Captive Reptile Mortality Rates in the Home and Implications for the Wildlife Trade

Janine E. Robinson, Freya A. V. St. John, Richard A. Griffiths, David L. Roberts

Response to: Robinson et al. (2015) Captive Reptile Mortality Rates in the Home and Implications for the Wildlife Trade. PLoS ONE 10(11): e0141460. doi:10.1371/journal.pone.0141460

Response to:

Robinson JE, St. John FAV, Griffiths RA, Roberts DL (2015) Captive Reptile Mortality Rates in the Home and Implications for the Wildlife Trade. PLoS ONE 10(11): e0141460. doi:10.1371/journal.pone.0141460

This survey-based study claims that the overall mortality rate for reptiles in UK homes is 3.6% in the first year following acquisition. The authors’ findings are at odds with a more extensive study that found that the mortality rate for reptiles in the home was 75% in their first year (Toland et al., 2012). There are numerous major problems with Robinson et al.’s study that explain their anomalous results and for which we provide examples as follows:

Statistical relevance: The survey involved asking questions of just 265 individual respondents at two reptile and amphibian markets in the United Kingdom by which the authors seek to reach a conclusion that applies to the 1.1 million reptiles in UK homes (PFMA, 2015). In our view, the 265 interviewees represent a disproportionate and unrepresentative sample on which to base an accurate estimate of mortality rate.

Respondent recall: The survey asked the respondents the following key question: “Of the X (reptiles) that you acquired over the last five years, how many died within the first 12 months?” The question not only implies that the respondents kept accurate temporal logs for mortalities within each 12-month period over five successive years, which is highly improbable and certainly not reported, but that the respondents were able to accurately recall this information from memory.

Response orchestration: the respondents were part of two events organised and promoted by reptile-keeping societies, both of which have public pro-trade and animal keeping agendas. Relatedly, vested interest parties utilised associated publicly accessible media in advance of the survey to express the value of ‘low mortality’ messages – effectively for propagandist reasons. Therefore, many of the respondents were likely aware of this Internet campaign (whether or not they were members of the special interest societies) and thus could have cultured (selectively appropriate) answers. No statistical formula is capable of correcting such data bias.

Respondent biases: the respondents reported that collectively they kept 6,689 reptiles over the preceding five years, which means that each respondent typically kept multiple animals. The authors report a ‘range’ of 1 – 1,003 animals per keeper, but manifestly the vast majority of respondents were keeping numerous animals. Therefore, the respondents were primarily multiple animal keeping hobbyists and breeders and not typical reptile-keeping members of the public. This means that the respondents were relatively specialised keepers compared with the average person and thus the base data are not representative of overall mortality rates in the home. Although the authors partially recognise this ‘subset’ issue, they do not factor it into their findings.

Respondent honesty: the authors repeatedly claim, based on the respondents’ indications, that they did not regard the questions about mortality in their animals to be ‘sensitive’ – ergo that they would provide honest answers. Relatedly, the authors also claim that their honesty test system (e.g. using playing cards to disguise the number of animals that may have died, to improve anonymity and to promote honest answers) verified the honesty of the respondents. However, the survey is wholly dependent on contributor declarations and honesty, and the ‘playing card test’ cannot ensure accurate reporting. This is because the cards can only assist anonymity and ‘face-saving’ by ‘disguising’ mortalities to a maximum of ten, and respondents who may have lost tens or hundreds of animals may be dissuaded from admitting mortality rates that exceed the card numerical ‘buffer’ (see previous comment regarding actual numbers of animals in captivity per keeper). The authors themselves admit that with increasing numbers of mortality (once 10 is exceeded) the level of protection afforded the respondent decreases; reporting a value of 20, indicates that 10 reptiles (at least) had died (see S2 Appendix). Oddly, the authors acknowledge possible (untested) reporting bias for ‘sensitivity’ regarding whether animals were genuinely captivebred versus wild-caught, but then expect the same respondents not to be sensitive about animals dying in their care. In addition, in the section ‘Evaluation of additive RRT’, 56% of respondents stated that they felt that their answers were protected, compared to 13% who did not feel that their answers were protected, which leaves a highly significant 31%.

The premature death of pet animals fundamentally questions the credibility of the owner, which increases risk of dishonesty, particularly if respondents felt they were in the presence of more experienced reptile keepers, or relatives (from whom they may have hidden past losses). Interviews were carried out in ‘breakout areas’ (e.g. cafeteria) and thus were not entirely private and potentially subject to the scrutiny of other keepers who may have been within audible distance. This aspect of ‘fear of losing face’ was not properly explored and would likely have had an impact on respondents’ answers. Regardless, if the respondents provided false information, then the data are skewed no matter what verification tests are applied to their given answers.

