Shem Unger, Ze Cline Hull, Lauren Diaz, John D. Groves, Lori A. Williams,Catherine M. Boinof Jahowski

aCarolina Headwaters, L.L.C., Mathews, NC, 28105, United States

bClemson University, Clemson, SC, 29634, United States

cNorth Carolina Zoological Park (Curator Emeritus), Asheboro, NC, 27205, United States

dNorth Carolina Wildlife Resources Commission, Raleigh, NC, 27606, United States

Keywords:

Captive propagation

Hellbender

Aquatic science

Freshwater conservation

Underwater camera systems

ABSTRACT

Captive propagation and translocation are becoming vital components of conservation and management strategies for Eastern hellbender (Cryptobranchus a. alleganiensis) populations. Zoos, aquaria, universities, and state and federal agencies are concomitantly collaborating on the protection, education, and maintenance of captive populations of this unusual, cryptic salamander. Conservation strategies include the use of artificial nesting structures, collection of eggs from the wild, and head-starting individuals in zoos or hatcheries. The effects of these strategies need to be monitored, however, traditional survey and monitoring methods for the species in the wild involves rock-lifting, which has the potential to both harm habitat and alter reproductive behavior.Therefore, there is a need to develop effective, non-invasive and non-destructive methods of monitoring both wild and captive populations of Eastern hellbenders. Herein, we compare two simple, affordable, underwater video and camera systems (borescope and Aqua-Vu cameras) in their ability to 1) facilitate detection of adults under potential cover items and 2) facilitate nest detection and monitoring in both wild and captive environments. Both cameras were successful in detecting individual hellbender presence, albeit with different resolutions and detection times. The borescope was better at accessing deep cavities given its large flexible attachment which allowed for greater flexibility of scanning crevices of adult shelters. However, search time increased and even low levels of suspended sediment reduced visibility with the borescope. The Aqua-Vu camera provided greater overall visibility and faster detection of individuals under both natural and artificial shelters. There was a significant difference in the amount of time required to detect hellbenders with each camera design when searching under natural rocks (borescope: median =67.8 s, Aqua-Vu: median =39.1 s; Kruskal Wallis Test H =15.62, p ＜0.001) and artificial shelters (borescope: median =30.9 s, Aqua-Vu: median =13 s; Kruskal Wallis Test H =25.23, p ＜0.0001). We detected 8 natural nests with actively guarding males and only one individual using a wild artificial shelter. We recommend hellbender researchers utilize a combination of underwater video cameras to suit their specific survey goals in both captive and field settings. Moreover, we recommend zoo staff incorporate these methods to not only monitor captive populations but also to potentially record breeding behavior in zoos and aquariums.

1.Introduction

Zoos provide excellent opportunities for active research and ex-situ conservation of amphibians and reptiles by establishing populations for reintroduction (Gascon et al., 2007; Gratwicke & Murphy, 2017).The Eastern hellbender (

Cryptobranchus a. alleganiensis

) is a large, fully aquatic habitat specialist which inhabits clear cool streams throughout the Southeastern, Northeastern, and Midwestern United States (Petranka, 1998, p. 587). This enigmatic species has undergone dramatic declines throughout its geographic range (Wheeler, Prosen, Mathis, &Wilkinson, 2003). Following the successful captive reproduction of Ozark hellbenders (

C. a. bishop

i) in 2011 (Ettling et al., 2013), the conservation of both Eastern and Ozark hellbenders has received increased attention by zoos and aquariums. A vital component of captive rearing and habitat augmentation efforts by zoos typically involves using a combination of natural and artificial shelters (Briggler &Ackerson, 2012). Increasingly, conservation agencies are collaborating with academic institutions, zoos, aquariums, and hatcheries to develop captive-propagation programs for the hellbender.

Presently, there are over thirty zoos and aquariums across fourteen states actively involved in hellbender conservation, research, and captive-rearing efforts (Burgmeier et al., 2017). In many cases, zoos and aquariums are actively engaged in either establishing a captive-breeding program for hellbenders or head-starting individuals that hatch from eggs collected from the wild in order to produce animals to release in the wild. One major challenge for these organizations is the difficulty of locating nests in the wild to serve as a source of eggs for captive propagation. For example, while rock-lifting surveys are the most effective method for locating all age classes of hellbender (Nickerson & Krysko,2003), they are time and cost-intensive and can destroy natural microhabitats that hellbenders rely on. Also, once captive hellbender populations are established, staff at these facilities must devise methods to monitor the health and behavior of individuals with minimal disturbance, which can pose additional challenges. Thus, there is a growing need to identify tools that can be used to minimize habitat disturbance associated with hellbender surveys in the wild and the invasiveness of monitoring hellbenders in captivity.

