Unified system describing factors related to the eradication of an alien plant species

System design

Based on our literature review we selected eradication related factors and ascribed them into four groups depending on the eradication context that matches the factors best (Fig. 1). The four contexts represent the four main aspects associated with the eradication of an alien species. The factors are further organized into groups combining factors of a similar nature. The rationale was to organize the factors in a way that facilitates work with the system. The first group, “location context” is therefore related to the spatial characteristics of the infestation and considers its size and location (factors 1–5) as well as ease of access for the necessary actions to be performed (factors 6 and 7). The next group is the “species context” which groups factors related to the biology of the considered target (invasive) species. Factors here are related to fitness and fecundity (factors 8–12) and detectability of the species (factors 13–14). In the next group, the “human context”, factors related to the human (personnel, managers, organizations) behavior influencing the eradication campaign are grouped. The included factor groups are knowledge (factors 15–17) and cognition and resources necessary for alien plant management (factors 18–23). The last context is the “reinvasion context”, which contains only one factor, ‘invasion pathways’ (factor 24).

The complex nature of the selected factors raised problems with classifying some of them into one group as they are relevant for at least two groups (e.g., factor 5 ‘monitoring rate’ and factor 8 ‘adaptation to new climate conditions’ depend on location and biology of managed population—they could be classified into both location and species context). In such instances we fitted the factors into groups that match the factor better. Regardless of the assignment of factors to different groups they should be considered jointly, as they all interact in shaping the success or failure of a specific eradication.

Selection of factors for the system

We considered 22 factors mentioned by Dodd et al. (2015). Some of them were redefined and expanded to improve their universal usage. In several instances it was necessary to further subdivide the factors into several components, which may counteract with each other and have a contrasting effect on eradication success (Table 1). Additionally we included ‘isolation of infestation’ (Table 1, factor 3), which describes geographical and ecological isolation of the target population. This factor was mentioned earlier as ‘insularity’ by Pluess et al. (2012a) and Pluess et al. (2012b).

Factors mentioned earlier in the literature but not incorporated in USEF are included in Table 1 only as similar or related terms for comparison purposes. Similar terms were differently defined in their original publications than in our system, but conveyed a similar meaning. Factors, which in our opinion had a lower significance for eradication success or their impact is difficult to assess, were evaluated as related terms. For example for the factor ‘number and distribution of propagules’ (i.e., size and spatial structure of propagule store, see definition of factor 9 in Table 1) the related terms include ‘propagule dispersal’ (Blood et al., 2019) and ‘mode of propagule dispersal’ (Panetta & Timmins, 2004). With no doubt propagule dispersal has a major impact on the invasion success. Seed dispersal is usually mediated through several dispersal mechanisms involving primary and secondary dispersal and natural or human assisted dispersal. Indicating a direct relation between predominant propagule dispersal mechanism of a given species and success of eradication of target population may be difficult to evaluate because of that. Blood et al. (2019) stated that “human-mediated dispersal should be easier to influence”. However evaluating the “potential for managing propagule dispersal” (Blood et al., 2019) in practice can be very difficult. The definition of an invasive species states that its appearance in a new site is mediated through human actions (see definition of invasive species by e.g., Blood et al., 2019). Another difficulty is demonstrating which of the dispersal mechanisms exploited by an invasive species was responsible for another site establishment. Such establishment instances are usually rare events, which are very difficult to observe and may be driven by many different dispersal mechanisms. Therefore we selected propagule distribution, which is a measurable consequence of dispersal and indicates individual recruitment location. In our system we did not include all the traits associated with species biology that may affect the success of the invasion. As separate factors we included only those traits that directly affect the success of the invasion in a significant way (see factors from fitness and fecundity group, Table 1). It is worth emphasizing here that the requirement of detailed study of the biology of the invasive population has been included in factor 5, ‘understanding of species biology’ (Table 1).

Description and interpretation of factors selected for the system

The synthetic characteristic of all the factors included in USEF is presented in Table 1. For some factors additional explanations/comments were necessary. In the text below we have analyzed in more detail only the most ambiguous and controversial factors leaving the more obvious ones without a comment. The text is organized in accordance with Table 1 and Fig. 1 and the subsections correspond to specific factor groups.

