Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Can post-capture photographic identification as a wildlife marking technique be undermined by observer error? A case study using King Cobras in northeast Thailand


Identifying individuals with natural markings is increasing in popularity to non-invasively support population studies. However, applying natural variation among individuals requires careful evaluation among target species, snakes for example have little validation of such methods. Here we introduce a mark-free identification method for King Cobras (Ophiophagus hannah) from the Sakaerat Biosphere Reserve, in northeast Thailand using both subcaudal scale pholidosis (scale arrangement and number) and unique ventral body markings to distinguish individuals. This project aims to evaluate the impact of observer error on individual identification. Observers of varying expertise, will distinguish between King Cobra individuals using identifying photographs from a previous study. We will ask randomly assigned observers to distinguish individuals via: 1) subcaudal pholidosis, 2) ventral body markings, and 3) combination of both measures. Using Bayesian logistic regression, we will assess the probability observers correctly distinguish individuals. Based on exploratory observations, we hypothesise that there will be a high probability of correct identifications using subcaudal pholidosis and ventral body markings. We aim to stimulate other studies implementing identification techniques for scrutinous assessment of such methods, in order to avoid subsequent errors during long-term population studies.


Correctly identifying individuals within populations is a fundamental assumption for a number of quantitative population analyses. With confident identification we can estimate abundance [1], density [2], behaviour [3], growth rates [4] and survivorship [5, 6]; thus generating the required information for both management and conservation actions.

Individual marking methods vary and depend upon the focal study species of a study. Non-invasive marking techniques include: neck collars [7], bands/rings [810] and external colour marks [11, 12]. Invasive techniques include tags [13], branding [14, 15], transponders [16, 17] and tissue removal [18, 19]. See Silvy et al. [20] for a comprehensive review of available marking methods. A fundamental assumption for many mark-recapture studies is that the mark does not influence survival or behaviour in marked individuals. Invasive marks can indeed be detrimental to long term survival in some species [21].

Despite various animal marking techniques, comparatively few options exist for snakes. Researchers commonly use techniques that cause some degree (though usually minor) of individual bodily harm. Such harm can raise ethical considerations [15, 22, 23]. Methods tailored to snakes include systematic scale clipping [2427] or branding with cauterising irons [15, 28, 29]. More recently, the ability for researchers to subcutaneously implant passive integrated transponder (PIT) tags efficiently and cheaply, have further increased capacity to identify individuals reliably [17, 30]. Major et al. [31] proposed the novel marking technique of visible implant elastomer (VIE) which can substantially aid in monitoring snake populations long-term, the method is particularly amenable to juvenile and small snake species.

Identification methods using solely morphological or phenotypic traits are increasingly sought after to overcome: the logistical costs of capturing study animals [32], the necessity to identify individuals via camera-trap images [33, 34], and avoiding legislation/ethics restrictions on marking animals [35]. Tiger research studies leverage the natural variation from contrasting black stripes to identify individuals [33, 34]. Other large felines, also have variable characteristics suitable for individual identification, such as bobcats (Lynx rufus) [36] and Serengeti cheetahs (Acinonyx jubatus) [37]; and alternative methods (variation in tail appearance) have been extremely successful in non-felines such as badgers (Meles meles) [38].

Marine studies have relied upon photographic identification tools for decades. Humpback whales (Megaptera novaeangliae), manatees (Trichechus manatus), and nurse sharks (Ginglymostoma cirratum) all demonstrate sufficient variation for individual photographic identification [5, 39, 40]. Meekan et al. [1] used pattern variation and scarring to identify individual whale sharks (Rhincodon typus), even a decade after initial sightings.

Attempts to introduce snake focused photographic identification technique however, have been limited. Head patch patterning can be a reliable character for identifying individuals in wild Mangshan pitvipers (Protobothrops mangshanensis) [41]. Dyugmedzhiev et al. [42] showed that nose-horned vipers (Vipera ammodytes) can be reliably identified using horn scale morphology. While, Bauwens et al. [43] photographed over 3200 individual European adders (Vipera berus) and showed that sufficient variation exists even within a large population. Furthermore, Carlström and Edelstam [44] showed that the black and white patterning on the ventral scales of a Swedish population of grass snakes (Natrix natrix), could be used to monitor individuals throughout their lifespans, which has led to further support and investigations into N. natrix and N. maura [4547].