Waterproof video cameras may offer a minimally invasive means of detecting hellbenders and their nests in the wild and monitoring individuals in captivity. Video monitoring of animals and their habitats in both natural and captive settings has emerged as an effective method in terrestrial systems (i.e., monitoring burrows of frogs (

Lithobates areolatus

; Heemeyer, Williams, & Lannoo, 2012) and lizards (

Crotaphytus collaris

; Santoyo-Brito & Fox, 2015)). Video monitoring has been increasingly used to document behavior of captive aquatic animals in zoos and aquariums (Hocking, Salverson, & Evans, 2015; Noer, Balsby,Anistoroaei, Stelvig, & Dabelsteen, 2017), and recent efforts have highlighted the use of underwater video for tracking behavior (Hartzell,Pitt, & Davis, 2017; Okada, Fukida, & Takahashi, 2015; Settle, Briggler,& Mathis, 2018) and occurrence (Santas, Persaud, Wolfe, & Bauman,2013) of aquatic salamanders in particular.

Hellbenders are well known for their high site fidelity to crevices,even when faced with disturbance. Underwater cameras such as the aqua vu have been used to visually conform locations of radio-tagged Ozark hellbenders in Missouri streams (Bodiniof et al., 2012) and observed no evidence that occasionally (e.g., monthly) probing crevices with cameras influenced movement of animals. In a separate study, researchers surveyed 180 artificial shelters monthly over a two year period and observed no obvious evidence that capture, processing, and subsequent replacement of hellbenders in artificial shelters reduced artificial shelter occupancy rates (Bodinof Jachowski et al., 2020). For example,this research showed that several hellbenders were captured from the same shelter on more than one occasion and one female was captured from the same shelter on at least 15 separate occasions despite the fact that she was handled each time. Based on these experiences, we are suspect that probing with a camera causes minimal disturbance to hellbender individuals and may become a part of monitoring programs or survey techniques given its non-invasive nature. While some hellbender experts have utilized various designs of underwater cameras, the efficacy of using underwater cameras to detect hellbenders and their nests in the wild and monitoring individuals in captivity has not been quantified.

The goal of our study was to investigate the efficacy of two alternative underwater camera systems for detecting and monitoring hellbenders in both captive and wild settings. Specifically, our objectives were to 1) compare the effectiveness of two underwater cameras(borescope and Aqua-Vu) at detecting adult hellbenders under artificial and natural shelters and 2) assess the potential of these cameras for detecting eggs (nests) in nest shelters as evidence of successful reproduction.

2.Methods

2.1.Equipment

For this research we utilized two readily available underwater camera systems to compare detection time of Eastern hellbenders, a Teslong NTS151RS-3M industrial borescope video inspection camera(www.teslong.com; typically utilized in plumbing or plumber's camera)and an Aqua-Vu Micro II fishing video camera (Nature Vision, Inc.;typically utilized by anglers and ice fisherman). The borescope was equipped with a 3-m, semi- flexible probe and six adjustable LED lights,while the Aqua Vu fishing camera was equipped with automatic infrared illumination and a 50-m cable. Both camera systems included a built-in,water-resistant monitor display. To allow for a more direct comparison of these two underwater viewing cameras, we modified the Aqua-Vu by fastening it to a semi-rigid 1-m gear tie made of bendable wire interior with durable, soft rubber exterior. Both cameras allowed for the ability to capture both video and photos and were readily available, ranging in price from ～$150 to $250 US.

2.2.Captivity camera surveys

To ensure that at least one of the cameras was effective at detecting hellbenders under shelter, we assessed camera performance in a captive facility known to be occupied by adult hellbenders before assessing camera performance in the wild. Specifically, our captive surveys took place in an 18 m long by 1.3 m wide artificial stream at the North Carolina Wildlife Resources Commission's Marion Fish Hatchery (Marion, NC) on August 15 and 16, 2018 (Fig. 1A). The artificial stream contained six known adult Eastern hellbender individuals and nine total shelters with water depth maintained at 0.4 m. Three of the shelters were artificial shelters constructed from concrete modified from the design of Mohammed, Messerman, Mayhan, and Trauth (2016), and six were natural rocks. Because we anticipated a learning curve associated with manipulating each camera to effectively search shelters, both surveyors (authors S.U. and Z.H) practiced using both camera designs to search similar but unoccupied shelters in an artificial stream prior to surveying the occupied artificial stream.

