Both state-of-the-art SBF-SEM microscopy  and two-photon excitation microscopy , combined with associated image-analysis technologies, yielded full 3D perspectives — at nano-scale resolution — of the glands and muscles in the mouthparts of the heterorhabdid species studied. Although these two methods are based on different principles, and use different types of fixative, the results were similar for both (Fig. 2). Sections from two different individuals of Mesorhabdus gracilis (Fig. 2) show planes of four pairs of glands (dashed outlines), and planes of one pair of muscles, which correspond nicely between the two pictures. Both imaging methods clearly show the same spatial relationships of glands and muscles. Gland contents, however, appeared to differ somewhat between methods. For example, gland lg1C2 in the SBF-SEM scan (Fig. 2a) appeared to be filled with tiny and flattened disc-shaped granules, whereas in the two-photon excitation microscopy scan (Fig. 2b) the granules appeared to be rather big and more rounded in shape. Curiously, gland contents also appeared to differ between sides even within a single specimen (compare contents of gland lg1C2 on the left and right side of Fig. 2b).
Numerous muscles and glands are associated with the mouthparts examined. All of the descriptive terms used here to refer to morphological units of muscles and glands do not imply any homology hypothesis (see  for a discussion of homology-free terminology in morphological description). Homology hypotheses for these descriptive terms are outlined explicitly in Table 1 and discussed in detail in the Discussion.
Gland morphology and arrangement
We adopted labral gland terms from Nishida and Ohtsuka , where gland cells were divided into three “Types” according to the arrangement of gland openings (Fig. 3a-d). We use the same terminology here, but apply these terms differently except for Heterorhabdus subspinifrons. The arrangement of gland openings is essentially the same as reported previously , but we found an extra opening of labral gland Type 2 in Disseta palumbii: two openings were reported earlier , but we found a third (Fig. 3a).
Even though gland openings were readily identified and easy to homologize among taxa, the size, shape and configuration of gland cells differed considerably among the four genera. In the particle feeding D. palumbii, gland cells are located postero-ventrally in the labrum, and are not associated with muscles (Fig. 3a, e: see Additional file 1: Figure S1 for viewing instructions for the interactive 3D-pdf images). In Mesorhabdus gracilis (intermediate feeding mode), the labrum is almost fully packed with labral gland cells and parts of these cells intercalate between the muscles lab-eso.dM3 and u-l.labM2 (Fig. 3b and f). In Heterostylites longicornis (intermediate feeding mode), labral gland cells are located at the posterior half of the labrum, and half of the cells are stacked between muscles lab-eso.dM3 and u-l.labM2 (Fig. 3g and c). Significantly, in the piercing carnivore, H. subspinifrons, all of the labral gland cells are highly extended anteriorly: a) Type 3 gland cells are enveloped by three muscles u-l.labM2, lab-eso.dM1–4 and for-eso.dM (Fig. 3h and d: click on the view “Labral Gland Type 3 and muscles” in the interactive 3D-PDF, Fig. 4d), b) Type 2 gland cells extend up to the posterior margin of the paragnath (Fig. 4d), and c) Type 1 gland cells are inflated, and posteriorly elongated into the paragnath (Fig. 4d). The total number of cells in gland Types 1–3 also differed among these genera (Table 1). Disseta palumbii has 15 pairs of cells, but M. gracilis, H. longicornis, and H. subspinifrons have only 8 pairs (Table 1). Type 1 and Type 3 glands were largest in the piercing carnivore, H. subspinifrons (Fig. 4d), but all three types were well-developed in the intermediate feeding-mode M. gracilis (Fig. 4b).
Our observations of cell numbers and orientation in each gland differ somewhat from Nishida and Ohtsuka . They reported “Type 1 and 3 labral glands have two secretory cells…Type 2 labral glands and the paragnathal gland have one secretory cell” in Heterorhabdus abyssalis, H. pacificus, H. papilliger, and H. spinifrons. However, our observation of H. subspinifrons confirmed two cells in Type 1 glands, but revealed three cells in each of Types 2 and 3 (Fig. 4d, Table 1). Regarding cell structures, Type 2 gland cells were previously considered to be anteriorly elongate cells along the labral wall, and Type 3 gland cells as small cells located within the posterior side of the labrum . However, our observations revealed that Type 2 gland cells extend toward the paragnath, and that the dramatically inflated Type 3 gland cells were directed anteriorly, reaching all the way to the forehead.
