Assessing and improving the veracity of international trade in captive-bred animals

Abstract

Captive breeding is often seen as a solution to sustainably increasing the supply of individuals in the wildlife trade. To be an effective conservation measure this requires robust systems to verify the authenticity of captive-bred species. The Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) monitors the international trade in Listed species—which for many taxa is dominated by captive-bred individuals—using the Review of Captive Trade (RCT) process. A key question is how best to identify laundered or inauthentic captive-bred trade and how has this changed over time and space. We develop targeted assessments based on multiple RCT criteria to identify probable instances of laundering and misuse of source and purpose codes in international trade records, and apply this to 39,167 records of captive trade from 2000 to 2020 spanning 53,674,762 individuals. We find a very low proportion of trade volume (1.8%, 37,835 individuals) misreported as originating from non-existent, registered Appendix I-breeding facilities, and low instances of exporter-reported captive trade being recorded by importers as wild-sourced (<4%) or ranched (1%). We also find that <2% of species-year-exporter records have abrupt shifts from wild to captive sources, potentially indicating laundering. Conversely, we find high incidences of exporter- and importer-reported trade differing in whether the trade was commercial or not – a phenomenon we attribute to differing definitions, not illegal activity. Our results indicate a low incidence of concerning international trade being reported, but we suggest this likely stems from reporting requirements that limit our assessments. We highlight additional trade data that, if embedded into Party’s annual reports, would vastly improve inferential potential, greatly increasing the number of records (Appendix II and III species) that could be verified with minimal effort for management authorities.

Keywords

Wildlife trade, CITES, Trade management, Trade policy, Captive breeding, Laundering

1. Introduction

Increasing global demand for animals and their products (Harfoot et al., 2018) can threaten species (Marsh et al., 2021) and reduce populations (Morton et al., 2021) if unsustainably managed. Captive breeding is a potential solution to meet demand for species while reducing pressures on wild populations, by breeding species in controlled environments without continuous input from wild populations (Tensen, 2016). Consequently, many heavily traded species are now primarily captive-sourced (Harfoot et al., 2018).

Captive breeding can additionally contribute to species conservation by generating incentives for individuals, communities, and businesses to engage in species conservation (Rivera et al., 2021). Supplying legal and ethically-sourced individuals can also undermine incentives to illegally source species if it is more cost-effective than extracting and laundering wild-caught individuals into trade (Aust et al., 2017). Tangentially, such enterprises can also have co-benefits for public education (e.g., zoos) or reintroduction programs.

Large-scale or commercial captive breeding is not without challenges and criticisms. Ethically, it may lead to the commodification of wildlife, where species are bred primarily for profit, with few welfare considerations (Green et al., 2020; Macdonald et al., 2021). Economically, an abundance of captive-bred individuals might create the perception that the species is abundant and unthreatened (Nijman et al., 2021), and for highly threatened species, if even a small portion of the supply is expanded to incorporate wild individuals, there could be significant risk to populations (Bergin et al., 2018). Captive-breeding operations can also act as fronts to launder wild-caught individuals. Where captive breeding is resource intensive, the input of wild individuals can reduce costs to facilities to the potential detriment of wild populations (Challender et al., 2019). Similarly, if strict quotas are enforced for wild harvests, labelling wild-caught species as bred in captivity can circumvent quotas, trade bans, or enhanced documentation (e.g., proof of sustainable offtake) (Lyons and Natusch, 2011; Nijman et al., 2021; Nijman and Shepherd, 2015).

The Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) regulates the international trade in Listed species between Parties, including the trade of captive-bred individuals. The CITES Secretariat and Animals Committee notes the potential benefits of captive breeding, offering incentives for such trade (CITES, 2010), but also acknowledges the potential scope for abuse. The Secretariat commissioned the Review of Captive Trade (RCT) process (CITES, 2020) to summarise and identify species-exporter combinations of potential concern by focusing on snapshots of recent years of trade data. The process uses criteria regarding trade volumes, trends, use of source codes, exporter-importer reporting, and likelihood that the species was legally acquired (for species traded above certain thresholds).

The RCT uses publicly available data to assess trade records against a range of criteria that could be indicative of fraudulent captive-bred trade. For example, the RCT checks whether exports originate from a species’ range countries (or from countries that border range countries or have legally imported live individuals of the species previously), yet it remains difficult to be certain that fraudulent captive breeding has occurred. The species could have been imported into the country prior to the Convention (Robinson and Sinovas, 2018), or stock could originate from unrecorded vagrants. Similarly, the process contrasts importer- and exporter-reported data to identify the misuse of source codes (e.g., the exporter reports trade as captive-bred, but the importer reports it as wild-sourced). But, again, an abundance of potentially valid explanations exist for such mismatches (Pavitt et al., 2019). This has been exemplified in the literature with CITES trade data being misinterpreted due to a misunderstanding over its limitations (e.g., discrepancies or differences between importer and exporter reported trade volumes) (Berec et al., 2018; Berec and Šetlíková, 2019; Challender et al., 2021; Pavitt et al., 2019; Slábová et al., 2021).

