Sexual selection and the evolution of male pheromone glands in philanthine wasps (Hymenoptera, Crabronidae)

Background Sexual selection is thought to promote evolutionary changes and diversification. However, the impact of sexual selection in relation to other selective forces is difficult to evaluate. Male digger wasps of the tribe Philanthini (Hymenoptera, Philanthinae) scent mark territories to attract receptive females. Consequently, the organs for production and storage of the marking secretion, the mandibular gland (MG) and the postpharyngeal gland (PPG), are subject to sexual selection. In female Philanthini, these glands are most likely solely subject to natural selection and show very little morphological diversity. According to the hypothesis that sexual selection drives interspecific diversity, we predicted that the MG and PPG show higher interspecific variation in males than in females. Using histological methods, 3D-reconstructions, and multivariate statistical analysis of morphological characters, we conducted a comparative analysis of the MG and the PPG in males of 30 species of Philanthini and three species of the Cercerini and Aphilanthopsini, two related tribes within the Philanthinae. Results We found substantial interspecific diversity in gland morphology with regard to gland incidence, size, shape and the type of associated secretory cells. Overall there was a phylogenetic trend: Ensuing from the large MGs and small PPGs of male Cercerini and Aphilanthopsini, the size and complexity of the MG was reduced in male Philanthini, while their PPG became considerably enlarged, substantially more complex, and associated with an apparently novel type of secretory cells. In some clades of the Philanthini the MG was even lost and entirely replaced by the PPG. However, several species showed reversals of and exceptions from this trend. Head gland morphology was significantly more diverse among male than among female Philanthinae. Conclusion Our results show considerable variation in male head glands including the loss of an entire gland system and the evolution of a novel kind of secretory cells, confirming the prediction that interspecific diversity in head gland morphology is higher in male than in female Philanthini. We discuss possible causes for the remarkable evolutionary changes in males and we conclude that this high diversity has been caused by sexual selection. Electronic supplementary material The online version of this article (doi:10.1186/s12862-017-0963-6) contains supplementary material, which is available to authorized users.


Definition and coding of aggregated characters of head glands of males and females
In order to test whether the interspecific diversity in the morphology of postpharyngeal gland (PPG) and mandibular gland (MG) is higher in male than in female Philanthinae, the characters and character states of both sexes had to be represented in the same coordinate system by a categorical principal components analysis (CATPCA). To accomplish this, we created an aggregate data matrix by combining and recoding 13 characters of female head glands derived from Weiss et al. However, four characters could not be combined since they were unique to males (characters 16 and 17, Table S4) or to females (characters 14 and 15, Table S4) (see section 3 below for details on the treatment of the resulting structural zeroes due to a missing character or lack of information on the character state). Some characters had been much more finely differentiated in females [1] than in males because in the latter the considerable variation in the same character was often so pronounced that fine-graded variations were not coded. Therefore, for some joint characters different original female character states were equaled to the appropriate state as defined for males.

[11] [-]
Location of the MG reservoir in the head capsule. Due to its relatively small size, the MG of female Philanthinae is always located anterior to the brain. Therefore, this character was not originally included in the investigation of female head glands. Here, all females were assigned to category (0). See Additional file 1 character 11 for definition of character states.  can bear cuticular hairs on their inner side, reaching into the lumen of the gland. In females, these hairs generally occur singularly, but in males of some species several hairs can also jointly originate from the top of or laterally from cuticular ridges.

[3] [2] [7]
Modifications of the PPG morphology. Since the PPG of female Philanthinae is less complex and shows much less interspecific variation, it was not originally recorded as in males (Table   S2, additional file 1). Here, we recoded the two characters that were used to define the shape of the upper and lower PPG of females [1] to match the definitions of PPG 'modules' established for males.

[7] [-]
Type of gland cells associated with the PPG. In females the PPG has never been found to be associated with secretory cells, hence this character was not originally defined in the morphological investigation of females. In the 'joint dataset', females are assigned to category (0).

Aggregated categorical principal components analysis of males and females
The aggregated dataset (Table S4) was subjected to a categorical principal components analysis as described for the data of males (main text: "Categorical principal components analysis" and section 2.1 of the additional file 1) to represent the head gland characters for males and females in the same coordinate system. Based on the results obtained from the initial CATPCA run including all 17 aggregated characters, the following adjustments of the analysis parameters were made: First, based on their relatively low Eigenvalues (= variance accounted for) in the initial CATPCA run, we decided to exclude the three characters (2) (Table S5). Second, to facilitate the interpretation of the resulting CATPCA plot, we also excluded character (4)   Philanthinae (supported by the maximum possible total Cronbach's α of 1, [2]). While most female data points are closely clustered, the data points for males are much more scattered and show two main aggregations similar to the CATPCA including only males (see Figure 4, main text). Vectors:

Shannon diversity indices for male and female gland morphology
To compare the diversity of gland morphology between the sexes the aggregated dataset was reduced to those characters that were shared by males and females. Therefore, two characters that were only defined in males and two characters that were only defined in females had to be omitted.
Following a method described by Tesfaye et al. [3], Shannon diversity indices of the resulting nine characters for males and females were calculated for each joint character for males and females separately, according to the formula:

= − ∑
With i being the number of categories of a given character, n being the total number of species for which this characters could be assessed, and n i being the number of species in a given category. The resulting nine indices for both sexes were compared using a Wilcoxon paired test in PAST (Version 2.08b, [4]).