A scientific note on the ant pitfall for quantitative diagnosis of Varroa destructor

Abstract

The mite Varroa destructor is a global challenge for apiculture and accurate quantification crucial for adequate and timely pest management. However, foraging ants are regularly found in hives and may interfere with mite diagnosis. Here, we quantify for the first time the impact of ants. We expect lower mite numbers on bottom boards with foraging ants and that estimates of phoretic mites are ant independent. From July to August 2007–2009, the experiments were conducted with 64 queenright honey bee colonies (predominantly Apis mellifera carnica, Table I). One apiary was used for 3 years, but each time, new colonies were monitored. All colonies (~11 frames of bees, 6–10 brood frames) were housed in Dadant hives (12 frames) with bottom board inserts for mite quantification (Imdorf et al. 2003) and placed in groups of four or five each on hive stands with four steel polders (50 cm aboveground). All colonies were treated in summer and fall using formic and oxalic acid (Imdorf et al. 2003). To quantify the impact of ants, we added traps (water-filled buckets [∅=20 cm]) to each of the four steel polders of the hive stands. At least once a month, water was refilled, and the surrounding vegetation was cut. The controls remained without traps. Quantifications were conducted weekly by removing the bottom board inserts and counting all mites and ants. In 2009, we also collected weekly 200 bee workers from the brood nests of 18 colonies for 8 weeks to evaluate the number of phoretic mites following Ritter and Ruttner (1980). For that purpose, nine colonies received traps (= treatments), and nine remained without (= controls) for 4 weeks; then, the groups were exchanged by relocating traps. Thus, we obtained from the same colonies mite infestation loads from both bottom board counts and phoretic mite estimates with or without ant traps. We also monitored all stands to investigate whether they were exposed to ant foraging. Ants were collected for taxonomic identification using morphometrics (Seifert 2007). We performed Kruskal–Wallis and Mann–Whitney U post hoc tests to compare ant and mite numbers. To test whether there is a regression between ant and mite numbers, we constructed a Linear Mixed Model (LMM), with log-transformed V. destructor numbers as the dependent variable, log-transformed ant numbers as fixed effect and colony as random effect, resulting in normally distributed residuals: