Vector analysis : a tool for preventing the introduction of invasive alien species into protected areas

Invasive alien species are the main agent of biodiversity loss in protected natural areas. Prevention is the most appropriate management tool for addressing this challenge, however, virtually all ongoing management efforts are focused on established populations. Although invasion processes include stochastic components, it is possible to compare the different vectors of introduction that operate in a particular area in terms of their potential to transport species of high risk of invasion efficiently and, once identified, to establish strategies of prevention, early detection and rapid action. This study proposes a system of prioritization of vectors of alien plant dispersal for optimizing the efforts for preventing invasion. The system was developed for the Ernesto Tornquist Provincial Park (province of Buenos Aires, Argentina), but it is directly applicable to other areas. Natural and anthropogenic vectors were evaluated and lists of the species potentially transported by each vector were elaborated according to the characteristics of their propagules. The system analyzes the relative importance of each vector according to: 1) the severity of the potential impact of transportable species, 2) the difficulty of controlling these species, and 3) the volume of transportable propagules. In the case under study, the maximum value of risk corresponds to cargo, followed by vehicles, streams, unintentional human transport, intentional human transport, wind and finally, animals. This analysis can lead to prevention strategies, mapping of dispersal routes and actions of early detection and rapid response.


Introduction
The impact of invasive alien species is a key component of global change and it is considered one of the main causes of biodiversity loss worldwide (Sala et al. 2000, Lövei and Lewinsohn 2012, Simberloff et al. 2013, Alexander et al. 2014).All protected natural areas contain alien species that are recognized as the main threat to their conservation objectives.Predictions indicate that their importance will increase in the future unless effective management measures are adopted (McKinney 2002, Pyšek et al. 2002).The effects of invasions can be manifested at different scales and in various ways, including reduction in the richness and abundance of species of the native biota, genetic changes in native populations through hybridization and interruptions in mutualistic networks (Pyšek et al. 2012).In some cases, the effects of the presence of one or more invasive species are so profound that they disrupt the functioning of entire ecosystems and interfere with their resilience and ability to provide ecosystem services (Vilà et al. 2011, Simberloff et al. 2013).
Invasion processes involve the successful overcoming of several challenges: a potential invader must survive transport from its place of origin, become established in the new site, persist and reproduce until a sustainable population is formed that eventually expands (Theoharides and Dukes 2007, Blackburn et al. 2011, Jeschke et al. 2013).The ability to successfully overcome these stages depends not only on the species' own characteristics, but also on the characteristics of the invaded habitat that determine its susceptibility to invasion, the number of propagules and introduction events, the establishment of effective relationships with local dispersal agents and other symbionts and the particular conditions at the time of the arrival of the propagules (Marco et al. 2002, Colautti et al. 2006, Dechoum et al. 2015, Amodeo and Zalba 2017).
The management of invasive alien species includes four basic components: prevention, early detection, eradication and control that coincide with each stage of the invasion process (Wittenberg and Cock 2001, Lodge et al. 2006, Davies and Sheley 2007).The best cost-effective method for dealing with invasive alien species is in the area of prevention, since the costs and impacts generated by an invasion process increase and sometimes the problems become irreversible (Leung et al. 2002, Ziller and Zalba 2007, Anderson et al. 2014).
Vectors are the transfer mechanisms responsible for the introduction and spread of invasive species in a certain area, including a wide variety of physical means or agents, from ballast water to horticulture, biological control and aquaculture (Ruiz and Carlton 2003).Vector interception or disruption has been identified as "the most vulnerable and directly manageable portion of the invasion sequence", as they allow to simultaneously avoid the delivery of whole sets of transportable species (Carlton and Ruiz 2005).
Many risk analysis associated to the probabilities of introduction by certain vectors has been developed, mostly at national or state borders (Gordon et al. 2012, Grosholz et al. 2012, Conser 2013, Kelly et al. 2013).Most of them consider the capacity of the vectors to safely transport propagules, the volume that can be carried and the frequency of operation, as well as the impacts associated to the transportable taxa.This is not the case for protected areas, where these kind of analysis are extremely infrequent.Despite the consensus on the disproportionate importance of prevention in the management of biological invasions, most management actions developed in nature reserves focus on the control or eradication of established populations (Schüttler and Karez 2008, Genovesi and Monaco 2013, Pauchard et al. 2015).This situation could be explained, at least in part, since the extent and seriousness of the problems attract the attention of those responsible for the management of the reserves disproportionately.Apart from the causes of this scenario, the consequences seem clear: the lack of effective preventive actions compromises the sustainability of protected areas that face the threat of invasive alien species.
