Aversion training wild kea with anthraquinone-treated repellent in non-toxic baits prior to 1080 operations does not improve survival outcomes

In New Zealand, the Nationally Endangered kea (Nestor notabilis) is a large endemic parrot susceptible to non-target mortality from 1080 (sodium fluoroacetate) which is used to control introduced mammals that threaten native biodiversity and livestock. Across all 1080 operations monitored since 2008, mean kea mortality is estimated to be 13%. Public concern around kea by-kill remains high and, consequently, significant effort is directed towards mitigating 1080 risk. Anthraquinone has previously shown potential as a suitable kea repellent, with captive kea learning and retaining an aversion to non-toxic baits containing 2.7% anthraquinone for at least ten months after repeated exposure. We tested whether wild kea could be aversion trained using these repellent baits prior to two large aerial 1080 operations during spring 2022 in central Westland (Otira-Taipo) and Arthur’s Pass (Arthur’s Pass West) and if this reduced kea mortality. Kea bait interactions and consumption were scored using trail camera video footage from 23 alpine sites where aversion training took place during the four-month period prior to 1080 baiting. Of the 40 radio-tagged kea present within the treatment area during one or both operations, 11 died from 1080 poisoning (7 of 24, or 29%, at Otira-Taipo and 4 of 18, or 22%, at Arthur’s Pass West). Aversion training with anthraquinone was therefore unsuccessful at reducing mortality to below the average. While rates of interaction with repellent baits were high overall, the mean quantity of baits consumed by 33 banded individuals was just 0.09 of a bait, well below the amount eaten by kea in the captive trials, and likely below the quantity required to create a lasting aversion. Very few adverse responses to baits were observed. A more desirable repellent for wild kea would induce an immediate and unequivocal negative effect upon first interaction with baits to create an immediate association. We recommend that no further attempts to aversion train wild kea prior to 1080 operations should be undertaken until such a repellent is developed or methodology is developed to mitigate sub-repellent doses. If wild kea do not consume enough repellent bait to induce an emetic response adequate to ingrain a strong aversion prior to a 1080 operation, it is possible that aversion training could increase the risk during toxic operations, because repeated exposure without ill-effect may unnecessarily cue birds onto baits.