The King Cobra (Ophiophagus hannah), although a large and easily recognisable snake in the field, does not exhibit much variability in phenotypic colour displays. Unless obvious scarring or deformities are present, these snakes are otherwise visually indistinguishable. However, we have now produced an identification method using subcaudal pholidosis and ventral body markings to distinguish between individuals from photographs. Our new method requires only basic photography skills and removes the need for typical invasive identification methods which usually require expertise and permits/licenses to perform.

Although photographic identification methods are increasing in their use, during long-term population monitoring [48], the ability for other researchers to accurately adopt such methods lacks information. Failing to account for observer error when using novel identification techniques can completely undermine the technique altogether leading to incorrect population estimates. Johansson et al. [49] evaluated observer for assessing individual snow leopards (Panthera uncia) during photographic surveys with 12.5% erroneous photograph classifications. The investigation by Johansson et al. [49] suggests that during wildlife camera surveys, or other photographic identification methods, observer error can compromise study findings.

Herein, we propose a study to assess the sensitivity of our novel identification method to observer error. Specifically, we aim to evaluate if observer error is too high that it may undermine our method in being applied to future King Cobra population studies, or potentially extrapolated to other snake species with similar morphological and phenotypical traits.


King Cobra samples

We discovered King Cobras as part of a long-term radiotelemetry study in northeast Thailand using public notifications, active trapping, active surveying and opportunistic captures between 2013 and 2020. Upon capture, we anaesthetised snakes to reduce stress, accurately record biometrics, sex individuals, collect samples and comprehensively photograph study animals. Further information regarding the location, capture and subsequent data collection can be found at Marshall et al. [6, 29, 50]. Photographs from prior capture events will be the main data for this proposed study; however, we only applied basic photography skills to record scalation and body patterning of individuals, and we were not able to standardise lighting conditions and positioning of animals (simulating actual lab conditions in most locations where King Cobra studies might occur).


We identified King Cobras captured between 2013 and 2018 (individuals 001–053), using systematic branding following the protocols of Winne et al. [15], and subsequently switched to using PIT-tags (individual 054 onwards); therefore, ensuring correct individual identification facilitating our identification method.

Snake species can have distinguishing subcaudal scale arrangements that is either described as divided or undivided. However, King Cobras in northeast Thailand can possess both divided and undivided scales [51]. Via exploratory analysis, we are confident that the arrangement of divided and undivided scales, and total number of subcaudal scales, is unique to individual King Cobras and can be used for identification, much like a fingerprint.

We preliminarily compared two individual King Cobras and highlighted the observed subcaudal scale arrangement using the software Inkscape (Fig 1), which gives an example of the differences between individuals.

Fig 1. King Cobra subcaudal patterning.

Subcaudal scale arrangements observed in two King Cobra individuals. Teal: ventral scales, light blue: anal plate, dark blue: undivided subcaudal scales and orange: divided subcaudal scales. Created using Inkscape.

We also investigated whether ventral body markings could further aid identification. King Cobras within the Sakaerat Biosphere Reserve (SBR) typically have yellow/orange colouration on the ventral portion of their hoods, often covered in black or grey peaks on some of the ventral scales. The ventral patterning differs sufficiently to distinguish individuals, highlighted in Fig 2; again, using Inkscape.

Fig 2. King Cobra hood patterning.

Ventral hood patterning observed in four King Cobra individuals. Light grey: outline of head, dark grey: dark colouration observed on hood and yellow: yellow/orange colouration observed on hood. Created using Inkscape.

We used photographs taken during each King Cobra capture and measurement to compare scalation and body patterning of each known individual. Photographs of each snake’s head, body and tail were examined visually, and subcaudal scales counted manually (MDJ). We repeated all counts twice for each individual, adding a third count if the first two counts disagreed (MDJ). From the photographs, we created a unique coding system for each individual based on the transitions from undivided to divided subcaudal scales, similar (though simplified) to a formula suggested by Shine et al. [52]. For example, an individual which has five undivided subcaudal scales, followed by three divided subcaudals, one more row of undivided scales and the remaining 80 scales are divided, would have a code of 5:3:1:80. However, we always start a count with the number of undivided subcaudal scales (the most common arrangement); therefore, if an individual has two divided scales, followed by six undivided and the remaining 90 are divided, the code would be 0:2:6:90.