The arrangement of gland openings also differed between the carnivore Heterorhabdus and the non-carnivore taxa. The openings line up nearly in a straight line in D. palumbii, M. gracilis and H. longicornis, but the opening for Type 1 lies far off the line in H. subspinifrons (Fig. 3a-d). Significantly, the opening for the Type 1 gland in H. subspinifrons lies directly at the posterior end of the hollow fang (Fig. 4d).
Secretory granules in the gland cells appeared to vary among taxa and among the three gland types (Fig. 5). Granules in homologous types of gland cells (based on location) were not similar in shape and size (e.g., compare “lg3c1”and “lg3c2” in Fig. 5a; “lg1c1” and “lg1c2” in Fig. 5b; “lg1c2” and “lg1c1” in Fig. 5f). However, granule form of homologous gland cells also differed between individuals of the same species (Fig. 2), and even between sides of the same individual (Fig. 2b). Therefore, these observations, combined with inconsistent resolution due to technical limitations of contrasting and resolution, greatly limited the utility of granule form as a tool for making any inferences about gland function or homology.
A small, fourth type of gland — termed here Epidermal Gland — was found by the ventral side of the epidermis, with the duct opening on the ventral side of both the labrum and paragnath in M. gracilis, H. longicornis, and H. subspinifrons (Fig. 4b, c and d: represented in yellow). No such cells were seen in D. palumbii. Cell numbers were lowest in M. gracilis (2 cells; but paragnath epidermal gland might have been overlooked because of the limited scanning field), greater in H. subspinifrons (7 cells), and highest in H. longicornis (14 cells). In addition, arrangement of the labral epidermal gland cells was erratic and not always symmetrical (e.g., Fig. 4c and d).
Epidermal gland cells in H. subspinifrons contained distinctive spindle-shaped secretory granules (“peg” cells and “leg” cells in Fig. 5g, h). Unfortunately, the contents of these epidermal gland cells were unclear in other genera due to limited contrast and resolution (Fig. 5c, d and e).
Muscle configuration and movement of mouthparts
Given the large differences in mandible form, the overall arrangement and attachment sites of muscles were surprisingly similar among the four genera examined (Fig. 4). These muscles are named based on their attachment sites or locations (Table 1). The only species-specific muscle we observed was in the highly derived carnivore Heterorhabdus subspinifrons, (“saggital labral muscle”, Fig. 4d, Table 1). This muscle was located at the posterior side of the labrum: one end attached just beside the opening of labral gland Type 1 and the other end attached near the esophagus opening (Fig. 4d: click on the view “Sagittal Labral Muscle insertions” in the interactive 3D-PDF).
In all four genera, masticatory movement of mandibles and cyclic muscular contraction within the labrum were synchronized soon after stimulation with a fine needle (Additional file 2: Movie SM1 A-D). In Disseta palumbii (particle feeder), cyclic contractions of the “Upper-Lower Labral Muscles 1” (u-l.labM1 in Fig. 3e) and the “Forehead-Esophageal Dilator Muscles” (for-eso.dM in Fig. 3e) were observed (Fig. 6a, Additional file 2: Movie SM1A). In Mesorhabdus gracilis (intermediate feeding mode), muscle bundles were not clearly recorded, but the “Forehead-Esophageal Dilator Muscles” (for-eso.dM in Fig. 3f) seemed to cyclically contract and lift up the esophagus area (Fig. 6b, Additional file 2: Movie SM1B). In Heterostylites longicornis (intermediate feeding mode), simultaneous cyclic contractions of the “Lateral-Esophageal Dilator Muscles 1” (lat-eso.dM1), the “Forehead-Esophageal Dilator Muscles” (for-eso.dM) and the “Labrum-Esophageal Dilator Muscles 1” (lab-eso.dM1 in Fig. 3g) created an expanding motion of the esophagus (Fig. 6c, Additional file 2: Movie SM1C). In Heterorhabdus subspinifrons (piercing carnivore), distinct muscles were not clearly recorded, but cyclic and coordinated contraction appeared to occur in the “Lateral-Esophageal Dilator Muscles 1” (lat-eso.dM1), the “Forehead-Esophageal Dilator Muscles” (for-eso.dM in Fig. 3h), the “Labrum-Esophageal Dilator Muscles 1” (lab-eso.dM1 in Fig. 3h) and the “Labrum-Esophageal Dilator Muscles 2” (lab-eso.dM2 in Fig. 3 h), which created an expanding motion of the esophagus (Fig. 6d, Additional file 2: Movie SM1D).