Prioritization of invasive alien species with the potential to threaten agriculture and biodiversity in Kenya through horizon scanning

Invasive alien species (IAS) rank among the most significant drivers of species extinction and ecosystem degradation resulting in significant impacts on socio-economic development. The recent exponential spread of IAS in most of Africa is attributed to poor border biosecurity due to porous borders that have failed to prevent initial introductions. In addition, countries lack adequate information about potential invasions and have limited capacity to reduce the risk of invasions. Horizon scanning is an approach that prioritises the risks of potential IAS through rapid assessments. A group of 28 subject matter experts used an adapted methodology to assess 1700 potential IAS on a 5-point scale for the likelihood of entry and establishment, potential socio-economic impact, and impact on biodiversity. The individual scores were combined to rank the species according to their overall potential risk for the country. Confidence in individual and overall scores was recorded on a 3-point scale. This resulted in a priority list of 120 potential IAS (70 arthropods, 9 nematodes, 15 bacteria, 19 fungi/chromist, 1 viroid, and 6 viruses). Options for risk mitigation such as full pest risk analysis and detection surveys were suggested for prioritised species while species for which no immediate action was suggested, were added to the plant health risk register and a recommendation was made to regularly monitor the change in risk. By prioritising risks, horizon scanning guides resource allocation to interventions that are most likely to reduce risk and is very useful to National Plant Protection Organisations and other relevant stakeholders.


Introduction
Invasive alien plant pests cause increasingly significant impacts on the economy and livelihoods in Sub-Saharan Africa (SSA) (Eschen et al. 2021). In the last decade, Kenya has been particularly affected by new introductions of invasive plant pests 1 which damage cultivated plants. For example, in 2011, a new disease of maize was reported in the Bomet and Naivasha districts of Kenya (Andae 2012;Makiche 2012). The disease was later identified as maize lethal necrosis disease (MLND) (Wangai et al. 2012). It is caused mainly by co-infection with maize chlorotic mottle virus (MCMoV), a virus first reported in South America (Xie et al. 2011;Braidwood et al. 2018), and other cereal viruses (Louie 1980;Adams et al. 2013;Stewart et al. 2017). Following the first report in Kenya, MLND spread to other countries in Eastern Africa (Lukanda et al. 2014;Adams et al. 2014;Kagoda et al. 2016;Mengesha et al. 2019;Kiruwa et al. 2020).
Since then, Kenya has reported other destructive invasive plant pests such as tomato leaf miner, Pthorimaea (= Tuta) absoluta in 2013 (Guimapi et al. 2016); potato cyst nematodes (PCN), Globodera rosotchiensis in 2015 (Mwangi et al. 2015) and G. pallida in 2016 (Mburu et al. 2020); papaya mealybug, Paracoccus marginatus in 2016 (Macharia et al. 2017); fall armyworm, Spodoptera frugiperda in 2017 (De Groote et al. 2020); and the spotted-wing drosophila, Drosophila suzukii in 2019 (Kwadha et al. 2021). All these, and other invasive plant pests have caused enormous strain on a sector that supports millions of livelihoods in Kenya. For instance, De Groote et al. (2020) demonstrated that S. frugiperda caused losses of about a third of the annual maize production in Kenya. A survey in 2018 in Kenya showed 41% of tomato farmers had lost a large proportion of their crop to P. absoluta, with a mean seasonal production loss of 114,000 tonnes of tomatoes, based on farmers' own estimates (Rwomushana et al. 2019). At continental scale, Eschen et al. (2021) recently estimated the annual cost of invasive alien species (IAS) to African agriculture. The two most "costly" invasive plant pests were P. absoluta and S. frugiperda, accounting for USD 11.4 Bn and USD 9.4 Bn per annum, respectively. Invasive plant pests have also been demonstrated to affect biodiversity. For instance, in Kenya, increased density of Parthenium hysterophorus correlated with a reduction in species diversity and richness (Murono et al. 2018) while Maundu et al. (2009) demonstrated immense impacts on semiarid and arid ecosystems caused by Prosopis juliflora.