We currently have a sample size of 81 King Cobras, comprised of 30 adult males, 23 juvenile males, 12 adult females, 4 juvenile females and 12 neonates collected from a single nest. However, research is ongoing and the sample size will likely increase; though we will cease collection and measurement of King Cobras on 01 September 2020. After examining these 81 individuals, we can distinguish all individuals using subcaudal scales alone; and observing ventral body markings when subcaudal counts match (which has occurred on one occasion). We have also compared the subcaudal pholidosis from 16 recaptured individuals (ranging from one to five recaptures) to assess any changes over time. We only observed a change in the total subcaudal scales in one individual (individual 018), resulting from damage to the tip of the tail. Despite this change, we were still able to distinguish this individual due to the observed transition from divided to undivided subcaudal scales. Individual markings can remain the same for at least six years (individual 007), as both subcaudal scale number/arrangement and ventral markings for this individual remained consistent between the two capture events.

As we captured study animals for a radiotelemetry study, we released all individuals back at their capture-site following data collection. We also discovered dead individuals opportunistically across our study site, due to mortality rates [6], which we mostly returned back to discovery site following data collection; however, we stored some individuals within 95% ethanol, currently maintained at the Sakaerat Environmental Research Station, northeast Thailand.

Assessing identification errors

To validate these measures as a potential identification method, we aim to confirm other observers—besides the authors—can accurately distinguish between individuals. We will create projects on the citizen science platform Zooniverse ( requesting the participation of both professional and non-professional citizens to identify individuals within subsets of our available photographs. We will maintain the availability of the project for six months, however, we will increase the duration of the project if a minimum sample of 500 participants, for each sub-project is not reached.

Prior to identifying individuals using our photographs, the site will ask each observer to specify their experience level in snake identification and/or distinguishing individuals within a population. Observers will then self-assign to one of three categories, based on the following criteria: beginner, the observer has no experience identifying snakes or individuals within a population; experienced, the observer has some applicable skill in identifying snakes/individuals within a population; expert, the observer has considerable experience identifying snake species based on morphological characteristics and/or has considerable experience identifying animals to the individual level during population studies.

We will design three projects for volunteers to attempt: 1) Subcaudal scale arrangement, 2) ventral body markings, and 3) a combination of both. Each project will have a tutorial explaining how observers should approach each task, before attempting the actual sets of photos. Each project will contain 50 sets of four images (resulting in 200 photos per project), and each set will contain between one and four King Cobra individuals. Observers will be asked to participate independently and only attempt each set within the project once. Correct answers will only be known to the authors and will not be shared with any participants, even after answers have been submitted. Following the six-month project period, we will share answers using the Talk section of the sub-projects, for interested participants.

To remove any influence of the background features of a photograph impacting identification, each snake will be presented on a blank background (Fig 3). Observers will be asked to independently identify how many King Cobras are represented by the four photographs (minimum one, maximum four), also highlighting which photographs represent the same individuals (if any). Observers will also qualitatively rate the average photo quality within the capture event, using the following criteria: Good, photos allowed for easy identification of individual subcaudal scales and/or ventral body markings; Acceptable, photos allowed for identification of individual subcaudal scales and/or ventral body markings but required careful examination to discern individual scales and/or distinct patterns; Poor, photo quality greatly hinder subcaudal scale identification and/or observing ventral body markings and in places did not allow for individual scales to be identified and/or ventral body markings to be observed. A preliminary outline of project workflow can be seen in S1 Fig.

Fig 3. King Cobra patterning comparison.

Subcaudal scales and ventral body markings of two King Cobra individuals, displayed on a clear background.


