20.1.1 Risk Assessment
Many tools have been developed to assess the risk of biological invasions, but most
focus on specific species as the hazard (Kumschick and Richardson 2013). Such tools
typically cover a variety of aspects related to the risks posed by a species, which
includes transport, establishment, abundance, spread, and impact (Leung et al. 2012).
However, there are also many tools that cover or assess only a specific part of the risk of
a species and which are often mistakenly put under the umbrella of “risk assessments”.
For example, much effort has been put into standardising the way we quantify impacts,
and many recent developments were made around impact assessment tools (e.g.
Blackburn et al. 2014; Bacher et al. 2018; Zengeya et al. 2020, Chap. 17). Similarly,
species distribution models and climate matching tools are improving the way we can
assess and map a species’ climatically suitable area in a new range (see Wilson et al.
2020, Sect. 13.5, Chap. 13). Van Wilgen et al. (2009), for example, present climate
suitability matches for amphibians and reptiles that can improve the way we assess risks
for this group. In combination with other tools, climate suitability models and impact
assessments can improve the assessment and analysis of risks of biological invasions in
a transparent, standardised, and repeatable manner. However, by definition, a risk
assessment needs to cover both the likelihood of the hazard occurring (i.e. the invasion
process) as well as its consequences (i.e. the negative impacts).
Despite the aspects of risk to be covered being rather straightforward, there are
many ways in which these can be assessed in practice. Several approaches to risk
assessment have been suggested, including the trait-scoring, statistical, decisiontree, rapid screening, mechanistic, and detailed approach (Keller and Kumschick
2017). Each of these has its own benefits and weaknesses and is based on different
premises:
Trait Scoring This assumes that species with specific traits have a higher chance of
becoming invasive, and of having higher impacts. In other words, the more “invasive
traits” a species has, the higher the risk it poses. This is the most common approach
used to date, including in the Australian Weed Risk Assessment model (AWRA;
Pheloung et al. 1999), which was developed as a pre-border tool to screen alien
plants to be introduced to Australia. The AWRA has been used around the world and
modified for different taxonomic groups including fish and freshwater invertebrates
(Copp et al. 2005; Tricarico et al. 2010; see Appendix in Kumschick and Richardson
2013). This approach of assessing risk requires information about the taxon as alien
and what makes some species more successful than others. It also assumes that the
main contributor to invasion success is inherent to the introduced species’ traits, and
it is generally limited to a taxonomic group with comparable traits.
Statistical Approach The statistical approach is very similar to trait-scoring insofar
as it uses traits of species to predict the potential for invasion and causing harm
(Keller and Kumschick 2017). In the trait-scoring approach (seemingly) important
traits are collated and rated by experts only. In the statistical approach the list of traits
is refined using statistical or machine learning algorithms. These algorithms find
576
S. Kumschick et al.
Many tools have been developed to assess the risk of biological invasions, but most
focus on specific species as the hazard (Kumschick and Richardson 2013). Such tools
typically cover a variety of aspects related to the risks posed by a species, which
includes transport, establishment, abundance, spread, and impact (Leung et al. 2012).
However, there are also many tools that cover or assess only a specific part of the risk of
a species and which are often mistakenly put under the umbrella of “risk assessments”.
For example, much effort has been put into standardising the way we quantify impacts,
and many recent developments were made around impact assessment tools (e.g.
Blackburn et al. 2014; Bacher et al. 2018; Zengeya et al. 2020, Chap. 17). Similarly,
species distribution models and climate matching tools are improving the way we can
assess and map a species’ climatically suitable area in a new range (see Wilson et al.
2020, Sect. 13.5, Chap. 13). Van Wilgen et al. (2009), for example, present climate
suitability matches for amphibians and reptiles that can improve the way we assess risks
for this group. In combination with other tools, climate suitability models and impact
assessments can improve the assessment and analysis of risks of biological invasions in
a transparent, standardised, and repeatable manner. However, by definition, a risk
assessment needs to cover both the likelihood of the hazard occurring (i.e. the invasion
process) as well as its consequences (i.e. the negative impacts).
Despite the aspects of risk to be covered being rather straightforward, there are
many ways in which these can be assessed in practice. Several approaches to risk
assessment have been suggested, including the trait-scoring, statistical, decisiontree, rapid screening, mechanistic, and detailed approach (Keller and Kumschick
2017). Each of these has its own benefits and weaknesses and is based on different
premises:
Trait Scoring This assumes that species with specific traits have a higher chance of
becoming invasive, and of having higher impacts. In other words, the more “invasive
traits” a species has, the higher the risk it poses. This is the most common approach
used to date, including in the Australian Weed Risk Assessment model (AWRA;
Pheloung et al. 1999), which was developed as a pre-border tool to screen alien
plants to be introduced to Australia. The AWRA has been used around the world and
modified for different taxonomic groups including fish and freshwater invertebrates
(Copp et al. 2005; Tricarico et al. 2010; see Appendix in Kumschick and Richardson
2013). This approach of assessing risk requires information about the taxon as alien
and what makes some species more successful than others. It also assumes that the
main contributor to invasion success is inherent to the introduced species’ traits, and
it is generally limited to a taxonomic group with comparable traits.
Statistical Approach The statistical approach is very similar to trait-scoring insofar
as it uses traits of species to predict the potential for invasion and causing harm
(Keller and Kumschick 2017). In the trait-scoring approach (seemingly) important
traits are collated and rated by experts only. In the statistical approach the list of traits
is refined using statistical or machine learning algorithms. These algorithms find
576
S. Kumschick et al.