Anatomical adventures in the fish auditory medulla^a) Free

In this contribution, I outline my journey into the field of comparative neurobiology, during which I focused on the comparative neuroanatomy of the inner ear and lateral line systems. My exposure to evolution began as a child, since as a native New Yorker I visited the paleontology exhibits at the American Museum of Natural History many times. This, along with numerous outings to the Bronx Zoo (New York Zoological Park, The Bronx, NY), made me excited about majoring in biology. Yet, I had no preconceptions about where this major would lead me.

During my first year at New York University, I had the great fortune of being a student in Dr. Douglas Webster's comparative anatomy and neuroanatomy courses. In subsequent years I participated in his comparative neuroanatomical studies on the auditory system in species of kangaroo rats, and carried out research on the projections of the eighth nerve in the tegu lizard. I was therefore exposed to the importance of understanding evolutionary history early on, a perspective that was reinforced and expanded by my graduate mentor, Dr. R. Glenn Northcutt at The University of Michigan. As a college freshman reading Romer's work (Romer, 1970) I was struck by the statement that the lateral line was the primitive vertebrate auditory organ, which would mean that it predated auditory receptors in the inner ear. Classical, non-experimental tract-tracing studies (studies in which specific nerve fibers could not be uniquely labeled) generally reported that the lateral line and eighth nerves had largely coincident projection areas in the “acousticolateralis area” of the dorsal medulla, where auditory nuclei would presumably be located based on their location in land vertebrates (Pearson, 1936a,b; Larsell, 1967). This supported the view that the lateral line system is an acoustic receptor accessory to those of the inner ear or, alternatively, that it forms the entire peripheral auditory system (e.g., Beard, 1884; Ayers, 1892; van Bergeijk, 1967). I wondered if this was really correct, and this led me to propose to Glenn that I use the experimental tract-tracing method of the time (staining for axonal degeneration, which I subsequently repeated by labeling nerve fibers with horseradish peroxidase, i.e., HRP) to be able to chart the central projections of the eighth and lateral line nerves separately. Glenn had the wisdom to guide me to the bowfin, Amia calva, which has a relatively accessible inner ear and, moreover, belongs to a clade (i.e., a monophyletic group; halecomorphs) that evolutionarily preceded the more abundant modern bony fishes (teleosts). As such, its lateral line and inner ear projections would potentially have a more primitive, or basal organization than those of teleosts. As discussed in Sec. II, I determined that the lateral line and eighth nerves in the bowfin have largely non-overlapping projections in the dorsal medulla, which would support the view that these nerves carry, on the whole, functionally distinct information.

In the same year that my manuscript reporting these projections appeared (McCormick, 1981), another study on the same topic was published by Bell (1981), but using two species of mormyrid teleosts: Peter's elephantnose fish, Gnathonemus petersii, and the elephant fish, Pollimyrus isidori. The results of the two papers appeared to be vastly different, suggesting to some (but not to the authors!) that one was “right” and the other “wrong.” However, the “divergence” between the two sets of results illustrates the importance of considering phylogeny and employing the comparative method. The results were not divergent at all, but instead reflected inconsistencies in nomenclature, differences in evolutionary history and function (i.e., character states of the primary lateral line nuclei and related electrosensory nuclei among anamniotes; McCormick, 1982; Kozloski and Crawford, 1998) and variations in the morphology of primary auditory areas/nuclei among teleosts (McCormick, 1992). The latter point is particularly unsurprising, given that teleosts, with over 26 000 extant species, is the most specious vertebrate clade and appeared roughly 300 × 10⁶ years ago, if not earlier. Since then, evolution's experiments have resulted in a remarkable diversity in teleostean morphology and lifestyle. Some of the variations in the peripheral and central auditory system are known (McCormick, 1992, 1999), but it is likely that some have yet to be discovered.

As is the case in all vertebrates, the fish brain is composed of common parts: telencephalon, diencephalon, midbrain, cerebellum, and medulla [Fig. 1(a)]. On the basis of cytoarchitecture (the structural arrangement of neurons) and the connections of the eighth and lateral line nerves to the medulla in the bowfin, I defined a dorsally located lateralis column that receives direct input from the lateral line nerves (nucleus medialis and nucleus caudalis) and a ventrally located octaval column which receives direct input largely from the eighth/octaval nerve (anterior, descending, magnocellularis, and posterior octaval nuclei). Nucleus magnocellularis alone had obvious inputs from both the octaval nerve and the lateralis nerves. The enlarged illustration of the medulla of the bowfin [Fig. 1(b)] illustrates the position of the six medullary nuclei comprising the dorsal and ventral columns. Because the inputs of these nerves to the medulla are largely non-overlapping (McCormick, 1981), the classically recognized “acousticolateralis area” does not exist.

