Biological invasions in Singapore and Southeast Asia: data gaps fail to mask potentially massive economic costs

The impacts of invasive alien species are well-known and are categorised as a leading contributor to biodiversity loss globally. However, relatively little is known about the monetary costs incurred from invasions on national economies, hampering management responses. In this study, we used published data to describe the economic cost of invasions in Southeast Asia, with a focus on Singapore – a biodiversityrich, tropical island city state with small size, high human density and high trade volume, three factors likely to increase invasions. In this country, as well as in others in Southeast Asia, cost data were scarce, * These two authors contributed equally NeoBiota 67: 131–152 (2021) doi: 10.3897/neobiota.67.64560 https://neobiota.pensoft.net Copyright Phillip J. Haubrock et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. RESEARCH ARTICLE Advancing research on alien species and biological invasions A peer-reviewed open-access journal


Introduction
Biological invasions cause significant ecological impacts around the world, posing profound impediments to conservation efforts and potentially driving marked socioeconomic costs (Hulme et al. 2009;Early et al. 2016;Seebens et al. 2017). Invasive species are amongst the main drivers of biodiversity loss worldwide (Malcolm and Markham 2000;Stigall 2010;Bellard et al. 2016;Haubrock et al. 2021). In a socioeconomic context, invasions can directly affect human health, damage goods and services, compromise public and social welfare and impact agriculture (Bradshaw et al. 2016;Paini et al. 2016;Jones 2017;Shackleton et al. 2019). Yet, disproportionately few economic resources are allocated to remediate the large-scale consequences of such invasions in different parts of the world (Hulme et al. 2009;Scaler, 2010;Early et al. 2016). One of the reasons underlying this discrepancy is undoubtedly related to the limited knowledge and societal awareness of their actual impacts (Courchamp et al. 2017).
Whilst the ecological impacts of invasive species are well-described (see, for example, Gurevitch and Padilla 2004;Didham et al. 2005;Cuthbert et al. 2019aCuthbert et al. , 2020Mofu et al. 2019), relatively few studies have synthesised monetary aspects associated with biological invasions (but see Pimentel et al. 2005 for the USA; Kettunen et al. 2009 for Europe; Oreska and Aldridge 2011 for the UK; Gren et al. 2009 for Sweden;Hoffmann and Broadhurst 2016 for Australia;Xu et al. 2006 for China). Yet, highlighting the economic costs of invasions can actually represent a key awarenessbuilding tool for both the general public and authorities, as well as an efficient way for motivating policies, guiding decision-making and prioritising management actions towards invasive species (Dana et al. 2014;McConnachie et al. 2016;Hiatt et al. 2019;Diagne et al. 2020a). Such economic costs might relate to a large variety of impacts, through damage directly or indirectly driven by invaders (e.g. Shwiff et al. 2010), to different types of expenditure dedicated to preventing, controlling or eradicating invasions (e.g. Hoffmann and Broadhurst 2016). Nonetheless, the scarcely reported economic costs are spatially, temporally and taxonomically fragmented (Diagne et al. 2020a), leading to a lack of a holistic understanding of the monetary aspects of invasions. This represents a major challenge for decision-making as invasions represent an ever-increasing trans-boundary socio-ecological challenge (Lovell et al. 2006;Marbuah et al. 2014;Diagne et al. 2020a). Particularly, while regional estimates have highlighted the diversity of costs (e.g. Pimentel et al. 2000Pimentel et al. , 2005Kettunen et al. 2009;Nghiem et al. 2013), limited spatial resolution has resulted in piecemeal financial commitments to tackle the growing economic problem of invasions at relevant scales. More detailed and comparable information on specific costs is urgently needed at the governmentlevel, where budgets are established and managed.
As an international travel and trade hub with numerous introduction pathways, Singapore is a country facing high risk of biological invasions (Yeo and Chia 2010;Seebens et al. 2013;Wong 2018) and may thus be a particularly useful example for such nationally-scaled cost estimation. Thus, Singapore is outstanding amongst other Southeast Asian countries due to its very dynamic economic connectivity, despite a relatively small surface area. Singapore is a highly urbanised and densely populated, but biodiverse, tropical island city state, centrally located within Southeast Asia Ng et al. 2011;World Bank 2019). The few publications reporting costs of invasive species in Singapore have suggested they might be important (Nghiem et al. 2013), yet costs have lacked synthesis. At least 142 non-native animal species have been reported in Singapore (Yeo and Chia 2010), including species listed on several 'worst invasive alien species' lists (e.g. IUCN).