Size and location of target population

One of the most important factors influencing eradication feasibility is the size of the target population. The most universal way of presenting this factor is reporting the area (m², ha, km², etc.) occupied by the invasive population, therefore we propose to name this factor ‘infestation size’ (see factor 1 in Table 1). Distinguishing ‘net infestation size’ and ‘gross infestation size’ (Rejmánek & Pitcairn, 2002; Dodd et al., 2015; Blood et al., 2019) enables including the spatial organization of the population, but if the population is not concentrated (individuals are sparsely scattered within the infested area) the term ‘gross infestation size’ may be misleading. We therefore propose to rename the ‘gross infestation size’ to ‘monitoring area size’ (factor 4, Table 1) as it does not necessarily represent the infested area, but an area which needs to be monitored for the species (re)appearance.

The aim of eradication is to diminish the population size lowering the ‘infestation size’ and ‘number of separate infestation sites’ until they reach “0” (Table 1, factors 1 and 2, respectively). Ideally the values of these factors should not remain constant during the eradication process. We thus propose to include information on the stage of eradication, e.g., time elapsing from the start of eradication (e.g., initial infestation size, number of separate infestation sites after 5 years of eradication campaign). We believe that if infestation size is accompanied by information on specific eradication stage (i.e., initial pilot actions, beginning of eradication, 5 years after eradication commenced, etc.) the analysis of such eradication process and its comparison with actions performed elsewhere will be much easier than without this information clearly presented. The addition of time dimension also allows reporting and comparing ‘infestation size’ in different time points during the eradication process. In contrast to the variable ‘infestation size’ the size of the monitored area should not change (with exception of an unsuccessful eradication only increasing the size). The whole area defined by ‘monitoring area size’ (Table 1, factor 4) needs to be constantly monitored for reappearance of the invasive species, as stated by Burgman et al. (2013) “total infested area can never decrease”. It is also advisable to add a “buffer zone” to the monitoring area, to contain the spread of the species (Fletcher et al., 2015).

Classifying ‘monitoring rate’ into one of the groups was not straightforward (factor 5, Fig. 1). We decided to include this factor in the location context and size and location group, because both components of this factor (annual monitoring period and recommended monitoring intensity) are mostly determined by environmental characteristics of the non-native range (Fig. 1 and Table 1), which may be perceived superior to species context factors.

Ease of access to target population

It is beyond all doubt that “Land use directly affects the invasion process because it modifies disturbance regimes and environmental conditions” (Pauchard & Alaback, 2004; Epanchin-Niell et al., 2010; Epanchin-Niell et al., 2012). However ‘land use’ as an eradication feasibility factor is difficult to measure, differently understood and used in different ways to predict the probability of eradication success. Dodd et al. (2015) distinguished two states of ‘land use’, i.e., residential land use and environmental land use, assigning respectively high and low expected relationship with eradication feasibility. Panetta (2015) mentioned that different eradication costs are associated with different land use. Panetta & Timmins (2004) mentioned land use in the accessibility to infestation context. Nevertheless neither of these authors attempted to define ‘land use’ in eradication feasibility context. In contrast, Pluess et al. (2012a) did not use the term ‘land use’, but distinguished three categories of accessibility to the infestation depending on the level of difficulties considering accessing the target area: “access to private properties problematic, remote areas concerned”; “some difficulties to access the target area”; “no difficulties to access the target area”. These authors also distinguished a factor connected with land use—‘man-made habitat’ including it in the ‘location specific factors’ group (Pluess et al., 2012a). Rohal, Cranney & Kettenring (2019) wrote about ‘landscape factors’ associated with human disturbance.

Additional problems and misunderstandings associated with the term ‘land use’ may have originated from the fact that Dodd et al. (2015) used the term ‘general accessibility’ to describe “distance to manager” and proposed another factor ‘number of separate land managers’ to describe the ownership relations of the infested area. It is worth mentioning, that this last factor is directly linked with possible difficulties with the access to the infestation resulting from ownership relations of infested area. Epanchin-Niell et al. (2010) stated that land subdivision into many ownerships impairs invasion control. They also introduced the term ‘management mosaics’ (“landscapes comprising many individually managed properties with a variety of uses”, Epanchin-Niell et al., 2010). Considering all the different perceptions of land use mentioned above we propose a longer, but more precise factor name—‘land use and ownership of infested area’. We further propose to subdivide this factor into four components describing different ways of its influence on the eradication process (factor 6, Table 1).

