Natives bite back: depredation and mortality of invasive juvenile Burmese pythons ( Python bivittatus ) in the Greater Everglades Ecosystem

Burmese pythons ( Python bivittatus Kuhl, 1820) are one of the world’s largest snake species, making them a highly successful and biologically damaging invasive predator in the Greater Everglades Ecosystem, Florida, USA. Though we have knowledge of python diet within this system, we understand very little of other interactions with native species. Effects native species have on invasive pythons, especially in the juvenile size class, are of particular interest as the prevalence of mortalities would inform potential population growth and trophic dynamics with native prey species. Native ophiophagous predators in Florida feed on smaller native snake species and it is unknown if they consistently recognize similarly sized juvenile invasive pythons as prey items. Using radiotelemetry, we found at least four native species within Big Cypress National Preserve that were implicated in juvenile python deaths, including three Florida cottonmouths ( Agkistrodon conanti Gloyd, 1969), five American alligators ( Alligator mississippiensis Daudin, 1802), one hispid cotton rat ( Sigmodon hispidus Say and Ord, 1825), and three mesomammals. One mortality was the result of an attempt to subdue a prey item 106% the size of the python, constituting the largest predator:prey size ratio ever reported in this size class. This finding may indicate that phenotypic variation in individual juvenile pythons includes behavior that could be maladaptive within the novel Florida environment. Here we describe some of the first confirmed cases of non-anthropogenic mortality in juvenile Burmese pythons in Florida and present evidence that invasive pythons in this size class are now being incorporated into the diets of native species in its invasive range.


Introduction
Following their establishment in Florida, USA, in the 1980s (Willson et al. 2011), Burmese pythons (Python bivittatus Kuhl, 1820) have successfully invaded a large portion of the Greater Everglades Ecosystem. The success of Burmese pythons in southern Florida is a result of their large body size, generalist dietary and habitat preferences, purported high fecundity, and rapid growth rates, with the number of reported pythons increasing annually (Reed et al. 2012;Guzy et al. 2023;Currylow et al. 2022c). Although pythons have been established in southern Florida for nearly four decades and evidence of predatory effects on native populations continues to grow (Taillie et al. 2021), little is known about how pythons and native species may otherwise interact.
Although data on natural causes of Burmese python mortality in their native range in southeast Asia are extremely limited (i.e., king cobras, Ophiophagus hannah Cantor, 1836; Smith et al. 2021), documented predators of the closely related Indian rock python (Python molurus Linnaeus, 1758) include crested serpent eagles (Spilornis cheela Latham, 1790) and Bengal monitor lizards (Varanus bengalensis; Bhupathy and Vijayan 1989;Krishna 2002;Goel et al. 2017). Additionally, Bhupathy and Vijayan (1989) documented python scales in jackal scat, a group of three jackals attacking an Indian rock python, and several pythons that were trampled by large ungulates in addition to possibly succumbing to injuries resulting from porcupine consumption (Bhupathy et al. 2014). Reports of mortality and depredation of Burmese pythons in the Everglades region are most common for the adult age class. Documentation of natural mortality or predators include American alligators (Alligator mississippiensis Daudin, 1802; Snow et al. 2006), American crocodiles (Crocodylus acutus Cuvier, 1807; Godfrey et al. 2021), bobcats (Lynx rufus Schreber, 1777), and an investigative or defensive attack by a Florida black bear (Ursus americanus floridanus Merriam, 1896;McCollister et al. 2021). Other sources of natural mortality include cold weather events (Mazzotti et al. 2016). Only very recently were the first instances of python nest depredation and defense described (Currylow et al. 2022b). Although it has been presumed that other native ophiophagous predators consume juvenile Burmese pythons, sparse records of depredation events have been made. These include one instance of depredation by an eastern indigo snake (Drymarchon couperi Holbrook, 1842; Andreadis et al. 2018), two instances by Florida cottonmouths (Agkistrodon conanti Gloyd, 1969;Bartoszek et al. 2021), and mention of three American alligators (Pittman and . The prevalence with which native predators consistently recognize and incorporate invasive pythons into their diet is unknown. In addition to limited depredation observations, there is a dearth of reported descriptions of any other causes of mortality of juvenile pythons other