There is a pressing need to scrutinise the growing international trade in captive-bred wildlife (Harfoot et al., 2018). As the diversity and scale of trade increase, manually doing so will prove increasingly difficult. A key research frontier is how best to use publicly accessible trade data to assess the veracity of captive-sourced trade at a global scale and to identify what additional data would overcome the current inferential limitations. We propose an assessment framework using bird and reptile trade records, two heavily traded groups (Marshall et al., 2020; Scheffers et al., 2019) with large proportions of their trade originating from captive sources (Bager Olsen et al., 2019; Harfoot et al., 2018). Firstly, we assess all historic captive trade records from 2000 to 2020 with an adapted version of the RCT to identify the most-commonly flagged criteria used to indicate potentially concerning trade. Secondly, we use specific combinations of criteria to make targeted assessments of the misuse of source codes and potential laundering of non-captive-bred species into reported-captive trade. Thirdly, we highlight inherent limitations in using trade data in their current form to assess the validity of captive-bred trade and suggest enhancements to Parties annual reports to greatly improve the retrospective inferential ability of processes like the RCT.

2. Methods

2.1. Data sources

The backbone of the assessment process is the CITES Database v2022.1 (maintained at https://trade.cites.org/by the UNEP WCMC), through which different summations and combinations are used to assess trade records (see Tables S1 and 2 for details). We define captive trade as trade under the source codes C, D, F, and R, as per the definition used in AC31 Doc 19.1 (C – captive bred in accordance with Resolution Conf 10.16 (Rev), D – captive-bred Appendix I species from commercial operations registered with Secretariat, F – born in captivity but does not meet the requirements for code C e.g. F1 generation, and R – ranched, defined as reared in a controlled environment after being taken from the wild where they would have had a low chance of survival). For specific criteria, we also compare and contrast C, D, and F with R to reflect the possibility of ranched individuals being mislabelled (laundered) as captive-bred (Table S1).

All re-exports and non-captive records were removed and only records from the target taxa were retained (birds and reptiles), then all records were converted to counts of individuals using the whole organism equivalents (WOEs) approach (Harfoot et al., 2018; Morton et al., 2022) (see Supplementary Methods for details on the benefits and limitations of this). The data were then summed as captive trade totals per exporter, per species, and per year, which we subsequently term records (the level of assessment). This process resulted in 39,167 records to be assessed (28,268 bird and 10,899 reptile records), totalling 53,674,762 WOEs (7,429,562 bird and 46,245,200 reptile WOEs) traded between 2000 and 2020. We note that while this Party-reported data forms the backbone of international processes like the RCT and RST (Review of Significant Trade), it is not without potential reporting biases (Heid and Márquez-Ramos, 2023) or without possible improvement.

2.2. Assessment criteria

Where possible, we followed the assessment criteria established by the UNEP-WCMC in AC29 14.1 following CoP17, but make several additions and changes, ultimately totalling 13 criteria (Table S1). We divide these criteria into four broad groups: volume trends, source or purpose code switching, species legal acquisition, and importer and exporter reporting inconsistencies (Table 1 and S1). Appendix I records are assessed using all the groups (criteria 1–13), whereas non-Appendix I species are only assessed using the first three groups (criteria 1–9). The reporting inconsistencies group (criteria 10–13) requires importer- and exporter-reported trade records, and while many Parties report both in all Listed species, it is only explicitly required for Appendix I species (Robinson and Sinovas, 2018). Failing any one of the criteria or even multiple criteria is not immediately indicative of fraudulent or illicit activity. The criteria serve rather as indicators of concern and a tool by which to prioritise more detailed examination of the records in question, although certain combinations of criteria are more likely to be concerning than others. Methods below detail the protocol for each criterion and Table S1 details the justification for each. All general data handling and plotting used the “tidyverse” ecosystem (Wickham et al., 2019) and global maps from the “rnaturalearth” package (Massicotte and South, 2023).

2.2.1. Volume trends criteria

For volume trends, we assess whether a record (summed captive trade totals per exporter, per species, and per year) represents a doubling (criterion 1) or a quadrupling (criterion 2) relative to the previous 5-year average. Criterion 3 assesses the absolute volume of the record, following the protocol used in the CITES Review of Significant Trade. We consider a volume to be substantial if greater than 50 for Least Concern species or greater than 12.5 for Near Threatened, Vulnerable, Endangered, Critically Endangered, Not Evaluation, Extinct in the wild, or Extinct species (as per those used in the RCT). We do not apply the threshold criteria used in the AC31 Doc. 19.1 to isolate particularly high volumes of a given Order, as we assessed every record since 2000.

