What’s in a Head? Comparative Morphology of Head Muscles in the Three Drusinae Clades (Insecta: Trichoptera)

The subfamily Drusinae (Limnephilidae, Trichoptera) comprises a range of species exhibiting differently shaped head capsules in their larval stages. These correspond to evolutionary lineages pursuing different larval feeding ecologies, each of which uses a different hydraulic niche: scraping grazers and omnivorous shredders sharing rounded head capsules and ltering carnivores with indented and corrugated head capsules. In this study, we assess whether changes in head capsule morphology are reected by changes in internal anatomy of Drusinae heads. To this end, internal and external head morphology was visualized using µCT methods and histological sections in three Drusinae species – Drusus bosnicus, D. franzi and D. discolor – representing the three evolutionary lineages. Our results indicate that Drusinae head musculature is highly conserved across the evolutionary lineages with only minute changes between taxa. Conversely, the tentorium is reduced in D. discolor, the species with the most aberrant head capsule investigated here. Integrating previous research on Drusinae head anatomy, we propose a fundamental Drusinae blueprint comprising 29 cephalic muscles and discuss signicance of larval head capsule corrugation in Trichoptera.


Introduction
Nature comes in manifold color, smell, shape, and taste. As in all other life forms a great variety of different morphologies can be observed in aquatic insects as well 1 . Differences in morphology typically are accompanied by a distinct ecological niche, related to -for instance -feeding mode 2 . Predators among aquatic insects may enjoy prehensile mouthparts, or a slender agile body that allows them to pounce on their prey 3 . Grazers have developed their mouthparts to an intricate array of brushes and bristles to scrape off benthic algae 3 . Passive lter-feeders, on the other hand, often have elongated antennae or legs equipped with bristles to collect food particles from the ow 3 . And those among aquatic insects feeding on larger detritus, the so-called shredders, often are rather bland but equipped with robust mandibles to masticate their food 3 .
Naturally, species of these functional feeding groups occur in different regions in a local habitatfollowing the distribution of food sources in the stream bed 2 . Predators can pick any spot if prey densities are high enough, but grazers, lter feeders and shredders need to select speci c habitat spots to maximize feeding e ciency [4][5][6][7] . These microhabitats are most importantly de ned by distinct ow velocities: at slowly owing sections detritus accumulates whereas benthic algae grow most densely at the upper sides of large stones in faster waters, where also food particle density in the drift is high [8][9][10][11][12] .
Consequently, astounding adaptations to hydraulic stress can be observed in aquatic insects in addition to those enforced by feeding modes. In caddis ies, behavioural adaptations include the use of silk as safety tether in Rhyacophilidae, Hydropsychidae and Brachycentridae, ballast stones in Goeridae 4 , and a silken stalk in Limnocentropodidae 3 . Conversely, morphological adaptations to hydraulic stress are rarely considered in caddis ies. The bodies of other aquatic insects are in contrast often modi ed to suit a particular purpose: the reduction of hydraulic stress. This is demonstrated by the attened, streamlined bodies of Heptageniidae that certainly are at the pinnacle of morphological adaptation to hydraulic stress 3 .
Recently, head capsule morphology was identi ed as potential adaptation to hydraulic stress and a particular feeding mode in Drusinae caddis ies 13 . The Drusinae are an intriguing group of Limnephilidae, and comprise three distinct evolutionary clades. Each of these clades comprises species sharing a particular larval feeding ecology: scraping grazers with toothless mandibles, and ltering carnivores and shredders with tooth-bearing mandibles. While evolutionary relationships between these clades remain to be clari ed, comparative morphological analyses based on adults indicate that Drusinae shredders may have retained ancestral characters 14 . Drusinae likely diversi ed in vicariant conditions under the impact of repeated cold ages and geological processes with scraping grazers representing the greatest radiation of the group. Extant Drusinae bear mark of their evolutionary background and develop adult and larval characters that indicate to which feeding group each species belongs 13 . In Drusinae larvae, head capsule morphology is of particular importance: head capsules of species of the ltering carnivore clade differ strikingly from the rounded head capsules of their congeners and other European Limnephilidae 13 . The signi cance of head capsule shape for their ecology remains unclear, but a link to feeding ecology was recently proposed. And while the cephalic anatomy of Drusus tri dus and D. monticola is known 15,16 , there is to date no information whether these scraping grazer species can serve as blueprint for all Drusinae, including shredders and the ltering carnivores with their peculiar heads (Fig. 1).
