Herpetology Notes, volume 14: 803-808 (2021) (published online on 26 May 2021) Photo identification of individual Blue Mountains Tree Frogs (Litoria citropa) Jordann Crawford-Ash1,2,* and Jodi J.L. Rowley1,2 Abstract. We used a Photo Identification Method (PIM) to identify individuals of the Blue Mountains Tree Frog, Litoria citropa. By matching the body markings on photographs taken in the field of the lateral, dorsal and anterior views of the frog, we were able to re-match two individuals; one after 88 days and the other after 45 days. We present the first evidence that photo identification is likely to be a useful tool in individual recognition of L. citropa and is one of a few studies on the use of PIM in Australian frogs. Keywords. Body patterning, Australian frogs, non-invasive, mark recapture, individual recognition, PIM Introduction and utilised across multiple study periods (Auger-Méthé and Whitehead, 2007; Knox et al., 2013; Marlow et al., In a time of unprecedented rates of biodiversity decline, 2016). PIM has been particularly successful in studies of a good understanding of age structures, habitat use, cetaceans and mammals (Würsig and Jefferson, 1990; and population fluctuations is necessary for effective Auger-Méthé and Whitehead, 2007; Urian et al., 2015) ecological and conservation management strategies and is increasing in use for reptiles and amphibians (Phillott et al., 2007; Kenyon et al., 2009; Marlow et world-wide (Bradfield, 2004; Knox et al., 2013; Šukalo al., 2016). This information is often obtained through various sampling and survey methods such as mark- et al., 2013; Sacchi et al., 2016). Although PIM is not recapture, where individuals are identified and marked a novel method for amphibian species, its utility in in some way, then released and recaptured at a later time. individual identification varies between species, and Techniques such as toe-clipping or Passive Integrated there have been very few studies utilising this method Transponder (PIT) tags are commonly used for marking in the highly species-diverse tree frog genus Litoria, amphibian species, but are invasive, costly and may not despite many having suitably distinctive body patterning be desirable to use with threatened species (Kriger et al., (but see: Clemas et al., 2009; Kenyon et al., 2009). 2006; Knox et al., 2013; Urian et al., 2015). We tested PIM for the first time on the Blue Mountains The Photo Identification Method (PIM) uses unique Tree Frog Litoria citropa (Duméril and Bibron,1841). body markings, patterning or scarring already on Litoria citropa is a medium-sized tree frog species the animal to identify individuals. It is non-invasive, that has declined in parts of its range in southeastern less time consuming in the field and inexpensive in Australia, is the focus of active research (Crawford-Ash comparison to other methods such as PIT-tagging, and and Rowley, 2021), and has distinctive body patterns, requires minimal equipment and/or training, with the making it a suitable candidate for testing the utility of added advantage of photographs being able to be kept PIM. Materials and Methods Our study site was at Dharawal National Park in 1 Australian Museum Research Institute, Australian Museum, southern Sydney, New South Wales, Australia. At two 1 William St, Sydney 2010, New South Wales, Australia. 2 Centre for Ecosystem Science, School of Biological, Earth and sites, we carried out visual encounter surveys on two Environmental Sciences, University of New South Wales, 100 metre transects every 5-6 weeks over a period of Sydney 2052, New South Wales, Australia. five months (October 2018 – April 2019) for a total * Corresponding author. E-mail: of ten surveys, or five surveys per transect. Each L. [email protected] citropa captured was gently held against a grey piece © 2021 by Herpetology Notes. Open Access by CC BY-NC-ND 4.0. of laminated card, and photographs were taken showing 804 Jordann Crawford-Ash & Jodi J.L. Rowley lateral and dorsal and anterior views of the individual. (Fig. 1). Due to the short-term nature of the study and The frog was then returned to its site of capture. permit conditions, we did not mark individuals during To match photographs, we used both manual this project using alternative methods, so we are unable and automatic pattern recognition methods. First, to determine the proportion of false negatives, however, photographs of L. citropa (Fig. 1) were manually we obtained identical results using pattern matching reviewed by four individuals, including two naive software (Hotspotter). observers (those that had not participated in the fieldwork) to avoid recognition of handled individuals Discussion or similar bias (Marlow et al., 2016). The black patterns Although time consuming when processing on the heads of the frogs were matched first, with photographs for matching both manually and via pattern any successful matches reconfirmed