The response variable from observer identification will be binary, and will exhibit a bernoulli distribution. These will classify as follows: ‘correct’, observer identified the correct number of individuals in an event; or ‘incorrect’, the observer identified the incorrect number of individuals in an event. We will classify observers’ ability to distinguish which individuals are the same using the same binary classification, i.e. ‘correct’: observers identified which individuals were the same and ‘incorrect’: observers were not able to identify which individuals were the same.

Using a Bayesian logistic regression model, we will address the two main questions of this study, 1) to what extent can observers identify the correct number of individuals in the capture events? and 2) to what extent can observers identify which individuals within a capture events are the same? We have opted to use a Bayesian logistic regression, as this approach treats model parameters as random variables, which differs to a frequentist approach which considers model parameters as fixed or unknown quantities, and also uses probability to model uncertainty [53]. Furthermore, we can also include prior information derived from literature review to inform the models. Using Bayes’ theorem, a probability distribution will be estimated for our parameters using prior information and observed data [54]. Despite Johansson et al’s. [49] study there are few available studies investigating photographic identification observer error (none for snakes), we will therefore be using uninformative priors for the Bayesian regression model (see proposed R code line 55 to 59, S1 File). The first Bayesian logistic regression model will investigate the observer proportion with correct number of individuals in the event. The second model will investigate the proportion of observers correct in showing which individuals were the same in each capture event. We will include three population effects: level of experience, true number of individuals in the capture event, and observer classification of photo quality. We will also include the photo set ID, and observer ID as a group effect.

To address our hypotheses 1) and 2), we will consider any error rate greater than 5% to undermine that particular aspect of the identification method (see proposed R code line 71 to 91 for details, S1 File). In order to validate this identification method for application in future studies, we would need to achieve a probability of observers identifying the number of individuals, and identifying which individuals are the same, of 95% or greater (see proposed R code line 71 to 139 for appropriate outputs, S1 File). Therefore, if either of the Bayesian logistic models show less than a 95% posterior probability point estimate, we will conclude that observer error does indeed undermine the photographic identification method. This is due to both the need to identify unique individuals within a population (hypothesis 1), and subsequently identify when a unique individual is recaptured (hypothesis 2).

All analyses will be implemented in R v.3.6.3 [55] and R Studio v.1.3 [56]. Bayesian logistic regression models will be performed using the brms [57, 58] package. Data will be manipulated and managed using the scales [59], tidybayes [60], reshape2 [61], and dplyr [62] packages, and subsequent visualisations of data will be created using the ggplot2 [63] and ggridges [64] packages. All package versions will be supplied following analyses.

Ethics approval and consent to participate

We had ethical approval for the prior data collection from the Suranaree University of Technology Ethics Committee (24/2560), with appropriate Institute of Animals for Scientific Purpose Development (IAD) licences held by C.T.S. Research commenced under the permission by the National Park, Wildlife and Plant Conservation Department, Thailand and the National Research Council of Thailand (98/59). Furthermore, we also had permission to conduct research from Thailand Institute of Scientific and Technological Research and Sakaerat Environmental Research Station. Data supplied by citizens via the platform, can be used, modified and redistributed according to the Zooniverse User Agreement and Privacy Policy (


We aim to assess the further applicability of the King Cobra identification method, by mitigating observer error induced uncertainty, having a baseline prior measure will allow observers to incorporate uncertainty into models.

Ferner and Plummer [65] suggest criteria for selecting suitable marking techniques during study design:

  1. “Marks should not affect the survivorship or behaviour of the organism.
  2. Marks should allow the animal to be as free from stress and pain as possible.
  3. Marks should identify the animal as a particular individual or a member of a cohort if desirable.
  4. Marks should last indefinitely or at least through the duration of the study.
  5. Marks should be easily read and/or observable by all informed individuals.
  6. Marks should be adaptable to organisms of different sizes.
  7. Marks should be easy to use in both laboratory and field, and use easily obtained material at minimal cost.
  8. Marks should be tested to meet these listed criteria before being put into wide-spread use.
  9. Marks should prevent marking application tools from being used without first being thoroughly disinfected and cleaned.”