Invasive alien species may be introduced to countries through various pathways. 2 Intentional or unintentional human-mediated activities are frequently involved, but natural spread also occurs from native or areas where they have been introduced, aided majorly by weather (Desneux et al. 2011;Nagoshi et al. 2018;Essl et al. 2019). International trade remains a major cause of spread of invasive pests (Westphal et al. 2008). Unfortunately, once biological invasions are recorded anywhere in Africa, the organisms can spread across the continent unabated as was observed with P. absoluta, S. frugiperda and MLND (Mahuku et al. 2015;Guimapi et al. 2016). While it is unrealistic to expect border security to stop the spread of invasive pests within SSA, the situation could be improved by availability of timely and adequate information at the national and regional levels on the highest risk species and enhanced ability to share this information to support planning and implementation of sustainable management strategies such as prevention of invasions through the early detection, containment and eventual eradication of invasive species (Roy et al. 2014(Roy et al. , 2017Essl et al. 2015;Faulkner et al. 2017Faulkner et al. , 2020. This information can also be utilised to constrict pathways by reducing and limiting the means of entry and spread, intercepting movements at border points, and assessing risk of planned imports (Simberloff et al. 2013).
Horizon scanning of (IAS) is an approach that can be used to generate information on possible biological invasions, and so support planning and management at country and regional level, as well as inform policy and practice (Caffrey et al. 2014). It is the systematic search for potential biological invasions and an assessment of their potential socio-economic impacts and potential impacts on biodiversity, considering possible opportunities for mitigating the impacts (Sutherland et al. 2008(Sutherland et al. , 2010aRoy et al. 2014). The approach has been used at country level such as in Cyprus to determine non-native species that could become invasive (Peyton et al. 2019), Great Britain (Roy et al. 2014), Spain (Gassó et al. 2009;Bayón and Vilà 2019) and at the regional level such as the European Union (Roy et al. 2019), Central Europe (Weber and Gut 2004), Western Europe (Gallardo et al. 2016), and the United Kingdom (Sutherland et al. 2008). Building on the above background, a study was conducted in 2018 to utilize horizon scanning to identify and assess alien species that are not currently recorded as present in Kenya but could be introduced and become invasive in future threatening the economy through impacting on agriculture and biodiversity. The assessment covered arthropod pests, nematode, and pathogenic organisms (bacteria, fungi, oomycetes, phytoplasma, viroid, and viruses).

Materials and methods
The prioritisation was carried out by a panel of 28 Subject Matter Experts (SMEs) convened from research and academic institutions in Kenya. The SMEs had experience in the following areas: entomology, bacteriology, mycology, nematology, and virology. The SMEs were allocated to three thematic groups based on their expertise: Entomology (18), Nematology (2) and Pathology which included bacteriology, mycology, and virology (8). An adapted version of the consensus method developed for ranking IAS (Sutherland et al. 2011;Roy et al. 2014Roy et al. , 2019 was used to derive a ranked list of invertebrates (arthropods and nematodes) and pathogenic organisms (bacteria, fungi/chromista, viroid, and viruses) that are harmful to plants and could possibly enter Kenya in the future and become invasive. The approach involved the following steps: Step 1. Preliminary horizon scanning At the first meeting in June 2018, the SMEs made a preliminary selection of pests that had not yet been recorded as present in Kenya. This exercise was carried out using the premium version of the horizon scanning tool included in the CABI's Crop Protection Compendium (CPC) (CABI 2021a). In this tool, information from the CPC datasheets is used to generate a list of species that are absent from the selected 'area at risk' (Kenya) but present in specified source areas. Initially, the source areas were all countries within and outside Africa, which produced a list of over 1700 species. The SMEs reviewed the list by removing all organisms that were not arthropods, nematodes and pathogenic organisms (bacteria, fungi/ chromista, viruses and viroid); species of arthropods, nematodes and pathogenic organisms known to already occur in Kenya (even though not reported as present, so appearing as absent in the CPC); and species of arthropods, nematodes and pathogenic organisms unanimously considered as not important pests for plants in Kenya, e.g., species that are specific to a plant genus that does not occur in the country.