We propose that the aberrant forms of ltering carnivore Drusinae heads are re ected in modi cations of the internal anatomy, which, in turn, correspond to different evolutionary trends within Drusinae. To address our hypotheses, we use a sample of three different species comprising a shredder (D. franzi), a ltering carnivore (D. discolor) and a scraping grazer (D. bosnicus), and integrate available information on D. tri dus and D. monticola. We hypothesize that location and number (or volume) of head muscles differs between shredder, scraping grazer, and ltering carnivore Drusinae. In particular, we posit that the shredder species will have the most complex internal organization, and that shifts in attachment sites as well as numbers of muscles occur in grazers and ltering carnivores.
Our second aim was to compare internal head anatomy between Drusinae clades. Despite the impressive differences in head capsule shape, each of the three species investigated here shares the same set of cephalic muscles, with only minute differences in the location of single points of origin of, e.g., frontal muscles ( Fig. 5, 6). In particular, the points of origin relative to the M. fronto-labralis and the number of individual muscle bundles of the M. fronto-epipharyngalis differs between the species as well as the points of origin of the M. fronto-pharyngalis relative to the M. fronto-labralis. In D. franzi, the points of origin are arranged sequentially along the dorso-ventral plane in the following order: M fronto-labralis, M fronto-pharyngalis, M. fronto-epipharyngalis. In D. discolor, the same order of points of origin can be observed, where the M. fronto-pharyngalis is located somewhat closer to the M. fronto-pharyngalis, and the M. fronto-epipharyngalis has more than one point of origin on either side. Drusus bosnicus displays a different con guration where the dorsalmost points of origin of the M. fronto-labralis, the M. frontopharyngalis and the M. fronto-epipharyngalis are in roughly the same dorsoventral plane, the M. frontopharyngalis has more than one point of origin on either side, and the M. fronto-epipharyngalis has several points of origin that are located obliquely in sequence from the dorsalmost point of origin. Further, D. discolor exhibits a doubled M. fronto-pharyngalis and the points of origin of some muscle bundles of the M. cranio-mandibularis medialis differs between D. discolor and the two other investigated species. (Fig. 2-4). In contrast, all other muscles including those of the alimentary canal and the maxillolabium are highly similar in all three species (Fig. 5-6). The only apparent internal change induced by the aberrant head morphology in D. discolor pertains to the tentoria, which lack a complete secondary supratentorial branch that is present in D. franzi as well as D. bosnicus (Fig. 5-7).

Discussion
Head anatomy of the Drusinae appears to be highly conserved. The number and arrangement of head muscles are virtually identical in all hitherto investigated Drusinae species 15,16 . The duplication of a muscle pair in D. discolor (M. fronto-pharyngalis ventralis) is be the only recognizable difference. Functionally, an engorging facultative predator such as D. discolor could bene t from greater mobility of the pharynx, but whether a single duplication or a somewhat larger volume of alimentary canal muscles can have that effect is doubtful. Differences in muscle volumes however could re ect feeding ecology of Drusinae shredders and scraping grazers. Drusus franzi relies on a recalcitrant food source may have relatively large mandible adductors that could enable stronger bites. Scraping grazers may be more limited in their food uptake by the number of scraping movements per unit time and larger mandible abductors in D. bosnicus may be an adaptation to this feeding mode by allowing for more scraping movements per unit time. While more comprehensive studies remain to be conducted, the preliminary data obtained here point towards the possibility that such volumetric differences can be observed: Mandible adductors make up for roughly 85% of the total reconstructed head muscle volume in D. franzi, 77% in D. discolor and 72% in D. bosnicus. At the same time our provisional summary found the greatest mandible abductor muscle volume in D. bosnicus (12% of the total head muscle volume), and D. discolor as having the greatest alimentary canal muscle volume (9% of total head muscle volume). These rough gures may be a rst indication for such a differentiation, but should not be trusted until veri ed in a larger, more standardized sample.
