Discussion

CHC divergence and variation in relation to social complexity

We compared chemical and associated transcriptomic complexity between Blattodea species with increasing levels of social organization. Overall, we did not find a consistent correlative pattern of CHC-based chemical complexity paralleling the different levels of social complexity in our representative Blattodea species. These results point to large categorical differences in CHC profiles across species, which appear to be independent from their level of social organization.
CHC divergence based on average chemical distances between the species neither reflects the different levels of complexity nor the phylogenetic divergence within the Blattodea (Fig. 3). M. darwiniensis , the most basal termite lineage (Inward et al. 2007b; Krishna et al. 2013) chemically clusters together with both a high (R. flavipes ) and a low (N. castaneus ) social complexity termite, and the solitary cockroach B. orientalis clusters with both a high (C. formosanus ) and a low (K. flavicollis ) social complexity termite. It has been hypothesized that chemical divergence clearly differing from an established molecular phylogeny indicates selection on chemical profiles for different functions overriding their phylogenetic information (Marten et al. 2009; Buellesbach et al. 2013). However, our chemical divergence does not display any pattern congruent with the social hierarchy of our study species, rendering any assumptions on selection for CHC functions reflective of the species’ respective social complexity highly unlikely. Concerning counts of gene transcripts stemming from orthologs of CHC biosynthesis genes from other insect species, these do neither quantitatively correlate with higher levels of social complexity nor with the total number of CHC compounds detected in each of our study species.
Future work could include more species as transcriptomes, genomes and chemical profiles become available, and utilize scalable frameworks such as phylogenetically-contrasted regression and Bayesian ancestral state reconstruction models (Simon et al. 2019). Additionally, caste-specific CHC variation in eusocial taxa could be taken into account in future studies as well, potentially adding another layer of complexity, despite the accompanying issues for direct comparability with solitary taxa.

CHC biosynthesis gene transcript variation across the studied Blattodea species

Acetyl-CoA carboxylase (ACC) catalyzes the first and rate-limiting step in CHC biosynthesis (Barber et al. 2005, see Fig. 1). In each of our analyzed cockroach and termite species, we found several distinct ACC transcripts (ranging from 3 in C. formosanus to 8 in B. germanica ) based on orthology to the Drosophila gene (Tab. 1). This rich abundance of ACC transcripts strengthens the argument for the universality of ACC as fundamental catalyst for the first step in CHC biosynthesis (see also Fig. 5).
For FAS genes, seven of them had already been identified inBlatella germanica , with five showing a significant effect on CHC compound quantities upon knockdown (Pei et al. 2019). In our analyzed transcriptomes, we were able to detect transcripts with strong orthology to two of these five FAS genes (BgFas 4 and 6). Transcripts with orthology to BgFas4 were detectable across all seven species, whereas BgFas6 transcripts were most abundant in the two cockroaches and only in three of the five analyzed termites (Tab. 1). Generally, it has been hypothesized that two types of FAS, cytosolic and microsomal, differentially impact CHC biosynthesis, with the former mainly governing non-methylated, straight-chain CHCs, and the latter being more specific for methylated CHCs (Chung et al. 2014; Wicker-Thomas et al. 2015, Fig. 1). Since we could not detect any transcript copies of BgFas6 in both M. darwiniensis andN. castaneus , the two species with the lowest amounts of methyl-branched alkanes (Fig. 2), it is possible that the BgFas6transcripts stem from a microsomal FAS associated with the production of methyl-branched alkanes. Intriguingly, for N. castaneus , in addition to showing both the lowest number and proportion of methyl-branched alkanes, it also shows the lowest transcript copy number of FASN2 orthologs, an oenocyte-specifically expressedDrosophila gene with a strong effect on methyl-branched CHCs and thus speculated to be microsomal (Chung et al. 2014; Wicker-Thomas et al. 2015).
Unsaturated compounds, whose biosynthesis is crucially dependent on desaturases (Wicker-Thomas et al. 1997; Coyne et al. 1999; Dallerac et al. 2000), occur in each of our studied Blattodea species, most abundantly in R. flavipes , M. darwiniensis and N. castaneus , intermediately in B. orientalis and B. germanica , but only in traces in K. flavicollis and C. formosanus (Fig. 2). This partially reflects the respective numbers of transcripts orthologous to the genes desat1 and desat2 , with the former being most abundantly represented in M. darwiniensis and N. castaneus , and the latter in R. flavipes (Tab. 1).
Genes coding for P450 Decarbonylases of the gene subfamily CYP4G have been shown to govern the final steps in CHC biosynthesis and have thus been suggested to be stable, highly conserved and particularly vital elements in this pathway (Feyereisen 2020; Holze et al. 2021, Fig. 1). Concordantly, at least one Cyp4g gene could be identified in all insect genomes screened to date (Qiu et al. 2012; Feyereisen 2020). We found transcripts orthologous to the Drosophila geneCyp4g1 as well as the migratory locust Locusta migratoriagene LmCYP4G102 in all our tested species (Tab. 1). However, their numbers vary largely with no apparent consistent pattern, hinting at more transcriptomic variety for these vital CHC biosynthesis elements than previously assumed.
The assessment of transcript counts as approximation of the actual genomic repertoire for CHC biosynthesis genes naturally has its limits and will remain speculative until targeted knockdown studies confirm the actual functions of the transcripts as well as their underlying genes. However, analyzing the abundance of unique transcripts per ortholog allows valuable insights into functional diversity potentially exceeding the information contained within whole-genome sequences (He et al. 2021; Sprenger et al. 2021). Correlating transcriptomic diversity directly with CHC profile variation has been attempted surprisingly rarely despite its potential to approximate the genetic control of CHC variation more accurately, as it has repeatedly been shown that the same genotype can produce different CHC profiles (Holze et al. 2021; Sprenger et al. 2021). Thus, our findings constitute promising first steps for future studies with the potential to corroborate our transcriptomic assessment with target gene counts once further genomic resources become available.