Recently, the available literature on economic costs of invasive species globally was compiled in the InvaCost database (Diagne et al. 2020b) with the aim of generating the means to fill knowledge gaps on invasion costs worldwide. Using data available from this database, we synthesised and described the available information on economic costs of invasions in Southeast Asia, focusing on Singapore in particular.
We specifically investigated (a) how recorded costs and species are characterised across Southeast Asian countries and (b) Singapore as a more detailed example or case study to describe recorded costs impacting its economy, according to (i) taxa, (ii) cost types and (iii) activity sectors. We also deciphered whether the level of reliability of estimates may impact the financial burden of invaders. Furthermore, we extrapolated additional costs for invaders reported in Singapore, but with unknown costs there. Finally, we correlated invasion costs with importation levels, surface area and population size amongst countries to assess the specificities of Southeast Asian countries. We hypothesised that the costs of invasive species in Singapore are underestimated and yet substantial, as are probably those of other Southeast Asian countries.

Data acquisition
Information on the economic cost of invasions in all the Southeast Asian countries (Brunei, Cambodia, East Timor, Indonesia, Laos, Malaysia, Myanmar, Philippine, Singapore, Thailand and Vietnam) was extracted from the InvaCost database (Diagne et al. 2020b;Angulo et al. 2021) concerning the global costs of invasive species, based on published literature, enabling comprehensive quantification of costs associated with invasive species at various spatio-temporal scales. The latest version of the database, as well as a summary of the whole procedure used to build and update it, can be directly accessed at https://doi.org/10.6084/m9.figshare.12668570. Briefly, the data in Inva-Cost were collected following (i) a series of literature searches using the Web of Science platform (https://webofknowledge.com/), Google Scholar database (https://scholar. google.com/) and the Google search engine (https://www.google.com/) and (ii) targeted searches through contacting experts and stakeholders to request potentially unpublished and/or publicly unavailable documents containing cost information. All the retrieved costs were standardised in an up-to-date currency (2017 USD), while also taking into account an inflation factor (Diagne et al. 2020b). We performed descriptive analyses of a subset of this database, by filtering data ('Official_country' column) to exclusively ascertain invasion costs in each country.

Cost calculation and description
We considered the total costs of invasions by amalgamating the recorded raw costs (column 'Cost_estimate_per_year_2017_USD_exchange_rate') per year from our subset. Due to the variability of temporal scales of cost estimates in InvaCost, we annualised the data, based on the difference between the "Probable_starting_year_adjusted" and "Probable_starting_year_adjusted" columns using the "summarizeCosts" function of the 'invacost' package (v.0.3-4) in R (v.4.0.2) (Leroy et al. 2020). Each expanded entry thus corresponded to a single year for which costs were available following this expansion process (i.e. costs spanning multiple years were divided amongst those same years). The resulting costs attributed to recorded species were examined according to different descriptive fields of the database (an updated description of these descriptive fields is openly available at https://doi.org/10.6084/m9.figshare.12668570): i. Method_reliability: illustrating the perceived reliability of cost estimates, based on the type of publication and method of estimation. Estimates in peer-reviewed publications or official reports or with documented, repeatable and/or traceable methods were designated as "High reliability"; all other estimates were designated as "Low reliability" (Diagne et al. 2020b); ii. Implementation: referring to whether the cost estimate was actually realised in the invaded habitat ("Observed") or whether it was only predicted to occur ("Potential"); iii. Type_of_cost_merged: grouping of costs according to the categories: (a) "Damage-Loss" referring to damages or losses incurred by invasion (e.g. costs for damage repair, resource losses, medical care), (b) "Management" comprising control-related expenditure (for example monitoring, prevention, management, eradication) and money spent on education and maintenance costs, (c) "Diverse/Unspecified" including mixed damage-loss and management costs (cases where reported costs were not clearly distinguished amongst cost types); iv. Impacted_sector: the activity, societal or market sector that was impacted by the cost (Suppl. material 2); note that individual cost entries not allocated to a single sector were classified under "Mixed" in the "Impacted_sector" column. A detailed summary of all descriptors can be found in Suppl. material 1 (see also Diagne et al. 2020b) and the final dataset in Suppl. material 2.

