Graham 1 Food finding using olfaction by scorpion mud turtles ( Kinosternon scorpioides ) Megan M. Graham Department of Ecology and Evolutionary Bi ology University of California, Santa Cruz EAP Tropical Biology and Conservation Program, Spring 2018 8 June 2018 ABSTRACT The mechanism used to locate food varies between animals, whether it be by olfaction, vision, hearing, touch, taste, or some combination of these. Many turtle species rely on vision and olfaction to find prey. I designed an experiment to test if scorpion mud turtles ( Kinosternon scorpioides ) can find food using olfaction by placing the turtles in Rio Cuajiniquil with three nonde script capsules containing sardine chunks and three empty, control c apsules . Observations of six turtles' interactions with the capsules suggested that they can locate food through olfaction; however, only one statistical analysis was significant which was that the turtles were more likely to bite full capsules than empty capsules. El descubrimiento de alimento por medio del olfato en las tortugas candado ( Kinosternon scorpioides ) R ESUMEN Los mecanismo s utilizado s para localizar alimentos varÂ’an entre las e species de animales , ya sea por olfato, visiÂ—n, audiciÂ—n, tacto, gusto o alguna combinaciÂ—n de estos. Muchas especies de tortugas dependen de la visiÂ—n y el olfato para encontrar sus presas. DiseÂ–ÂŽ un experimento para probar si las tortugas candado ( Kinost ernon scorpioides ) podÂ’an encontrar comida usando el olfato . ColoquÂŽ las tortugas en el RÂ’o Cuajiniquil con tres cÂ‡ psulas que contenÂ’a n trozos de sardina y tres cÂ‡psulas vacÂ’as de control. Las observaciones de las interacciones de seis tortugas con las cÂ‡p sulas sugirieron que pueden localizar los alimentos a travÂŽs del olfato; sin embargo, solo un o de los anÂ‡lisis estadÂ’stico s fue significativo, la cual fue que las tortugas tenÂ’an m ayor probabilidad de morder las cÂ‡psulas completas que las cÂ‡psulas vacÂ’as. INTRODUCTION The ability to locate food is important for survival in all species. Different organisms rely more heavily on certain senses to find prey items and often use more than one sense at a time. Turtles typically depend on vision and olfaction to locate food (Savage 2002). Since many semi aquatic turtles, including Kinosternon scorpioides , live in murky water, olfaction may be one of the more reliable methods to detect objects (Bowden 2010). Scientists are unsure of exactly how turtles smell underwater. One possibility is that both i n and out of water , turtles pulsate their throats to move air or water through their mouth and nostrils where different chemicals can be detected on the olfactory tract and taste buds ( Savage 2002; Bowden 2010). Additionally, scorpion mud turtles , along with other mud and musk turtles, have barbels on their chins. Barbels are whisker like protrusions that are thought to function as sensory receptors (Ramdial 2014 ; Fergus 2007 ). Most turt les are omnivorous, but scorpion mud turtles ( Kinosternon scorpioides ) are primarily carnivor e s and scavengers , feeding on insects, spiders, fish, shrimp, frogs, snails, and
Scorpion mud turtle olfaction Graham 2 worms (Berry and Iverson 2011) . They will also feed on aquatic plants, algae, frui ts, and seeds, but given the choice, they prefer the former foods ( Ramdial 2014). While these turtles can survive on land, they spend most of their time in the water, particularly in ponds , swamps, marshes, and slow moving rivers ( Ramdial 2014 ). Another species of turtle in the same genus , Kinosternon sonoriense , has been reported to forage by moving slowly along the bottom of a body of water with its head and neck fully extended, swinging back and forth. Once prey is detected, the turtle retracts its hea d slightly and lunges. When eating, small prey items will be swallowed whole and larger items will be torn apart and chewed (Hulse 1974). Little is known about how Kinosternon scorpioides locates prey. One study was done in which two scorpion mud turtles were captured and one was observed for three months. The researchers believe that these turtles locate prey visually because the turtle did not eat prey while their tank was in a light tight bag for 46 hours but did eat the prey once exposed to light ( Mong e NÂ‡jera and Morera Brenes 1987). This leads into the question my study attempts to answer: can Kinosternon scorpioides locate food through olfaction? MATERIALS AND METHODS I caught six turtles in Rio Cuajiniquil in Cuajiniquil, Guanacaste, Costa Rica w ith the help of several people . I kept the turtles in captivity outside for the duration of the experiment. The turtles were kept in the shade in separate clear plastic tubs with about two centimeters of river water and a rock to climb onto . I took measure ments of each turtle's carapace length and width and plastron length and width to the nearest millimeter and their mass. I also took note of identifying marks on the turtles such as scars, shell damage, and pronounced features to differentiate them. After the last night of testing, I returned the turtles to the exact location I had found them in Rio Cuajiniquil. I constructed small capsules made of two black bottle caps per capsule, secured