than direct human affects (e.g., road and levee vehicle strikes or captures). Yet, the vast majority of the Greater Everglades Ecosystem is contiguous wildlands with limited human-accessibility. Information on how pythons are interacting with, and adapting to, their novel wildland habitats can help researchers estimate survival and dispersal. For example, Figure 1. Photographic evidence and representation of some of the variety of confirmed and potential causes of mortality found for invasive Burmese pythons (Python bivittatus Kuhl, 1820) in the Greater Everglades Ecosystem in 2021 in Big Cypress National Preserve, Florida, USA. From left to right: American Alligator (Alligator mississippiensis Daudin, 1802) depredations, mesomammal depredations (felid prints in muddy substrate; e.g., bobcat, Lynx rufus Schreber, 1777), Florida cottonmouth (Agkistrodon conanti Gloyd, 1969) depredations, potential avian depredations, mishandling/misidentification of appropriate prey (e.g., large hispid cotton rat, Sigmodon hispidus Say and Ord, 1825). Images by U.S. Geological Survey. some captive hatchling pythons have been known to refuse food to the point of death from starvation (see Josimovich et al. 2021 and citations therein). If phenotypes exhibited in the wild also encompass such maladaptive traits, the potential ecological affects that a clutch of juveniles represent could be drastically altered. The causes and prevalence of juvenile python mortalities are key factors to understanding population growth potential, a topic of particular interest for wildland managers.
Identifying the influence native species have on the survival of invasive pythons is valuable for informing restoration efforts and management actions. Additionally, understanding natural causes of juvenile mortality informs demographic estimates, population size, growth rates, and structure. Here, we describe and discuss confirmed and suspected causes of natural mortality (Figure 1) in juvenile Burmese pythons observed within their invasive range in Florida, USA.

Materials and methods
Our study site was within Big Cypress National Preserve, Florida USA, managed by the U.S. National Park Service. The 295,000-ha preserve is characterized by a mosaic of pine grasslands, cypress swamps, hardwood hammocks, disturbed edge habitat, prairies, and marshes, all of which are bisected by gravel roads, a single paved highway, and water management levees and canals. Higher elevation pinelands are dominated by slash pine (Pinus elliottii Little & Dorman) and saw palmetto (Serenoa repens [W. Bartram] Small). Limestone depressions often support cypress swamps dominated by bald cypress (Taxodium distichum (L.) Rich.). Live oaks (Quercus virginiana Mill.) are representative of hardwood hammocks. Prairies and marshes are dominated by grasses and sedges, but plant communities vary significantly based on the hydroperiod of the area.
As part of another study, juvenile Burmese pythons (ranging from 55-90 cm snout-vent length; Currylow et al. 2022c) were opportunistically collected from June through September of 2020 and 2021. Only females were retained, surgically implanted (e.g., Hale et al. 2017) with very high frequency (VHF) radio transmitters (Holohil Systems Ltd.,Carp,Ontario,Canada;, and released at the point of capture. Pythons were subsequently tracked to their exact location at least twice per week. During tracking, we pinpointed the location of the radio signal within three meters, aiming to obtain a visual observation of the python whenever possible while minimizing disturbance. Visual verifications were obtained at least once per month to note the external appearance of the snake's body condition. The transmitters were equipped with mortality sensors that were triggered by non-movement for ≥ 24 h. For two weeks following the surgery, we checked the signals daily for active mortality sensors. If a mortality sensor was active, we immediately tracked the signal to the exact location. We inspected and photo-documented each mortality event for any causal evidence (e.g., predator prints, disturbed vegetation, presence of fur, scat, or scales, scrape or bite marks, use of underwater photography, etc.; Figure 1). If any portion of the python carcass was still present, we collected it for further evaluation in laboratory necropsy. In some instances, we obtained measurements of animal prints found at the scene from field photographs using freely available image processing software (Rasband 1997(Rasband -2016. Any bite marks found on the recovered carcasses or transmitters were evaluated for size, bite pairs, and intercanine distance to identify possible predators/scavengers (e.g., Lyver 2000). We also identified depredation events by assessing whether the transmitter had been ingested and was emitting a signal from within another animal.

Results
During radiotelemetry efforts, we documented 19 juvenile python mortalities occurring between 31 May 2021 and 11 February 2022 (Table 1). Of the 19 events, we were able to determine the causes of mortality for 63% (12 of 19). We attribute eight deaths (67%) to native reptilian predators, one (8%) to injuries sustained while handling the largest relative prey item ever reported in this age class (also a native species), and three (25%) to mammalian predators. The remaining seven mortalities could not be confidently attributed to any specific cause, but some possibilities could be excluded. Each of these observations are described in detail below. Potential avian depredation. Transmitter recovered hanging in prairie grasses 1.5 m above the ground. Closest potential perch ~ 10 m away

Cottonmouth Depredation (CD; n = 3)
We documented three Florida cottonmouth depredations on juvenile pythons, the first of which was discovered 31 May 2021 and confirmed by radiography ( Figure 2; further described in Bartoszek et al. 2021). The two subsequent instances occurred on 01 and 08 October 2021. We tracked both telemetry signals to cottonmouths and confirmed each signal was emanating from the predators as they moved upon being disturbed. We continued to regularly track the signals twice a week to the cottonmouths until both transmitters were expelled and recovered six and nine days later. In both instances, the python was confirmed alive eight days prior to first observations of the cottonmouths.

Alligator Depredation (AD; n = 5)
Throughout the study, we confirmed five American alligator depredations on juvenile pythons. Depredation events were discovered on 10, 28, and 30 October 2021, 07 December 2021, and 11 February 2022. Each of the 5 juveniles were visually confirmed to be alive 4 to 16 days before we confirmed depredation. We tracked each telemetry signal to the location of each alligator across several consecutive visits and monitored the movements of transmitter signals with the alligators (occasionally using underwater photography; e.g., Figure 1). We estimated the sizes of four of these alligators to range 1.5-2 m, with one larger animal estimated to be 3 m in length. We continued to monitor the alligators following the depredation events, but none of the transmitters were ever detected expelled or recovered before expiration.

Foraging Death (FD; n = 1)
On 17 October 2021, we tracked the python (last visually confirmed alive seven days prior) to a flooded hardwood habitat where we found its carcass with a large food bolus. The transmitter was not emitting a mortality signal, indicating the python had perished within 24 h of being discovered. We noted the python to have several wounds on both sides of the body, and we collected the carcass for necropsy. At necropsy, the combined weight of the python and food item was 251 g. Once removed, we identified the food bolus to be a 129 g hispid cotton rat (Sigmodon hispidus Say and Ord, 1825). The remainder of the snake carcass weighed 122 g (28 g less than upon release). The rat was consumed headfirst and sustained minimal degradation, consistent with it being consumed less than 24 h prior to discovery (see Secor 2008). Evidence from the necropsy (size and shape of wounds) indicates that the injuries may have been inflicted by the rat while the python was attempting to capture and subdue the prey (Figure 3). The posterior puncture wound may have punctured the non-vascularized portion of the right lung (Figure 3c). We found that the anterior puncture wound had penetrated the stomach lining ( Figure 3d).

Mesomammal Depredation (MD; n = 3)
On 12 September 2021, we tracked the signal of MD1 (last visually confirmed alive one day prior) into disturbed edge habitat near a road where we located the transmitter body. The transmitter body had sustained numerous small puncture marks and abrasions to the waterproof epoxy coating which we determined to be indicative of carnivorous mesomammal teeth (Figure 4 left panel). The antenna was severed at the base, which we also found consistent with mesomammal mastication, and was not recovered. Approximately 7 m from the transmitter location, we observed a group of partial mammal tracks, with the most complete print measuring 3.0 × 3.6 cm in area. The shape and size of these tracks were consistent with a small felid such as a young bobcat, but we are unable to rule out a feral domestic cat (Felis catus Linnaeus, 1758) considering proximity to private homesteads. On 15 September 2021, we tracked the signal of MD2 (last visually confirmed alive five days prior) into dense vegetation where we located the transmitter body resting atop leaf litter. The antenna was severed near the base, consistent with mesomammal mastication. We discovered several small clumps (approx. 1 × 2 cm) of regurgitated fur within 10 cm of the transmitter, but the flooded leaf litter at the site prohibited finding further evidence of predator identity.
On 19 September 2021, we tracked the signal of MD3 (last visually confirmed alive seven days prior) into disturbed edge habitat around five meters from a gravel road where we found the carcass and transmitter. The anterior half of the carcass appeared to have been consumed and the transmitter body and antenna were outside the body cavity, adjacent to the remains (Figure 4, right panel). The area was partially flooded scrub habitat with indication of heavy wildlife traffic. Three partial mammal prints were discovered within two meters (one < 5 cm) of the carcass in a wildlife corridor through the vegetation to the road. The size (approx. 4 cm 2 ) and shape of these partial prints appeared to be felid but complete tracks/prints were not observed.

Unattributed Mortalities (UM; n = 7)
On 08 October 2021, we tracked the signal of UM1 (last visually confirmed alive eight days prior) into disturbed edge habitat approx. 5 m from a gravel road. At the point of strongest transmitter signal, a faint smell of decay was detected emanating from a patch of tall grasses and woody debris. We located the transmitter under vegetation along a network of small mammal trails or tunnels in the long grasses. The transmitter lacked any damage to the waterproof epoxy coating and the antenna was still attached but was damaged and deformed. No remains of the juvenile python were present. We last tracked the python 3 days prior to a location 251 m from the transmitter discovery site.
On 19 October 2021 we tracked the signal of UM2 (last visually confirmed alive five days prior) to scrub habitat approx. 11 m from a nearby road. The mortality sensor activated during the track, indicating that the transmitter had been at rest at the location for exactly 24 h. We found the transmitter on the ground in dense grasses, antenna and body clean and intact without any marks or sign of the python carcass ( Figure 5A). We discovered some ventral and keeled dorsal snake scales scattered approx. 1 m from the transmitter, but based on the size, shape, and characters, they likely belonged to a cottonmouth shed. Our last tracking event 3 days prior placed the python 3 m from the mortality site.
On 23 October 2021 we tracked the signal of UM3 (last visually confirmed alive 12 days prior) to a flooded, disturbed hardwood hammock and recovered the carcass. The intact carcass was wrapped around the base of several small woody plants ( Figure 5B) with only one section of skin (2.5 × 1.2 cm) missing on the left side and exposing ribs. A small portion of the transmitter body was visible between the ribs. Five superficial wounds were noted along the posterior half of the carcass. No other explanatory evidence was found. We last tracked UM3 3 days prior to a location 10 m away from the mortality site.
On 29 October 2021 we detected a mortality signal from the transmitter of UM4 located in a difficult to access point at the edge of a spoil pile near a canal. On 30 October we gained access to the location and discovered the carcass in a state of advanced decomposition. This python had only been in the field for eight days and was always associated with the spoil pile at the release location. We were unable to obtain a visual of the snake between the time of release and retrieving the carcass.
On 29 October 2021 we tracked the signal of UM5 to pineland with a moderately open palmetto understory. The carcass was found in a clearing roughly 50 cm in diameter ( Figure 5C). It was partially decayed, appeared compressed, and was supporting invertebrate decomposer larvae. The transmitter antenna was severed and found outside of the carcass. Some dried palmetto fronds and live ground cover plants were matted down in the immediate vicinity but the midstory vegetation surrounding the carcass was undisturbed. The transmitter was not emitting a mortality signal indicating that the transmitter was recently jostled, but we detected no animal tracks in the area. We observed this python twice in the week prior to discovering the carcass (most recent was four days prior, 5 m away) and noted that the python appeared to be in poor body condition.
On 19 November 2021 we tracked the signal of UM6 (last visually confirmed alive 20 days prior) to a flooded pineland with some snags and sparse palmetto understory. The transmitter was found near the base of a pine snag in shallow water (approx. 2 cm deep; Figure 5D). The transmitter was intact with no signs of claw or tooth marks and not emitting a mortality signal. Potential perch sites were available above the transmitter. There were several hairs collected from the surface of the transmitter body. The stiff hairs, appearing to be whiskers, were black with white roots and tips ranging from approx. 1-2 cm. No other explanatory evidence was found. We last tracked this python 4 days prior at 5 m from this site.