Fitness and fecundity of the target population

The adaptation of the invasive organism to the climatic conditions in a new environment (e.g., Galera, Chwedorzewska & Wódkiewicz, 2015) is as important as land use. Dodd et al. (2015) proposed a factor named ‘climate match of species with site’. This factor could be used to prioritize eradication efforts of especially dangerous species (candidate species), which have not reached the status of invasive species in a specific area (Cunningham et al., 2004; Branquart et al., 2016) however its use for the evaluation of eradication feasibility is in our opinion doubtful. If the target species has already established a naturalized and/or invasive population in an area the analysis of this factor is in our opinion not necessary. The ability to exploit specific climatic conditions may not be uniform across the whole species and may be population specific, due to intra- and inter-specific admixture (Lachmuth, Durka & Schurr, 2010; Lawson Handley et al., 2011). The species has to be highly climate adapted to be considered naturalized and even more to be declared invasive (e.g., Richardson et al., 2000; Richardson, Pyšek & Carlton, 2011; Colautti & MacIsaac, 2004). Additionally, climate match is difficult to assess as many different climatic factors may interplay with different species life history traits (Kriticos & Brunel, 2016; Galera et al., 2018). The use of several map types illustrating the distribution of climate types in the species native and non-native range may be an acceptable solution (Branquart et al., 2016). Nevertheless such traits of invasive species as phenotypic plasticity (Galera, Chwedorzewska & Wódkiewicz, 2015), tolerance to a broad range of environmental conditions (Higgins & Richardson, 2014) and rapid adaptation to climate conditions (Colautti & Barrett, 2013), that all blur the climatic match of the species, have to be considered as well. Therefore we decided to name this factor ‘adaptation to new climate conditions’ and to distinguish its two components (see factor 8, Table 1).

Another factor from the group aggregating species traits is ‘vegetative propagation’ (factor 10, Table 1). Dodd et al. (2015) and Blood et al. (2019) named this factor ‘vegetative reproduction’, but we propose to change its name to ‘vegetative propagation’ as no new genetic combination arises through asexually driven population number increase. As a result there is no chance of new combination of alleles to arise (save mutations) with a new invasion potential. If the alien species invades a habitat differing much from the home habitat, change in genetic information due to new allele combination through intra- or inter-specific admixture usually triggers a rapid spread of the invasive species in the non-native range (e.g., Buhk & Thielsch, 2015).

Detectability of target species

The factor describing the detectability of the focus species, ‘detection possibility’ (Table 1, factor 13), also needs to be clarified as it has different aspects addressed in the literature. This factor describes the feasibility of correctly identifying the focus species. The important aspect here is the ability to distinguish the invader from other, local species to avoid mistakes. Many means of such detection are possible, depending on the species. Besides identification with naked eye or microscope (Pluess et al., 2012a), detection using remote sensing (Müllerová et al., 2016 and literature cited therein) or identification by detector dogs (Cherry et al., 2016; Williams et al., 2019) may be considered.

Reinvasion possibility

In a separate group, the “reinvasion context”, we have included one factor, the ’invasion pathways’ (factor 24, Fig. 1) as repeated invasion brings the necessity to repeat the eradication action. For this factor it is important to consider and analyze the invasion pathways and vectors which may reintroduce the unwanted species after successful eradication or reinforce the population being eradicated (e.g., Faulkner et al., 2020; Shackleton et al., 2020).

The inclusion of other factors influencing eradication success into USEF is possible. However in every instance we suggest giving a precise definition and discussing the differences from previously used terms.

Evaluating eradication feasibility

Different values/states of each factor may either promote or restrict eradication. We analyzed each factor separately and within it distinguished from 3 to 5 categories (depending on factor) potentially influencing the eradication success in a different way. We assumed that the distinguished categories should be universal (applicable in the assessment of each invasive species management action) and relatively easy to classify (categorization should be clear and mutually exclusive). We used an ordinal scale to score the factor’s category impact on eradication success (Table 2). For all factors the lowest score values (1) reflect factor states promoting eradication and highest score values (maximum for each factor) reflect factor states hindering eradication most.