Surveys were conducted by randomly selecting a specific camera design (Aqua-Vu or borescope) for the first survey, and the entire raceway was searched initially using only one camera system. Then after a period of 15 min, the survey was repeated with the other camera system. This repeated survey was done to limit bias and control for any difference in surveying when individuals were previously confirmed under a specific shelter. In addition, surveyors followed the same protocol for searching shelters with either camera system. Prior to the initial survey, none of the six hellbender individual locations was known by surveyors. In brief, one surveyor searched each of the nine potential shelters by walking upstream outside of the raceway, while another surveyor timed search effort using a stopwatch. The surveyor carefully placed each camera in any crevice large enough to serve as a shelter entrance for an Eastern hellbender adult (～10 cm diameter; Rossell et al., 2013). Next, the surveyor searched all available area under the shelter rock by inserting the camera, moving the camera slowly enough to visually monitor on the camera system screen. The entire shelter was searched and the presence of an individual hellbender was noted if the camera picked up the outline of the head, tail, or other anatomical features. Each shelter was searched once with each camera system.

Fig. 1.Monitoring captive adult Cryptobranchus alleganiensis in artificial shelters at the Marion Hatchery, Marion North Carolina using the Aqua-Vu camera system(A) and searching natural shelters (wild) in North Carolina using the borescope (B). Yellow arrows indicate typical search image obtained for each camera system.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

2.3.Wild camera surveys

Once we determined that at least one of the camera designs could be used to detect hellbenders in a captive setting, we conducted surveys for wild hellbenders in four natural stream reaches. Each stream reach consisted of an approximate 100-m length of stream from four separate tributaries of the French Broad River in western North Carolina. We choose not to report the precise locality of stream reaches to avoid illegal collection of hellbenders and disturbance of sensitive habitats (information on file with North Carolina Wildlife Resources Commission). To maximize the potential for encountering active nests during our surveys,we conducted wild surveys during the breeding season (late August-mid September 2018 and end of August 2019). Two stream reaches (FB1 and FB2) contained only natural rock shelters, while two stream reaches(FB3 and FB4) contained natural rock shelters and artificial shelters(concrete nest huts modified from Mohammed et al., 2016) that had been installed to supplement nesting habitat for wild populations. FB4 Artificial shelter surveys were conducted in 2019 following their deployment in later summer of 2019.

Search effort during wild surveys was similar to that of captive camera surveys, with minor modifications. To evaluate whether cameras(borescope and Aqua-Vu) were useful for detecting hellbenders in the wild, we initially snorkeled each stream reach to determine 1) if any hellbenders were visibly present (males actively guarding shelters) and 2) to ensure that stream reaches included at least 15 potential shelters.We defined a potential shelter as a natural rock ≥45 cm in length with at least one usable entrance (10 cm diameter; Rossell et al., 2013). To standardize search effort among stream reaches we randomly selected 13 potential shelters per reach to survey using cameras. We then randomly selected one of the two camera models to use during the first survey and used the alternative camera for the second survey. Each shelter was surveyed twice (once with each camera system) during two separate stream reach surveys. During each survey, one observer used the camera to scan each shelter for all possible rock entrances/crevices while a second observer used a stopwatch to record the time until either a hellbender was detected or an exhaustive search of the shelter was completed (no individual detected). We allowed a period of 15 min to lapse between successive surveys of the same stream reach (Fig. 1B).Each time a hellbender was found occupying a shelter, we recorded the shelter size (length, width, and height), distance to shore, and stream depth in meters. Our survey protocol was identical for each stream reach with the exception of FB3, where we only surveyed artificial shelters due to logistical constraints and did not measure shelter size for any shelters,as artificial shelters did not vary in size.