2.2.2. Source code-switching criteria

The code-switching criteria flag where Parties may be misusing or abusing the CITES source codes. Criterion 4 assesses whether captive trade records (codes C, D, F, and R) show a rapid switch to or from wild sources (W, U, X, or NA). First, the record is assessed for increasing trend – has captive trade (𝐶𝑦) in a given year (𝑦) increased to more than double the average of the previous 5 years (2𝐶‾𝑦−5:𝑦−1, see Eq. 1.1). The trends were then extracted for both captive and wild trade time series (𝐶Δ,𝑊Δ Eq. 1.2). If the trend from one source was complemented by opposing trends from another source, then 𝐶Δ+𝑊Δ=0, e.g., an average decrease of −100 individuals from captive sources is matched by an average increase of +100 wild-sourced individuals. The final step assesses the explicit difference between sources, e.g., if captive volumes have doubled from an average of 1000–2000, and wild volumes have halved from an average of 2 to a volume of 1, then there has been a doubling from one source coinciding with a halving from another. However, the volumes are so disparate to be of no relevance. Therefore, the final step assesses whether the summed volume change from both sources (|𝐶Δ+𝑊Δ|, Eq. 1.3) compensates at least 50% of the change from whichever source has the greatest change (0.5(max(|𝐶Δ|,|𝑊Δ|), Eq. 1.3). Similarly, criterion 5 assesses whether captive trade shows a switch to or from explicitly captive (C, D, or F) to ranched (R). The process outlined in Eq. (1) remains the same but now 𝐶 denotes codes C, D, F and 𝑊 denotes changes in ranched trade (code R).Eq. 1𝐼𝑓𝐶𝑦>2𝐶‾𝑦−5:𝑦−1[1.1Checkcaptivetrendsareclearlyincreasing]𝐶𝑦−𝐶‾𝑦−5:𝑦−1=𝐶Δ[1.2Getchangeforcaptiveandwild/ranched]𝑊𝑦−𝑊‾𝑦−5:𝑦−1=𝑊Δ|𝐶Δ+𝑊Δ|≤0.5(max(|𝐶Δ|,|𝑊Δ|)[1.3Checkproportions,seeexplanation]

Criterion 6 looks at whether the species is traded in that year as source code C (animals bred in captivity in accordance with the Convention) and whether it has ever previously been exported from that Party under source code F (animals born in captivity that are the F1 generation or otherwise do not meet the Convention criteria for bred in captivity). Code C assumes the more stringent CITES definition of ‘bred in captivity’ as per Conf. 10.16 (Rev CoP19), which stipulates the stock has produced F2 or subsequent offspring in the environment or in a manner demonstrated to be able to.

2.2.3. Legal acquisition criteria

The legal acquisition criteria assess the likelihood that an exporter has captive populations of a species. The checks assess firstly whether the exporting country is a range country for a traded species (criterion 7), then whether the exporter shares a border with any range countries (criterion 8), and finally whether the exporting country has previously imported the species live (criterion 9). Records that pass criterion 7 are referred to as within range; records that fail criteria 7 but pass either 8 or 9 are referred to as probable occurrences as they offer potential pathways for the species to have populations in the country; and records that fail all legal acquisition criteria (7–9) are termed improbable occurrences. A caveat is that countries bordering range countries may also be hotspots of laundering trade out of range countries.

2.2.4. Reporting inconsistencies criteria

The reporting inconsistencies criteria were applied on a per exporter-species-year basis. Due to potentially differing practices in reporting, e.g., trade late in the year may be reported that year or potentially in the subsequent year’s reports by different reporters, it is possible that importer- and exporter-reported data differed solely due to the time of reporting. Including two or three summed years of records can partially avert this issue but increases the risk of flagging differences that are years apart, when this is likely just due to temporal trade fluctuations. Therefore, we assess per-year differences. For criteria 10, 11, and 12, we first checked whether the total importer- (CDFR𝐼+WUX𝐼, Eq. 2.1) and exporter-reported volumes (CDFR𝐸+WUX𝐸, Eq. 2.1) for that species from that exporter in that year differed by less than 25%. Only records that met this requirement were assessed (records failing to meet it could not be assessed so were not scored as failing). The equivalent reporting requirement ensures we focus on discrepancies in the proportion of each source reported rather than the absolute volume.

Criterion 10 assesses whether the volume of captive (C, D, F, and R codes) to wild (W, U, X, and NA codes) trade differs between exporter- and importer-reported records (Eq. 2). Firstly, calculating the ratio of captive trade (CDFR) reported by exporters (𝐸) and importers (𝐼, Eq. 2.2) and then repeating this for wild trade (WUX, see Eq. 2.3). Next, we confirm that exporter- and importer-reported captive and wild ratios both differ by at least 10% (Eq. 2.4) and that the reporting ratios differ in different directions (e.g., captive ratio is greater from the exporter, and the wild ratio is greater from the importer, Eq. 2.5).