In order to define and score factor states we tried to use the suggestions given in the literature, possibly clarifying and adjusting the sets of categories to all possible cases (Table 2). Nevertheless, only for three factors (‘infestation size’, ‘monitoring area size’, ‘propagule longevity’) we decided to adopt a scale proposed by other authors. For three other factors (‘number of separate infestation sites’, ‘pre-reproductive period’, ‘reaction time’) we extended the scales and proposed new factor value definitions. We proposed completely new definitions for the categories distinguished for the following twelve factors: ‘isolation of infestation’, ‘monitoring rate’, ‘adaptation to new climate conditions’, ‘number and distribution of propagules’, ‘vegetative propagation’, ‘detection possibility’, ‘knowledge of current location of infestation sites’, ‘understanding of species biology’, ‘eradication achieved elsewhere’, ‘applicable control methods’, ‘personnel awareness’ and ‘invasion pathways’. In all other cases, our concepts and/or definitions of categories are partly based on the literature (Table 2).

Some factors were readily dividable into ranges, which could be easily ranked. Other factors were more problematic, raising the necessity to develop a more complicated approach. Below we focus on the more demanding factors in the sense of categorization and/or scoring.

For ‘isolation of infestation’ (Table 2, categories for factor 3) we used a very simple categorization proposed by Pluess et al. (2012a) and Pluess et al. (2012b). In addition we took into account the importance of geographic (distance of islands from the nearest continent) and ecological barriers within the mainland (i.e., continent or large island). We compiled the ranges of the islands’ size and distances based on the variability of islands in terms of these characteristics (according to data in UN Islands Directory, Dahl, 1998). In the case of islands located in the Southern Ocean, we propose to take into account their distance from a continent other than Antarctica (this continent is a marginal source of plant propagules, Galera et al., 2018).

In the case of ‘monitoring rate’ we assumed that a typical search attendance is once a month. If the recommended monitoring frequency is higher than once a month, and the monitoring season lasts the entire year, then a very high monitoring intensity with over 12 visits per year is needed (Table 2, factor 5).

For ‘land use and ownership of infested area’ we found it impossible to create clear categories with the use of all four components (compare definition of factor 6. in Table 1 and definitions of categories of this factor in Table 2). We decided to include land ownership relations, because without obtaining the appropriate approvals, the action would be illegal. The second argument is the fact that the data on the number and attitude of owners of the area under consideration for species eradication is relatively easy to obtain. For the factor of site ‘accessibility’ (factor 7, Table 2), we have tried to include both the distance from the headquarters that workers have to travel to the infestation site and possible difficulties resulting from landform.

The factor ‘adaptation to new climate conditions’ (factor 8, Table 2) is an interplay between the climatic conditions in the native and nonnative range and species adaptability to new climate conditions. We decided to measure both species adaptability and climate distance based on five major climate zones distinguished in a popular and constantly updated Köppen-Geiger classification (e.g., Peel, Finlayson & McMahon, 2007; Beck et al., 2018; MacLeod & Korycińska, 2019). For simplicity we categorized species climate adaptability into: low—species present in one climatic zone, medium—present in two zones, high—present in three or more climatic zones. In the case of climate adaptability, both primary and secondary geographical range of the species should be taken into account. For new alien taxa, e.g., hybrids/cultivars having a relatively small range, it should be assumed that their climatic adaptation abilities are unknown. We categorized the climatic distance between the site of the analyzed infestation and the most similar climatic conditions within the continuous range of the species as: short—species present in the same climatic zone, medium—species present one climatic zone away, high—species present two or more climatic zones away. Subsequently we scored the possible combinations of these variables. Our rationale for scoring was that the more the species is adaptable and the shorter the climatic distance the more difficult it will be to successfully eradicate the species.

In ‘number and distribution of propagules’ (categories for factor 9 in Table 2), both seeds and vegetative propagules should be taken into account. We assume that if propagule number is lower and they are concentrated in specific locations it is easier to deplete the propagule bank. The cutoff values for propagule number were based on the assessment of the size of soil seed banks of different plant communities presented in Baskin & Baskin (2001), as seeds may retain viability much longer than vegetative propagules.

