Research on translocation, and particularly reintroduction, is extensive. As a result, there is much knowledge to be drawn from the literature for assisted colonization.
In the following section, we present and discuss guidelines on how to plan and implement assisted colonization as an adaptation tool to climate change. By relying on concepts that have been developed for reintroduction or translocation, we bridge the gap between our knowledge of translocation and performing successful assisted colonization.
Recommendations on planning and implementing assisted colonization
The guidelines for planning and implementing successful assisted colonization under climate change take the form of a list of questions along with a list of methods to answer them (Table 2). At the planning phase, the first priority is to identify whether a species is threatened by climate change and is thus a candidate for assisted colonization (Q1 in Table 2). The decision frameworks proposed by Hoegh-Guldberg et al. (2008) and Richardson et al. (2009) have been partly designed to address this issue by suggesting questions to investigate, but do not explicitly propose a method to do so. To assess whether or not a species is a candidate for assisted colonization, we suggest exploring whether the species is experiencing (or is projected to experience) an increased extinction risk associated with range contraction driven by climate change. Habitat suitability models (Hirzel & Le Lay, 2008; McRae et al., 2008; Elith & Leathwick, 2009; Osborne & Seddon, 2012) have been put forward as a way to identify conditions promoting species’ survival and project changes in their distribution under different climate-change scenarios (obtained from sources like the IPCC 2007 or other global or regional climate models: Wolf et al., 2010; Barbraud et al., 2011). To quantitatively assess changes in species’ risk of extinction associated with likely range contraction, habitat suitability models can be combined with a spatially explicit population viability analysis (see e.g. Keith et al., 2008; Brook et al., 2009). Habitat suitability models will also highlight populations that are not immediately threatened and could potentially be used as source populations for translocations (Q2 in Table 2), as well as potential translocation sites where the environmental conditions are projected to remain stable as climate changes (Q3 in Table 2). However, once potential translocation sites have been identified, the risk of the introduced species becoming invasive should be assessed (Q3 in Table 2). A risk assessment could be based on which of the traits known to promote invasiveness the species possesses (see e.g. Sakai et al., 2001; Van Kleunen, Weber & Fischer, 2010). Loss, Terwilliger & Peterson (2011) suggested using laboratory and field tests as a way to assess invasiveness likelihood. Quantitative models of community interactions could also be designed on a case-by-case basis to predict the impact of a new species assemblage on the introduced species’ behavior.
Table 2. Table summarizing the questions that need to be answered, and the methods that can be used to do so, to maximize the success of assisted colonization under climate change
|Planning||Q1. Is the species threatened by the impact of climate change?|| |
- Decision frameworks
- SDM identifying future range contraction potentially followed by a spatially explicit PVA
|Q2. Which population (if n > 1) can be the source for the translocated individuals?|| |
- SDM identifying populations not threatened by the range contraction
- Scenario-based population dynamics modelling to project the source population's abundance under different harvesting scenarios (scenarios are defined by numbers of individual harvested e.g. 0, 10, 20, etc.)
|Q3. Where can the species be translocated?|| |
- SDM to locate where the identified suitable environmental conditions will be spatially distributed in the future
- Risk assessment of the likelihood the introduced species will become invasive using knowledge on intrinsic traits that promote invasiveness, laboratory and field tests, or community-based modelling
|Q4. How many individuals and what sex ratio should be translocated?|| |
- Scenario-based population dynamics modelling to project the translocated population's dynamics under different founder population scenarios (scenarios are defined by e.g. founding numbers, sex and age composition, genetic composition, etc.)
|Q5. What management should be applied to the translocated population?|| |
- Scenario-based population dynamics modelling to predict the abundance under different management scenarios (scenarios are defined by different management options, e.g. supplemental feeding, vaccination, doing nothing, etc.)
|Implementation||Q6. Is the source population negatively affected by the removal of individuals?|| |
- Monitoring to determine source population's abundance and demographic parameters
- Population dynamics modelling to project the source population in the future
|Q7. Are the projections made for the translocated population correct?|| |
- Monitoring of the translocated population's abundance and demographic parameters
- Comparison between projection and observed abundance
- Population dynamics modelling to project the source population in the future
|Q8. What adaptive management decision, if any, should be made?|| |
- Scenario-based population dynamics modelling to predict the abundance under new management scenario (scenarios are defined by different management options, e.g. supplemental feeding, vaccination, doing nothing, etc.)
The second priority at the planning stage is defining the logistics of the translocation by identifying the best source population (Q2 in Table 2), the optimal founder population (e.g. size, age composition, sex ratio and genetic composition; Q4 in Table 2) and the best management, if any, to be applied to the newly translocated population (Q5 in Table 2). We recommend a scenario-based approach in order to find the optimal answer to these questions. This involves building a population model and investigating the impact of different management decisions (i.e. the scenarios) on the population trajectory (Armstrong & Reynolds, 2012). The outcomes of the different runs are assessed against a specific objective set for the population. For example, managers may want to know the optimal founding population in order to maximize the growth rate of the translocated population and to minimize costs. When implementing the translocation, they would select the strategy given by the scenario that performed best against that objective, that is, that gave the most cost-efficient strategy. This scenario-based approach allows the identification of a case-specific optimal translocation strategy for both the source and translocated populations. In addition, it can yield predictions for the future dynamics of both populations if this optimal strategy is implemented. One caveat to this approach is the amount of data required to parameterize the models. In some cases, it may be possible to obtain estimates of expected demographic parameters from published meta-analyses (Brawn, Karr & Nichols, 1995; Falster, Moles & Westoby, 2008; McCarthy, Citroen & McCall, 2008).