2.4.Egg surveys

Immediately following the breeding season (middle of September 2018), we conducted separate nest surveys by only searching shelters where individuals (typically guarding males) had been observed previously with either the Aqua-Vu or borescope cameras in the same sites(captive and wild, both hatchery and natural streams). After the initial survey for eggs with both camera systems and due to the observation of several shelters that contained males actively guarding, we ceased using the Aqua-Vu, as it is larger in size (～0.5 cm borescope versus ～2.5 cm Aqua-Vu diameter of camera) and was deemed too obtrusive. It is possible the larger size diameter could potentially interfere more with breeding or guarding behavior, and thus we wanted to minimize its use while males were guarding nests. Therefore, we primarily report on the feasibility of the borescope for detection of eggs, as the borescope camera was deemed less invasive due to its smaller size and less likely to interfere with salamander guarding behavior.

2.5.Data analysis

Our data did not fit the assumptions for normality according to a Shapiro-Wilk test for normality, so we performed a Kruskal Wallis test to test the assumption that search time was equal between the two camera systems (borescope vs. Aqua-Vu). We performed the Kruskal Wallis test in R to check the null hypothesis that the distribution of detection time was the same for both camera systems. We used MS Excel® to determine averages for rock size, water depth, and distance to shore.

3.Results

3.1.Captivity camera surveys

During captive surveys, we detected all six hellbenders present in the hatchery raceway using both camera models. However, the distribution of detection time was significantly difference, where the Aqua-Vu camera provided the quickest detection (borescope: median =28 s,Aqua-Vu: median =15.8 s; Kruskal Wallis Test H =5.66, p =0.017).Anecdotally, we noted that the Aqua-Vu also provided images of higher resolution than the borescope. Also, unlike the Aqua-Vu (equipped with automatic infrared lighting), the light equipped on the borescope illuminated any suspended sediment causing backscatter which often prevented capture of a clear image.

3.2.Wild camera surveys

In natural streams, we detected hellbenders using both camera models. Notably, only one of the artificial shelters that we surveyed were occupied by hellbenders during our surveys. The Aqua-Vu method allowed for greater ease and less time searching both artificial and natural shelters occupied by hellbenders. The distribution of detection time required to exhaustively search artificial shelters was significantly different (Kruskal Wallis Test H =25.23,

p ＜

0.0001) for the Aqua-Vu(median =13 s) relative to the borescope (median =30.9 s). Similarly, the distribution of detection time required to exhaustively search occupied and unoccupied natural rock shelters was significantly different (Kruskal Wallis Test H =15.62,

p ＜

0.001) when using the Aqua-Vu (median =39.1 s) relative to the borescope (67.8 s). When searching natural shelters, we detected about twice as many hellbenders using the Aqua-Vu (

n

=10 hellbenders) relative to the borescope (

n

=4 hellbenders; Table 1). In almost all instances of detections using cameras, individuals were not visible during preliminary snorkel surveys.Natural shelters that we observed adults using averaged 1.5 mL ×1.02 m H ×0.21 m W, were located in water ranging from 4 to 60 cm deep(mean =26.6 cm), and averaged 3.41 m from shore.

Table 1Comparison of median detection time (in seconds) of Cryptobranchus alleganiensis (hellbender) individuals during captivity and wild camera surveys, number of detections per stream, and number of successful egg masses observed. Hatchery refers to individuals maintained at Marion Hatchery, Marion, North Carolina (captivity surveys, nine shelters), while FB1, FB2, FB4 (each 13 shelters) refer to two streams with natural shelters surveyed and FB4 both artificial and natural shelters surveyed for hellbenders using borescope (B) and Aqua-Vu (Av) underwater viewing system. FB3 (only artificial shelters) not shown as no individuals were detected.

3.3.Egg surveys

Using the borescope, we were able to confirm the presence of egg masses (i.e., active nests) under eight natural shelter rocks across three stream reaches (FB1, FB2 and FB3; Fig. 2). The entrances of all the natural nests we detected were guarded by male hellbenders. No nests were detected in artificial shelters (FB3 or FB4) or in the captive (Marion Hatchery) setting (with both natural and artificial shelters present and occupied).

Fig. 2.Images of wild Cryptobranchus alleganiensis eggs taken with Aqua-Vu (A) and borescope (B) underwater viewing systems. Note Aqua-Vu image provided increased overall visibility, while borescope field of view was limited to close fields of view (within ～2—5 cm), and images only showed a close-up of one or two individual eggs whereas the Aqua-Vu allowed for viewing of multiple eggs.