For ‘propagule longevity’ (factor 11, Table 2) we decided on the simple 3-point scale, taking into account the concept of three types of seed banks (Thompson, Bakker & Bekker, 1997). We are also aware that it is very difficult to determine the seed bank longevity greater than 5 years. In order to determine this factor, longevity of both generative and vegetative propagules should be considered. When classifying into a category, the longest period should be taken into account. In practice it is usually seed longevity.

Similarly to factor 11, both generative and vegetative propagules must be considered in determining the ‘pre-reproductive period’. When classifying into categories, the shortest period should be taken into account (in practice, vegetative reproduction usually occurs faster). The scale of ‘pre-reproductive period’ (factor 12, Table 2) was extended by us to include species, that start producing propagules after more than 10 years (e.g., some trees).

The four-point scale used by Panetta & Timmins (2004), Blood & James (2018) and Blood et al. (2019) for the detectability period has been adopted by us to categorize ‘annual period of detectability prior to seed set‘ and extended to six levels. We classified species which are “always inconspicuous” as plants with annual period of detectability of “less than 1 month” (factor 14, Table 2). Only sexual reproduction has to be considered when categorizing, as vegetative propagules are often produced all year round.

To assess the impact of ‘eradication achieved elsewhere’ on the target eradication it is important to know if any of them proved to be successful. Gains from successful eradications depend on the climatic difference between the location of those actions and the location of eradication attempt being analyzed. This is reflected in our evaluation of the factor categories on eradication success (Table 2).

The evaluation of ‘applicable control methods’ includes a two-step procedure. The first step should be to assess if any of the known eradication methods is successful. The second step is scoring the factor based on the most successful method. The rationale is that at least one effective eradication method suffices to remove the species.

In the case of ‘knowledge of current location of infestation sites’ and ‘understanding of species biology’, we assume that the managers have a basic knowledge of the location of the target infestation and biology of the species (compare Table 2, categories for factors 15 and 16). When assessing the level of ‘personnel awareness’ and ‘sufficient allocation of resources’ (Table 2, factors 20 and 22), only three categories were proposed, taking into account the fact that the precise definition of categories is not possible. We are also aware that objective self-assessment is difficult, if not impossible sometimes. We assume the good will and common sense of workers rejecting the possibility of a total lack of awareness and financing. In the case of ‘coordination between monitoring agencies’ (Table 2, categorization for factor 21), we assumed that the situation in which the action is carried out by one institution is the most favorable (Dodd et al., 2015).

For ‘invasion pathways’, we assumed that the risk of reinvasion always exists. When categorizing, we took into account the knowledge of potential introduction pathways and the possibility of blocking or controlling them (compare Table 2, categories for factor 24).

Indications for the system usage

It is tempting to sum up all the score values to depict the overall effectiveness of an eradication action. Unfortunately because of the different nature and varying impact of factors on eradication success this cannot be done. Even if all the factors are scored, the resulting comparison between actions may be misleading. To evaluate the overall feasibility of an analyzed eradication action the factors should be weighted according to their importance in obstructing the eradication success. This can be achieved only after sufficient data accumulation. This data should be reported according to a unified reporting scheme. Our work is a suggestion of such a scheme.

At first the quality of the assessment should be performed. The higher the number of factors, which cannot be evaluated due to the lack of information the lower the quality of the assessment. The lack of accumulated data acquired in accordance with a unified scheme is the main obstacle to comparing eradication actions (e.g., Wilson et al., 2014). Cunningham et al. (2004) noted: “The eradication feasibility measure reported in this study needs further testing against actual eradications”. Despite the passage of almost twenty years since their publication, this remark is still relevant. The number of cases of eradication actions, the effectiveness of which has already been proven, remains small. For example Gardener, Atkinson & Rentería (2010) out of 30 analyzed eradication projects involving 23 species evaluated only four species as eradicated. Dodd et al. (2015) out of 17 analyzed infestations deemed only one species as successfully eradicated.