A list of 194 species (108 arthropods, 9 nematodes, 15 bacteria, 48 fungi/chromista, 1 viroid, and 13 viruses) was obtained. This was further refined by selecting only arthropods, nematodes and pathogenic organisms for which a full datasheet 3 was available in the CPC and the Invasive Species Compendium (ISC) (CABI 2021b) leaving 100 species (62 arthropods, 9 nematodes, 7 bacteria, 16 fungi/ chromista, 1 viroid, and 5 viruses). Species with a full datasheet in the CPC but not in the ISC were eliminated because without a full datasheet, there was not enough information to evaluate the species for characteristics of invasiveness; however, it also could indicate the species had not been invasive anywhere.
The SMEs from the National Plant Protection Organisation (NPPO), Kenya Plant Health Inspectorate Service (KEPHIS) added an additional 20 species not reported as present in Africa, but which were adjudged as potentially being of phytosanitary concern because they affect value chains key to the Kenyan economy. For instance, fruit flies such as Anastrepha ludens, A. obliqua, Bactrocera tryoni, and D. suzuki were included because of their effect on the fruit industry; the bacterial ring rot causal agent, Clavibacter sepedonicus and blackleg and soft rot pathogens, Dickeya dadantii, D. dianthicola, D. solani, D. zea and Pectobacterium atrosepticum because of their effect on the potato (Solanum tuberosum) value chain where zero tolerance in seed is emphasized. The 20 species included 8 arthropods, 8 bacteria, 3 fungi/chromista and 1 virus (indicated in Supplementary file S2). This resulted in a list of 120 species which included 70 arthropods, 9 nematodes, 15 bacteria, 19 fungi/chromista, 1 viroid, and 6 viruses which were scored as described below.
Step 2. Description of the scoring system The risk scoring system used was based on that described by Roy et al. (2019) but was modified during the first meeting so that there was an agreed and common understanding of the terminology used, applicable to the Kenyan context. Roy et al. (2019) assessed the likelihood of arrival, establishment, spread, and magnitude of potential negative impact on biodiversity and ecosystem services while in this assessment, the likelihood of entry (arrival), establishment, and potential magnitude of socio-economic impact and potential magnitude of impact on biodiversity were assessed. The likelihood of spread was not considered because once an alien species arrives ashore, exponential spread within and between countries in SSA has been observed assisted majorly by human-mediated activities especially if the criteria for entry and establishment are met and the key pathways are available (Mahuku et al. 2015;Kansiime et al. 2017;De Groote et al. 2020). A 5-score system for the four parameters (entry, establishment; socioeconomic and biodiversity impact) where 1 suggested unlikely to enter or establish; or minimal impact and 5 suggested very likely to enter or establish; major impact was used. A detailed scoring system version is provided in Supplementary file S1 but briefly outlined below.
To assesses likelihood of entry, 1 suggested absent from Africa and unlikely to be in the imported commodity; 2, absent from Africa but likely to be infrequently imported on a commodity; 3, present in Africa (not in neighbouring countries) and spreads slowly; or absent from Africa but: recently spreads very fast on several continents, or often associated to a commodity commonly imported, or frequently intercepted in Kenya; 4, present in Africa (not in neighbouring countries) and spreads fast, or in a neighbouring country and spreads slowly; and 5, present in a neighbouring country (Ethiopia, Somalia, South Sudan, Tanzania, and Uganda) and spreads fast. To assess the likely pathways of arrival, three likely pathways as defined by Hulme et al., (2008) were considered, unaided (UN) which comprised mainly natural dispersal; commodity (CO) which comprised intentional introduction as a commodity or with a commodity or unintentional introduction with a commodity (contaminant); and stowaway/hitchhiker (ST) which comprised vectors. Pathogenic organisms especially viruses and viroid which could be carried by vectors spreading through natural dispersal (UN), the stowaway pathway was considered although the commodity (CO) pathway was also considered if they were seed-borne, hence could come with the commodity as contaminants. The stowaway pathway was also considered for soil-borne pathogenic organisms which could unintentionally be introduced with soil.