Drusinae head anatomy was rst investigated in D. tri dus, a representative of the Drusinae grazer clade 15 . Data on another Drusinae grazer species, D. monticola, suggested high congruence of this species with the previously described situs 16 but did not cover other evolutionary lineages of Drusinae.
Here, we present evidence contrary to our initial hypotheses, suggesting largely identical head muscle number and arrangements in all three major evolutionary lineages of Drusinae with minor deviations in the ltering carnivore clade. Interestingly, the con guration of frontal muscles in the Drusinae grazer clade observed in D. bosnicus was also observed in D. monticola. While it is probable that D. tri dus exhibits the same pattern, the available data do not allow for an assessment. Whether this con guration is typical for Drusinae scraping grazer remains to be evaluated, but this notion is conceivable because of the close relationships within this clade 13,17 .
Concerning the internal head skeleton, the tentorium, we posit that the changes of head shape in D.
discolor and other ltering carnivorous Drusinae 13 induce modi cations such as the increasing simpli cation of the tentorium. In this regard, we assume that the modi ed head capsules of the ltering carnivore Drusinae offer greater mechanical stability due to their structured surface as compared to the rounded head capsules of the other Drusinae -thus, the second branch of the tentorium is super uous and can be reduced. We base this interpretation on the observed mechanical properties of corrugated bodies, that are capable of withstanding greater forces 18,19 . In-eld measurements indicate that ltering carnivore larvae occupy microhabitats where hydraulic stress is higher compared to other Drusinae 20 . Adaptations increasing stability of particularly exposed body parts such as a corrugated head capsule may prove bene cial under such circumstances. However, head capsule shape in ltering carnivore Drusinae was previously interpreted in relation to ow modi cation around the larval head and feeding ecology. Flow patterns around Drusinae larval heads are the focus of ongoing research (Vieira et al. unpubl.), but comparative analyses of mechanical properties of Drusinae head capsules remain to be conducted.
From a systematist point of view, the reduced second arm of Drusinae tentoria could be, pending further studies, a synapormorphy accompanying head capsule modi cation in this clade, opposed to a potentially plesiomorphic biramal tentorium of the Drusinae common ancestor.
The lack of differences in the internal anatomy of Drusinae heads that differ strongly in their outer head capsule morphology is surprising. We present the rst data on a caddis y larva with an aberrant head capsule shape, but if our ndings apply to other taxa as well remains to be investigated. A wide range of Trichoptera taxa develop larvae in which the head capsules are not in a simple round shape. Amongst the European Trichoptera, Lithax niger is certainly one of the most distinctive forms, but to date no comparative morphological studies are present of this species. Likewise, there are no anatomical treatments on other species with aberrant head shapes. A suite of potential model taxa of Brachycentridae (Micrasema), Beraeidae (Beraea), Goeridae (Silo, Goera, Lithax), Limnephilidae ( ltering carnivorous Drusus, Philocasca, Pseudostenophylax), Rossianidae (Goeriella), Apataniidae (Allomya) and Hydropsychidae (e.g., H. tabacarui) develop larval heads distinctly different from the rounded ones sported by their congeners. Whether the minor impact of head capsule modi cation on internal head anatomy can be con rmed in other taxa as well will be subject of future studies. The absence of major changes however suggests that head capsule modi cation is not a costly means of adaptation to speci c habitats. Conversely, any change in cephalic musculature will inevitably affect feeding, gut movement or size and con guration of the central nervous system. Modi ed head capsules as observed in some Drusinae but also in other groups can therefore probably evolve quickly and at low evolutionary costs if less important areas of the cephalic exoskeleton are involved. Intriguingly, anecdotal evidence suggests that some other Trichoptera larvae with modi ed head capsules use high-stress hydraulic niches (e.g. Allomya, pers. comm. J.J. Giersch).