Criterion 11 assesses the same but for ratios of explicitly captive (C, D, and F codes) and ranched (R codes) trade (same principle as Eq. (2), but C, D, F, R are replaced with C, D, F; and W, U, X are replaced with R). Likewise, criterion 12 focuses on all captive trade (C, D, F, and R codes) and assesses the exporter- and importer-reported ratio of commercial to non-commercial trade (contrasting purpose rather than source).Eq. 2𝐶𝐷𝐹𝑅𝐸+𝑊𝑈𝑋𝐸≈𝐶𝐷𝐹𝑅𝐼+𝑊𝑈𝑋𝐼[2.1𝐸𝑥𝑝𝑜𝑟𝑡𝑒𝑟𝑎𝑛𝑑𝐼𝑚𝑝𝑜𝑟𝑡𝑒𝑟𝑟𝑒𝑝𝑜𝑟𝑡𝑠𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑑𝑏𝑦≤25%]𝑅𝑎𝑡𝑖𝑜𝐶Δ=𝐶𝐷𝐹𝑅𝐸𝐶𝐷𝐹𝑅𝐼[2.2Getcaptivereporterratio]𝑅𝑎𝑡𝑖𝑜𝑊Δ=𝑊𝑈𝑋𝐸𝑊𝑈𝑋𝐼[2.3Getwildreporterratio]|1−𝑅𝑎𝑡𝑖𝑜𝐶Δ|≥0.1&|1−𝑅𝑎𝑡𝑖𝑜𝑊Δ|≥0.1[2.4Checkthedifferenceis≥10%]𝑠𝑔𝑛[1−𝑅𝑎𝑡𝑖𝑜𝐶Δ]≠𝑠𝑔𝑛[1−𝑅𝑎𝑡𝑖𝑜𝑊Δ][2.5Checkoneisincreasing,theotherdecreasing]

Criteria 13 assesses if exporter-species-year records contain any trade volume under the source code D, concerning Appendix-I animals bred in captivity for commercial purposes in operations that are included in the Secretariat’s Captive Breeding Register. If any volume for that year was traded under code D, we checked the register to see whether that species-country combination is present.

We do not explicitly consider raw volume differences between importer and exporter reports as importers are not required to report trade in Appendix II or III species, and the use of exemptions for Appendix I species also removes this reporting requirement. However, of the 39,167 records, in 76 cases importer-reported values were ten-fold higher than exporter-reported values (where volumes were >10). This covered 30 bird and 25 reptilian species (see Table S4 for a breakdown of the 25 instances concerning the highest importer-reported volumes). We hypothesise these are likely due to importers reporting planned trade and exporters reporting realised trade.

2.3. Targeted assessments

We focused on three targeted assessments where specific criterion or combinations of criteria can provide additional insight into potentially concerning trade:

• (1)The first focuses on source code D misuse (uses criterion 13, Table 1), where Parties report trade under source code D without having registered an Appendix I breeding operation for that species to the Secretariat. Such Parties are misreporting their trade, potentially to gain the benefits associated with being a registered breeder.Table 1. Criteria schematic (See Table S1 for details). The second row shows the broad category each criterion falls into.All (App I, II and III)Appendix I onlyVolume trendsCode-switchingLegal acquisitionReporting inconsistencies1. Captive volume doubled
  2. Captive volume quadrupled
  3. High volume4. Captive to wild
  5. Captive to ranch
  6. Traded as F17. Native range
  8. Exporter borders range countries
  9. Previously imported stock10. Captive-wild ratio
  11. Captive-ranch ratio
  12. Commerical – non-commercial ratio
  13. Registered breeder
• (2)The second concerns potential laundering incidences identified through reporter mismatch (uses criteria 7, 10–12). Records from range countries were selected where they failed any of the first three reporting inconsistencies criteria. For example, this would highlight trade where the exporter reports a native species as captive-bred but the importer reports it as wild-sourced. Such trade is potentially concerning as it could represent the laundering of wild individuals into captive trade. CITES authorities (often in collaboration with species experts such as IUCN Specialist Groups) provide specific guidance to aid importers in differentiating captive and wild individuals (e.g. not solely reliant on exporting paperwork). The guidance includes absence of closed leg rings or other physical marking, non-natural colour morphs, condition, presence of scars and parasitic load, etc. (Lyons and Natusch, 2015). As criteria 10–12 can only be applied to Appendix I records, this assessment was only applied to these records. Similarly, we assess instances where the same occurs for ranched individuals, e.g., species taken from the wild as young and reared in captivity (ranching) are reported as captive-bred. We also explore where importers and exporters report purpose codes differently as this potentially would highlight instances where commercial trade is being misreported as non-commercial to avoid national quotas and legislation specific to commercial exports.
• (3)The third concerns potential laundering identified through source switching (uses criteria 1, 4 and 7). We identify records from species range countries with a rapid rise in captive volume coinciding with a commensurate decrease in wild volumes. As this focuses on temporal shifts in source rather than discrepancies between reporters, all Listed species were assessed.

3. Results

3.1. Criteria failed 2000–2020

Between 2000 and 2020, there were 39,167 species-country-year records, comprising 5230 records for Appendix I, and 33,937 records for Appendix II/III (and not listed) species. Only 4.21% (197 bird and 23 reptile records) Appendix I and 2.44% (625 bird and 204 reptile records) Appendix II/III records failed no criteria. Conversely, no Appendix I failed all 13 criteria or all 9 criteria for Appendix II/III records (Fig. S1C). Appendix I records failed a mean of 3.38 criteria for birds and 3.49 for reptiles, while Appendix II/III records failed a mean of 3.59 criteria for birds and 3.27 for reptiles.

