Aspects of the ecology of the native frogs Leiopelma Archeyi and L. Hochstetteri, and the impact of compound 1080

Abstract

A study of Archey's frog (Leiopelma archeyi) and Hochstetter's frog (L. hochstetteri) was undertaken in the Tapu area, Coromandel, in 1995 to assess the impact of 1080 poison. Frog populations within and outside the poisoned area were monitored by repeated monthly counts over strip transects. Frog numbers were assessed using generalised linear models. Test power (β, the probability of correctly identifying the presence of a 1080 effect on frog numbers), was determined using the non-central chi-squared distribution. Vegetation was assessed by two-way indicator species analysis (TWINSPAN) to establish the similarity of habitat in 1080-treated and untreated areas. Poison application did not cause a decline in L. archeyi. The low numbers of Hochstetter's frogs found were insufficient to determine with statistical certainty whether any population decrease occurred. Theoretically the native frog populations are unlikely to be affected by 1080 used in pest control; however, the applicability of results from this field study to other 1080 operations may be compromised by high rainfall following bait distribution, and subsequent rapid poison breakdown. The necessity of analysing test power (i.e. probability that a decline will be detected) in any future studies of 1080 impact is stressed. L hochstetteri density varied from 0-22 frogs per 100m length of stream (0-15 per 50m strip transect), and 0-8 frogs/100m2 (0-8 per 50m strip transect) on terrestrial forest floor transects. Hochstetter's frog appears to be reasonably mobile on land, and to utilise terrestrial habitat more commonly than previously indicated. L. hochstetteri individuals found in terrestrial locations were significantly larger (p≤0.005) than those discovered along creeks. L. archeyi density ranged from 1-35 frogs/100m2 (1-35 per 50m strip transect) on forest floor transects. A scarcity of L. archeyi sightings along creeks (3 frogs found) confirms reports of near-exclusive terrestrial habitat use by this species in Coromandel. Both species were found predominantly under rocks (96% of L. archeyi sightings, 93.3% of L. hochstetteri sightings). Pairs or groups of frogs were found on numerous occasions. Social or environmental factors are at least partially responsible for multiple sightings in L. hochstetteri; the occurrence of L. archeyi pairs and groups appears density-dependent. Monthly counts of L. hochstetteri along creeks were significantly correlated (p≤0.05) to maximum and minimum temperatures leading up to and including the search day, and rainfall over the preceding ten days. No consistent correlation was found between L. archeyi numbers and temperature or rainfall variables. Diurnal searches for L. archeyi must cover potential retreat sites in vegetation, as well as the rocks and logs usually examined, if relative or absolute density estimates are to be obtained. L. hochstetteri along Tapu River headwaters showed no preference for retreat types (i.e. the proportion of retreats occupied was not influenced by retreat type), indicating normal searches of only rocks may be sufficient to gauge relative density in this area. These results should not be extended to other L hochstetteri populations without confirming a similar lack of preference in other areas. Frog density indices commonly used in native frog research were compared for reliability by Pearson's correlation coefficient (r) using search effort data collected during field work. Search effort indices (frogs per number of retreat sites searched, or frogs per hour of search effort) are subject to mild or extreme bias from variation in search speed, terrain, and retreat site density, and are unsuitable for comparison between different studies. Indices of sample densities, based on a defined search area (m2) are not unduly affected by these factors although subject to more controllable influences. Consistent application of search effort by experienced searchers, and conducting searches under comparable recent climatic conditions and local moisture availability, are likely to minimise extraneous variation in area-based indices.