By implementing the identification for King Cobras in the Sakaerat Biosphere Reserve, we can satisfy all of the suggested criteria by Ferner and Plummer [65]. Despite the need for snakes to be captured for subsequent identification, 1) and 2) are achieved through the lack of mutilation or physical tags; however, the need for only a few distinguishing photographs can substantially decrease handling times, which is specifically important when working with large venomous snakes. We aim to satisfy 3) and 4) by showing that natural marks are unique for each sampled individual in our subset of the population, and recaptures will help fortify this method for long-term population studies. If snakes are required to be captured as part of an ongoing study, 5), 6) and 7) are also achievable as subcaudal scales counts are easy to perform, researchers only need basic skills to photograph appropriate areas of the snake species of interest and standard photography equipment permits the photographing of very small subjects. The purpose of this proposed study is to determine if the methodology is a reliable method for identification and can satisfy the abovementioned criteria (9). Lastly, this methodology allows us to identify individuals without the need for any specialised marking equipment, therefore reducing the need for equipment sterilisation within a field or laboratory setting, outside of standard sterile working environments (10).

The potential to use subcaudal scale anomalies was first suggested by Shine et al. [52], where the authors investigated the subcaudal patterning of 53 blacksnakes (Pseudechis porphyriacus) and 115 water pythons (Liasis fuscus). Their study showed 41 (77%) P. porphyriacus and 56 (49%) L. fuscus exhibited unique subcaudal patterning. Shine et al. [52] concluded to say that “subcaudal scale formulae may significantly improve the investigator’s ability to recognize specific animals on recapture”, however, we are yet to find any further studies which have attempted to adopt this methodology.

We hope that the results of this study can be implemented in other photographic identification studies, particularly those focusing on snake population monitoring. This may help to expand the scope of studies which are limited by expertise or licensing to perform the currently accepted marking techniques; namely branding, scale clipping and PIT-tagging. For example, another long-term King Cobra study is being conducted within the Western Ghats, India. In Shankar et al. [35], the authors report on some of the findings of their investigation, specifically involving the capture and biometric data of King Cobras. A quote from their paper states “We could not permanently mark released snakes because we were unable to obtain permission from the Forest Department”. This statement alone supports the need for alternative methods of identification for not only King Cobras, but other snake species where permissions and licensing implement road blocks in ongoing research efforts. In the investigation by Shankar et al. [35], snakes were required to be captured and moved as part of a conservation initiative lead by the authors to safely remove unwanted snakes from people’s homes. Therefore, since snakes were already being manipulated as part of this service, our proposed methodology could have allowed the researchers to take even the most basic of photographs of the King Cobras for subsequent identification. This could have broadened the scope of their research to investigate if individuals would move back towards removal sites during subsequent re-captures, or even aided towards population estimates and detectability using villager notations; to name a few examples.

We are confident that the proposed methodology can be used as a reliable tool for identifying individuals within our population of interest. However, through the use of external observers, we can evaluate if the wider applicability of our method is undermined by observer error, resulting in misidentifications and subsequent errors in population estimates [49]. We hope to inspire other studies investigating novel identification techniques to assess the resulting observer error before distributing their techniques to the wider scientific community; leading to a more rigorously tested scientific standard for identification methods for wildlife.

If we identify observer error to be too great that it ultimately invalidates our identification strategy, we propose that methodology outlined in Sreekar et al. [66] could be used. Specifically, we could use the Interactive Individual Identification System (I3S Manta), to evaluate if hood patterning can supplementary, or additionally, be used to identify individuals, through a computed measure of similarity, removing the bias of human error from observations.

Supporting information

S1 Fig. Zooniverse workflow example.

A preliminary design for the subcaudal scale arrangement project workflow as seen by Zooniverse volunteers.


S1 File. Simulated Bayesian logistic regression model.



