A state-space modelling approach to wildlife monitoring with application to flying-fox abundance

Abstract

Monitoring flying-foxes is challenging as their extreme mobility produces highly dynamic population processes, considerable logistic difficulty, and variability in estimated population size. We report on methods for inferring population trend for the population of the spectacled flying-fox (Pteropus conspicillatus) in Australia. Monthly monitoring is conducted at all known roost sites across the species’ range in the Wet Tropics Region. The proportion of animals in camps varies seasonally and stochastic environmental events appear to be influential. We develop a state-space model that incorporates these processes and enables inference on total population trends and uses early warning analysis to identify the causes of population dynamics. The model suggests that population growth rate is stable in the absence of cyclones, however, cyclones appear to impact on both survival and reproduction. The population recovered after two cyclones but declined after a third. The modelling estimates a population decline over 15 years of c. 75% (mean r = − 0.12yr−1 and belief of negative trend is c. 83%) suggesting that conservation action is warranted. Our work shows that a state-space modelling approach is a significant improvement on inference from raw counts from surveys and demonstrates that this approach is a workable alternative to other methods.

Monitoring hunted species of cultural significance: Estimates of trends, population sizes and harvesting rates of flying-fox (Pteropus sp.) in New Caledonia

“…Counts of animals typically suffer from high variance, especially when considering highly gregarious and mobile species such as flying-foxes [43–45]. The large variance observed here (CV of global average: 151%) certainly reduced the power to detect a significant trend over seven years.…”

“…The two species cannot be distinguished during fly-out and counts refer thus to total abundance. Fly-out counts realised from a vantage point were preferred over counts realised below the roost [38–42], despite this technique has sometimes been judged as less accurate [43–45]. However, we believe this method was the most appropriate for counting flying-foxes in New Caledonia as: i) roosts are located in dense tropical forests where the canopy often hides a large proportion of animals, ii) flying-foxes are highly sensitive to disturbance when roosting, possibly because of intensive hunting, and do not tolerate observers walking close to roosting trees (typically <300m) and iii) maximal roost size recorded in New Caledonia (19,000 individuals) did not exceed observers’ counting capability [38,39].…”

Extreme mobility of the world’s largest flying mammals creates key challenges for management and conservation

“…poliocephalus , and P . scapulatus are estimated through Australia’s National Flying-Fox Monitoring Program [ 66 ], and roosting away from known roosts is identified as the major contributor to uncertainty around flying-fox population trend estimates [ 67 , 68 ]. We suggest that the accuracy of the monitoring could thus be substantially improved by the annual inclusion of tracked individuals to help reveal previously unidentified roosts.…”

Using weather radar to monitor the number, timing and directions of flying-foxes emerging from their roosts

“…Further, Westcott et al . 1,46 identified roosting away from known roosts as the major contributor to uncertainty around current flying-fox population trend estimates. Future work could include automatic detection by radar of unknown flying-fox colonies based on the ‘signature’ of reflectivity and velocity 47,48 , and so help improve the accuracy of existing monitoring programs.…”