Moreover, the scarcity of tools for organizing actions that reduce the risk of introduction and establishment of new species is daunting (Davies and Sheley 2007).Although invasion processes include stochastic components, like the co-occurrence of propagule arrival and appropriate environmental conditions for establishment (Radford 2013), it is likely to anticipate which species are most likely to arrive in an area, the severity of their potential impacts, the most likely means of arrival, and which sites are most likely to be colonized.In particular, it is possible to compare the different vectors of introduction operating in a given area in terms of their potential to transport highly invasive species efficiently.
Vectors also travel through more or less predictable routes known as pathways (Mack et al. 2003).The combination of knowledge about vectors with higher chances of transporting high risk species and the routes that they travel to and within a particular area leads to the organization of preventive actions, early detection and rapid action (Lodge et al. 2006, Ziller andZalba 2007).This alternative also has the advantage of simultaneously addressing the risk of introduction of complete sets of species sharing the same means of transport and / or pathways of introduction and dispersion.
The objective of this study is to create a system of risk analysis for the introduction of invasive or potentially invasive alien plants by identifying the vectors of the highest priority for control.We selected the Ernesto Tornquist Provincial Park, a nature reserve located in the southern part of the Pampas Biome, in the Argentine Republic, as a case of analysis for the elaboration and application of this system.The park is dominated by grass steppes and surrounded by an agricultural landscape.Vectors of plant dispersal in the area include physical means like wind and watercourses, dispersal by birds, mammals and invertebrates, and human mediated spread in association to footwear and clothing, vehicles and cargo (Zalba and Villamil 2002, Loydi and Zalba 2009, Amodeo and Zalba 2013).
The reserve undergoes intense invasions by alien species, including different species of trees and shrubs (Zalba andVillamil 2002, Zalba et al. 2009).Apart from this problem, there is a high number of introduced plant species in the region that have not yet become established in the reserve (Long and Grassini 1997), and preventing their entry should be a priority in the management of the area.The analysis of routes and vectors is an appropriate response to reduce the impact of invasive species by minimizing the risks of introduction, as well as lowering the very high costs associated with the control.

Study area
The Pampas biome is one of the most characteristic landscapes in southern South America, as well as being one of the most greatly transformed ecosystems by anthropogenic actions, with only a very small area that is protected effectively (Bertonatti et al. 2000, Bilenca andMiñarro 2004).The grasslands of South America face a serious and increasing challenge associated with the progress of invasive alien species, particularly woody plants (Fonseca et al. 2013).The Ernesto Tornquist Provincial Park (ETPP) represents one of the few protected areas of Pampas grassland in Argentina (Bilenca andMiñarro 2004, De Villalobos andZalba 2010).The reserve covers an area of approximately 6700 ha in the central area of Sierra de la Ventana, in the province of Buenos Aires, Argentina (38°3.90'S,61°58.33'W).The climate in the region is temperate and rainfall varies between 500 and 800 mm annually (Burgos 1968).The vegetation is dominated by grass steppes, including species of Stipa, Nassella, Piptochaetium, and Festuca, as well as herbs and shrubs of Asteraceae.The flora of the park includes some 550 species of native plants and some 140 alien species (Long andGrassini 1997, Long et al. 2004).Damiani (2007) cites a total of 324 alien plant species growing within the ETPP and in an area of about 20 km around it, including agricultural and livestock fields, paved roads, secondary roads, and parks and gardens in small villages.Twenty-three species that behave as invasive in the area, extensively growing over natural and semi natural environments, and 23 others that can be considered to be of high risk on account of their biological characteristics and previous invasive behavior, have not yet been detected in ETPP, or are restricted to intensive use zones (Damiani 2007, Long andGrassini 1997, María Andrea Long, Systematic Botany, Universidad Nacional del Sur, pers. comm.).All these species can therefore be considered as high priority in a prevention strategy (Appendix 1).

