Development and application of novel ornithological survey methods for the detection of cryptic avian indicator species that predict grassland health.

Abstract

The anthropogenic pressure on South African grasslands to meet the needs of the burgeoning human population has exposed them to extensive permanent transformation and degradation. Indicator species may identify changes in the grassland ecosystem. One such indicator species for natural sourveld grassland condition in South Africa is the red-winged francolin (Scleroptila levaillantii), whose population density is negatively correlated to grazing intensity and annual burning. Pointing dogs (Canis lupus familiaris) have been used extensively to aid ecologists in detecting these and other cryptic gamebirds to establish abundance. Here, a reliable method was developed to count cryptic gamebirds in the Greater uMgeni Vlei Expansion Area, KwaZulu-Natal Midlands, South Africa, where the route through a survey site was flexible. A variation to the existing distance sampling technique was proposed where the dog global positioning system (GPS) track was the transect line. The study investigated the effect of varying environmental conditions on the distance from which a pointing dog could reliably and consistently detect a bird and allow calculating a detection distance based on influential environmental variables. Between March – October 2021, using pointing dogs fitted with GPS devices, controlled and uncontrolled trials were conducted on Japanese quail (Coturnix japonica) and red-winged francolin in their natural habitat, respectively, to establish the environmental conditions that affect detectability and the detection distance from which a dog can detect a bird of known location. A total of 21 surveys were conducted (August 2020 – October 2021), on four survey sites by one or two pointer dogs fitted with GPS devices, to establish the population densities and territory of red-winged francolin. Individual area of search, established from the detection distance based on nominal wind speed and GPS track, was calculated. The redundancy in area of search enabled the evaluation of relative proficiency of detection of red-winged francolin. Of the environmental variables monitored, only nominal wind speed significantly influenced detection distance, where an increase of one-knot wind strength resulted in an increase in detection distance by 0.64 m. This enabled an area of search, considerate of influential environmental conditions, to be derived and the probability of detection within that search area = 1. Results showed significantly better precision and accuracy when surveying with two dogs when compared with one dog. The calculation of detection distance, where the probability of detecting a bird at this distance = 1, addresses the bias of varying scenting conditions. The established area of search, where the probability of detecting a bird within this area = 1, addresses the situation where known coveys in an area of known size remain undetected. Since the area of search is independent of time spent searching and normalised for redundancy, the bias introduced by varying physical aptitude is mitigated. Consideration for the application of this method should be given to the environmental conditions under which the surveying is planned since the detection distance function is derived for conditions at the present study sites. These techniques, based on a variable survey route through the survey site, may be used by citizen scientists to assist land managers, conservationists, and ecologists in establishing the abundance of red-winged francolin, contributing to burning and grazing regime management to enhance conservation efforts for the species.