against all other recognition software (approximately 5 hours for manual body markings (Fig. 2). We then ran our anterior-view identification, and 20 hours for preparing and processing photographs through a pattern recognition software, for the pattern recognition software), we found that PIM Hotspotter (Crall et al., 2013) to confirm our matches. was efficient to complete in the field, taking less time This was completed by creating a directory for the head than more traditional mark recapture methods, likely markings of the frogs and then defining the region of causing less stress and disturbance to the frog. It was interest (ROI) of the photograph, referred to as a “chip” also an easily transferable task between team members, in the HotSpotter software. This software then compares without the need for prior training or technical skills. each photograph and a selected chip to all other chips PIM does not require heavy or expensive equipment, in the directory and ranks the chips from most similar with our study completed on a compact camera, so is to least, with an accompanying similarity score. We additionally suitable for remote fieldwork. established a similarity score threshold of 5000 for PIM provides a permanent record of the individuals this study (Sacchi et al. 2016). We took the top ranked body condition and physical appearance over time, matches (Rank 1) above our similarity threshold, and which could be invaluable if there are changes to other then reconfirmed the matches visually. identification methods such as damage to toes after toe clipping due to predators or injury (ARAV, 2009). This Results technique may also offer an identification method for In total, 47 individuals were captured during the study juvenile frogs or small adult species that may be too (38 at site 1 and 9 at site 2), resulting in 188 photographs small for other mark recapture methods (Morrison et al., (4 per individual; lateral, dorsal and anterior). Two 2016), however further research would be required. individuals from site 1 were identified as recaptures A potential downfall in PIM use is the current lack of using manual recognition and using HotSpotter. data on the integrity of body patterning in frogs over Through our manual matching, we visually confirmed time. We saw no noticeable changes in the expression that one adult male (JR13-1 or Individual 1) was found of body patterning in our study, however this is not 88 days after the initial capture, and the other adult conclusive due our sample size and the short-term male (JCA5-2-1 or Individual 2) after 45 days (Fig. 2). nature of our study. In the few studies that have been Distinctive patterns that were matched were the nasal conducted on the distinguishability of body patterns stripes, upper lip speckling, and spots on the flanks and over time in Australian Litoria species, the results have top of the thighs (Fig. 1–2). Hotspotter ranked JR13- been variable. Litoria genimaculata juveniles remained 1 (Individual 1) most similar to JCA5-3-1 with a score distinguishable as they developed to an adult (Kenyon of 39708.7 when matching the facial markings photos. et al., 2010) whereas, in the same frog species it was The next highest score for facial markings was between found that dorsal patterns had changed within a 2 month JCA5-2-1 (Individual 2), and JCA 4-3-1 with a score period as adults (Kenyon et al., 2009). Colour change of 9006.1, these two scores were dramatically higher has additionally been found to conditionally affect the than all other comparisons (1677.2-13747.8). Both expression of body patterns (Withers, 1995; Kang et matches were then confirmed once again with their al., 2016), and therefore the suitability of using photo flank patterning, with Individual 1 scoring 32288.7, identification methods should remain species-specific, and Individual 2 scoring 22264.9. These were again the or used in addition to other methods until the integrity highest ranking in similarity (3579.4-14158.9). All other of the body patterning is further investigated. comparisons did not have visually similar patterning Despite these limitations, we present preliminary Photo identification of individual Blue Mountains Tree Frogs 805 Figure 1. Examples of non-matches and the variation in the unique body patterning on the head and right flank of the Blue Mountains Tree frog, Litoria citropa. Photos by Jordann Crawford-Ash. 806 Jordann Crawford-Ash & Jodi J.L. Rowley Figure 2. Two adult male Litoria citropa individuals manually and through pattern recognition software re-matched 88 days after initial capture for individual 1, and 45 days after initial capture for individual 2. Photos by Jordann Crawford-Ash. Photo identification of individual Blue Mountains Tree Frogs 807 evidence that photo identification is likely to be a useful dynamics in three co-occurring frog species of southern Sydney, tool in individual recognition of L.