Proportions of records failing each criterion (excluding the reporting-inconsistencies group) were consistent across taxa and Appendices (Fig. 1A and B). The most commonly failed criterion for Appendix I species was trade from a non-range country (criterion 7), with 73.6% (3850/5230) of records failing this check. Similarly, trade from countries that do not share a border with range states (criterion 8) occurred in 69.5% of records. However, only 9.1% fail the final criteria for legal acquisition of whether a species has ever been imported live into the country of export prior to the current export (criterion 9), leaving a potential avenue for Parties outside of a species’ native range to possess captive populations.

Criterion 6 was flagged by 55.7% of records, indicating the majority of individuals of a species are traded for the first time from captive breeding facilities in accordance with the Convention (post-F1) as opposed to trade from the F1 generation. Criteria 1, 2, and 3 were the next most commonly failed criteria and concerned increasing temporal trends (criteria 1 and 2) and relatively high absolute volumes (criterion 3). Criteria 4, 5, 10, and 11 were the least commonly failed criteria (each with <1% of records failing), with few records showing switches between captive and wild or ranched sources, and few records having differing sources reported by the importer and exporter. Patterns for Appendix II records were broadly similar with criteria 7, 8 (legal acquisition), 1 and 2 (volume trends), and 6 (never traded as F1) being the most frequently failed (Fig. 1B).

3.2. Potential misuse of the register of captive-breeding operations (assessment 1)

Of the 673 records (species-year-exporter combinations) that were reported using the source code D, 21.1% (142) concerned species-exporter combinations that are not present in the Secretariat’s register of captive-breeding operations, which amounted to 37,835 WOEs (1.8% of the 2,061,673 WOEs traded using code D). Thirty-three Parties misused code D, although only 20 Parties did so more than once (Fig. 2A). South Africa had the most incidents (29), with 27 concerning parrots traded in volumes ranging from 1 to 30 in a single year. These included 4 incidences of Critically Endangered blue-throated macaw (Ara glaucogularis) and 30 WOEs in 2019 of the now Critically Endangered yellow-naped amazon (Amazona auropalliata). The total incidences and volume through time remained highly variable with incidences peaking in 2000 but with relatively low volumes, while volumes peaked drastically in 2015/16 (Fig. 2B). The sporadic high volumes of potential misuse is driven solely by voluminous trade in saltwater crocodiles (Crocodylus porosus) from Malaysia and Viet Nam, Siamese crocodiles (C. siamensis) from the Republic of Korea and Malaysia, and Nile crocodiles from Namibia (Fig. 2).

3.3. Potential laundering from range countries – reporter mismatch (assessment 2)

The trade of captive-bred species from range countries offers an avenue for wild-caught species to be laundered into captive trade (Fig. 3A). Of the 514 Appendix I records that were reported in similar total volumes by the exporter and importer from range Parties, 17 (6 with volumes >10) showed evidence of a mismatch between captive- and wild-sourced individuals (Fig. 3B). For example, in 2015 the United Arab Emirates (UAE) reported the export of 5734 captive-bred Asian houbara (Chlamydotis macqueenii, Vulnerable, Appendix I species) and no wild-caught individuals, whereas importers reported 5012 captive-bred and 962 wild-caught individuals from the UAE (Fig. 3B). One interpretation is that between 722 and 962 wild-caught individuals were misreported as captive-bred by the UAE. In this instance, the trade concerned individuals bred for reintroduction and the importer (Jordan) failed to report a source (NA), which is conservatively assumed to mean wild in analyses for the Secretariat (CITES, 2020). Multiple other legitimate inferences exist for this discrepancy: 1) data entry errors by the exporter or importer; 2) Parties differentially reporting permits issued and actual trade; and 3) year-end trade is reported by one Party late in the year while the other Party reports it in the subsequent year.

Only five (four with volumes >10) records showed mismatches between captive-bred and ranched trade. In 2017, again the UAE reported solely captive exports of Asian houbara (7943 WOEs), but importers reported 5183 captive-bred WOEs and 3034 ranched WOEs, a mismatch of 2760–3034 WOEs (Fig. 3C). Such a discrepancy could arise through deceit (laundering wild-caught eggs/juveniles) or through the parties applying conflicting definitions of captive and ranched. Far more common was the mismatch of purpose codes, with 223 (148 with volumes >10) records showing evidence of this (Fig. 3D). In a small number of cases (10/148), this was due to the importer reporting an NA purpose (e.g., high-volume Siamese crocodile exports from Thailand in 2000, 2008

10). In the majority of cases, it was instances where the exporter reported trade as commercial, but the importer reported it as another purpose (black dots, Fig. 3D), with the importer for many relatively low-volume avian species (e.g., falcons) reporting the purpose as personal. Potentially more concerning are the far smaller number of incidences where trade is exporter reported as non-commercial, but reported as commercial by the importer (such incidences could indicate the avoidance of national laws on commercial trade or through conflicting definitions). Examples include the export of 100 Hispaniolan rhinoceros iguana (Cyclura cornuta) from the Dominican Republic in 2004 reported as a zoo transaction, whereas the importer (Spain) reported them as commercial.