To assess likelihood of establishment, (1) suggested Kenya is climatically unsuitable or host plants are not present; (2) only few areas in Kenya climatically suitable; or host plants rare; (3) large areas in Kenya climatically suitable and host plants rare; or only few areas in Kenya climatically suitable but host plants at least moderately abundant; (4) large areas in Kenya climatically suitable and host plants moderately abundant; and (5) large areas in Kenya climatically suitable and host plants very abundant. For the potential magnitude of socio-economic impact, (1) suggested the species does not attack plants that are cultivated or utilised; (2) the species damages plants that are only occasionally cultivated or utilised; (3) the species damages plants that are regularly cultivated or utilised but without threatening the cultivation, utilisation, or trade of this crop; (4) the species has the potential to threaten, at least locally, the cultivation of a plant that is regularly cultivated or utilised; or to regularly attack a crop that is key for the Kenyan economy without threatening the latter; and (5) the species has the potential to threaten, at least locally, a crop that is key for the Kenyan economy. For potential magnitude of impact on biodiversity, (1) suggested the species will not affect any native species; (2) the species will affect individuals of a native species without affecting its population level; (3) the species has the potential to lower the population levels of a native species; (4) the species has the potential to locally eradicate a native species or to affect populations of a protected or keystone species; and (5) the species has the potential to eradicate a native species or to locally eradicate a keystone species.
Step 3. Scoring of species After a group training at the first workshop, the scoring of species was done independently by all SMEs as assigned to respective thematic groups. Except for entomology and nematology thematic groups, SMEs in the pathology group assessed species based on their expertise. For instance, bacteria were assessed by bacteriologists; fungi and oomycetes by mycologists; and viruses and viroid by virologists. However, some SMEs were knowledgeable in all disciplines. Scores below three were considered low risk because of the low effect on likelihood of entry, establishment, economic as well as biodiversity damage; scores of three were considered moderate while scores above 3 (4 and 5) presented a high risk because they had an opposite effect to the low scores. For each species, confidence was estimated for each score for, the likelihood of entry; establishment; potential magnitude of socio-economic impact; and potential impact on biodiversity; likely pathway of arrival; and for the overall score following Blackburn et al. (2014). The rating proposed by Blackburn et al. (2014) was originally modified from the EPPO pest risk assessment decision support scheme (OEPP/EPPO 2012). The information to support the scores and confidences and the likely pathways was obtained from compendia (CPC and ISC), published (journal articles and reviews), and grey literature (conference papers and proceedings; dissertations and theses; government documents and reports and newspaper articles). The SMEs also relied on their existing knowledge for assessing the species. The likely pathway of arrival and associated confidence levels were used to help focus discussions on the possibility of entry and establishment but did not contribute to the overall score. The assessments were compiled and returned to all assessors before the consensus workshop.
The overall score was obtained by the following formula: Step 4. Consensus workshop In December 2018, a consensus workshop was held for all assessors (SMEs). The assessments for each thematic group were reviewed one by one, and any discrepancies between scores were discussed among the assessors. The assessors had the opportunity to modify their scores according to the opinions of the other SMEs, but reaching consensus was not imperative. The overall risk score for a species was calculated using the median score for the four parameters, rather than the mean. This overall score was validated by the assessors in each thematic group through consensus, and in cases of disagreement, the individual scores, and the evidence on which they were based were re-discussed. The assessors who could not attend the second workshop were able to comment on the scores by email. The overall score was used Likelihood of entry × likelihood of establishment × (magnitude of socio -economic impact + magnitude of impact on biodiversity).
to rank species according to their potential threat for Kenya.
Step 5. Finalising the ranked list After the workshop, discussions were carried out among the SMEs to suggest the next course of action for the highest ranked species. A high ranked species was any species with a minimum score of 54 (according to the formula used) and above, because such a species scored a three for all the assessable attributes or more than a three in at least three or more attributes. This score of three suggested the situation turned more towards the possibility of entry, establishment, and higher impact (social-economic or biodiversity). However, only the top 20 species were selected for follow-up action as not all could be addressed in case the NPPO (KEPHIS) decided to consider preventive action due to limited resources. For arthropods, 21 species were selected because the 20th and 21st species tied on score (90) while all nematodes were selected because they were few yet the least had a risk score of 54. The suggested actions included considering whether a full pest risk analysis was merited; whether any changes to or additional preventive measures might be required; possible establishment of surveillance and early warning activities; and development of contingency plans to allow a rapid and effective response if a particular species was detected. The type of action suggested for a particular species was based on the importance of the value chain and the status of the alien species in Africa and neighbouring countries. Species for which immediate action was not suggested (not in the top 20 prioritised species), a plant health risk register (Baker et al. 2014) was recommended to which they were added. This will allow regular monitoring through reassessment to determine possible change in their risk status. The presence/absence in Kenya was again checked for all species in June 2021 by searching published documents.

