Our results clearly indicate that the vast majority of the male offspring in the neotropical bumblebee B. wilmattae is worker produced. Given that at least four out of five colonies were definitely queen right at the point of male sampling, matricide, which occurs in several annual species of social insects [28, 14], can be excluded as cause for the high percentage of worker produced males. Thus apparently in B. wilmattae, the workers win the queen-worker conflict over male production, with about 85% of all genotyped males estimated to be worker derived.
Offspring of up to eight workers could be detected among the genotyped males of a given colony and both patrilines participated in male production in the colonies with doubly mated queens. Since the average relatedness was not significantly below the threshold of r < 0.5 in any of the analyzed colonies worker policing would not have been expected to be adaptive based on kin selection theory alone. The presence of up to eight reproductive workers in each colony indicates that there is no monopolization of reproduction by single workers, but in the absence of behavioral data there is no definitive evidence against the presence of worker policing. The few dominant male producing workers might well police male eggs of other subordinate workers or of the queen. Based on the overall number of workers present during dissection of the colonies C1 and C3-C5, an estimated 3.91% of the workers of these colonies successfully produced male offspring that reached adult stage. In three queen right B. terrestris colonies, that were studied by van Doorn and Heringa , 13.1% of the workers (224.3) became egg-layers during the competition phase, but almost all eggs were eaten by the queen. Thus, just a very small percentage of males reach adult stage and a small amount of workers could successfully reproduce. Worker policing also occurs in monandrous and monogynous social hymenoptera like B. terrestris[14, 15] or Polistes chinensis antennalis. Besides relatedness at least five other factors , among them the cost of worker reproduction  and worker dominance behaviour , can favor the evolution of worker policing also in species with single mated queens and resulting high worker relatedness.
Our results are in accordance with another neotropical species, B. atratus, where Zucchi  found 90% of the males to be worker produced based on the observation of egg laying behaviour. Such worker dominance over male production seems absent in the various temperate Bombus species, that have been studied so far. Although Van Honk et al.  claimed that the queen loses her reproductive dominance after the competition point and workers start male production in B. terrestris, a subsequent study  based on a much larger sample of colonies showed that most males were actually queen produced. These results gained further support by studies applying microsatellite markers and genotyping [13, 11], showing that 95.7% of the males were queen produced. Although 38% of the workers laid eggs, only very few of these eggs developed into adult males, showing that the queen wins the conflict over male production in spite of laying workers in queen right B. terrestris colonies. Also in other temperate bumblebee species the workers seem to lose the conflict over male parentage. For example, in four queen right colonies of B. impatiens only 9% of workers had developed ovaries and only two workers were observed to lay eggs which were later on destroyed by the queen . Similar in B. melanopygus just 19% of males were worker derived , in B. hypnorum 20%  and in B. ignitus 5% . In contrast worker reproduction was higher in B. deuteronymus with between 30% to 50% worker produced males . Although workers do produce males after the death of the queen, this seems to be insufficient to offset the large number of queen produced males over the full season [15, 26].
In our study sampling of males took place in the second half of the colony cycle two to four weeks before the death of the queen, where sampling ended and the colonies were dissected. As a consequence the eggs from which our sampled males had emerged, were laid five weeks to two months prior to the death of the queen and thus in her full presence. The only exception might be colony C2, where we cannot exclude the absence of the queen at the point of sampling; however also in colony C2 the males sampled had been produced three to four weeks earlier depending on the estimated development time of drones in bumblebees [11, 32, 30].
Clearly B. wilmattae deviates from the typical male production pattern found in many temperate bumblebee species. Four of the five analyzed colonies (C2-C5) showed remarkable high numbers of worker produced males (80% - 97%). Only in one colony (C1) the percentage of worker male offspring was less with 60%. Also, this was the colony with the lowest number of successfully reproducing workers, but still the worker produced males outnumbered those produced by the queen.