On 20 December 2021 we detected a mortality signal from the transmitter of UM7 (last visually confirmed alive 24 days prior) and tracked the signal to an ecotone between bald cypress and prairie. The transmitter was found hanging in tall prairie grasses approx. 1.5 m above ground ( Figure 5E). There was no mid or overstory within 10 m of the suspended transmitter. The transmitter was complete and intact with no signs of damage. There was no flesh adhered to the transmitter, but a number of ants were crawling on and around the unit. We last located the signal 3 days prior at a location 20 m from the transmitter discovery site.

Discussion
Native species of the Greater Everglades Ecosystem contributed to at least 63% (12 of 19) of the documented mortality events of invasive juvenile Burmese pythons, whereas we were unable to assign any predator or cause to the other 37%. We found that 72% of the known depredation events (8 of 11) were attributable to reptilian predators. Native reptiles' ability to effectively compete and capitalize on this pervasive python invader may simply be due to microhabitat overlap of these semi-aquatic species increasing the chances of their interactions. In another telemetry study, Pittman and  also reported reptilian depredations (alligator and indigo snake) on several juvenile pythons. These reptile predators ingest their prey whole and digestion is slow, affording the opportunity to track to, and positively identify, the predator. Transmitters were eventually recovered from every cottonmouth depredation. In contrast, we continued tracking alligator predators for up to 21 weeks post-depredation event without a single transmitter recovery. Alligators are known to retain indigestible spheroid objects in their digestive tracts far longer than digestible materials (Wings 2007;Grigg and Kirshner 2015), and we suspect that the transmitters are being similarly retained as pseudo-gastroliths.
The single foraging death we discovered is of note for two reasons. One, the prey item (native hispid cotton rat) was 106% of the study animal's body weight, the largest predator:prey size ratio ever recorded for pythons of this size class. The prey item was on the smaller end (129 g) of the average weight of adult cotton rats (100-250 g; Curlee and Cooper 2012) but there are few other mammalian species of appropriate size that may be found in unknown densities in the Preserve ( Pifer et al. 2011). Therefore, it could have been that the python had not yet happened upon any alternatives and had depleted its yolk stores. The carcass of the python alone was found to weigh 19% less than the animal's weight at release 18 days earlier. This could have been one of the first meals the hatchling python attempted, leading to the second noteworthy aspect involving prey recognition. In the native range of other pythons, native rat species are the smallest food item documented in the smallest size classes, with maximum predator:prey size ratios peaking around 60% (e.g., Shine et al. 1998;Madsen and Shine 2002). In our study, the naivete of the young python in identifying and handling appropriate prey items resulted in mortality (e.g., Kornilev et al. 2022). Thus indicating the possibility that there could be some maladaptive behavioral phenotypes exhibited by this invasive species in the novel environment where prey types and sizes may differ from those in the native range. The idea of maladaptive behavior in this Florida population is not new. Mazzotti et al. (2011) found that sub-adult and adult pythons inappropriately responded to a 2010 prolonged cold weather spell in southern Florida. Some pythons attempted to bask and succumbed to intolerably low temperatures, instead of retreating to thermally-stable refugia. This was notably unlike temperate native large reptiles (Mazzotti et al. 2016). However, that 2010 cold event is thought to have driven the Florida python population to rapidly adapt to colder temperatures in the region (Card et al. 2018).
For the mesomammal depredations, the prey items are masticated in bites, allowing the predator to consume the carcass without ingesting the transmitter. This prevented the option of tracking to, and indisputably identifying, the mammalian predators. In most cases where evidence of mammalian depredation was apparent, some prints were found and most often were felid. However, several potential small-and medium-sized mammal predators are present in Big Cypress National Preserve including the North American river otter (Lontra canadensis Schreber, 1777), coyote (Canis latrans Say, 1823), bobcat, Everglades mink (Mustela vison evergladensis Hamilton, 1948), raccoon (Procyon lotor Linnaeus, 1758), gray fox (Urocyon cinereoargenteus Schreber, 1775), and Virginia opossum (Didelphis virginiana Kerr, 1792;Pifer et al. 2011). In the vicinity of homesteads and private inholdings, domestic dogs (Canis lupus familiaris Linnaeus, 1758) and domestic cats may also present threats, though we have never encountered these domestic animals near tracked snakes. While not used in this study, DNA sampling could aid in identifying predatory species associated with recovered carcasses or transmitters. However, this method of predator identification is lacking application on remains of herpetofauna, and it is likely that hair and feathers are better vectors for DNA as opposed to skin or muscle tissue (Imazato et al. 2012) as we found here. For this study, the larger problem is being able to swab study animal remains before the DNA degrades or is diluted during decomposition (Imazato et al. 2012) in this semi-aquatic system.