At present to circumvent the necessity of weighting factors we propose first to divide the scored value by the maximum value for each factor. In such a way the evaluation of each factor’s input is represented by a number between 0.20 and 1.00. The sum of factors’ inputs results in an overall eradication action feasibility assessment. The greater the number the more difficult it is to eradicate the species.

Another problem associated with eradication evaluation is missing information. This is understandable, especially in the case of recent actions as it is sometimes difficult at first to ascribe a value to an eradication related factor. These gaps however impair our ability to evaluate the action. An earlier proposed method is to ascribe a mean value (Weiss & Iaconis, 2002) or a default value in absence of information (Panetta & Timmins, 2004; Blood & James, 2018; Blood et al., 2019). This attempt however yields data for comparison, but the quality of such an assessment or comparison is low. Therefore we propose to restrain from comparison until some assessment of data is available. The drawbacks associated with lack of data are further discussed in the next section.

An example of USEF usage

To illustrate how our system works, we prepared an exemplary comparison of three different invasions (Table S1 in Supplemental Data). It is a thorough comparison of a P. annua invasion in the Antarctic on King George Island, and Stellaria media (L.) Vill. and P. annua invasions on Macquarie Island. We chose these invasions to show how our system can help to compare invasions of two different species at the same site, or invasions of the same species but in different locations. We assessed each factor using our proposed scale and added descriptions and explanations for each assessment based on available data published in the literature.

We summed the factors’ inputs for all three eradications. The P. annua eradication on King George Island was evaluated at 13.75, P. annua on Macquarie Island at 8.38 and S. media—at 12.86. While we were able to evaluate all the 24 factors for P. annua on King George Island, we found no information for evaluating 10 factors for P. annua on Macquarie Island and five factors for S. media on this island. This stresses the importance of missing data. If we scored each of the factors with missing data as 0.50, the results for P. annua and S. media on Macquarie Island would be respectively 13.38 and 15.36, but at this level of missing data the eradication assessment has a low quality and the whole eradication action comparison needs to be treated with caution as stated earlier. We can further compare only separate and known factors between the selected eradication actions.

One could assume that in every eradication instance managers possessed information on the infestation size, but the lack of reporting of this crucial information disables further sound comparisons. Even in the example of relatively well documented invasions on Macquarie Island we were forced to mark some factors as unknown, because specific information was insufficient with just vague descriptions published or even lacking completely (Table S1, factors 1, 2, 4). The absence of almost any information concerning infestation size and distribution of P. annua on Macquarie Island severely affected the eradication success probability assessment (see also factor 15 in Table S1). The only published descriptions are very general and no studies about the topic were published to inform about the exact area.

This shows how remote assessment is difficult and highlights how much data is still needed. For the comparison of different eradication campaigns, comparable datasets organized in accordance to commonly defined driving factors are needed. Unfortunately in up-to-date literature even the most evident and necessary data is not presented. For example Pluess et al. (2012b) observed that “precise quantitative measurement of the infestation area (in km²) was rarely available”.

For the invasions compared here as well as for other islands, the distance from the closest continent/landmass (e.g., Tye, Soria & Gardener, 2002; Caujape-Castells et al., 2010; Kueffer et al., 2010) seems to be a good measure of the degree of isolation. However, even in the case of such an easily measurable factor, the assessment must take into account the specificity of the location of a given infestation. Macquarie Island lies 1,800 km form Antarctica (nearest continent, Dahl, 1998), 1,100 km from New Zealand and 640 km from Auckland Islands and is surrounded by a few small islets (Macquarie Island Nature Reserve and World Heritage Area, 2006). King George Island belongs to the South Shetlands archipelago and is situated only 160 km from Antarctica. For this reason, the isolation index (based on distances to nearest island, group and continent) proposed by Dahl, is high for Macquarie Island (112), while for King George Island it is only 26 and the island was not considered “more isolated” (Dahl, 1998). However, Antarctica is a marginal source of propagules, and the ecological isolation of the Antarctic region is enhanced by the atmospheric Polar Vortex and oceanic Polar Frontal Zone (Galera et al., 2018). In line with our earlier proposal (see chapter Evaluating eradication feasibility), we took into account the distance of two analyzed islands from another continents other than Antarctica. Taking into account that the terrestrial ecosystems of King George Island are environmental islands (Table S1), it can be concluded that this invasive population of P. annua is one of the world’s most isolated infestations. In this situation, we considered the isolation of both analyzed islands to be a very favorable factor for eradication and scored them in the same way (Table S1, factor 3).