4.Discussion

Our study is the first that we are aware of to evaluate the efficacy of using non-invasive cameras to detect hellbenders and their nests in either captive or wild settings. We found that both the Aqua-Vu and borescope cameras were useful for detecting hellbenders occupying natural and artificial shelters, although the Aqua-Vu camera was more efficient than the borescope. However, once hellbenders were detected,only the borescope was useful for confirming presence of eggs in a guarded cavity. We attribute differences in efficacy for detecting hellbenders primarily to differences in the depth of field for the two camera models we evaluated. The Aqua-Vu camera had a greater depth of focus with increased ability to see objects further away and more of the frame of the monitor in focus, whereas the borescope had to be within 2—3 inches of an object (individual or egg) to be in focus. The increased depth of field afforded by the Aqua-Vu allowed surveyors to visually scan more of the area under each shelter within each frame of view, thus reducing the time to complete an exhaustive survey of any given shelter.However, because of the relatively large size of the Aqua-Vu camera head, we found it to be of poor utility for detecting eggs that were hidden behind guarding males with relatively small shelter entrances. In contrast, the relatively small size of the borescope camera and flexibility of the camera wand facilitated insertion of the camera behind guarding males (which were left undisturbed) and access to deep recesses of rock cavities where hellbenders deposit their eggs.

Monitoring captive individual animals while concomitantly minimizing disturbance, is a goal of many zoos and aquaria. Moreover, we observed no negative aspects of using our underwater cameras, with the exception that on a few occasions hellbenders occupying rock shelters moved slightly forward to investigate camera. In addition, the use of waterproof cameras may minimize habitat disturbance aid in obtaining other vital information on breeding and behavior in Cryptobranchids(Browne et al., 2011). Previous studies have utilized the Aqua-Vu camera to record fish behavior, but primarily only in marine (Boeger,Pie, Ostrensky, & Cardoso, 2006; Spencer, Stoner, Ryer, & Munk, 2005)or lake ecosystems (Schaner, Fox, & Taraborelli, 2009). However, our study is the first that we are aware of to evaluate use of the cameras in a stream environment. While previous studies have utilized video camera systems (Okada et al., 2015) and emerging action cameras (Hartzell et al., 2017) to observe behavior of aquatic salamanders, these video systems are not compatible with monitoring hellbender shelters that are flush with substrate and provide inadequate internal space for camera mounting. Furthermore, shelters that we observed in this study typically had only one entrance, in many cases only ～10 cm in diameter, which would have been too small for a typical larger action camera (i.e.,GoPro®, etc.) to access. Based on our observations, typical action cameras are probably better suited for monitoring breeding activity at rock entrances. In contrast, we showed that both Aqua-Vu and borescope cameras have the potential to function as valuable tools for zoo personnel to less invasively detect individuals and eggs in both wild and captive settings.

While the future holds much promise for the continued captive propagation of Eastern hellbenders in zoos and other collaborating agencies, much work remains to be done for developing less invasive survey methodologies for wild populations and monitoring captive populations of this enigmatic salamander. Therefore, zoo personnel should incorporate both underwater cameras, depending on research goals, either detecting and monitoring individuals or collecting eggs for captive rearing. For example, when goals require detecting or monitoring adult hellbenders, we recommend use of the Aqua-Vu camera.When locating nests is of interest, we recommend a combination of the two cameras as their use in the field may depend on whether there are active males guarding potential nest or specific habitat is surveyed,including larger shelters. Moreover, the underwater camera system we used for monitoring hellbenders could be utilized to monitor populations of other aquatic species both in zoos and in the wild.

CRediT authorship contribution statement

Shem Unger: Conceptualization, Data curation, Formal analysis,Visualization, Methodology, Supervision. Zeb Cline Hull: Data curation, Investigation, Methodology. Lauren Diaz: Investigation, Data curation. John D. Groves: Investigation, Methodology, Conceptualization. Lori A. Williams: Conceptualization, Methodology. Catherine M.Bodinof Jachowski: Supervision, Data curation, Formal analysis.

Acknowledgements

We thank both Peter Lamb and Cody Patton from the North Carolina Wildlife Resources Commission's Marion Fish Hatchery and Lorie Stroup from the Forest Service for their assistance in this project. The North Carolina Wildlife Resources Commission provided the necessary permits for this project (NCWRC 19-ES00286).