  1. 1. Meekan MG, Bradshaw CJA, Press M, McLean C, Richards A, Quasnichka S, et al. Population size and structure of whale sharks Rhincodon typus at Ningaloo Reef, Western Australia. Mar Ecol Prog Ser. 2006;319:275–285.
  2. 2. Efford M, Dawson DK, Borchers D. Population density estimated from location of individuals on a passive detector array, Ecology. 2009;90(10):2672–2682.
  3. 3. Merrick MJ, Koprowski JL. Should we consider individual behavior differences in applied wildlife conservation studies? Biol Cons. 2017;209:34–44.
  4. 4. Madsen T, Shine R. Silver spoons and snake body sizes: prey availability early in life influences long-term growth rates of free-ranging pythons. J Ani Ecol. 2000;69(6):952–958.
  5. 5. Chaloupka M, Osmond M, Kaufman G. Estimating seasonal abundance trends and survival probabilities of humpback whales in Hervey Bay (East coast of Australia), Mar Ecol Prog Ser. 1999;184:291–301.
  6. 6. Marshall BM, Strine CT, Jones MD, Theodorou A, Amber E, Waengsothorn S, et al. Hits close to home: repeated persecution of king cobra (Ophiophagus hannah) in northeastern Thailand. Trop Cons Sci. 2018;11:194008291881840.
  7. 7. Senar JC, Carrillo-Ortiz J, Arroyo L. Numbered neck collars for long-distance identification of parakeets. J Field Orn. 2012;83(2):180–185.
  8. 8. Koronkiewicz TJ, Paxton EH, Sogge MK. A technique to produce aluminum color bands for avian research. J Field Orn. 2005;76:94–97.
  9. 9. Pierce AJ, Stevens DK, Mulder R, Salewski V. Plastic colour rings and the incidence of leg injury in flycatchers (Muscicapidae, Monarchidae). Ring Migr. 2007;23:205–210.
  10. 10. Griesser M, Schneider NA, Collis M-A, Overs A, Guppy M, Guppy S, et al. Causes of ring-related leg injuries in birds–evidence and recommendations from four field studies. PLoS ONE. 2012;7(12):e51891. pmid:23300574
  11. 11. Baird P, Robinette D, Hink SA. A remote marking device and newly developed permanent dyes for wildlife research. Wild Soc Bull. 2017;41(4):
  12. 12. Lü H, Fu M, Zhang Z, Su S, Yao W. Marking fish with fluorochrome dyes, Rev Fish Sci Aqua. 2020;28:117–135.
  13. 13. McFarlane GA, Wydoski RS, Prince ED. External tags and marks. Amer Fish Soc Sym. 1990;7:9–29.
  14. 14. McMahon CR, Burton HR, Van Den Hoff J, Woods R, Bradshaw CJA. Assessing hot-iron and cryo-branding for permanently marking southern elephant seals. J Wild Man. 2006b;70(5):1484–1489.
  15. 15. Winne CT, Willson JD, Andrews KM, Reed RN. Efficacy of marking snakes with disposable medical cautery units. Herp Rev. 2006;37(1):52–54.
  16. 16. Gibbons JW, Andrews KM. PIT tagging: simple technology at its best. BioScience. 2004;54:447–454.
  17. 17. Burger J, Zappalorti R. Effects of handling, marking, and recapturing pine snakes (Pituophis m. melanoleucus) from the New Jersey Pine Barrens. J Environ Indic. 2011;6:17–32.
  18. 18. Schmidt K, Schwarzkopf L. Visible implant elastomer tagging and toe-clipping: effects of marking on locomotor performance of frogs and skinks. The Herp J. 2010;20(2):99–105.
  19. 19. Grafe TU, Stewart MM, Lampert KP, Rödel M-O. Putting toe clipping into perspective: a viable method for marking anurans. J Herp. 2011;45(1):28–35.
  20. 20. Silvy NJ, Lopez RR, Peterson MJ. Wildlife marking techniques. In Silvy NJ, editors. The Wildlife Techniques Manual, 7th ed. Baltimore, MD: John Hopkins University Press; 2012. p. 339–376.
  21. 21. Guimarães M, Corrêa DT, Filho SS, Oliveira TAL, Doherty PF Jr, Sawaya RJ. One step forward: contrasting the effects of toe clipping and PIT tagging on frog survival and recapture probability. Ecol Evol. 2014;4(8):1480–1490.