Results
The full results of the assessments are provided in the Supplementary file S2 while the 21 arthropods, 9 nematodes and 20 pathogenic species prioritised with follow-up actions are presented in Tables 1, 2 and 3 respectively. All assessments are presented as agreed in December 2018, but in the discussion, we note where risks are likely to have changed since then, for example due to recent spread making introduction more likely.

Arthropods
In total, 70 arthropod species were assessed which included 69 insects and 1 mite. Six of the species had not been recorded in Africa at the time of the assessment while 13 species were recorded in neighbouring countries (Ethiopia, Somalia, Sudan, Tanzania, and Uganda). The overall risk scores ranged from 6 to 160 with 11 of the 70 species that scored a minimum of 54 and above (high scorers) had already been reported in neighbouring countries (Supplementary file S2). Only one mite, Brevipalpus lewisi, had an overall score of 60. The prioritised species (21) for followup actions included eight Hemiptera, four Coleoptera, three Diptera, three Lepidoptera, two Hymenoptera, and one Thysanoptera. The top five species included two species of the whitefly Bemisia tabaci complex, MEAM1 and MED, the peach fruit fly Bactrocera zonata, the yellow crazy ant Anoplolepis gracilipes, and the Southern armyworm Spodoptera eridania. Most of the arthropods (89%) were adjudged likely to arrive as contaminants on commodities, i.e., on their host plants, 16% were likely to arrive as stowaways (also referred to as hitchhikers), while 6% are good fliers so likely to enter unaided (Supplementary file S2). Because some species could enter the country through multiple pathways, that's why the percentages do not add up to 100. Suggestions for the most needed actions for the 21 prioritised species included mostly full PRAs and surveys or surveillance for their presence or introduction in the country in particular, species that have been reported in neighbouring countries (Table 1). It is recommended that the rest of the species (not prioritised) are added to the plant health risk register to regularly assess change in risk.

Nematodes
Nine nematodes were assessed with least risk scores of 54 obtained for Heterodera goettingiana and a maximum score of 90 obtained for the groundnut testa nematode, Aphelenchoides arachidis (Table 2 and Supplementary file S2). All assessed species     (Talwana et al. 2008;Lesufi et al. 2015). All are most likely to arrive with their host plants as contaminants (seed-borne) although A. arachidis could arrive from across the border in Uganda in soil (stowaway). Since all the nine assessed species received risk scores of 54 (minimum) and above, suggestions for the next steps have been made which include conducting a full PRA and surveillance (Table 2). This exception was made because the assessed nematodes were very few and all met the minimum requirement (risk score of 54 and above) for action. and Xanthomonas fragariae. They all scored more than the 54 minimum risk score with all also scoring more than the 3 minimum score for likelihood of entry and establishment indicating possible entry and establishment in Kenya (Table 3 and Supplementary file S2). The majority (85%) of pathogens were likely to arrive as contaminants on commodities, especially as seed-borne pathogens, and/or as stowaways (68%) if the pathogen could be carried in soil (soil-borne) or by a vector (virus and viroid). One of the viruses assessed that could arrive either in seed (contaminant) or in soil (stowaway) was PMTV (Calvert and Harrison 1966;Jones and Harrison 1969;Latvala-Kilby et al. 2009). Unlike many viruses that are vectored by insects, PMTV is vectored by the fungus, Spongospora subterranea f.sp. subterranea (Jones and Harrison 1969;Kirk 2008), the causal agent of powdery scab in S. tuberosum (Harrison et al. 1997;Merz and Falloon 2009), and having a wide global distribution (Merz 2008). The first course of action suggested for the prioritised pathogenic species was to survey for potential presence in Kenya, particularly those that have been reported in neighbouring countries while for others especially those known to affect value chains prioritised by Kenya (MoALF 2019), full PRAs are proposed (Table 2). It is recommended that the rest of the species (not prioritised) are added to the plant health risk register to regularly assess change in risk.