together by two black zip ties. Half of the capsules had sardine ch unks inside, and the other half were empty and served as controls. All of the capsules had the same number of holes cut in them to allow the sardine scent to better flow into the water. Tests were performed in an approximately 2.5 meter by 2.5 meter pool i n Rio Cuajiniquil ; three boundaries of the pool were marked by rows of rocks and the fourth was marked by a barbed wire fence (Figure 1) . The water was about 20 centimeters deep at the deepest point. Three pairs of capsules were used for each test , except for the tests conducted on the first night where only one pair was used and placed in the Ã”left' position (Fig. 1) . A pair consisted of one capsule with food (full) and one capsule without (empty) . The capsules in each pair were placed approximately 20 cen timeters apart. The pairs of capsules were placed in the river pool in a triangle formation. One pair was placed upstream by the barbed wire fence, and the other two pairs were placed in the downstream corners of the pool (Figure 1). Each capsule was held down by a small rock so that the capsule was still accessible to the turtle. If a turtle tried to escape under the fence or over the rocks , I picked it up and placed it back at the starting point , roughly halfway between the downstream capsules . One at a time, I would place a turtle roughly halfway between the downstream capsules, facing upstream (Figure 1) . Once the turtle was in the water, I started the timer. Each t est lasted 30 minutes unless a turtle bit one of the capsules at which time I ended the t rial . I recorded the times of when a turtle would touch a c apsule with its nose , bite a c apsule , and when a turtle had
Scorpion mud turtle olfaction Graham 3 to be reset after trying to leave the pool . With this, I took note of what the first contact was . Tests were performed nightly from 15 Ma y 2018 through 18 May 2018 , usually starting around 8:00 pm . I tested the turtles at night because they are primarily nocturnal (Forero Medina and CostaÂ–o Mora 2011) , partially evidenced by the fact that it took roughly six hours of searching during the da y to find one turtle and 30 minutes at night to find four turtles . Each turtle was tested once per night. Four of the turtles were tested three times each, and the other two were each tested twic e . !"# $ $ $ %&'($)*++ $ ,*-."/$)0-"$1"2'" $ $ 3456# $ '*578+" $ 98++$'*578+" $ $ $ :0-"'60&2$&1$ )*6"-$1+&) $ ;8-6+"$76*-602<$ 5&7060&2 $ $ $ $ $ Figure 1. Set up of testing pool in Rio Cuajiniquil . The pool was approximately 2.5 by 2.5 meters. Three pairs of capsules were placed in a triangle formation. A pair consisted of one capsule with food (full) and one without food (empty). Capsules within a pair were approximately 20 centimeters apart. Not to scale. After each t est on the last two nights , I replace d the sardine chunks in the full capsules and swi tch ed the locations of the empty and full capsules within a pair . I replaced the bait to ensure that the scent wasn't being washed away between t ests , so each turtle would have the same strength olfactory cue. The capsules were switched between trials to s omewhat randomize the location and help control for any other factors that may have been causing a turtle to go to a particular capsule such as other cues already present in the river . I performed a chi squared test on the total number of times full and empty capsules were touched , the number of times full and empty capsules were touched first during a test , and the
Scorpion mud turtle olfaction Graham 4 number of times full and empty capsules were bitten. I also did a Wilcoxon signed rank test on the first contacts per turtle and the total co ntacts per turtle to test the statistical significance of these data . An alpha level of .05 was used for all statistical tests. RESULTS Even though the majority of my data was not statistically significant, my observations provide evidence that scorpion mud turtles may use olfaction to find food. Each turtle bit a full capsule at least once across all of its trials. In the seven instances that an empt y capsule was touched, five of those times a full capsule was contacted after ward, during the same trial. F our of those five full capsule contacts were bites (Figure 2) . I define a touch as a turtle touching a capsule with its nose, and I define a contact as a turtle touching a capsule with its nose or biting a capsule. $ $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$ = $ $$$$$$$$$$$$$$$$> $ $$$$$$$$$$$$$$$$?=$$$$$$$$$$$ $ $$$$?>$$$$$$ $ $$$$$$$$$@=$$$$$$$$$ $ $$$$$@>$$$ $ $$$$$$ $$$ $$$A= $ $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ ;04"$B40286"7C $ $ Figure 2. Each turtle's performance across all nights and the times that they contact ed capsules are shown . Not all turtles were tested e ach night. Trial number corresponds to date with 15 May being one and 18 May being four. Full capsules contained sardine chunks, empty capsules did not. Four of the six turtles contacted full capsules more often than empty capsules. One turtle contacted empty capsules more often, and another turtle contacted both capsules equally (Figure 3 : W value = 1.5 ) .