In most of the unattributable mortality cases, the flooded nature of the habitat surrounding the mortality site prevented discovery of any potential predator prints. Because decomposition rates are generally correlated with temperature, with higher temperatures leading to more rapid decomposition (Carter et al. 2007), the decomposition rates in the hot, humid conditions of South Florida can quickly obscure any remaining evidence on the carcasses. Concordantly, we could not reliably attribute a cause of death for two cases where carcasses were recovered. Even less informative were those cases where only the transmitter was found. In the case of UM1, the undamaged condition and protected placement of the transmitter within small mammal tunnels in the vegetation diminish the likelihood of mesomammal or avian depredation, but not death followed by rodent scavenging.
Starvation followed by scavenging may be the most likely cause of some unattributed mortalities, but further fine-scale work would be required to confirm and determine the frequency of incidence. In the case of UM5, we documented the python's poor body condition one week prior to recovering the carcass, indicating the python possibly starved, perished, and was scavenged. Similarly, due to the short period that UM4 was alive in the field, lack of movement from the release point, and no evidence of any disturbance to the area, we believe that this mortality was also attributable to starvation. Though this age class is known to refuse food for the first several weeks of life as they absorb yolk stores, some individuals appear to never acquire any appetite in captivity (e.g., Josimovich et al. 2021 and references therein) but the prevalence of this in free-ranging wild individuals is unknown. These cases might also indicate that phenotypic variation could include some maladaptation to the novel ecosystems experienced by this invasive species (e.g., inappropriate cover, prey recognition/availability, behavioral response to environmental changes, predator avoidance strategies, etc.; see also Mazzotti et al. 2011;McCollister et al. 2021;Anderson et al. 2022;Currylow et al. 2022a). However, we also acknowledge that handling and captivity along with transmitter surgery as part of these research activities could have altered the behavior of the study pythons.
In the case of UM7 where the transmitter was recovered 1.5 m above the ground, the placement and condition of the transmitter along the open habitat location adjacent to bald cypress indicates avian depredation. Avian depredation has long been hypothesized as a threat to young pythons in southern Florida, as avian predators have been recorded to take pythons in their native range (Goel et al. 2017). Although we confirmed no avian depredation events in this study, there are several native avian species in Big Cypress National Preserve that theoretically could incorporate juvenile pythons into their diets. These ophiophagous species include swallow-tail kites (Elanoides forficatus Linnaeus, 1758), red-shouldered hawks (Buteo lineatus Gmelin, 1788), red-tailed hawks (Buteo jamaicensis Gmelin, 1788), ospreys (Pandion haliaetus Linnaeus, 1758), great blue herons (Ardea herodias Linnaeus, 1758), white ibises (Eudocimus albus Linnaeus, 1758), and wood storks (Mycteria americana Linnaeus, 1758;Gehring 2003;Meyer et al. 2004;Roble 2013;Durso et al. 2017).
Although we anticipated that there could be high mortality in this age class, the variety and prevalence of mortality events we found was somewhat unexpected. Understanding how these invasive snakes interact with their environment through ontogeny is only recently being investigated and described (e.g., Taillie et al. 2021;Anderson et al. 2022;Currylow et al. 2022b). Successful invasive predators pose outsized threats to naive native prey species, but the naivete of juvenile pythons may, in turn, make them susceptible to native Everglades predators (sensu Cliff et al. 2022). Though we cannot rule out that the study methods themselves played a role in any of the mortalities, our findings support hypotheses that invasive pythons in Florida may possess some maladaptive traits within this novel environment and that some native predator species (perhaps other reptiles in particular) may have started to reliably recognize juvenile pythons as prey. Although much more work is needed, our observations contribute to limited but growing indications of native species' resilience in southern Florida's Greater Everglades Ecosystem.