Embryonic development of insect heads involves the formation of parietals and the frontoclypeus following a "bend and zipper" model 21 . Head appendage tissue (with the exception of the labrum) is not involved in this process, and the corresponding muscles are mesodermal derivates that make contact with the epidermis during embryonic development 21,22 . Developmental gene expression regulates head capsule formation and shape, where gnathal appendages are formed under in uence of pair-rule and Hox genes 21 . Processes and developmental genes controlling head capsule shape in Trichoptera are not known. Evidence from other insects with head capsule modi cations, such as Scarabaeidae, suggest that sets of developmental factors are co-opted to act as controlling agents in horn formation 23,24 .
Assuming that the same or highly similar molecular controls of head capsule shape are used across the more homogeneous Trichoptera is therefore plausible. However, exact patterning and developmental mechanisms, and how development of species-speci c head capsule shapes is maintained over time, remains obscure. In particular comparative assessments within Drusinae as well as between different caddis y families should be made to clarify the genomic background of head capsule corrugation and indentation. Most probably the same genes are involved in different families, but how exactly head capsule shapes take form during development and which ecological function head capsule shape and corrugation have is enigmatic. Available data on Drusinae hydraulic niches suggest that head capsule corrugation may be linked with high-stress microhabitats optimal for lter-feeding 20 . In other taxa (e.g. Goeridae, Brachycentridae, Apataniidae, etc.) head capsule corrugation and indentation may be the result of similar ecological constraints.

Sample preparation
Three Drusinae specimens of three different feeding and evolutionary clades, Drusus bosnicus, D. discolor and D. franzi, were used for µCT analysis. µCT scanning For µCT analysis, larvae were stained for 21 days in 1% (w/v) phosphotungstic acid (PTA) in 70% ethanol and washed in in 70% ethanol to remove unbound PTA from tissue. Afterwards, the larvae were mounted vertically in 70% ethanol in the tip of a plastic pipette, and sealed in with para lm. Larvae were scanned on an XRadia MicroXCT-400 (Carl Zeiss X-ray Microscopy, Pleasanton, CA, USA) at 80kVp/ 100µA using the 4X detector assembly. Projections were recorded with 15s exposure time (camera binning = 1) and an angular increment of 0.225° between projections over a 360° rotation. Tomographic slices were reconstructed with a voxel resolution of 2.87 µm (reconstruction binning = 1) using the XMReconstructer software provided with the µCT system.

Image processing
The merged volume was exported as *.TXM le into Amira 2019.1 (FEI SAS, Mérignac, France (part of Thermo Fisher Scienti c™)). A 3D bilateral lter was used to lter the image volume for noise reduction. Image segmentation was achieved in Amira 6.5.0 (Visage Imaging, Inc., San Diego, CA, USA). Internal head anatomy (head muscles, tentoria, central nervous system including cerebral ganglion mass, gnathal ganglion mass, frontal ganglion and innervation patterns) were manually segmented and assigned to different "materials" within the segmentation editor. Three-dimensional surface renderings were created based on this manual segmentation using the Amira Surface Generate tool.
Histology, computer-based 3D reconstruction and post processing Heads of D. bosnicus and D. discolor were cut off from the remaining body for histological processing. First samples were dehydrated with acidi ed dimethoxypropane followed by three rinses with acetone before being in ltrated and embedded in Agar LVR resin (Agar Scienti c, Stansted, UK). Cure resin blocks were serially sectioned with a Diatome HistoJumbo diamond knife (Diatome, Nidau, Switzerland) at 1µm section thickness on a Leica UC6 ultramicrotome (Leica microsystems, Wetzlar Germany). Sections were stained with 1% toluidine blue and sealed in epoxy resin. Analysis and photography of the serial sections was conducted on Nikon NiU compound microscope with a Nikon DsRi2 microscope camera (Nikon, Tokyo, Japan).