Scorpion mud turtle olfaction Graham 5 $ Figure 3. Contact counts per turtle. Number of times each turtle made contact (bi te or touch) with a container across all nights. Full containers had sardine chunks inside, empty ones did not. Every turtle bit a full capsule at least once, and no turtles bit an empty capsule. Two turtles touched full capsules more than empty, one turtl e touched them equal amounts, and three turtles touched empty capsules more than full capsules (Figure 4) . In t otal, full capsules were touched nine times, empty capsules were touched ten times, full capsules were bit ten nine times, and empty capsules were never bit ten ( Table 1 ) . Figure 4. Interaction (touching with nose and biting full and empty capsules) counts per turtle across all trials. Full capsules contain ed sardine chunks, empty capsules d id not. =$ ?$ @$ A$ D$ >$ E$ F502#$G0("$ H&$H*4"$ G"<*+&/&2$ F"-"2/0506#$ I*++06*$ ,&..*JK&8$ I&826$ ;8-6+"$ I&26*'6$18++$ I&26*'6$"456#$ =$ ?$ @$ A$ D$ >$ F502#$G0("$ H&$H*4"$ G"<*+&/&2$ F"-"2/0506#$ I*++06*$ ,&..*JK&8$ I&826$ ;8-6+"$ ;&8'L$18++$ ;&8'L$"456#$ ,06"$18++$ ,06"$"456#$
Scorpion mud turtle olfaction Graham 6 Table 1. Interaction counts across all turtles across all trials. Number of times full and empty capsules were touched by a turtle's nose and number of times they were bitten. Full capsules contained sardine chunks, empty capsules did not. Interaction Count Touch full capsule 9 Touch empty capsule 10 Bite full capsule 9 Bite empty capsule 0 Three turtles first contacted full capsule s more often than empty capsules. One turtle first contacted empty and full capsules equally often, and one turtle first contacte d empty capsules more often (Figure 5: W value = 6). Figure 5. First contact counts for each turtle across all trials. The first capsule a turtle made contact with (touched with nose or bit) once placed in the testing pool. Full capsules had sardine chun ks inside, empty capsules did not. No Name and Bobba Lou each had two trials, and the rest of the turtles each had three trials. Serendipity didn't make contact with any capsules during two of her trials. Twenty eight contacts (touches and bites) were performed across all trials for all turtles, 18 (64%) of which were contacts with a full capsule and ten (36%) of which were with an empty capsule (X 2 = 2.286 , p = .131 ) . Of the 14 trials where capsules were contacted (to uched with nose or bitten), eight (57%) of the initial contacts were with full capsules and six (43%) were with empty capsules. These values only differ by two points so are not statistically significant . There were two additional trials where no capsules were contacted. D ISCUSSION The turtles' propensity for biting full capsules is strong evidence that they can locate food using olfaction . Every turtle bit a full capsule; h alf of the turtles bit full capsules once across all of their trials, and the other half bit full capsules twice , for a total of nine full capsule bites . Empty capsules were never bitten ( Fig. 4 ) This result was found to be statistically significant ( X 2 =$ ?$ @$ A$ D$ F502#$G0("$ I*++06*$ F"-"2/0506#$ H&$H*4"$ G"<*+&/&2$ ,&..*JK&8$ I&826$ ;8-6+"$ I&26*'6$18++$ I&26*'6$"456#$
Scorpion mud turtle olfaction Graham 7 = 9, p < 0 .05) . Biting is a feeding behavior, not a foraging behavior. Biting suggests that the turtle has located the food item , identified it as food, and decided to eat it. I t is unlikely the turtles were using vision, touch, or hearing t o determine that the capsules contained fish because the black color of the capsules didn't stand out, the ridged plastic texture doesn't feel like the turtles' usual prey, and the capsules were not moving or making noise . Taste does not play a significant role in turtles' ability to l ocate objects (Savage 2002). This leaves olfaction as the most likely explanation of how they found the sardine filled capsules. Additionally, in five of the eight trials where a turtle touched a full capsule with its nose, the same full capsule was later bitten. In four of these five instances, the two interactions were 45 seconds or less apart. This suggests that the turtles were smelling the capsules first to determine if