Once the initial assisted colonization has taken place, in situ monitoring becomes the backbone of future decisions (Ewen & Armstrong, 2007; Armstrong & Seddon, 2008; Nichols & Armstrong, 2012). It should be done for both the source and the translocated populations. A key requirement, however, is that monitoring programs should be designed with specific questions in mind (Armstrong & Seddon, 2008). The monitoring data collected from the source population can, for example, be used to verify that the initial conclusion that the source population would not be negatively affected by the removal of individuals was correct (Q6 in Table 2). Similarly, the monitoring data collected for the translocated population can be used to match the projected population trajectory with the actual translocated population growth (Q7 in Table 2) and to increase the understanding of the translocated population dynamics to support future management decisions through adaptive management (Q8 in Table 2; McCarthy et al., 2012; Nichols & Armstrong, 2012).
By following the proposed systematic guidelines in Table 2, decision makers could be one step closer to securing the future of biodiversity with assisted colonization. Nevertheless, we have identified several issues at the implementation and research levels that have to be resolved before assisted colonization can become the most cost-efficient adaptation tool for species threatened by climate change.
Issues regarding implementation
There is an untold number of species that will be potential candidates for assisted colonization, that is, for which the answer to Q1 (Table 2) is ‘yes’ because they are projected to have an increased extinction risk due to climate-change-related range contraction. One pressing question is how to choose which species to move first? There are several prioritisation schemes and methods that currently attempt to guide conservation choices, for example, cost-benefit analysis (Bottrill et al., 2008), biodiversity hotspots (Joppa et al., 2011) and phylogenetic uniqueness (Isaac et al., 2007). These could all be useful to prioritize species for assisted colonization. However, once assisted colonization has been implemented (i.e. a species has been introduced to a new area), undoing this action and its consequences is very difficult, thus leaving little room for error when making decisions. As a result, concerns like the risks associated with species introduction, such as diseases or potential for invasiveness for example, could play a large role in deciding for which species to take action first.
Secondly, one of the possible future impediments to implementing assisted colonization will be the question of who will make the critical decisions of prioritization and implementation. Today, depending on the project, conservation decisions are made by various groups, from small organized groups of people (e.g. the Torreya Guardians; McLachlan et al., 2007) to Non-Governmental Organizations (e.g. Durrell Wildlife Conservation Trust, BirdLife International; Corry et al., 2010) and governments (e.g. New Zealand Department of Conservation; Cromarty et al., 2002). However, the need for assisted colonization will grow alongside the impact of climate change on everything else, increasing the chance of conflicts between groups. While no one entity can be in charge of making decisions regarding all assisted colonization projects, we hope that by using the same decision-making framework at least the decisions made will be transparent.
Further research required
There are also a number of challenging issues for research. Firstly, there are the issues of the accuracy and scale of climate-change projections that are to be used in habitat suitability models (e.g. to answer Q1–3 in Table 2). There are several models that can be used to project climate change (‘the multi-model dataset at PCMDI’; www-pcmdi.llnl.gov/ipcc/data_status_tables.pdf); these are mostly global climate models (GCMs) that originally had coarse resolution but have been downscaled to resolutions as small as 30 arc-second (http://www.worldclim.org). The issue is that GCMs do not always agree on predictions at the global scale, let alone at the scale needed for conservation, which is often much smaller than a country and can be sometimes as restricted as a field.
Secondly, there are four possible responses of species to climate change: thrive, adapt, shift or die. Although they are clear-cut, little is known of the true potential of species to tolerate or adapt to climate change. Often, the problem will lie in identifying which alternative outcome the species are moving towards, for example, species with long generation time and limited dispersal like the Florida torreya (McLachlan et al., 2007) will be both unable to adapt and unable to shift range. Unless there is evidence for range shifting, the only way to be really sure about the species’ fate would be to wait and observe. However, this is not an option for those species that are declining rapidly. Because conservation money is limited, it is inefficient to move species that would have survived had they been left in their original range. On the other hand, some species could be saved with assisted colonization but may be ignored because the skills necessary to identify their lack of adaptation are missing (Hill, Griffiths & Thomas, 2011). Although studies have started investigating the problem for plants (see Jump & Peňuelas 2005 for a review), one of the biggest challenges yet is to learn to predict plasticity and adaptation capacity for all species.
Thirdly, the relationship between environmental conditions and species’ demography is still not fully understood. So far, most published research has focused on identifying the relationship between some climate variables and the survival and reproduction of species based on current or past conditions (see e.g. Borrego et al., 2008; Frederiksen et al., 2008). However, our ability to predict how changes in baseline environmental conditions will affect future demographic rates is still limited. For example, difficulties arise when the future environmental conditions are outside the range of conditions experienced by species in the past, and models based on current conditions cannot be trusted to be informative (Berteaux et al., 2006). As a result, the quality and accuracy of predictive population models in the face of climate change, such as those that would be built to answer Q2 (Table 2), may be compromised. Lessons from historical introductions for reasons other than conservation may help in these circumstances (cf Cassey et al., 2008).
Finally, Armstrong & Reynolds (2012) suggest five topics related to population modelling that warrant further research. While those are offered in the context of reintroduction, advances on those subjects would benefit assisted colonization or indeed any form of translocation. In particular, they suggest focusing on the long-term genetic effect of translocating populations. Inbreeding and loss of genetic diversity could have a significant impact on the long-term viability of any population, but especially those established through translocation as they are generally small. However, genetic issues are seldom included into predictive population models (Armstrong & Reynolds, 2012), which may be an impediment to our ability to accurately predict the dynamics of translocated populations.