3.4. Potential laundering from range countries – source switching (assessment 3)

Sudden shifts from wild-sourced to equivalent captive-bred volumes can be inferred as either a potential positive shift in supply as captive-breeding displaces wild capture, or an opportunity for laundering as wild-caught individuals are reported as captive-bred (Fig. 4A). Of the 9163 records of captive trade from countries within the traded species range, only 178 (1.9%) showed evidence of wild-to captive-source shifts (89 records with traded volumes >10). Incidences are concentrated in Indonesia (34 incidences), with three times as many incidents as the next Party (Malaysia, 10 incidents), while more than half of the flagged Parties only having a single incidence (16/27 Parties, Fig. 4B). Similarly, few species were involved in incidents more than once.

Turquoise-fronted amazon (Amazona aestiva) exported from Argentina in 2011, experienced rapidly declining wild-sourced volumes concurrent with a rapid increase in captive volumes (Fig. 4C). The rapid but sequential increase in captive volumes and the sporadic nature of the wild-sourced trade in the species from Argentina likely means there was an ongoing shift from wild to captive sources. For reptiles, exports of Oriental ratsnake (Ptyas mucosa) from Indonesia in 2011 (the highest captive-bred increase relative to the 5-year mean), suddenly increased from none in 2006–2008 to 50,000 WOEs in 2011, whereas the wild-sourced trade collapsed from >80,000 in 2006 to 15,015 in 2011. The sudden appearance of high volumes of captive-bred (code C) individuals even for a species amenable to captivity, is potentially suspicious and calls into question whether any wild-sourced individuals were laundered into the captive breeding program (Nijman, 2014). The distinction between the two examples above cannot be ascertained from data alone, and even with biological context, it is difficult to determine whether laundering occurred.

3.5. Inferential issues for trade from non-range countries

For birds, 90.5% of traded individuals reported to be bred in captivity were exported from non-range countries (6,721,902 of 7,429,562), with 2.8% coming from countries of improbable occurrence (Fig. 4C). The opposite is true for reptiles, with 95.5% of captive-bred individuals originating from range countries, mostly consisting of commercial caiman, alligator, and iguana species (Fig. 5). Species on average were exported by 7.4 countries (avian mean = 8.9, reptilian mean = 5.2).

Trying to identify potentially concerning incidences of captive trade from non-range countries (e.g., species in trade from Parties with no plausible explanation for obtaining self-sustaining captive populations), is especially challenging given the range of ways species could come to be in non-range countries. Highly threatened in the wild, African grey parrots (Psittacus erithacus) are exported from 115 different exporters, the majority being non-range countries that have imported live individuals prior to trade and could feasibly possess captive populations (Fig. 5B). The only improbable country of export for the species was Estonia in 2003 for exporting 1 captive-bred (code F) individual for personal purposes, although given the species popularity in private ownership this remains feasible. Similarly, Java sparrow (Lonchura oryzivora) is exported from 36 different Parties, including Czech Republic where it is exported in large numbers despite no record of live individuals being imported (improbable occurrence). Given this species is so profligate in the cage bird trade, abundant in private ownership, and commonly naturalized, there likely are commercial breeders in the country. This exemplifies the issue with using range data to determine where captive species of dubious origin are occurring without greater knowledge of established captive populations.

We suggest improbable trade could be identified by focusing only on the first instances of trade in species from improbable countries that are only exported from that one country. For species with such limited trade presence it is harder to rationalise historic explanations for the species presence in sufficient numbers for feasible captive breeding. Examples include commercial exports of masked shining-parrot (Prosopeia personata, a Fiji endemic) from the Netherlands to Qatar and Indonesia in 2015 (150 live individuals) and 2016 (75 live individuals), and the commercial export of three Madagascan endemic gecko Uroplatus sp. from Ukraine in 2015 and 2016. It is likely impossible to know whether a facility or individual legally possessed Madagascan endemics in the Ukraine, but building this process into permitting at the time of trade by flagging non-range species that are improbable occurrences in the exporting country would be a useful step to ensuring the provenance of traded species. In this case, multiple EU Scientific Authorities requested clarification from the exporting company resulting in some negative decisions and some instances where the trade was accepted due to a lack of evidence (Nijman pers. observ.). In some flagged cases, such as Goniurosaurus hainanensis (an endemic cave gecko to Hainan Island), sudden appearance can be explained by recent listing such that prior imports and exports are largely unknown.

4. Discussion

Insight into the potential for laundering in the trade of CITES-listed captive-bred individuals is limited, and abundant with nuanced caveats and alternative interpretations. Our results show that targeted assessments can reveal instances of captive trade worthy of further scrutiny. Over 37,000 WOEs, 1.8% of the total exported trade volume under source code D, were reported to come from registered breeding facilities from Parties with no registered breeders for that species. The total incidence of potential laundering through mismatching reporter sources was very low (<5% of records), and while mismatching proportions of commercial trade are far more common (40% of records) they can potentially be attributed to inconsistent definitions for commerciality. Rapid source switching was uncommon, occurring in <2% of records. We then highlight a potential method for screening improbable records based on how widely the species is exported. Overall, we highlight a relatively low proportion of trade potentially in breach of the Convention, a phenomenon we partly attribute to the limited data CITES Parties are required to report.

