USGS - science for a changing world

South Florida Information Access (SOFIA)

publications > paper > a field test of attractant traps for invasive Burmese pythons in southern Florida > discussion

A field test of attractant traps for invasive Burmese pythons (Python molurus bivittatus) in southern Florida


Literature Cited
Figures and Tables

We successfully captured the target species in traps with both entrance types, and our various detection methods yielded an estimate of python density in the study area. However, an experimental trap trial with a capture rate <0.05% per trap night cannot be considered particularly successful in terms of potential control of an invasive species such as the Burmese Python.

Standardised visual surveys yielded no python observations, but two pythons were opportunistically observed during trap checks. These results confirmed our suspicion that diurnal visual surveys are unlikely to be effective for pythons in summer, during which pythons appear to engage in movements primarily during nocturnal hours; warm-season visual surveys conducted by driving along roads at night have resulted in capture of several hundred pythons in southern Florida (Snow, Mazzotti, unpubl. data). We also doubt that nocturnal pedestrian surveys in summer using lights would be very effective, because a pedestrian can cover only a small area in an evening and the thick vegetation present in much of southern Florida would reduce detection rates. In contrast, diurnal surveys in the winter months can yield much higher catch per unit effort along cleared levees and other areas, because pythons emerge from concealing vegetation to bask. Continued accumulation of insights on python activity and detection probability will allow land managers to focus control efforts, including trap deployment, on times and places that will maximise catch per unit effort.

Most large-bodied pythons (and other giant constrictor snakes) appear to employ ambush predation as their primary means of capturing prey (Branch 1988; Dirksen 2002; Reed et al. 2007; Reed and Rodda 2009), although there are a few observations of active foraging (Martins and Oliveira 1998; Alexander and Marais 2007). Ambush predation implies relatively infrequent movement, which could reduce rates of encounters with control devices such as traps (Shiroma and Akamine 1999). Moreover, an ambush predator might approach the vicinity of an attractant traps, detect the attractant scent, and settle into an ambush posture rather than actively investigating the trap and encountering the entrance. For these and other reasons, attractant traps may be generally ill-suited to capturing large numbers of ambush-foraging snakes. However, our results indicate that at least some pythons entered traps, and we did not observe any pythons in ambush postures adjacent to traps. Radiotelemetry and other research avenues will be needed to determine daily and seasonal movement behavior of pythons, and results may allow prediction of the likelihood of pythons encountering attractant traps in various habitats and trap densities. Such information would be a vital component of planning for integrated python control efforts.

Native and introduced rodents were abundant by the end of the study, likely supplying a ready prey resource for pythons. Indeed, high rodent biomass associated with high primary productivity and vegetable availability may have drawn pythons into the Frog Pond area in previous years when the site was in intensive agricultural production, and apparently high python density was a primary reason for selecting this site for the trap experiment. High rodent abundance may have adversely affected our python capture rates. Capture rates in snake traps often decline when prey are abundant (Gragg et al. 2007; Tyrrell et al. 2009), possibly because snakes reduce foraging frequency in a prey-rich environment and thus have fewer opportunities to encounter traps. Necropsies of pythons collected from the Frog Pond in previous years revealed high predatory success, with as many as 14 rats identified from a single python's gastrointestinal tract. In prey-rich environments, an ambush predator may not need to change ambushing locations frequently, thus reducing encounter rates with attractant traps. Scent from abundant prey may also mask the scent from attractant traps, further reducing encounter rates.

We had expected that post-trapping disc-harrowing operations would kill any pythons over which the harrow ran, but instead we observed a high apparent survival rate (>60%) of 11 pythons found during these operations. None of the 11 pythons had previously been captured in a trap. Lack of recaptures could be due to behavioral responses in previously-trapped pythons, including the tendency to avoid traps (trap shyness) or because they fled the study site after capture. Alternatively, trapped pythons may have emigrated from the study site during the experiment as part of their regular pattern of activity. Based on their body sizes (1740 - 2240 cm SVL), the lack of subterranean refugia in the area that was harrowed, and the thoroughness of our searching behind the harrow, we believe it unlikely that the three marked pythons were killed by the harrow and missed by observers.

Disc-harrowing results have implications for previous estimates of python density in the Frog Pond area. In previous years, python density estimates ranged from 0.044 to 0.109/ha across a large (505 ha) area, but these were largely based on finding dead pythons in the fields, often several days after harrowing, with python presence indicated by the presence of vultures. If our finding that 7 of 11 pythons actually survived harrowing is a reasonable estimate of survival rate, adjustment of results from previous years in the Frog Pond would yield a range of 0.069-0.171 pythons/ha. This range is comparable to the estimate of 0.136/ha we derived from harrowing 81 ha in 2009. However, these are minimum density estimates; disc-harrowing occurred over a number of days and some pythons may have left the study site in response to mechanical disturbance and vibration resulting from nearby harrowing activities, or we may not have detected some pythons killed and buried by the harrow.

It is possible that both rodent and python densities were relatively low in April 2009 when the fields were allowed to go fallow, and rodent densities probably increased exponentially in ensuing months in response to rapid growth of vegetation in the fields. If pythons select habitats with high prey abundance, then it follows that python density would have increased in the Frog Pond area in response to increasing rat density. Python densities may thus have been low during early parts of the trap experiment, rendering difficult any attempt to estimate trap success over the duration of the trap experiment as a function of the number of pythons present.

Our traps captured relatively few non-target organisms, and survival of non-targets was 100%. Conservation concerns associated with deaths of native non-targets captured in python traps may be minimal, with the caveat that results are likely to vary across habitats with different animal communities. If conservation concerns are allayed, then it may be possible to check traps at less-frequent intervals; labor is typically the most expensive component of trapping budgets and fewer trap checks per unit time would greatly reduce costs associated with operational python trapping.

There were major discrepancies in the perception of non-target species composition and relative abundance resulting from visual surveys, trap captures, and disc-harrowing in the Frog Pond area. Disc-harrowing exposed 18 rat snakes (Pantherophis), for example, but none were captured in traps or observed during visual surveys. Conversely, 24 amphibians were captured in traps but only 6 were observed during visual surveys and none were observed during harrowing. The innate biases associated with any detection or capture method must be considered when attempting to assess species composition or relative abundances, and any single method will likely yield biased results.

Our field experiment resulted in few captures of the target species, despite a moderately large effort and a pool of available targets that at any given time probably exceeded the number captured. Behavioral (e.g., ambush predation) or environmental (e.g., high prey density) factors may have reduced the likelihood that a python would encounter, be attracted to, and enter our traps. Alternatively, our traps may be suboptimal in their design or choice of attractant, potentially failing to attract pythons or frustrating their efforts to enter. Additional research would be invaluable in helping to distinguish between these hypotheses. However, developing a snake trap that maximises capture rates can be a lengthy process. For example, field tests of 49 different trap designs and >24,000 trap-nights were required to settle on an optimal design for capture of brown treesnakes on Guam (Rodda et al. 1999b), yet even the less effective designs often had capture rates approximately two orders of magnitude higher than we achieved for pythons. We anticipate that a series of additional experiments in both field and captive settings will be required for development of an optimal python trap. As an example, unbaited traps placed along drift fences (Jackley 1943; Gibbons and Semlitsch 1981) and other methods designed to intercept moving pythons may capture individuals that are not motivated by prey (e.g, aphagic individuals during the mating season), as well as reducing bycatch of non-target predators attracted to bait in attractant traps.

There are now three species of exotic giant constrictors established in Florida (Snow et al. 2007a; Reed et al. 2010) and suitable climatic conditions for other species of giant constrictors appear to exist in the United States (Reed and Rodda 2009), so land managers will continue to request effective control tools for these taxa. Our results suggest that traps are unlikely to result in eradication of pythons at landscape scales, but that additional testing may result in traps that are a useful component of management efforts that include a range of control tools. More generally, our results highlight the difficulty of developing control tools for invasive top predators that are cryptic and which may have behaviors that make them unlikely to regularly encounter control tools. Overall, we conclude that refinement of trap technologies and further assessment of various environmental contributors to trap success will be necessary before it will be possible to produce a thorough assessment of the utility of traps for invasive giant constrictor snakes.

< Results | Acknowledgments >

| Disclaimer | Privacy Statement | Accessibility |

U.S. Department of the Interior, U.S. Geological Survey
This page is:
Comments and suggestions? Contact: Heather Henkel - Webmaster
Last updated: 04 September, 2013 @ 02:04 PM (KP)