Methods
The characteristics of the propagules (presence of wings, pappus, hooks, sweet pulp, etc.) and dispersal strategies of the 46 species considered to be of high priority for prevention were analyzed from the literature and the vectors that might intervene in their dispersion were identified.
In order to analyze the relative importance of each vector, the severity of the potential impact and the difficulty of controlling each transportable species were taken into account, as well as the volume of propagules that the vector could carry.
The potential impact of the vector index (PIV) was defined as the weighted sum of the number of species transportable by a vector for each category of potential impact: PIV = 100 * number of species with high PI + 10 * number of species with medium PI + number of species with low PI.
The values of high, medium and low potential impact were taken from Damiani (2007), who established an impact index considering the risk of establishment of the species based on fourteen criteria: previous invasive behavior, niche width, density of growth, hybridization risk, allelopathy, toxicity for humans, toxicity for wildlife, flammability (capacity to increase fire frequency or intensity), palatability, capacity to host parasites and pathogens, life cycle, reproductive strategy, seed production and dispersal.Each criterion has different alternatives associated with corresponding numeric values that are combined in a final estimation of potential impact of the species.
The control difficulty index of the species transported by the vector (CDV) was defined as the weighted sum of the number of species transportable by the said vector corresponding to each category of control difficulty: CDV = 100 * number of species with high CD + 10 * number of species with mean CD + number of species with low CD.
The values of high, medium and low control difficulty were also extracted from Damiani (2007), who calculated them considering six species features: presence of spines and stinging hairs, generation time, ability to regrow after cutting, response to grazing, response to fire, and persistence in the seed bank.Numerical indexes for each criterion were combined to assess the difficulty to control each species.
The severity of impact of each vector (SI) was calculated from the values of the potential impact and control difficulty indexes of the species transported by the vector, according to: Where SImax represents the maximum severity of impact obtained among the considered vectors.
The Transportable Volume (TV) was estimated by analyzing both the number of propagules available for transport (TP) and the carrying capacity of the vector (CC).
The number of available propagules (TP) for each vector was calculated by combining the information related to the abundance of the species in the area with the production and temporal availability of transportable propagules by that vector.
The abundance of each species in the study area was estimated on a relative scale, assigning a value of 1 to the rare species (few populations of a few individuals), the value of 2 to the abundant species (few populations with many individuals or many populations with few individuals) and the value of 3 to very abundant species (many populations with many individuals).This information was obtained from literature (Long and Grassini 1997) and from consultations with specialists of the regional flora.The number of propagules produced by each species was classified as low (1), moderate (2), high (3) or very high (4), considering the ability of an adult plant to produce seeds and / or vegetative reproduction structures (bulbs, rhizomes, stolons, tubers and plant cuttings).This data was extracted from the bibliography.The proportion of months in the year during which the propagules of each species are available for eventual transport by each vector was also determined.Thus, for example, a plant producing fleshy fruits available for consumption and dispersal by vertebrates for two months each year would obtain a value of 2/12 = 0.17 for the animal vector; whereas we could expect an availability of 12/12 = 1 for vector loads, if their seeds remain viable in the soil.
These three variables were multiplied by each other to calculate the abundance of propagules for each species.The abundance values of propagules for all transportable species were added to obtain the total number of propagules available for transport by each vector (TP).
Two variables were considered for estimating the carrying capacity of each vector (CC): 1-the volume transported in each potential introduction event, defined in relative units: 1 small; 10 medium; 100 large; 1000 very large, and 2-the frequency of vector activity throughout the year in the study area, expressed in relative units: 1 low; 10 medium; 100 high; 1000 very high.
These two variables were multiplied to calculate the carrying capacity (CC) of each vector.
The transportable volume (TV) per vector was calculated by adding the propagation availability and carrying capacity: TV = (TP +CC) / TVmax Where TVmax represents the volume of transportable propagules by the vector with the greatest transport capacity.
Finally, the values of impact severity (SI) and transportable volume (TV) were combined to calculate the risk associated with each vector (RV): RV = (2 * SI+ TV) / 3 The impact severity value was multiplied by 2 to reflect its relative importance when analyzing the risk associated with each vector.