4.1. Improving the inferential potential of trade data and CITES policy

A 2016 TRAFFIC Briefing (TRAFFIC, 2016) used seven case studies to illustrate the issue of fraudulent captive breeding in international trade, five identified due to doubt that the exporter could feasibly breed that quantity of individuals. However, the RCT and, specifically, national trade reports cannot be used to identify feasible numbers because there is no centralised list of facilities for nearly all species, and no public data on individuals and vital rates at facilities. The only way to identify these instances in the current system is through expert identification including visits to facilities, an expensive and time-consuming process that scales poorly with the ever-increasing diversity of traded species, particularly as newly described species enter trade in greater numbers (Marshall et al., 2020). Improving the quality and type of data collected on trade (e.g., Party annual reports) is the only scalable way to evaluate captive trade to the accuracy originally called for (TRAFFIC, 2016). A further challenge is retrospectively ground-truthing potentially fraudulent records. A caveat to our methodology and the RCT itself is while these assessments are based on logical reasoning, there is no definitive way to determine whether historic captive trade was fraudulently reported. Likewise, we specifically address potential fraudulent international trade in CITES-Listed species, but the wider international and domestic illegal trade remains much harder to quantify (Fukushima et al., 2021). Furthermore, current reporting requirements only allow certain criteria to be applied to certain records (e.g. many require dual importer and exporter reporting and can only be applied to Appendix I species) and it is possible or even likely that the vast majority of unassessed records would show little evidence of fraudulent reporting, but without improved reporting this will remain impossible to quantify.

Changes to CITES policy are time-consuming to evaluate and, even if agreed, take time for Parties to fund and implement. However, that does not mean the type of data submitted in annual reports cannot be improved. Currently, the range of alternate explanations for captive species volumes in trade that cannot be disentangled limits interpretations of our targeted assessments and the RCT. While some caveats are unavoidable (e.g., honest reporting errors), many could be reduced with enhanced reporting. We make five additional data collection considerations – ordered by decreasing feasibility – that would achieve this (Table 2): (1) reporting the use of exemptions under Article VII of the Convention; (2) declaring whether national trade reports are based on realised or permitted trade; (3) declaring in national reports when trade occurred in the last month(s) of a year (year-end trade – where trade is at risk of being reported by the exporter and importer in different years); (4) importer and exporter reporting for all Listed species; and (5) developing an expanded captive-breeding registry database (Table 2). The first three recommendations only require one additional data field to be included in Party’s annual reports. By augmenting the annual reports with these data, retrospective checks (like the RCT) would have enhanced ability to discern where reasonable explanations exist for discrepancies in reported trade. Such additions will lead to a level of disaggregation compared to previous records in the database, but this would deliver much greater insight.

Table 2. Long-term potential for the more effective use of CITES trade data to identify concerning incidences of captive trade.

Reporting/Policy change	Justification	Feasibility
1. Noting when an exemption is used	Greater ability to unpick mismatches in importer and exporter-reported trade.	High – Requires management authority cooperation and additional information in annual reports of exports. Records may be disaggregated.
2. Distinction between permitted and realised trade	Greater ability to unpick mismatches in importer and exporter-reported trade.	High – No additional data collection for Parties bar confirming that their annual reports concern realised or permitted trade.
3. Distinction for year-end trade	Greater ability to unpick mismatches in importer and exporter-reported trade.	High – Management authorities would need to note in their annual reports if trade occurred/was permitted in the first/last month of the year. Records would be disaggregated.
4. Dual reporting for Appendix II/III species	Allows reviews of significant/captive trade processes to contrast two sources for all Listed taxa not just Appendix I.	Medium-low – Already completed by some Parties, but a systematic change for others needing additional national-reporting infrastructure for management authorities. Not necessarily for scientific authorities if it only involves recording the import of CITES-listed species not issuing an import permit (for App II/III species).
5. Expanded captive breeding registry to Appendix II/III	Building up a global picture of where and in what numbers Listed species are bred would enable a more comprehensive understanding of feasible source countries and realistic outputs.	Very low – Systematic change and significant additional workload for scientific and management authorities and the Secretariat.

The final two recommendations we wish Parties to consider represent far greater systematic changes to CITES processes, and require major amendments to the existing articles of the Convention. Dual reporting (importer and exporter) for all Listed species (Appendices I, II, and III) (Table 2, row 4) would enable the full scope of the RCT criteria. Currently, several of the more insightful RCT criteria are only applied to Appendix I species as they rely upon contrasting data sources and cannot be reliably applied to the majority of traded species or volume. Many Parties already report both imports and exports (e.g., EU Parties), for others this would require substantial changes in how they discharge their commitments to the Convention, increasing workload and funding needs of national management authorities. But by only requiring the reporting of imports for Appendix II/III species, not the additional checks required for import permits of Appendix I species (i.e., assessing non-detriment Article III, P.3), the burden would be greatly reduced. Only requiring dual reporting for threatened species (as per IUCN Red List assessments) is another potential way of focusing labour and funding to those taxa most likely to benefit. However, aligning dynamic IUCN assessments with reporting obligations would require compromise and agreement on which species Parties are obliged to report each year, creating potential implementation issues.