Temporal dynamics and cost extrapolations
To investigate the temporal dynamics of invasion costs, we used the "summarizeCosts" function implemented in the R package 'invacost' (Leroy et al. 2020). With this method, we calculated the observed cumulative and average annual costs covering the period for which costs were recorded, displaying the changes in invasion costs over time.
As cost information for invasive alien species in Singapore, which we used as an example, was limited (three species; see Results for more details), we also extrapolated potential costs for a few additional known invasive species present in Singapore, but which had recorded costs outside Singapore. For this, we used the most recent comprehensive list of alien animal species in Singapore (n = 142; Yeo and Chia 2010). With this information, we first estimated the mean annual cost of the species listed in Yeo and Chia (2010) outside Singapore (at the "country" scale) that was available in InvaCost, assuming the InvaCost database contained recorded cost information for Singapore over the same period . We then applied a correction factor that considers the cost difference between the average costs of all invasive alien species in Singapore and the average costs of all invasive alien species outside Singapore (excluding extreme val-ues, i.e. the upper and lower 12.5% when implementing the correction factor to cost data). The corrected mean cost of each of these species was then summed to obtain an additional cost of biological invasions not directly available from records in Singapore.

Southeast Asia and national comparisons
Given Singapore is an economic centre, we compared the available cost information of Singapore -in terms of cost entries and number of recorded species ) -to other available information on invasive alien species costs in Southeast Asian countries recorded in InvaCost (via the aforementioned data processing methods). Furthermore, we compared invasion cost entries with other countries worldwide using a linear regression, based on import value (collected from the International Trade Centre (https:// www.trademap.org/tradestat/Country_SelProduct_TS.aspx) to (i) see how the lack of available data can affect the estimated economic costs and (ii) examine the relationship between trade volume and economic activities with the cost recording of invasive species. We focused on the 50 countries ranking highest in import value, but with recorded data in InvaCost. Further, we collected the data of species that have been introduced in all countries in Southeast Asia (see Results for more details) from the Global Alien Species First Records Database (Seebens et al. 2018; accessed in June 2020).
Finally, we examined the relationships between invasion costs (observed and high reliability costs only) and (i) land area and (ii) human population size using linear regressions (log-transformed) and examined how Singapore compared to other countries globally and in Southeast Asia particularly. Land area and population size per country were obtained using 2020 data from worldometer (https://www.worldometers.info/ world-population/population-by-country/).

Costs across taxa, types and sectors in Singapore
Cost data originated from seven records from six different published sources (n = 34 expanded entries). The recorded costs were found to have occurred after 1975 and amounted to US$ 1.720 billion in total (Figure 1).
At the taxonomic level, cost estimates were available for species from two families, Culicidae (n = 6 estimates) and Corvidae (n = 1). Within Culicidae, Aedes spp. drove all of the recorded costs, with four records attributed to A. aegypti alone and two as a combination of A. aegypti and A. albopictus. Although A. albopictus is native to Singapore, it was not possible to separate joint cost estimates, which accounted for < 0.05% of total Culicidae costs. For Corvidae, the single cost estimate was associated with Corvus splendens.
The overall estimated cost was mainly caused by Aedes spp. with a total of US$ 1.72 billion split between damage-losses (US$ 1.14 billion) and management costs (US$ 578.01 million). For C. splendens, the single cost estimate reached US$ 765.24 thousand and concerned costs attributed to control-related management efforts (Figure 2a). With respect to the impacted sector, all Aedes spp. costs were associated with a combination of impacts on authorities-stakeholders, health and public and social welfare. The single recorded costs for C. splendens impacted authorities-stakeholders ( Figure 2b). The reported economic costs were associated with terrestrial systems alone and, thus, no costs were documented in aquatic invasions.