A diagram of this analysis is presented in Fig. 1.

Results
The analysis of the propagules and dispersal strategies of the species of high priority of prevention in the PPET allowed us to associate them with a total of three natural and three anthropogenic vectors.The natural vectors identified were streams, wildlife and wind.The anthropogenic vectors included transport by vehicles (in mud attached to the chassis and tyres), movement directly associated with people (unintentional: in footwear and clothing, food, camping equipment, and intentional: ornamental plants and vegetables) and the movement associated with cargo (soil, sand, debris, and dry plant material).
Of the 46 species evaluated, 25 have propagules with structures that facilitate their dispersion by wind (e.g.small and light seeds, winged diasporas, feathery organs), 7 show seeds with traits that promote their dispersal by water (light seeds or floating vegetative structures) and 13 fruits are potentially dispersed by animals (edible or with hooks, barbs or awns that adhere to fur).We also concluded that all the propagules of the analyzed species could be transported in loads of materials (earth, debris, sand), whereas 39 show traits that would facilitate their transport by cars, trucks and other vehicles (small seeds, adherent propagules).Twenty-eight species could be easily dispersed directly and unintentionally by people (on footwear and clothing, such as fruits of food plants or associated with camping equipment).Finally, 23 species could be intentionally mobilized by the people for their ornamental value or cultivation for other human purposes (Appendix 1, Fig. 2A).
The analysis of the different vectors, combining the potential impact of the transportable species (Damiani 2007), resulted in an index of the potential impact of the vector that varied between 240 and 1693.On the other hand, the index of the difficulty of control of species transported by the vector takes values that go between 321 and 1891.In both cases, the maximum value corresponds to cargos and the smaller one to streams.Thus, the severity of impact of the vector index was maximized for cargo (1), followed by vehicles (0.87), unintentional human transport (0.56), intentional human transport (0.48), wind (0.47), wildlife (0.31) and streams (0.16) (Fig. 2B).
Regarding the transport capacity of the different vectors, the transportable propagules index varied between 11 and 226, again reaching the maximum value for cargo and the minimum for streams.
Twenty species were evaluated as very abundant, 16 as abundant and 10 as rare.A high number of propagules were produced by 30.4% of the species under study, moderate production by 50% and a low number of propagules by eight species (17.4%).Only one species (Melia azedarach) was considered as having a very high production of propagules.
It was defined that propagules of all plants that can be transported in association with cargo or intentionally by humans are available for these vectors for 12 months per year.Vehicles and unintentional human transport might transport species with available propagules for periods of two to five months per year; whereas animals and streams could transport species with available propagules between one and 12 months per year.The wind vector could disperse species with available propagules between one and three months per year.
The carrying capacity, for its part, was considered maximum for the cargo, stream and wind vectors, whereas the minimum value was for the vehicle and unintentional human transport vectors.
The volume transported at each potential introduction event was considered to be very large for cargo, streams, wind and intentional human transport; medium for vehicles and animals and small for unintentional human transport.
Only intentional human transport was considered to have a very low frequency of activity.For unintentional transport by humans and mediated by animals, the frequency is considered high, whereas it is classified as medium for cargo, wind, vehicles and streams.
Thus, the transportable volume index resulted maximum for cargo (1), followed immediately by streams and wind (0.98), whereas the rest of the vectors received values of ten to one hundred times lower in terms of their relative transport capacity (Fig. 2C).

Discussion
In this study, we designed and applied a risk analysis system associated with vectors responsible for the introduction and dispersal of plant species, which constitutes a simple and novel alternative of high potential value for decreasing the risk associated to invasive species by reducing propagule pressure in a variety of ways: improving detection measures and border policies, limiting vector contamination, controlling invasive populations in source regions, helping to raise public awareness of problems to find alternatives for invasive species (Pyšek and Richardson 2010).As we previously mentioned, there are many antecedents aimed at reducing unwanted introductions by assessing the risk associated with vectors and pathways, most of them applied at national or state borders (Gordon et al. 2012, Grosholz et al. 2012, Conser 2013, Kelly et al. 2013).The main differences of our approach include it local focus, primarily designed for individual reserves, what can result in an improvement of the precision of the analysis.It is also based on a context-specific perspective that drives the attention of the administrators to real threats posed by potentially invasive species that are present in the surroundings.