Discussion
Horizon scanning, a method used in prioritization of IAS (Roy et al. 2014(Roy et al. , 2017(Roy et al. , 2019Bayón and Vilà 2019) was applied in this study to identify plant pests that could be introduced in Kenya, become invasive and cause enormous socio-economic and biodiversity damage. The species were ranked according to their potential threat and appropriate actions suggested for some prioritised species. Since the exercise was conducted in 2018, four pest species have since been found to be present in Kenya. This demonstrates that for some of the assessed species and probably some that were eliminated because they were presumed absent in Africa and in Kenya, could have been present in Kenya at the time of assessment. This was caused by a gap in reporting which is mainly attributed to lack of or limited resources to update pest lists, organise horizon scanning to prioritise likely pest incursions, and conduct regular surveillances on prioritised pests (Kansiime et al. 2017). This results in most countries in SSA reacting to pest incursions rather than proactively stopping them. The four pest species include two arthropods (red gum lerp psyllid and the spotted wing drosophila) and two pathogenic organisms (the bacterial species, P. parmentieri and the viroid, potato spindle tuber viroid) (Kwadha et al. 2021). The first arthropod, red gum lerp psyllid (Glycaspis brimblecombei) was found in Kenya in 2018 but reported in 2020 (Wondafrash et al. 2020) indicating it was already present at the time of the assessment. Its risk was scored as high judging by the likelihood scores for entry, establishment, and economic impact (Supplementary file S2). This is because it had recently invaded eucalyptus production areas in different parts of the world, including several African countries. In all these areas, it quickly became a pest of commercially important eucalyptus species, such as Eucalyptus camaldulensis. It remains to be seen how damaging the species becomes in Kenya, as in several regions, the intentional or accidental introduction of its specific parasitoid, Psyllaephagus bliteus entirely or partially controlled the pest (Caleca et al. 2018). If the economic impact in Kenya is significant, biological control should be considered. The second arthropod, D. suzukii is a fruit pest of Asian origin that had only been reported in Morocco and Réunion in Africa at the time the assessment despite its rapid worldwide spread in recent years hence the moderate overall risk score of around 60. It is mainly a pest of temperate climates that does not cope well with high temperatures (Ørsted and Ørsted 2019). Its preferred hosts include strawberry (Fragaria ananassa), blackberry (Rubus fruticosus), blueberry (Vaccinium corymbosum), and raspberry (Rubus idaeus) (Garcia 2020). These hosts are not yet major crops in Kenya hence the likelihood of establishment and potential economic consequences were both scored as moderate (3). Kwadha et al. (2021) identified D. suzukii on R. idaeus, F. ananassa, V. corymbosum and Punica grantum (pomegranates) at one farm in Nakuru county but not the other five counties were the sampling was also done. Although Kwadha et al. (2021) did not indicate the likely pathway, the pest could probably have been introduced through commodities. Berry (2020) indicated that D. suzukii lays its eggs and feeds internally as immatures within fruits making spread through fruits the most important pathway. The second but unlikely pathway could be soil where pupation occurs (Berry 2020). The main hosts which are also indicated above are becoming increasingly grown in the Kenyan highlands raising the likely potential economic impact (Kwadha et al. 2021). Therefore, possibilities for containment and/or management measures such as biological control should therefore be assessed (Kwadha et al. 2021;Seehausen et al. 2021).