From a methodological point of view, all reported costs were classified as "Observed", i.e. considered as actually occurring and not based on predictions or extrapolations from outside the invaded area. Every documented Aedes spp. cost was obtained from accessible peer-reviewed literature and thus deemed "High reliability". Conversely, the single cost estimate of C. splendens was deemed to be of "Low reliability" (Figure 2c). Accordingly, more than 99.9% of costs were deemed "High reliability".
Comparing the costs of recorded species in Singapore with their average annual costs per country outside of Singapore, after excluding extreme values (removing 25% extreme values, i.e. the top and bottom 12.5%), costs and expenditure in Singapore were around three times lower than those in the rest of the world. From the 142 species recorded in Yeo and Chia (2010), only an additional 11 were recorded in the InvaCost database (Suppl. material 4). Applying the average annual monetary cost discrepancy as a correcting factor to the average annual costs of the 11 invasive species, using the InvaCost data from outside Singapore, resulted in an additional projected annual average cost of US$ 22.33 million per year and a total of US$ 893.13 million additional costs considering the period 1975-2015.

Southeast Asia and national comparisons
The monetary impact of invasions recorded in Southeast Asia totalled US$ 16.89 billion between 1960 and 2020. Amongst these, Singapore ranked fifth relative to other countries in terms of reported costs, with two recorded invasive alien species and seven recorded cost entries in InvaCost. Notably, Brunei had the lowest number of recorded entries (1), species (1) and costs (US$ 6.7 million), while Thailand had the highest costs (US$ 5.2 billion) and most recorded entries (13) according to InvaCost (Table 1), suggesting considerable spatial heterogeneity in the region (Figure 3). In countries where lists of known invasive alien species were available , all had reported costs for 10% or less of known invasive alien species, with Singapore having the lowest proportion of aliens with costs (< 3%). We further identified a significant correlation between trade volume and the number of recorded entries in InvaCost (Suppl. material 5). When the number of records from Singapore is related to the volume of trade imports (Figure 4), which has been shown to be strongly related to cost entries (Haubrock et al. 2021b;Kourantidou et al. 2021), the relationship highlights a number of entries 40 times lower than expected. The under-reporting of cost entries in Singapore was considerably more apparent than other high-ranking Southeast Asian countries (i.e. amongst top 50 globally in terms of imports), with Thailand, Vietnam, Malaysia and Indonesia also having fewer records than expected based on imports, but the Philippines having more cost records than expected (Figure 4).
Considering all countries, invasion costs related significantly positively to both land area and population size (Supplement 5). When compared to other countries with costs, Singapore displayed considerably greater costs relative to those variables, even relative to other Southeast Asian nations which mostly clustered together (Figures  . Relationship between the import value and the number of records in InvaCost, focusing on the 50 countries ranking highest in both GDP and import values, but with recorded data in InvaCost. Note that all variables are displayed on a ln-scale. Singapore shows a large deficit of records related to expectations from its import value. 5a, b). Indonesia, Myanmar and Vietnam (and Laos in the case of surface area) had lower invasion costs than expected, based on surface area and human population.

Discussion
The recorded invasion costs in Singapore over the past 40 years have reached US$ 1.72 billion in total which represents about ⁵⁄ ₆ of the Ministry of the Environment and Water Resources (S$ 2.83 billion; US$ 2.12 billion), ⅔ of the Ministry of Trade and Industry (S$ 3.68 billion; US$ 2.76 billion) or more than ⅓ of the Ministry of National Developments (S$ 4.8 billion; US$ 3.67 billion) annual budgets in 2017 (https://www.singaporebudget. gov.sg). Despite these costs being high, our study shows that the available entries in the da-tabase were highly fragmentary, with the majority of documented alien animal species in Singapore being absent from the cost estimation (Yeo and Chia 2010). This further puts into perspective overall costs that are already surprisingly high for such a small area, especially when actual costs are expected to be more numerous and thus overall higher than the few recorded costs. Indeed, we show not only that Singapore has few cost entries, but also that it has about 40 times fewer than expected from its trade volume. Contrastingly, comparisons, based on costs relative to land area and human population size, evidenced considerably higher costs in Singapore pro rata, based on those variables, with costs comparable to countries approximately 600-times larger and 10-times more populous. These trends were even more marked when compared to relationships amongst other Southeast Asian countries, which were more in line with the global cost pattern.
