Abstract
Identification of the resources an animal is selecting allows us to understand how animals use what is available to meet their requirements for survival, growth, and reproduction. Understanding the process by which selections are made is essential for successful conservation management as it allows us to investigate the impacts of modified landscapes and to develop informed policies and remedial measures to mitigate current and future negative impacts. However, despite the important role for conservation management, the quantification of resource selection is not always straight forward. Obtaining and interpreting resource selection information can be problematic for many species (Hebblewhite and Haydon, 2010, Aarts et al., 2008), potentially leaving important management questions unanswered.
The New Zealand pāteke/ brown teal (Anas chlorotis) are a cryptic, predominantly nocturnal dabbling duck, that inhabits wetland areas. Formerly widespread and numerous throughout New Zealand, pāteke have undergone a decline described as ‘arguably the most dramatic of any New Zealand species’ (Williams, 2001). Pāteke are ranked as ‘nationally endangered’ by the Department of Conservation (O'Connor et al., 2007) and have been listed as ‘endangered’ on the International Union for the Conservation of Nature (IUCN) Red List since 2000 (IUCN website http://www.iucnredlist.org). Of particular concern, is the alarming juvenile starvation rate within the pāteke population at Great Barrier Island (Watts et al., 2016). This starvation has not been recorded in any other pāteke population. These recent declines in both range and number of pāteke, as well as the unprecedented juvenile starvation rates for Great Barrier pāteke, mean that identification of the resources used by pāteke is of particular importance.
Traditional methods of studying habitat and diet selection can be limited when the study species is widely dispersed, nocturnal, or utilises habitats such as wetlands and ponds with restricted safe access for researchers (Beyer and Haufler, 2008). To date, the most widely used tool for habitat selection studies is Very High Frequency (VHF) radio-telemetry (Aebischer et al., 1993). However, despite its significant application in habitat studies, radio telemetry can be intrusive, requires a high number of field-work hours, and often has a high sampling bias (Aarts et al., 2008). These limitations can result in incomplete or inaccurate estimates of habitat use (Beyer and Haufler, 2008). Traditional methods of dietary study (i.e. faecal samples or foraging observations) can often also be limited for the same types of species. For example, faecal samples can be difficult to obtain from some habitat types, and can be incomplete with only parts of some prey being recoverable (Barrett et al., 2007). Additionally, traditional methods tend to reflect ingested rather than assimilated items, resulting in biased dietary estimates due to differential digestion.
Pāteke are an example of a species for which VHF monitoring is of limited efficacy. Radio-telemetry is the main monitoring tool currently being used in pāteke management areas, however, as well as being invasive and expensive in terms of staff hours, (O’Connor et al 2007), their predominantly nocturnal behaviour (Williams 2001) and use of habitats which are dangerous and difficult to access at night-time, are additional barriers to gaining a comprehensive understanding of habitat selection. Despite extensive and long-term VHF monitoring of pāteke, and several pāteke habitat use studies (Parrish and Williams 2001, Barker and Williams 2002, Worthy 2002, Fraser 2005, Dumbell 1987, MacKenzie and Fletcher unpublished, Fraser and Beauchamp in preparation), as well as a thorough diet study using conventional techniques (Moore et al., 2006), there remain significant gaps in our knowledge, leaving major management questions unanswered (Moore et al., 2006, B.T.C.T., 2009, Fraser and Beauchamp, 2009, Willoughby, 2004, Weller, 1974).
In this study, I use techniques not previously applied to study pāteke resource selection. I used new generation GPS/GSM units to collect location data from two Pāteke populations, Okiwi Basin on Great Barrier Island and Mimiwhangata Coastal Reserve in Northland. I applied Maxent models to the GPS data to identify and compare the use of habitats between the two study populations (Phillips et al., 2004, Ghisla et al., 2012). I then identified correlations between temporal (e.g. cattle presence, rainfall) and spatial variables (e.g. distance to other features, size of feature) in relation to the key habitat features. To identify and compare the diet selection of the two populations, I used stable isotopes from pāteke blood and feather tissue and applied MixSIR mixing models to the collected data.
A total of 109 individual pāteke were captured for this study, 51 at Mimiwhangata, and 58 at Okiwi. There were 36 GPS tags deployed on pāteke at Mimiwhangata, of which 12 were successful, collecting a total of 430 GPS data points. There were 28 GPS tags attached at Okiwi, 14 of which were successful and collected a total of 750 GPS location points. Habitat selection was found to be the same between the two populations. However, the relative availability of preferred habitats between locations was different, with less of the preferred habitats being available at Okiwi. Ponds were found to be the most strongly selected habitat type, with paddocks being the most commonly used. There was a strong preference for ponds with at least 50% of the bank area vegetated, over ponds with less vegetation cover. Pāteke at both locations showed a preference for paddocks that had either currently or recently had cattle present. Forest interior was the least likely habitat type to contain pāteke for either population, with forest edges having a slightly higher probability than interior forest for the Okiwi population.
There was a significant difference in body mass between the two pāteke populations, with Mimiwhangata birds being 19% heavier than their Okiwi conspecifics. The δ15N ratios were significantly higher for the Mimiwhangata birds, suggesting more animal sources, possibly invertebrates, in the diet of these pāteke. The Mimiwhangata females had a different diet to the males, with higher δ15N ratios, suggesting more invertebrates in the female diet. The diet of Mimiwhangata pāteke was more varied between individuals than the diet of the Okiwi population.
Okiwi juvenile pāteke had lower δ15N levels than those of the adults, and their dietary range was more restricted than that of the adults. Of all variables tested, age of pāteke had the most significant correlation with probability of survival for Okiwi pāteke, with adults pāteke at Okiwi being significantly more likely to survive than juveniles. Use of ponds within the home range of an individual was correlated with an increased probability of survival; so too was use of wetlands, but to a lesser extent.
The findings of this study can be applied to help direct the management of pāteke habitats, and aid in enabling conservation managers to make decisions for effective, positive habitat modifications in areas of use for pāteke. Specifically, in regard to establishing ponds, and managing cattle access and vegetation around pond areas within pāteke occupied areas. This study also provides at least a preliminary indicator of the cause of the Okiwi juvenile starvation, ultimately contributing to the recovery of the species.