CHC-based communication mechanisms and outlook for future studies

It has long been hypothesized that the complexity of CHC profiles reflects the complexity of chemically encoded communication mechanisms necessary to maintain more socially complex insect societies (Korb and Thorne 2017; Kronauer and Libbrecht 2018; Holland and Bloch 2020). However, how CHC profiles actually encode biologically relevant information such as nestmate affiliation or task allocation has remained largely elusive so far (Buellesbach et al. 2018; Menzel et al. 2019; Heggeseth et al. 2020). Further elucidation on the exact encoding mechanisms in CHC profiles and which compound combinations actually convey chemical information will be instrumental in gaining a better understanding on how insect populations and societies are maintained at different levels of social complexity. Furthermore, although insects generally synthesize the majority of the components in their CHC profiles themselves (Nelson and Blomquist, 1995; Blomquist and Bagnères, 2010), several studies have demonstrated the impact of several biotic and abiotic factors such as diet, habitat and microbiome on CHC profiles as well (Fedina et al. 2012; Rajpurohit et al. 2017; Teseo et al. 2019). Thus, disentangling these factors from the conserved CHC profile functionalities represents an additional challenge in future studies that will nevertheless be indispensable to fully comprehend and explore CHC-mediated communication mechanisms.
Future studies should also consider the occurrence of very-long chained CHC compounds of up to C58 in Blattodea surface profiles as recently demonstrated with non-standard analytical methods (Golian et al. 2022). However, neither exact compound quantifications nor identifications (e.g., discrimination between n -alkanes and methyl-branch alkanes) are so far possible in this higher chain length range extending beyond the CHC compounds traditionally accessed and identified through gas-chromatographic separation (Schnapp et al. 2016; Bien et al. 2019). Therefore, to include very long-chain CHCs in future analyses, novel methods are necessary to reliably assess their exact quantities and compound classes. Moreover, despite CHCs clearly constituting the dominant, most investigated compounds in chemical signaling (Blomquist and Bagnères 2010; Chung and Carroll 2015) and have been implied to have convergently evolved, unifying communication modalities in eusocial insects (Leonhardt et al. 2016; Funaro et al. 2018), we cannot exclude the additional potential of non-CHC compounds also used in signaling which might reflect social complexity (e.g. Hanus et al. 2010; Smith et al. 2016; Steitz et al. 2019). Lastly, CHC metabolic networks could be amenable to various kinds of more elaborate complexity analyses such as metabolic network pathfinding (Kim et al. 2017), identification of critical connectors (Kim et al. 2019), determination of topological characteristics (Goryashko et al. 2019), and flux balance analysis (Beguerisse-Díaz et al. 2018). These types of complexity analyses have, to our knowledge, not been attempted so far in this context at all and might largely aid in obtaining a more holistic correlative view on complexity on a chemical, genetic and social level.