As discussed in detail below, the ranking obtained in this work is consistent with particular features of our case study, including heavy transit of vehicles associated to tourism and cargo, strong and frequent winds (particularly during plant dispersal seasons), and a dense network of water courses.This situation will clearly change in other reserves, but the framework should still be useful to calculate a specific scoring of dispersal vectors.
The development of an index of the relative importance of vectors of introduction and dispersal presents some challenges, such as comparing vectors as different from each other as the wind and the sole of a shoe.Another weakness associated with this index is related to its need of information about the presence of invasive or potentially invasive species in the area surrounding the reserve that could be not available in some cases.On the other hand, data on previous invasive behavior of the species of interest is becoming easier to obtain with growing regional and national databases on invasive species.Something similar occurs with the characteristics of the species that permit to associate them to dispersal vectors, as most of the potentially invasive plants are regionally or even globally shared (Randall 2017).It is also important to recognize that the invasion process is dynamic and that some of the species that are classified as non-invasive at one time could become aggressive invaders if there are changes in the environmental conditions or the invasive population itself (Davis et al. 2000, Jiménez et al. 2011, Dechoum et al. 2014, Schrama and Bardgett 2016), possibly affecting the relative importance of the different vectors under analysis.It is therefore advisable to update the lists of species to be included in the analysis periodically.
Apart from the specific function of this analysis, the structure of the proposed indexes allows us to separate the different components associated with the potential impact of each vector and this could guide actions for reducing their potential impact on the area (Davies and Sheley 2007).Thus, management actions could be oriented, alternatively or complementarily, towards reducing the frequency or capacity of the individual transport of a vector, controlling its effects during periods of availability of transportable propagules and avoiding the transport of high risk species (e.g., through the elimination of the foci of invasion at the origin or in the path that a vector travels), etc.The structure of this system would also enable to evaluate more specific dispersal vectors (for example bicycles vs. walking or horseback riding), opening up interesting opportunities for the zoning and management of protected areas against the challenge of invasive alien species.
The vectors analyzed in our case study are clearly separated into two groups: on the one hand the anthropogenic agents (cargo, vehicles and intentional and unintentional transport by people) and, on the other hand, the natural means of dispersal (water, wind and animals).Due to their intrinsic characteristics, these two sets of vectors are associated with different and complementary management strategies, while the former allow and justify control and preventive actions; the latter are more naturally associated with early detection, since it is difficult or directly not feasible to reduce their transport capacity.
The results of the analysis place the vectors of cargo and transport associated with vehicles among the highest risks of entry of potentially invasive plant species in the study area.A number of studies have shown that unintentional transport by vehicles, either associated directly to the vehicle, or with cargo, is an important mechanism of seed dispersal (Clifford 1959, Lonsdale and Lane 1994, Von Der Lippe and Kowarik 2007, Ansong and Pickering 2013).The climatic conditions, the season of the year, the place where it is driven and the parts of the vehicle exposed to the environment affect this type of dispersal; as well as the weight and size of the seeds and the place where it is loaded (Zwaenepoel et al. 2006, Von der Lippe and Kowarik 2008, Veldman and Putz 2010, Taylor et al. 2012).While the relative importance of vehicles and transported freight is likely to vary between reserves, their particular relevance has an encouraging aspect, considering that the points of entry of freight vehicles and passenger cars are often few in number and are well defined, and that the same is true for the dispersal routes of these vectors within the reserves (internal roads and parking areas).The cleaning of vehicles before entering the area has proven to be an efficient measure for reducing the amount of propagules transported.The duration and type of washing will depend on the size and shape of the vehicle (Rew and Fleming 2011).Other preventive measures could include restricting vehicular traffic or creating invasive species free zones along roadsides (Davies and Sheley 2007).The handling of cargo allows specific actions, including quarantine systems (temporary deposit of the material entered in safe places that allow the detection and elimination of species that could germinate and settle there).There is also the option of evaluating the sites of origin of the materials, avoiding those affected by invasions of species transportable by this vector, in addition to thoroughly cleaning the containers before loading.These preventive measures should be complemented with periodic surveys along the internal roads in search of plants that might have entered these pathways, and their immediate removal (Lee and Chown 2009).