The study surveys were conducted at two events known to be associated with significant criminal activity, described in a recent High Court ruling to involve ‘impermissible sales of animals’ (McGowan, 2013). Both formal traders and so-called ‘hobbyist-sellers’ of reptiles have for several years been made aware of serious legal repercussions for those found to be selling reptiles at the two surveyed events (‘International Herpetological Society Breeder’s Meeting’, Doncaster and ‘Federation of British Herpetologists’ Kempton Park Reptile Breeders’ Meeting’). These legal dangers have recently been highlighted through successful prosecutions of trade-hobbyist-sellers, and one of these prosecutions was in progress throughout the study, presenting a substantive deterrent against respondent honesty. The adverse publicity surrounding these two events has likely stimulated a collective interest amongst stallholders and visitors in presenting their hobby or trade in a more favourable light. Traders prosecuted at these events, including the Doncaster market, claimed to be hobbyists selling their own surplus pets but have been deemed by the courts to be commercial traders. These well-publicised cases have further encouraged other breeders and sellers to adopt a default self-definition of ‘hobbyist breeder’.

Furthermore, the survey questionnaire was incidentally designed in such a way that data regarding the nature of the respondents are inevitably skewed. The second question of the questionnaire asks the reptile keeper to categorise him or herself as either a private keeper (who keeps reptiles for pleasure), a private breeder (who keeps and breeds reptiles for pleasure), a private breeder (who keeps reptiles mainly for monetary gain) or a commercial enterprise. The distinction drawn between the latter two categories, however, is unclear and even if respondents had answered honestly, artificial separation between these two categories probably led the authors to class those breeders keeping reptiles for monetary gain as noncommercial. Also, the positioning of this question so early in the questionnaire may have added to its apprehension quality and affect response accuracy. It is reasonable to propose that a proportion of respondents, who constituted at least part of the data, were commercially motivated breeders and sellers, but wished to avoid declaring their activities. The events’ associations with illegal activities probably decreases further the likelihood of honest reporting by surveyed individuals.

A further problem relates to respondent honesty in terms of a perceived risk of prosecution. Should respondents admit to significant mortality rates then they may be liable to consideration for prosecution in respect of committing offences under the Animal Welfare Act (2006). Whether or not such respondents may be formally identified during the survey, they may nevertheless feel exposed to risk of identification and conceal both their mortality rates and their concern about them. Accordingly, it is likely that some or all respondents would wish to misrepresent their losses to avoid risk of prosecution.

Given the controversy and history around these itinerant selling events, it is improbable that the event organisers would have permitted a study where the potential outcomes would be likely to undermine the already low credibility, security and longevity of such events. The study authors, therefore, may have made a fundamental error in involving the politically motivated event organisers in the promotion of the survey. The capacity for propaganda-cultured responses referred to above was likely a major factor in the ‘acceptability’ of the study at the two events in anticipation of favourable results. Other issues: The results also cited an indication that captive-bred reptiles have lower mean mortality rates than those of wildcaught or captive-farmed origin. However, the authors reported that the “differences were not considered significant given overlapping confidence intervals” (italics added). If the data shows no statistical significance, then no difference exists. The discussion also claims that “captive bred individuals appeared to have lower mortality rates”, but then proceeds to contradict itself by saying “this was just a non-significant trend”. Other scientifically weak claims are present in the conclusions, for example phrases such as ‘Cases of high mortality in the trade are reported [references cited], but these cases do not appear to be frequent”. The expression “do not appear” has no place in empirical scientific research.


In our view, the survey offers highly misleading conclusions and exemplifies the dangers of using selective small-scale subset material to assess a more complex superset issue. The cohort interviewed was unrepresentative, and it is doubtful that even the ‘right’ cohort, based on actual mortality logs and not distant recall, would be both willing and truthful as to losses where to do so may expose respondents to embarrassment and/or prosecution. The very nature of this study’s approach suggests that respondents were likely to provide false answers. The study’s anomalous results are therefore highly questionable and unreliable, and serve to draw attention away from the true impact of captivity on reptile mortality rates.

Phillip C Arena BSc(Hons) PhD
Robert Laidlaw CBiol MRSB
Angelo J L Lambiris NHED MSc PhD CBiol FRSB
Thomas E S Langton BSc(Hons) CBiol FRSB
Anthony Pilny DVM DABVP
Catrina Steedman BSc(Hons) MRSB
Elaine Toland BSc(Hons) MRSB FRSPH
Clifford Warwick PGDipMedSci CBiol CSci EurProBiol FOCAE FRSB


McGowan, J. (2015) Kent v Arun District Council. Neutral Citation Number: [2015] EWHC 2295 (Admin)
PFMA (2015) Pet Food Manufacturers Association. Pet Population 2015

Toland, E., Warwick, C., & Arena, P.C. (2012) The exotic pet trade: pet hate. The Biologist 59(3);14-18.

No competing interests declared.