they had food, and once they detected the scent, they would try to eat it. When a t urtle bit a capsule , I ended the trial and removed it from the pool. This limited the number of times a capsule could be bitten as only one bite could occur per trial whereas multiple touches could occur in a 30 minute trial. Usually, a turtle would not sp end more time at an empty capsule than it took to touch it but would often stay near or repeatedly touch a full capsule (in the data, these were counted as one touch unless the turtle left the capsule and came back to it) . Additionally, in five of the seve n t rials an empty capsule was touched, the turtle later contacted a full capsule within the same trial (Fig. 1). This could indicate that the turtles need to be close to the source to pinpoint where a smell is coming from. From a distance, the turtles may have only been able to tell that the sardine smell was coming from a general area of the pool. As they got closer to the pair of capsules, they could have gone to the empty capsule first because the source of the smell (the full capsule) was only 20 centim eters away, and they needed to get close to an object to determine if it was the source. Four of the six turtles made contact (touch or bite) with the full capsules more often than the empty ones; however, this tendency was not statistically significant (X 2 = 2.286 , p =.131 ). Callita contacted empty and full capsules an equal number of times, and Bobba Lou contacted empty capsules more often than full capsules. All but one of the turtles touched an empty capsule at least once (Fig. 3) . The propensity for touching empty capsules can be explained in several different ways. First , switching the position of the capsules could have left a residual scent on the substrate or the rock holding the capsule th at the turtles could detect. Second , the turtles may have been able to detect by olfaction the disturbance in the substrate when a capsule was placed and went to investigate i t. Third , the pool may have been inundated with sardine scent, especially as the night went on and sardine was present in the pool longer . Four th , the bottlecaps may have had trace amounts of Powerade on them , and, if the turtles' sense of s m ell is sensitive enough, they may have been able to detect it and investigate d it because it smells similar to fruit or out of curio sity for the new smell . I expected that the turtles would go to a full capsule as their first contact during a test more often than an empty capsule; however , this was only the case with three turtles and eight out of 14 first contacts. This phenomenon can also be explained by the four previously mentioned reasons why the turtles may have gone to investigate empty capsules. It seems likely that Kinosternon scorpioides can detect food only using olfaction . Based on my observations of the turtles foraging an d contacting the capsules, it appeared that they c ould find food only using olfaction . While observing the turtles during the tests, they clearly seemed to be foraging, moving underwater with their heads and necks extended and moving in
Scorpion mud turtle olfaction Graham 8 what appeared to be a searching motion , very similar to what Hulse described (1974) . The turtles did not typically linger at an empty capsule when they touched it. In addition to spending more time at the full capsules, in five of eight tests where a turtle touched a full ca psule , it later bit the same full capsule (Fig. 2) . Observing these turtles biting only full capsules (often biting the capsules enough to dislodge them from the rock that was holding them) and only spending time at the full capsules , rather than the empty ones, indicate d that they could locate food with olfaction. The statistical analysis of th e biting data was also statistically significant. However, given my modest sample size, the statistical analyses of the rest of my data do not illustrate that the tu rtles contact full capsules more often than empty capsules in a statistically significant way. Future studies could repeat this study over a longer time frame with more turtles and keep capsules in the same locations across t rials . My study also raises