  22. 22. McMahon CR, Bradshaw CJA, Hays GC. Branding can be justified in vital conservation research, Nature. 2006a;439:392–
  23. 23. Wilson RP, McMahon CR. Measuring devices on wild animals: what constitutes acceptable practice? Front Ecol Environ. 2006;4:147–154.
  24. 24. Brown WS, Parker WS. A ventral scale clipping system for permanently marking snakes (Reptilia, Serpentes). J Herp. 1976;10(3):247–249.
  25. 25. Fabien A, Bonnet X, Maumelat S, Bradshaw D, Schwaner T. Diet divergence, jaw size and scale counts in two neighbouring populations of tiger snakes (Notechis scutatus). Amphibia-Reptilia. 2004;25:9–17.
  26. 26. Bell SLM, Herman TB, Wassersug RJ. Ecology of Thamnophis sauritus (Eastern ribbon snake) at the northern limit of its range. Northeastern Nat. 2007;14(2):279–292.
  27. 27. Dubey S, Brown GP, Madsen T, Shine R. Male-biased dispersal in a tropical Australian snake (Stegonotus cucullatus, Colubridae). Mol Ecol. 2008;17:3506–3514.
  28. 28. Madsen T, Shine R. Conflicting conclusions from long-term versus short-term studies on growth and reproduction of a tropical snake. Herpetologica. 2001;57(2):147–156.
  29. 29. Marshall BM, Strine CT, Jones MD, Artchawakom T, Silva I, Suwanwaree P, et al. Space fit for a king: spatial ecology of king cobras (Ophiophagus hannah) in Sakaerat Biosphere Reserve, northeastern Thailand. Amphibia-Reptilia. 2019;40:163–178.
  30. 30. Chevalier HL, Calvez O, Martínez-Silvestre A, Picard D, Guérin S, Isselin-Nondedeu F, et al. Marking techniques in the marbled newt (Triturus marmoratus): PIT-tag and tracking device implant protocol. Acta Herp. 2017;12(1):79–88.
  31. 31. Major T, Alkins DR, Jeffrey L, Wüster W. Marking the un-markable: visible implant elastomer in wild juvenile snakes. Herp J. 2020;30:173–176.
  32. 32. Speed CW, Meekan MG, Bradshaw CJA. Spot the match–wildlife photo-identification using information theory. Fron Zool. 2007;4(2). pmid:17227581
  33. 33. Karanth KU. Estimating tiger population from camera-trap data using capture-recapture models. Biol Cons. 1995;71:333–338.
  34. 34. Karanth KU, Nichols JD. Estimation of tiger densities in India using photgraphic captures and recaptures. Ecology. 1998;79(8):2852–2862.
  35. 35. Shankar PG, Ganesh SR, Whitaker R, Prashanth P. King cobra Ophiophagus hannah (Cantor, 1836) encounters in human-modified rainforests of the Western Ghats, India. Hamadryad. 2013;36(2):62–68.
  36. 36. Heilbrun RD, Silvy NJ, Tewes ME, Peterson MJ. Using automatically triggered camera to individually identify bobcats. Wild Soc Bull. 2003;31(3):748–755.
  37. 37. Kelly MJ. Computer-aided photograph matching in studies using individual identification: an example from Serengeti cheetahs. J Mamm. 2001;82(2):440–449.
  38. 38. Dixon DR. A non-invasive technique for identifying individual badgers Meles meles. Mamm Rev. 2003;33(1):92–94.
  39. 39. Kendall WL, Hines JE, Nichols JD. Adjusting multistate capture-recapture models for misclassification bias: manatee breeding proportions. Ecology. 2003;84:1058–1066.
  40. 40. Castro ALF, Rosa RS. Use of natural marks on population-estimates of the nurse shark, Ginglymostoma cirratum, at Atol das Rocas Biological Reserve, Brazil. Environ Biol Fishes. 2005;72:213–221.
  41. 41. Yang D, Chen S, Chen Y, Yan Y. Using head patch pattern as a reliable biometric character for noninvasive individual recognition of an endangered pitviper Protobothrops mangshanensis. Asian Herp Res. 2013;4(2):134–139.