As part of my thesis work, I was also able to compare the cytoarchitecture of the dorsal and ventral columns in the bowfin with those of other species. This was possible because Glenn Northcutt had, over time, assembled a large collection of serially-sectioned vertebrate brains stained with cresyl-violet, which reveals neuronal somata. Analyzing the cytoarchitecture of the medulla of the ray-finned fishes in the collection, along with further study of classical and more recent literature, led me to several conclusions. First, I found that the ventral octaval column is fundamentally similar among ray-finned (actinopterygian) fishes although in teleosts (modern bony fishes) there is an additional tangential nucleus. In later studies, however, I found morphological variations in the descending nucleus that were not initially obvious to me (McCormick, 1982, 1992). These variations are discussed in Sec. IV. Second, it was apparent that the dorsal, mechanosensory lateral line column is organized similarly among ray-finned fishes. Last, my thesis work revealed how the complex evolutionary history of the electrosense in jawed vertebrates relates to the location of first-order electrosensory nuclei/lobes in the medulla (McCormick, 1982).

As I continued my post-graduate career at Georgetown University, first as a postdoctoral fellow in Dr. Arthur Popper's lab, then as a research assistant professor in Dr. Ann Butler's lab, and later as a faculty member at Oberlin College, a main objective of my work was to combine cytoarchitectural and tract-tracing studies in fish species chosen on the basis of taxonomy. I initially returned to the bowfin, separately tracing the neural connections of the saccule, lagena, and utricle (the otolith end organs), the macula neglecta, and the three semicircular canal cristae to their termination areas in the medulla (McCormick, 1983). These connections are remarkably similar to those reported by Barry (1987) in a cartilaginous fish (the clearnose skate, Raja eglanteria; Fig. 2), and thus likely are ancestral for jawed fish.

Dr. Mark Braford and I also found that the connections of the descending octaval nucleus to the midbrain in the bowfin (Braford and McCormick, 1979) were similar to those in the clearnose skate (Barry, 1987) suggesting a common plan of ascending connections. Since then, the combined results of many investigators have shown that the central circuitry of the fish auditory system is remarkably similar to that of other vertebrates. That is to say, that auditory circuits in vertebrates share a common pattern of organization (Fig. 3).

Throughout my career, I have been fascinated by the varying results of studies of inner ear endorgan function. Except for the three sensory cristae of the semicircular canals, each of the remaining four otic end organs (the saccule, lagena, utricle, and macula neglecta) have been implicated in hearing in one species or another (reviewed in McCormick, 1992). The macula neglecta of cartilaginous fish, which has peripheral anatomical features unique to this taxon, appears to be auditory, although a vestibular role has not been investigated. In the ray-finned fishes, this organ has been “neglected” save for reports of its central connections in the bowfin (Saidel and McCormick, 1985) and in the goldfish Carassius auratus (McCormick and Braford, 1994), and its function is unknown.

The ancestral and ubiquitous mode of hearing among fish is the detection of acoustic particle motion by one or more of the otolith organs (Popper and Hawkins, 2018). Some species have evolved another mode of hearing, sound pressure detection, enabled by coupling the inner ear to a gas filled structure, such as the swimbladder (Popper , 2022). At the time of my thesis work, the teleost saccule was described as the main auditory end organ based mainly on single fiber recordings and saccular microphonics. The goldfish lagena, however, was discovered to have both auditory and vestibular regions (Furukawa and Ishii, 1967). Moreover, whereas the fish utricle is universally a gravistatic end organ, its morphological specializations in clupeids were known to impart an additional auditory sensitivity (Denton and Gray, 1979, 1980; Plachta , 2004). Platt and Popper (1981) noted that the morphological characteristics of the sensory maculae of the otolithic organs and their physical position relative to the swimbladder likely determine their specific sensory functions. Significantly, they speculated at the time that otolithic organ sensory hair cell maculae might be “receptor mosaics,” such that a given end organ might have areas responsive to the higher frequencies of sound and other areas responsive to lower frequencies of the gravistatic sense. This speculation was borne out by Dr. Richard Fay's seminal study in the goldfish on the responsivity of individual otolithic end organs to acoustic particle motion (Fay, 1984; Fay , 2023). Dick found that each otolithic end organ encodes unique directional information due to the directional sensitivity of its various hair cell populations as well as the plane of the populations relative to the vertical and horizontal. He therefore proposed that the central computation of this directional information, potentially from all three of the otolith end organs, could contribute to sound localization (see also Popper , 1988; Rogers , 1988,; Lu , 2004, 2010).