a n ew question : how sensitive are scorpion mud turtles' olfactory receptors? A CKNOWLEDGEMENTS Thank you to Frank Joyce for being such an amazingly patient primary advisor , driving me to the river every night, and helping me figure out how to do statistics. Thank you to AndrÂŽs Camacho for being my secondary advisor. Thank you to Michael Spaeth for standing in the dark watching turtles with me for nine plus hours and never complaining, to Jenna Sparks for consistently putting aloe on my back , lending me that a wesome shirt, and sacrificing her hair tie, and to the rest of my fellow Cuajiniquillas for their help, support, and lack of complaining when asked to stay up late and watch turtles with me . Thank you to Freddy Ampie for finding my first turtle . Thank you to Sydney Schalk for editing my paper and thinking my turtles were cute. Thank you to --for translating my title and abstract into Spanish. Last but not least, thank you to Flory Castro for being an amazing homestay mom and taking care of me during this endeavor. LITERATURE CITED Berry, James F., and John B. Iverson. " Kinosternon scorpioides (Linnaeus 1766) Scorpion Mud Turtle." 31 Dec. 2011. Accessed 27 May 2018. Bowden, R. M. (2010). Turtles: Freshwater. In Encyclopedia of Animal Behavior (pp. 462 468). Retrieved from https://www.sciencedirect.com/topics/agricultural and biological sciences/kinosternidae Fergus, Charles. Turtles . 1 ed., Stackpole Books, 2007. Accessed 7 June 2018. Forero Medina, German, and CostaÂ–o Mora, Olga V. " Kinosternon sc orpioides albogulare ( Dumeril and Bocourt 1870 ) Ã White Throated Mud Turtle , Swanka Turtle ." 31 Dec. 2011. Accessed 2 9 May 2018. Hulse, A. C. (1974, July 15). Food Habits and Feeding Behavior in Kinosternon sonoriense (Chelonia: Kinosternidae. Journal of H erpetology , 8 (3), 195 199. Retrieved from
Scorpion mud turtle olfaction Graham 9 https://www jstor org.oca.ucsc.edu/stable/pdf/1563164.pdf Monge NÂ‡jera, J., & Morera Brenes, B. (1987). Notes on the feeding behavior of a juvenile mud turtle, Kinosternon scorpioides. Herpetological Review , 18 (1), 7 8. Ramdial, K. (2014). Kinosternon scorpioides (Scorpion Mud Turtle). In The Online Guide to the Animals of Trinidad and Tobago . Savage, J. M. (2002). The Amphibians and Reptiles of Costa Rica (pp. 738 750). Chicago, IL: The University of Chicago Press . A PPENDIX Appendix 1 . Measurements of turtles and other physical features. Table 2 . Physical features of turtles . Identifying turtle features including carapace width, carapace length, plastron width, plastron length, mass , and sex . Name Carapace wi dth Carapace length Plastron width Plastron length Mass Pronounced beak hook Sex Callita 16.6 cm 19.5 cm 11.8 cm 14.0 cm .613 kg Yes Male Megalodon 16.5 cm 20.9 cm 11.4 cm 15.0 cm .662 kg Yes Male Bobba Lou 18.5 cm 22.3 cm 12.5 cm 16.8 cm .841 kg No Female No Name 18.5 cm 22.5 cm 12.7 cm 16.0 cm .827 kg No Female Spiny Mike 17.6 cm 20.0 cm 11.6 cm 14.9 cm .731 kg No Female Serendipity 17.3 cm 21.3 cm 12.0 cm 16.8 cm .712 kg No Female I determined sex by looking at carapace width and turtle mass (both are larger in females), the curve of the plastron (males have a slightly concave plastron), and the upper beak hook (more pronounced in males) ( Forero Medina and CostaÂ–o Mora 2011 ; Savage 2002). Appendix 2 . Photos of turtles captured with date captured, identifying features, and personality.
Scorpion mud turtle olfaction Graham 10 Turtle: Callita . Caught on 8 May 2018 around 4:00 pm. Smallest of the six turtles. Most eager to be free/let out of tank. Usually quick to eat. Turtle: Megalodon. Caught on 9 May 2018 at 8:53 pm . White, U shaped scar on top of nose and dents on front of carapace . Skittish.
Scorpion mud turtle olfaction Graham 11 Turtle: Bobba Lou. Caught on 9 May 2018 at 8:40 pm . Heaviest turtle. Turtle: No Name. Caught on 9 May 2018 at 8:47 pm. Shy. Slow to come out of shell.
Scorpion mud turtle olfaction Graham 12 Turtle: Spiny Mike. Caught on 9 May 2018 at 9:00 pm. Pronounced spine along posterior half of carapace. Most relaxed. Turtle: Serendipity. Caught on 14 May 2018 at 9:45 pm. Many white scars on top of nose. Skittish. Would fight against being held with her c laws.