Behavioral differences following ingestion of large meals and consequences for management of a harmful invasive snake: A field experiment

Abstract

Many snakes are uniquely adapted to ingest large prey at infrequent intervals. Digestion of large prey is metabolically and aerobically costly, and large prey boluses can impair snake locomotion, increasing vulnerability to predation. Cessation of foraging and use of refugia with microclimates facilitating digestion are expected to be strategies employed by free-ranging snakes to cope with the demands of digestion while minimizing risk of predation. However, empirical observations of such submergent behavior from field experiments are limited. The brown treesnake (Serpentes: Colubridae: Boiga irregularis) is a nocturnal, arboreal, colubrid snake that was accidentally introduced to the island of Guam, with ecologically and economically costly consequences. Because tools for brown treesnake damage prevention generally rely on snakes being visible or responding to lures or baits while foraging, cessation of foraging activities after feeding would complicate management. We sought to characterize differences in brown treesnake activity, movement, habitat use, and detectability following feeding of large meals (rodents 33% of the snake's unfed body mass) via radio telemetry, trapping, and visual surveys. Compared to unfed snakes, snakes in the feeding treatment group showed drastic decreases in hourly and nightly activity rates, differences in refuge height and microhabitat type, and a marked decrease in detectability by trapping and visual surveys. Depression of activity lasted approximately 5-7 days, a period that corresponds to previous studies of brown treesnake digestion and cycles of detectability. Our results indicate that management strategies for invasive brown treesnakes need to account for cycles of unavailability and underscore the importance of preventing spread of brown treesnakes to new environments where large prey are abundant and periods of cryptic behavior are likely to be frequent. Characterization of postfeeding behavior changes provides a richer understanding of snake ecology and foraging models for species that consume large prey.

Keywords: automated telemetry; behavioral ecology; brown treesnake (Boiga irregularis); detectability; foraging ecology; submergent behavior.

Figure 1

Brown treesnake ingesting a rodent meal (photograph by Michael Hogan, taken during a separate study)

Figure 2

Nightly activity plots from the same snake during low (a) and high (b) activity periods. Shaded gray areas indicate hours of darkness from 1800 to 0700. Blue and red dashed lines represent ranges of pulse interval states. Black lines at y = 1200 indicate state transitions (“tips” of the transmitter) identified by the postprocessing algorithm

Figure 3

Hourly mean transmitter tipping rates over the seven days following the beginning of trials

Figure 4

Hourly predicted activity rates based on post hoc pooling of treatment group data by days postfeeding, and pooling all control group days into a single level. Shaded areas represent ± 1 standard error of the estimate

Figure 5

Mean hourly tipping rates by night for treatment (fed) snakes compared to the baseline hourly rate for unfed control snakes (all nights pooled). Shaded areas represent ± 1 standard error of the estimate. “***” p < 0.001; “**” p < 0.01; “*” p < 0.05; “NS” not significant at α = 0.05

Figure 6

Fed snakes (red) had higher predicted refuge sites (±1 SE) than unfed snakes. Predictions based on the fixed‐effects model. Mixed‐effects significance values: meal effect, p ≪ 0.001; snout‐vent length effect, p = 0.001. Shaded areas represent ± 1 standard error of the estimate

Figure 7

Logistic regression fits and standard errors for differences in proportion of refuge type use between fed and unfed snakes. Fits are from fixed‐effects versions of top mixed‐effects models (constrained to include the MEAL term for sake of comparison). Significance values from top mixed‐effects model (see Table 2). “***” p < 0.001; “**” p < 0.01; “*” p < 0.05; “NS” not significant at α = 0.05

Figure 8

Estimated capture probability (detectability functions) for fed and unfed treatment groups by effort type and snake size, as predicted by fixed‐effects logistic regression. Shaded areas represent ± 1 standard error of the estimate. Note that the detection function for trapping may be depressed by simultaneous visual searching (see Discussion)