The very few recorded costs were linked principally to the human health sector and mainly driven by mosquitoes, with large incurred costs listed for healthcare and their control. This is mostly related to costs arising from limiting the risk of infectious human diseases, such as Zika, dengue or chikungunya, which are caused by pathogens, vectored principally by A. aegypti and A. albopictus, as well as losses through direct healthcare costs (Beltrame et al. 2007;Zammarchi et al. 2015). Indeed, mosquitoes are considered as a severe problem in Singapore, underlined by the considerable costs on control and the medical field (Carrasco et al. 2011). These total costs relating to human health in Singapore are significant, considering previous estimation of annual costs on human health and environment in the entirety of Southeast Asia (US$ 1.85 billion; US$ 1.4-2.5 billion per year) estimated by Nghiem et al. (2013). Moreover, our extrapolations for species known to be present in Singapore, but with no reported costs there, indicated further economic impacts summing to US$ 893.13 million over 1975US$ 893.13 million over -2014 Although this figure has to be taken with caution, it underlines the magnitude of potentially occurring costs which are not accounted for in published literature. These numbers are still likely underestimated , given that these additional costs stem from just 11 of the 142 known animal invaders in Singapore that were available in InvaCost, with plant species missing entirely. Indeed, information on plant invasions in Singapore and, particularly, with regard to their monetary impacts, are scarce (Meyer 2000), with Yeo and Chia (2010) listing only relatively few invasive examples, such as the water hyacinth Eichhornia crassipes, which entered Singapore's waterways and proliferated to a damaging extent. As such, most invaders lack cost information at the Singapore scale, yet also internationally. Nevertheless, this lack of information, although striking, is neither surprising nor different from what is found in similar studies elsewhere. First, we showed that this is a general pattern in the region, with Singapore amongst the countries with most cost entries in Southeast Asia. Second, national or regional studies on the economic costs of biological invasions outside this region also consistently reported only between 2% and 10% of invasive alien species having recorded costs, for example, Argentina (Duboscq-Carra et al. 20201), Asia , Australia (Bradshaw et al. 2021), France , Germany (Haubrock et al. 2021c), Mexico (Rico-Sánchez et al. 2021 and United Kingdom (Cuthbert et al. 2021a).
In the context of Southeast Asia, this national bias is even more pronounced; amidst differences in economic activities amongst countries (note that Singapore has the highest GDP per capita in Southeast Asia), the lack of cost information for invasive alien species more broadly across Southeast Asian countries is striking. Singapore had the lowest proportion of known invasive alien species with reported costs, while all Southeast Asian countries had costs for 10% or below in terms of listed invasive alien species. This is also noteworthy in an all-Asia context , as shown by a lack of cost information in, for example, South Korea (only one 'Unspecified' record), Saudi Arabia (no records), Turkey (no records), Thailand (only records considering A. aegypti and A. albopictus) and Iran (no records), which are all amongst the 10 countries with the highest GDP in Asia (International Monetary Fund 2019; https://www.imf.org/). This suggests that lower economic wealth is likely not to be a determinant of how biological invasions -and their monetary costs -are documented (Nghiem et al. 2013).
Regarding the overall cost estimation, it is possible to overestimate costs if one assumes that the costs repeatedly occurring over time are repeated for a longer duration than it actually occurs (if total duration is not reported). To stay conservative, we assigned a single duration year for cost entries for which such information was missing and the cost was potentially ongoing. Furthermore, it is possible that the annual monetary burden increased over the years due to frequent descriptions of new invaders. In addition, the spatial scale for estimating costs in InvaCost reflect 'site' and/or 'country' level estimates, meaning that the national burden could be higher as some 'regional' costs may not have specified specific countries. Additionally, we show that the relatively large number of alien species present in Singapore (see Yeo and Chia 2010) potentially contributes further costs exceeding those that were recorded in InvaCost. However, one should consider that a) Yeo and Chia (2010) presented detailed information only for animals, excluding plants and microbes in this assessment; and b) the difficulties in quantifying certain types of economic impact -especially concerning ecosystem services and the many forms of damage that occur indirectly (Spangenberg and Settele 2010). For all these reasons, it could be assumed that the presented costs may represent potentially a massive underestimation of the real economic costs of biological invasions in Singapore and Southeast Asia.
