Trends in mammalian predator control trapping events intended to protect ground-nesting, endangered birds at Haleakalā National Park, Hawaiʻi: 2000–14

Predation and habitat degradation by non-native species are principal terrestrial threats to the federally endangered Hawaiian Petrel (ʻuaʻu, Pterodroma sandwichensis) and Hawaiian Goose (nēnē, Branta sandvicensis) within Haleakalā National Park (HALE), Maui, Hawaiʻi. Since 1981, HALE has maintained a network of live traps to control invasive mammalian predators and protect these endangered birds. To evaluate trapping efficiency in HALE, we evaluated four types of trap outcomes for the years 2000–14: Bait Lost (62 percent), No Event (23 percent), Trap Triggered (10 percent), and Predator Event (Rat Caught, Cat Caught, or Mongoose Caught; 4 percent). We used a multinomial logistic regression model to explore trends in the probabilities of broad outcomes (No Event, Other Event [Bait Lost or Trap Triggered], or Predator Event [Rat Caught, Cat Caught, or Mongoose Caught]). Temporal variations in the probabilities of No Event, Other Event, or Predator Event were best explained by ʻuaʻu season (off-season, pre-laying, incubation, or nestling), month, year, and seasonal rainfall with greater probabilities of Predator Event during the ʻuaʻu nestling period (July–October). The probability of Predator Event or Other Event decreased with increased rainfall. Spatial analysis showed that percent vegetative cover and vegetation type best explained variations in the probabilities of trapping outcomes with the probability of Predator Event being greatest in developed and tree covered areas. The proportion of trapping events that resulted in Rat Caught was at least 20 times greater than the proportions of events resulting in Cat or Mongoose Caught throughout the 15-year management period. Temporal analysis showed that season, year, and maximum temperature best explained variations in probabilities of Predator Event; the probability of Rat Caught was greatest during the ʻuaʻu pre-laying and incubation periods (February–June), was greater during periods of warmer maximum temperatures, and overall, increased over the 15-year management period. The probability of Mongoose Caught was greatest during the ʻuaʻu offseason (November–January), decreased through time (2000–14), and decreased with increasing weekly maximum temperatures. Trends in Cat Caught were hard to detect because of small sample sizes, though slight trends indicated cat captures were most frequent during the ʻuaʻu off season and less frequent through time (2000–14). The probability of a Cat Caught event was also negatively correlated with weekly temperatures. Spatial analysis showed elevation best explained variations in probabilities of capture for rats, cats, and mongoose. Overall, predator catches were fewer at higher elevations, and of predators caught at higher elevations, the clear majority were rats. Our results are being used by HALE Endangered Wildlife Management staff to evaluate existing methods for predator control and efficacy of existing trap-based control strategies intended to protect ʻuaʻu and nēnē.