Trade-offs between complementarity and redundancy in the use of different sampling techniques for ground-dwelling ant assemblages

Sampling ground-dwelling ants in the field is relatively fast, but sorting and identifying in the laboratory is costly and time-consuming. Using sub-samples collected in 25 km(2) grids, we surveyed ant assemblages using sardine baits, pitfall traps and Winkler extraction in three Amazonian vegetation types. Combining all three techniques detects the greatest number of species, but may be inefficient. Therefore, we compared the pooled results from the three techniques to results using one or two techniques combined. We evaluated whether the extra information acquired by adding a sampling technique compensates for the time and money associated with the extra processing. We also evaluated the consequences of the reduced effort on the retention of ecological information captured by the three techniques, using soil clay content, terrain slope and altitude as predictor variables in an ecological analysis. Pitfall traps captured the largest number of species and had the highest congruence with ant assemblages recorded by other techniques. Redundancy analysis indicated that pitfall-trapping is the most efficient technique, allowing reduction of 48% in cost and 43% in time. The loss of information about species richness when using only pitfall traps is apparently compensated by the saving of cost and time in the field and laboratory, because use of this technique alone was sufficient to detect all the responses of the ant assemblage to environmental variables that were detected by other techniques. These results indicate that considerable gains in efficiency can be obtained in most Amazonian-forest monitoring programs for ants by using only pitfall traps. (C) 2012 Elsevier B.V. All rights reserved.