One of the pathogenic organisms, the bacterial species P. parmentieri originally belonged to the soft rot Enterobacteriaceae (Ma et al. 2007) but has since been transferred to soft rot Pectobacteriaceae (Adeolu et al. 2016;van der Wolf et al. 2021). Pectobacterium wasabiae, a species originally transferred to P. parmentieri (Khayi et al. 2016) was confirmed in Kenya through surveys conducted in 2016/2017 (Kamau et al. 2019). Pectobacterium parmentieri was scored as having a relatively low likelihood of entry because it was only reported in South Africa at the time, a country with less trade in S. tuberosum with Kenya offering limited or no pathways (contaminants) however, a high likelihood of establishing was suggested resulting in significant economic damage because other Pectobacterium species have been reported in Kenya (Kamau et al. 2019). Similarly, PSTVd, the second pathogenic organism judged absent in Africa and Kenya at the time of assessment, has since been reported in Ghana and Kenya in Solanaceous uncultivated species (S. anguivi, S. anomalum, S. cerasiferum, S. coagulans, S. dasyphyllum, S. incanum, S. macrocarpon and S. virginianum) and in Kenya in tree tomato (S. betaceum) (Skelton et al. 2019;Kinoga et al. 2021). Although the viroid also naturally infects pepper and chili (Capsicum annuum), pepino (S. muricatum), eggplant (S. melongena), tomato (Lycopersicon esculentum), and S. tuberosum (Mackie et al. 2002;Verhoeven et al. 2004Verhoeven et al. , 2010, no report of its presence in any of the aforementioned crops is available although this does not necessary confirm absence as lessons have demonstrated (Wondafrash et al. 2020;Kwadha et al. 2021).
Potato spindle tuber viroid is designated a quarantine pest in many countries therefore, detection surveys are suggested to confirm presence in key export crops especially chilli and flowers as well as food security crops especially S. tuberosum, which is widely grown by smallholder farmers in Kenya, yet this viroid is known to cause major loses in yield and reduce tuber quality (Owens 2007;Kochetov et al. 2021). The viroid is known to be transmitted through contact (Verhoeven et al. 2010) indicating it may be moved between plant species through crop handling which suggests probable presence in C. annuum, S. muricatum, S. melongena, L. esculentum, and S. tuberosum since S. betaceum is also widely grown in the similar areas in Kenya like the mentioned crops (Waswa et al. 2020;MoALF 2021;Kinoga et al. 2021). Although studies demonstrate the viroid can also be transmitted by the aphid species, M. persicae, this is only possible if the source plant is infected with both the viroid and potato leafroll virus (PLRV) (Syller et al. 1997;Syller 2001). The capsid protein of PLRV encapsulates the viroid resulting in transmission by M. persicae. However, PLRV and M. persicae are not known to infect any of the hosts in which PSTVd was reported in Kenya leaving transmission between species through crop handling the main possible mechanism; within species through true seed if it is present in the mentioned crops (Verhoeven et al. 2010) or by aphids if present in S. tuberosum (Syller et al. 1997;Syller 2001).
The above examples suggest there may be other species present that have not yet been detected. Thus, one of the follow-up actions (Tables 1, 2 and 3) for prioritised species, particularly those recorded as present in neighbouring countries, is detection surveys.
Examples include B. tabaci (MED species) (Misaka et al. 2020), A. gracilipes (Löhr 1992), Bruchus pisorum (Mendesil et al. 2016), Pseudaulacaspis pentagona (CABI 2021b), Trogoderma granarium (EPPO 1981); the nematode A. arachidis (Talwana et al. 2008;Lesufi et al. 2015); and the pathogenic species such as C. fimbriata (Rouxa et al. 2001), X. citri (Balestra et al. 2008;Derso et al. 2009;Ference et al. 2018), P. atrosepticum (CABI 2021b), and Faba bean necrotic yellows virus (Abraham et al. 2000). Other species were adjudged as having a moderate likelihood of entry but had a high overall score due to the high scores obtained for likelihood of establishment and magnitude of socio-economic impact. Kenya has prioritized a number of value chains under the Agricultural Sector Transformation and Growth Strategy, 2019-2029 as key to improving livelihoods and supporting economic growth (MoALF 2019). Therefore, species which could affect the prioritised value chains are appropriate targets for conducting a full PRA. The outcome of the PRA will advise implementation of import controls and the preparation of contingency plans. Species in this category included Thrips palmi (melon thrips), Cacoecimorpha pronubana, Euwallacea perbrevis (tea shot-hole borer), and Peronospora sparsa (cause of downy mildew of roses) among others.