Since Dick's initial study, similar results have been demonstrated for each of the otolith end organs in a number of teleost species (e.g., Lu , 1998; Lu , 2003, 2004; Maruska and Mensinger, 2015; Vetter , 2019; Rogers and Sisneros, 2020) and for the saccule and lagena of a sturgeon (Meyer , 2010). Moreover, portions of the saccule and lagena are reported to be sensitive to sound in a clade that predates the origin of bony fishes; cartilaginous fish (Lowenstein and Roberts, 1951; Budelli and Macadar, 1979; Corwin 1981). The variability of otolith end organs sensitive to sound supports the speculation that the distribution of acoustic sensitivity among all, or a subset, of the otolith end organs is a ubiquitous and ancestral condition for jawed vertebrates and could explain why evolution could alter the relative contributions of the otolithic end organs participating in hearing in different species (McCormick, 1992). For example, the specializations of the utricle for hearing ultrasound in the Clupeiformes (for example herrings and shad) could evolve because the ancestral utricle had hair cell populations that encoded whole-body accelerations induced by sound, no matter how minor relative to those supporting gravistatic sensitivity.

This speculation brought three questions to mind, which spurred research in my lab for the rest of my career:

1. Is the hypothesized ancestral pattern of otolithic organ input to primary nuclei (i.e., the pattern in the clearnose skate and the bowfin, Fig. 2) retained in teleosts?

2. Do species that augment acoustic particle motion with pronounced sensitivity to sound pressure have concomitant changes in the central auditory pathways, at least in the primary nuclei?

3. Are the otolith organ maculae ancestral sensory mosaics, such that each macula has both auditory and vestibular regions? Furthermore, can the central inputs of the otolithic end organs provide an anatomical basis for the suggestion of Dick and others that the directional sensitivity of the three end organs might be combined centrally?

In order to pursue these questions from an anatomical standpoint, it would be necessary to trace the branches of individual otolithic end organs to their first-order projection sites. This presented a technical problem because of limitations in accessing these branches in vivo.

The in vivo tract-tracing experiments performed in the bowfin and the clearnose skate were possible because the octaval nerve branches to the otolithic end organs are relatively accessible, and because these fish are relatively hearty and can thus survive for the several-day period required for the tracer to reach the termination point of the nerve axons. In many teleosts, the more ventral location of the inner ear makes it relatively inaccessible to surgical manipulation consistent with post-surgical survival. Two notable exceptions, the oyster toadfish, Opsanus tau, and the plainfin midshipman, Porichthys notatus, have inner ears positioned in the head similar to that of the bowfin, and have become model species in studies of hearing (Feng and Bass, 2017; Bass, 2023).

Thus, in the absence of an in vitro method for tracer studies, I was greatly constrained in the choice of teleost species I would be able to use in my comparative studies. During our time at Georgetown University, the only in vitro tract tracer known was cobalt chloride. This tracer works well in invertebrates, but poorly/not at all in fish. Nevertheless, Mark and I would have a “yearly attempt” to try the cobalt chloride in vitro method in a goldfish. After yet another unsuccessful try, I thought it could do no harm to substitute HRP for the cobalt chloride, label the posterior lateral line nerve in a goldfish (since we already knew from in vivo studies where it projected), refrigerate the head, and then process the brain as if the experiment had taken place in vivo. To our great surprise, the in vitro HRP results were identical to the in vivo results! With some further adjustments, what came to be known in our labs as “The Frankenstein Method” was born (McCormick and Braford, 1984).

Mark and I became faculty members at Oberlin College in 1986. Mark was focused on the evolution of the fish telencephalon, but we collaborated from time to time on the teleost inner ear using our original (McCormick and Braford, 1993, 1994) and subsequently refined (McCormick, 2001) in vitro method.

I will focus the rest of this contribution on the descending octaval nucleus (DO), the largest of the octaval nuclei and the nucleus that in fishes universally projects to the midbrain torus semicircularis. The DO is composed of dorsal and intermediate regions that receive input almost exclusively from the otolithic end organs, and a ventral region that receives input from the semicircular canal cristae. In all fish species that have been studied, neurons occupying the medial-most area of the dorsal and intermediate regions send axons to the known/presumed auditory area, torus semicircularis of the midbrain (Fig. 3). In the goldfish, neurons in the medial-most area of the dorsal region send axon collaterals to a second order nucleus (Fig. 3) (Yamamoto and Ito, 2005), the secondary octaval population of McCormick and Hernandez (1996). The dorsomedial neurons in goldfish have connections through a dorsal commissure (McCormick and Hernandez, 1996), whereas in the oyster toadfish, identified auditory neurons in the lateral DO have commissural connections (Edds-Walton and Fay, 1998; Edds-Walton and Fay, 2008). In general, similar connections in other teleosts, along with physiological studies, support the conclusion that auditory regions in the dorsal descending nucleus are analogous (but not homologous) to the amniote/mammalian cochlear nucleus (McCormick, 1992, 2011).

