgins et al., 2003).

With the above caveats in mind, we feel that the approach set out here is valuable as a practical way of assessing the level of habitat compatibility, as a first step in screening alien species for their invasive potential. Many countries now have objectively mapped biomes and/or vegetation types, and many have excellent databases of species distribution. Global databases of climate, land use, and topography at fine resolution are now freely available, making it possible to undertake the type of analysis presented in this paper for many parts of the world. This would certainly help in making screening and decision-making more objective.