A further group of species scored low on likelihood of entry, but high on likelihood of establishment and socioeconomic impact, so they were not near the top of the overall ranking. Several of the species in this group are potential pests of S. tuberosum, including D. solani, D. dadantii, D. dianthicola, C. sepedonicus, Phoma exigua f.sp. foveata, and Synchytrium endobioticum). Solanum tuberosum is an important crop in Kenya which has recently suffered from new pests such as PCN (Mwangi et al. 2015;Mburu et al. 2018Mburu et al. , 2020. For several of these species, adjusting the score for likelihood of entry upwards by 1 would have put the final overall score in the top 20 prioritised pests. This suggests that follow-up activity should not necessarily be limited to the highestranking or prioritised species. Secondly, the scores need to be regularly reviewed, thus implementing a plant health risk register such as the one implemented by the United Kingdom (Baker et al. 2014) to which all assessed species and also those identified through interceptions at border points should be considered.
The methodology for horizon scanning described here is an adaption of previously used methods (Sutherland et al. 2011;Roy et al. 2014Roy et al. , 2019, adjusted for the Kenyan context. Such pest prioritisation schemes have emerged particularly in the last decade to support plant health decision making by risk managers and policy makers in prioritising the large number of potential invasive species (MacLeod and Lloyd 2020). Most of these prioritisation or risk ranking systems have been deployed in high income countries, so the use of this approach in Kenya was novel. The approach adopted worked well, bringing together SMEs from a range of organisations. They achieved consensus on modifying the criteria for the scores, as well as in the final scoring when there were discrepancies between the experts' scores. As invasive plants were excluded from the exercise, further work would be required to adjust the definitions to cater for invasive plants. The area in which the SMEs felt least comfortable was in scoring of potential impact on biodiversity. Few species scored highly on impact on biodiversity; three arthropods and two pathogens had a risk score of 4. This may be because socioeconomic impacts of invasive arthropods and pathogens are generally better known than their impact on biodiversity, and confidence in the impact on biodiversity score was often low. Invasive ants are an exception because they are considered to have serious effects on biodiversity worldwide (Mikissa et al. 2013;Mothapo and Wossler 2017;Mbenoun Masse et al. 2017). Two ants in our assessment had a high score for their potential impact on biodiversity, A. gracilipes and Linepithema humile. Many non-native herbivores and plant pathogens are a serious global threat to native biodiversity and ecosystems and in some areas, they are already a threat (Kenis et al. 2009;Ghelardini et al. 2017).
It was noted that a change of 1 point in a score could move a pest many places up the list of priorities, and pest prioritisation schemes are not without shortcomings (MacLeod and Lloyd 2020). Rather than being a one-off activity, the results should be reviewed regularly, particularly in the light of any new information that might arise. Thus, part of the outcome of a horizon scanning process such as this could be systematic monitoring of information sources to detect possible changes to risk, which can be recorded in a plant health risk register. Kenya does not have a plant health risk register or list of prioritised pests for prevention, and thus the horizon scanning process adopted here could provide the basis for such a register. Given the practicality of the approach and the widespread lack of pest prioritisation in SSA, we propose that the approach reported here could benefit many other countries on the continent if adopted. It could also be implemented at a sub-regional level, such as the East African Community (Eastern Africa), Southern African Development Community (Southern Africa) or Economic Community of West African States (Western Africa), and possibly by the African Union, for which a new Plant Health strategy is currently under development. Subregional-or regional-based assessments are indeed more important than country-based assessments because lessons have demonstrated that once an invasive pest lands on the African continent or subregion, control of spread across countries is virtually impossible due to weak or non-existent border biosecurity and porous borders.
Authors' contributions All authors participated in the assessments of likely IAS. Joseph Mulema, Roger Day and Marc Kenis put together the initial drafts of the manuscript while all other authors read, contributed further edits and approved the final manuscript.
Funding This work was financially supported by the Foreign, Commonwealth and Development Office (FCDO), UK, the Directorate-General for International Cooperation (DGIS), Netherlands, the European Commission Directorate-General for International Cooperation and Development (DEVCO) and the Swiss Agency for Development and Cooperation (SDC) through CABI's Action on Invasives and PlantwisePlus Programmes. CABI is an international intergovernmental organisation, and we gratefully acknowledge the core financial support from our member countries and lead agencies. See https:// www. cabi. org/ about cabi/ who-we-work-with/ key-donors for details.
Availability of data and material All additional relevant data and materials have been submitted as supplementary materials.
Code availability Not applicable.

Conflicts of interest None.
Ethics approval Not applicable.

Consent for publication Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.