Finally, establishing a registry of breeding operations that trade any Listed species internationally (not just a fraction of Appendix I species; Table 2, row 5) would shift towards a more proactive than retrospective system. Initially, all organisations internationally trading Listed species could be required to register their existence. The ultimate aspiration would build on the protocol outlined in Resolution Conf 12.10 Annex I (CITES, 2010), in which breeders provide details of their operation and estimates of stock, mortality, and production to their management authority. This would create an invaluable resource that could link with existing registries (Ex. EU Database on Captive Production Systems) and feed into the RCT to assess instances where facility production rates differ markedly from what would be expected or trait values diverge from published estimates (Janssen and Chng, 2018).

Verifying where Listed species are truly captive bred would enable much faster and accurate verification of acquisition. For example, Cuba exports large volumes of threatened Java sparrows, it is not a range country, nor does it border one, and has no reported history of live imports, yet it is highly likely there are legitimate breeding operations in country for this widely and easily bred species. But conversely, the small numbers of captive-bred Papuan hornbills (Rhyticeros plicatus) exported from the Solomon Islands (range country) are unlikely to be captive bred given the lack of known captive breeding success for the species (Nijman and Shepherd, 2015). If breeding operations were required to register and provide minimal evidence to national management authorities, exports of Javan sparrows from Cuba would likely have gone ahead, but of Papuan hornbills would have been flagged. Further examples include the export of relatively large numbers of reportedly captive-bred dendrobatid frogs from Kazakhstan despite the lack of live import and the taxa’s infrequent captive-breeding history (Nijman and Shepherd, 2010).

4.2. Evidencing sustainability and conclusions

While ratifying the authenticity of international trade records is key to the Convention, it cannot answer a more fundamental question: whether volumes taken from the wild or reportedly bred in captivity are sustainable or feasible. Our understanding and knowledge of reproductive traits for most species has rapidly increased (Etard et al., 2020). To effectively use this vast data pool, there must be a similar increase in the type and nuance of trade data collected. Currently, we lack an accurate understanding of the realistic output for many captive-bred species (Janssen and Chng, 2018; Morton et al., 2023). At minimum, national-level records for the number of CITES-Listed individuals in captive-breeding facilities should be reported. Likewise, there is a scarcity of data concerning domestically traded captive-bred individuals, hampering our wider understanding of what species and quantities are feasibly produced. Such data is vital in estimating realistic outputs for captive production and identifying instances where captive trade is merely a front for laundering. Aspiring to such a standard must be the minimum to ensure captive breeding does not endanger wild populations.

As CITES celebrates 50 years since its inaugural signing, we can look forward and imagine what the next 50 years could achieve. While retrospective assessments of the likely validity of captive trade can be insightful, they are currently fraught with uncertainties of interpretation. Removing some of these uncertainties through additional reporting requirements would greatly enhance the utility of retrospective reviews of trade and further safeguard the sustainable use of species.

CRediT authorship contribution statement

Oscar Morton: Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Writing – original draft, Writing – review & editing. Vincent Nijman: Conceptualization, Methodology, Supervision, Writing – review & editing. David P. Edwards: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Vincent Nijman reports a relationship with IUCN SSC Groups that includes: non-financial support. VN is a member of the IUCN/SSC Asian Songbird Trade, primate, bear, and mollusc specialist groups and was a previous member of the Dutch CITES Scientific Authority (2005–2017), giving him privileged information about the import and export of CITES-listed species into the EU. However, all data collected were derived from publicly available sources. This manuscript represents the views of the authors and does not necessarily reflect that of the CITES Scientific Authorities or IUCN/SSC groups.

Acknowledgements

The authors would like to thank UNEP-WCMC staff for early discussions of CITES processes and David Wilcove for comments on an earlier draft of this manuscript.

References

1. P.W. Aust, N. Van Tri, D.J.D. Natusch, G.J. AlexanderAsian snake farms: conservation curse or sustainable enterprise?Oryx, 51 (3) (2017), pp. 498-505, 10.1017/S003060531600034XView at publisher View in ScopusGoogle Scholar
2. Bager Olsen et al., 2019M.T. Bager Olsen, J. Geldmann, M. Harfoot, D.P. Tittensor, B. Price, P. Sinovas, K. Nowak, N.J. Sanders, N.D. BurgessThirty-six Years of Legal and Illegal Wildlife Trade Entering the USAORYX (2019), 10.1017/S0030605319000541View at publisher Google Scholar
3. Berec and Šetlíková, 2019M. Berec, I. ŠetlíkováImportant step to understanding the CITES Trade Database: a reply to Pavitt et alBiol. Conserv., 230 (December 2018) (2019), pp. 197-198, 10.1016/j.biocon.2018.12.018View PDFView articleView in ScopusGoogle Scholar