  42. 42. Dyugmedzhiev A, Tzankov N, Natchev N, Naumov B. A non-traumatic multi-operational method for individual documentation and identification of nose-horned vipers (Vipera ammodytes (Linnaeus, 1758) (Squamata, Viperidae)) allows reliable recognition of recaptured specimens. Biharean Biol. 2018;12(2):92–96.
  43. 43. Bauwens D, Claus K, Mergeay J. Genotyping validates photo-identification by the head scale pattern in a large population of the European adder (Vipera berus). Ecol Evol. pmid:29531711
  44. 44. Carlström D and Edelstam C. Methods of marking reptile for identification after recapture. Nature. 1946;158:748–749.
  45. 45. Hailey A and Davies PMC. ‘Fingerprinting’ snakes: a digital system applied to a population of Natrix maura. J. Zool. Lond. (A). 1985;207:191–199.
  46. 46. Zuiderwijk A and Wolterman R. Tellen en fotograferen van ringslangen bij Amsterdam. De Levende Natuur. 1995;96:72–81.
  47. 47. van Roon J, Dicke I, Brinks R, Zuiderwijk A and Janssen I. Capture and recapture of grass snakes near Amsterdam. Proceedings of the 13th COngress of the Societas Europea Herpetologica. 2006;191–192.
  48. 48. Gore MA, Frey PH, Ormond RF, Allan H, Gikes G. Use of photo-identification and mark-recapture methodology to assess basking shark (Cetorhinus maximus) populations. PLoS ONE. 2016;11(3):e0150160.
  49. 49. Johansson Ö, Samelius G, Wikberg E, Chapron G, Mishra C, Low M. Identification errors in camera-trap studies result in systematic population overestimation. Sci Rep. 2020;10:6393. pmid:32286438
  50. 50. Marshall BM, Crane M, Silva I, Strine CT, Jones MD, Hodges CW, et al. No room to roam: king cobras reduce movement in agriculture. Movement Ecology. 2020;8(33).
  51. 51. Cox MJ, Hoover MF, Chanhome L, Thirakhupt K. The snakes of Thailand, Chulalongkorn University of Natural History, Thailand; 2012
  52. 52. Shine C, Shine N, Shine R, Slip D. Use of subcaudal scale anomalies as an aid in recognizing individual snakes. Herp Rev. 1988;19(4):79–80.
  53. 53. Gelman A, Carlin JB, Stern HS, Dunson DB, Vehtari A, Rubin DB. Bayesian data analysis, third edition. CRS Press; 2013.
  54. 54. Nalborczyk L, Batailler C, Lœvenbruck H, Vilain A, Bürkner P-C. An introduction to bayesian mulltilevel models using brms: a case study of gender effects on vowel variability in standard Indonesian. J Spe, Lang, Hear Res. 2019;62(5):1225–1242.
  55. 55. R Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. 2020.
  56. 56. R Studio Team. RStudio: integrated development environment for R. Boston, MA: RStudio, Inc. 2020.
  57. 57. Bürkner P-C. brms: an R package for bayesian multilevel models using stan. J Stat Soft. 2017;80(1):1–28.
  58. 58. Bürkner P-C. Advanced bayesian multilevel modelling with the R package brms. The R Journal. 2018;10(1):395–411.
  59. 59. Wickham H, Seidel D. Scales: scale functions for visualization, R package version 1.1.0. 2019.
  60. 60. Kay M. tidybayes: tidy data and geoms for bayesian models. 2019
  61. 61. Wickham H. Reshaping data with the reshape package, J Stat Sof. 2007;21:1–20.
  62. 62. Wickham H, François R, Henry L, Müller K. dplyr: a grammar of data manipulation. R package version 0.8.5. 2020.
  63. 63. Wickham H. Elegant graphics for data analysis. Springer-verlag New York. 2016.
  64. 64. Wilke CO. ggridges: ridgeline plots in “ggplot2”. 2018.
  65. 65. Ferner JW, Plummer MV. Marking and measuring reptiles. In: Dodd CK, editors. Reptile ecology and conservation. a handbook of techniques. Oxford: Oxford University Press; 2016. p. 45–49.
  66. 66. Sreekar R, Purushotham CB, Saini K, Rao SN, Pelletier S, Chaplod S. Photographic capture-recapture sampling for assessing populations of the Indian gliding lizard Draco dussumieri. PLoS ONE. 2013;8(2):e55935.