Our scoring of the accessibility factor differed between the compared eradications. The topography of Macquarie Island makes it difficult to reach the invasive populations, and access may be also restricted by dense seal colonies and penguin rookeries. Such problems are non-existent on King George Island, as the target populations are easily reached from the buildings of the Polish Antarctic Station (Table S1, factor 7).

Unfortunately we were not able to escape all ambiguity by using our system, even if precise information had been published. This is depicted e.g., in our assessment of the number of separate infestation sites in the case of S. media on Macquarie Island. Williams et al. (2019) wrote about “eight loosely defined, high density populations which varied in area and S. media density” and “six isolated plants […] detected outside these populations”. We considered this infestation to consist of eight sites, disregarding information on separate individuals (Table S1, factor 2).

In comparison to the two infestations on Macquarie Island, the P. annua invasion on King George Island has either a greater or equal chance of being successfully eliminated. Even with the simplified scoring system we used, the analysis of Table S1 allows specific conclusions to be drawn. The compared cases are similar for eight factors as the same factors’ inputs were obtained for all three infestations (factors 3, 6, 8, 13, 16, 18, 23 and 24 in Table S1). This is understandable because the two islands are located in the polar climate zone, and both P. annua and S. media are small annual weeds with a similar life strategy.

The comparison of two P. annua populations according to “one species/different regions” scheme shows that there is a greater possibility of successful eradication of the species from King George Island in terms of three factors: ‘number and distribution of propagules’, ‘knowledge of current location of infestation sites’ and ‘eradication achieved elsewhere’ (factors 9, 15 and 17 in Table S1, respectively). On the other hand, ‘propagule longevity’ (factor 11) reduces the likelihood of successful eradication of P. annua from this Island. It is worth noting that in terms of ‘propagule longevity’ and ‘number and distribution of propagules’, the two compared populations of annual bluegrass differ significantly. The invasive P. annua population on King George Island can theoretically be considered relatively easy to eradicate. However, in practice, even this so far turned out to be difficult (Galera et al., 2021).

The two invasions taking place on Macquarie Island being compared to “one region/different species” scheme differ in three factors (factors 10, 15 and 19, Table S1) and only one of them is related to the “species context”, i.e., ‘vegetative propagation’. However, the results of this comparison are uncertain due to the lack of published data.

Our comparison shows that despite King George and Macquarie Island invasions are similar in environmental conditions (polar climate) and the target species (annual plants creating seed banks), they differ in eradication feasibility. Out of the two invasions on Macquarie Island the S. media invasion is better documented than P. annua, therefore its eradication feasibility can be more easily estimated. This shows that the usage of our system makes it easy to reveal the overlooked aspects of invasion. On the other side of the spectrum, a proper assessment of a very well researched invasion at King George Island would be an important tool to start a definite eradication plan and actions.

Benefits of USEF usage

USEF offers a simple tool for general and easy eradication assessment and comparison, by covering and organizing a broad range of terminology. Our system is intended to be used by researchers on site, to take under consideration as many factors as possible into eradication preparations and help them share the data further in a universal and accessible manner. Eradication prioritization on islands is a good example where our system can be used. Islands scattered among different climatic zones can be exposed to vastly differing conditions and therefore comparing locally devised frameworks can be difficult and not informative (e.g., Dawson et al., 2014; Helmstedt et al., 2016; Stanbury et al., 2017).

Defining a universal scoring system for eradication associated factors is challenging. So far published papers using prospective scoring systems, simple scales or categorization for evaluating eradication impedance considered local specificity of the eradication course, but lacked universality (see e.g., Cunningham et al., 2004; Panetta & Timmins, 2004; Blood et al., 2019). We propose a simple ordinal evaluation scale, which can be easily amended, altered or adapted to the local conditions, or reported next to other scales used in ongoing actions. In our example (Table S1) we attempted an eradication feasibility scoring that would not relay on relativity to other compared invasions. The developed universal scoring system of eradication associated factors proposed by us should be reevaluated after accumulation of experience reported in accordance with the unified description system.