Our work also reveals a considerable taxonomic bias in the reported economic impacts of the 142 reported alien animal species in Singapore. The weighting of costs towards taxa in the database does not reflect the 'true' taxonomic composition of alien species in Singapore. Freshwater fishes and reptiles together make up the majority of alien species in this country (61%) (Yeo and Chia 2010), but no cost data were found for any of these taxa here. Yet, Yeo and Chia (2010) present anecdotal information that several non-native plant species (e.g. the South American water hyacinth, Eichhornia crassipes) are likely to have necessitated regular management at various scales, sometimes at considerable (yet unquantified) financial cost. This information, however, mostly relied on Wee and Corlett (1986), who, although most likely being outdated, listed 34 potentially invasive plant species present in Singapore. Nevertheless, these two accounts together are only about one quarter of the 648 species listed by GRIIS (Kwek et al. 2020), underlining the gap of cost reporting for invasive species in Singapore.
Whilst we cannot exclude that some existing cost data may have not been captured by the InvaCost database, this taxonomic discrepancy should be discussed. Singapore has a history of freshwater species introductions (Yeo and Chia 2010;Ng et al. 2010;Liew et al. 2012;Ng and Yeo 2012;Kwik et al. 2013;Ng et al. 2015Ng et al. , 2016a. Accidental releases/escapes aside, key drivers of intentional releases can often be cultural (e.g. for aesthetic, recreational or religious reasons; Yeo and Chia 2010). Usually, impacts on aquatic habitats or native communities are less obviously perceived by the public and authorities or are perceived as beneficial for local municipalities (Selge et al. 2011;Kilian et al. 2012). This could partially explain the overall bias towards costs on terrestrial habitats and the lack of information regarding aquatic habitats (Cuthbert et al. 2021b). Yet, as Singapore and many other countries of Southeast Asia are (or include) islands and, in many cases, have extensive and economically-important inland water systems, it is striking that no cost exists here for aquatic invasions. Furthermore, birds are known to be commonly released for religious purposes (Su et al. 2016); however the present study contained costs for just one species, indicating additional knowledge gaps.
Given that management and control costs usually outweigh the costs of prevention and surveillance (Leung et al. 2002), the presence of various introduction pathways in Singapore (Yeo and Chia 2010;Jaafar et al. 2012) raises the concern about how economic costs are related to pathways (Liu et al. 2019). Indeed, this should be evaluated for framing management policies by relevant stakeholders, because currently, Singapore does not have specific management plans in place that address threats from major invasive alien species, but has implemented various surveillance/monitoring programmes (National Parks Board Singapore 2015).
Despite most of the economic costs in Singapore being related to the control of invasive species and the costs of healthcare, it can be assumed that other damage or losses have not yet been estimated. For example, similarly data-poor studies found major costs for agriculture in Argentina or the UK (Duboscq-Carra et al. 2021;Cuthbert et al. 2021a) or forestry in Sweden (Haubrock et al. 2021b). In each case, it seemed clear that these trends were driven by few records, suggesting that a richer cost record might, in each case, reveal costs for other activity sectors, substantially raising the overall estimates. In Southeast Asia, biological invasions could exert a very significant toll on major economic sectors, such as forestry in Indonesia, agriculture in Vietnam, fisheries in the Philippines or tourism in Thailand. In the case of many invasive species, only with more costs being described can we get a better understanding of the cost distribution for each descriptor. Furthermore, without information on the financial pressures that invasive species apply to an economy, efforts to tackle these, whether through prevention, surveillance or applied control and monitoring efforts, might fail at an underestimated monetary value due to inadequate investments. Given the likely underestimated costs identified for biodiversity-rich Southeast Asian countries and illustrated with Singapore, alongside their rapidly growing population densities, trade volumes and GDP, the need for effective invasive species management and cost reporting is paramount.