The wind vector represents a particular challenge (Davies and Sheley 2007) and preventive actions could be aimed at eliminating nuclei of transportable species located on the windward side of the reserve.If this were not possible, areas of high risk of invasion could be defined depending on the location of these nuclei and the prevailing winds during the months of seed production, which should be subject to regular monitoring and control tasks.
Streams as vectors follow in the order of risk.In this case the preventive measures are more complex and the effort should be directed at monitoring of the banks in search of points of entry of species (Cabra-Rivas et al. 2014).In general terms, the search actions should focus on streams that correspond to watersheds originating outside the reserve, concentrating the training efforts of personnel dedicated to detection on the set of species transportable by this vector, which clearly increases the chances of an efficient identification.In addition, resources could be devoted to the detection of nuclei of these species in sectors of the watershed located outside the reserve, where eradication would act as an efficient preventive measure that would save efforts and resources for the detection and control of internal foci of invasion (Säumel and Kowarik 2010).
The management of intentional and unintentional anthropogenic transport vectors includes a significant component of education and awareness.In the case of the former, it is essentially a question of avoiding the use of potentially invasive plant species in the staff residences and in the recreation areas (parks, gardens, shade trees) and replacing high risk plants in these sites.The unintentional transportation in clothing, footwear, backpacks, or other personal items have been documented in numerous studies (e.g.Whinam et al. 2005, McNeill et al. 2008, Pickering and Mount 2010, Auffret and Cousins 2013).Some reserves regulate the number of visitors and the period of access to reduce the unwanted introduction of propagules.There are natural protected areas in U.S.A. and New Zealand that require footwear, clothing, vehicles and equipment to be cleaned prior to entry (Genovesi and Monaco 2013).Researchers and park rangers pose a particularly high risk as they go to areas that are not accessible to the public, including areas of special conservation value (Chown et al. 2012, Huiskes et al. 2014).
The control of dispersal by animals leaves an even smaller space for prevention tasks, but could motivate monitoring tasks at sites with greater frequency of use by agents of high dispersal efficiency (e.g., wire fences or trees used as perches by frugivorous birds, Gosper et al. 2005, Buckley et al. 2006, Amodeo and Zalba 2013).
Making a list of high-risk species for each place and adapting the vectors that transport them, the analysis developed in this paper can be applied to other protected areas, political units or as a basis for the allocation of prevention efforts, early detection and early control of invasive species, translating the prevention premises frequently seen in the literature on biological invasions into concrete actions.

Aknowledgements
This work was supported by the National Scientific and Technical Research Council (CONICET), Argentina and Universidad Nacional del Sur, Bahía Blanca, Argentina.We are grateful to the Ernesto Tornquist Provincial Park rangers and authorities and to Rosemary Scoffield and Nicolás D´Onofrio for language revision.

Figure 1 .
Figure 1.Vector analysis schema.Diagram of the analysis of the relative importance of vectors associated with the introduction and dispersal of invasive alien plants in Ernesto Tornquist Provincial Park (Buenos Aires, Argentina).

Figure 2 .
Figure 2. Vectors of introduction and spread of invasive and potentially invasive alien plants present in intensive use zones of the Ernesto Tornquist Provincial Park and it's surroundings (Buenos Aires, Argentina).A Number of species associated to each dispersion vector according to the characteristics of their fruits and seeds and their human use B Severity of impact of vectors depending on the potential impact of transportable species and the difficulty of their control C Relative capacity of vectors to transport propagules D Risk associated with vectors depending on the potential impact of transportable species, the difficulty of their control and the transport capacity of the vector.

Table 1 .
Vectors characterization.Potential impact (PIV), control difficulty (CDV), severity of impact (SI), transportable propagules (TP), individual transport capacity, activity frequency, carrying capacity (CC), transportable volume (TV) and resulting risk (RV) for vectors capable of transporting invasive and potentially invasive alien plants present in intensive use zones of the Ernesto Tornquist Provincial Park and it's surroundings (Buenos Aires, Argentina).