Evaluation of Pugmark Census Technique

Abstract

Effective conservation strategies hinge on reliable population estimation technique. The potential of pugmark as a tool for population estimation of tigers was statistically examined. Eleven variables were selected and they were found effective in discrimination between individual tigers. These variables also discriminated between male & female in a population of 10 known tigers with 100% accuracy. This sex discrimination modal was found 97.4% accurate during validation over three new known tigers. The above said variables were also robust for pooling left and right pugmark. The soil depth of range 0.5 cm to 2 cm was found to be ideal for PI P preparation, since the 11 predictor variables were examined to be insensitive towards shape and size change of pugmark due to substrate condition. Classification of 17 individual tigers was achieved with 97% accuracy using stepwise discriminant function analysis. The statistical protocol developed in this study was found to be 100% accurate in predicting the correct number of tigers and correct classification of pugmarks, during validation in a blind set simulated 'census-exercise' from a known number of tigers (some of the track sets used for this analysis were not used for developing the DFA model). Variability in pugmark tracings due to tracer's effect was found to be significant. ANOVA (p-value 0.002 with df-1, df-53) results suggest a significant difference between variables measured from tracings and those from photos. However, DFA was unable to discriminate between tracings and photographs suggesting that either could be used in the model. Seventy five percent of the "experts" (n=8) were 100% correct in deciphering individual tigers in a blind test involving 15 tiger pugmarks from 7 individual tigers. The remaining 2 experts overestimated the tiger numbers by 1. The experts had an accuracy of 92% (sd = 7.8) in correct classification of tiger pugmarks. The other groups of respondents, ecologists (n=11) and lay persons (n=15) overestimated the tiger numbers and had an accuracy rate of 670/0 (sd = 24), and 640/0 (sd = 6) respectively in classifying pugmarks correctly. This study suggests that pugmarks can be used as a tool for population estimation of tigers, if subjected to analytical protocol developed here in within the constraints outlined.