I spent most of my years at Oberlin, in collaboration with Mark and Oberlin College undergraduates, addressing the first two of the three questions listed in the previous section. First, in all of the teleosts we studied except for one (the gizzard shad, Dorosoma cepedianum; McCormick, 1997) (Fig. 4), otolith end organ inputs had essentially similar patterns of distribution in the descending nucleus (McCormick, 1999, 2001; McCormick and Braford, 1993, 1994, O'Marra and McCormick, 1999), and are consistent with findings in other species (Kozloski and Crawford, 1998; Bass , 2000, 2001; Maruska and Tricas, 2009). The termination areas of the saccular, lagenar, and utricular branches are arranged medial to lateral and ventrolateral, with some overlap between neighboring zones. Apparently, the answer to question one is that the ancestral pattern of otolith organ input to the DO (i.e., the pattern in the clearnose skate and the bowfin, Fig. 2) is retained in almost all teleosts studied to date. Although the teleostean anterior octaval nucleus is not the focus of this contribution, it too has inner ear inputs that resemble those in the ancestral condition, but with some species-specific variations (McCormick, 1999).

Otolith end organ projections to the DO also provided insight into question two: namely, whether pronounced sound pressure sensitivity enabled by the connection of the inner ear to a gas-filled structure is reflected by morphological specializations within this nucleus. The answer is not straight-forward because a particular morphological variation in this nucleus relative to the ancestral condition was revealed in all of the teleosts I studied (McCormick, 1998, 1999), whether they possessed a specialized auditory periphery (i.e., an otophysic ear) or not! This variation, which I did not appreciate during my earlier comparative analysis (McCormick, 1982), involves the anatomical relationship of the dorsal-most neurons of the DO to a molecular layer of fibers originating from higher levels of the brain that caps part of the medulla—the cerebellar crest (CC). The CC provides input to dorsally extending dendrites of these neurons, while their ventral dendrites receive input from one or more of the otolith end organs. The functional significance of CC input to DO has not been studied. However, physiological studies of CC input to first-order lateral line neurons reveal that this input is a component of an adaptive filter circuit that cancels the sensory reafferent or “self-motion” component of incoming information (Montgomery and Bodznick 1993, 1999, 2016; Bell , 1997).

In the clearnose skate and the bowfin, the roughly crescent shaped DO is not adjacent to the cerebellar crest, and its dorsal-most neurons have no obvious input from it, although this apparent lack of input should be re-visited [Fig. 5, Pattern A (A)]. In teleosts, however, the DO has a medial extension that protrudes dorsally and lies adjacent to the CC and the neighboring mechanosensory nucleus medialis (Fig. 5, Patterns B and C). This dorsomedial extension, referred to as the dorosomedial zone, (McCormick and Braford, 1993, 1994; McCormick, 1999) may be an evolutionary specialization of the dorsal neurons in the DO of the bowfin, or may represent a new cell component in the DO. Within Pattern B, there can be discontinuity [Fig. 5, Pattern B (B)] or continuity [Fig. 5, Pattern B (C)] between the dorsomedial zone and the remainder of the dorsal DO. Species exhibiting either type of Pattern B do not have a swimbladder-inner ear (otophysic) connection. However, species exhibiting Pattern C, in which the dorsomedial zone is noticeably larger than in Pattern B, all have an otophysic periphery.

The association of a specialized, otophysic auditory periphery with a dorsal DO possessing a large dorsomedial zone appears to be an example of how specializations of the auditory periphery are reflected in morphological characteristics of the dorsal DO. An even more dramatic example of this point is present in clupeids, in which the utricle and cephalic lateral line uniquely respond to sound pressure by virtue of their mechanical coupling to a specialized portion of the swimbladder—the auditory bullae (Allen , 1976; Denton and Blaxter, 1976). The dorsal DO of the gizzard shad exhibits Pattern C and unique inner ear inputs that reflect the unique auditory periphery characterizing clupeids. The clupeid utricle has three separate maculae, two of which are entirely or partially responsive to sound pressure (Popper and Platt, 1979; Best and Gray, 1980). Nerve fibers from these two maculae project, at least in part, to the medial most region of the dorsomedial zone, thereby supplanting the typical projection of saccular fibers in non-clupeids (Fig. 3 and Fig. 6; McCormick, 1997; also see Meredith, 1985). It would be fascinating to know how this connectional difference is reflected at higher order brain levels.

It was not until I had access to a confocal microscope at Oberlin College that I was able to begin to address my third question: the hypothesis that the otolith organ maculae may be ancestral sensory mosaics, such that each macula has both auditory and vestibular regions. My goal was to determine whether individual neurons in the dorsal DO that project to the auditory midbrain have inputs from more than one otolithic end organ. The answer to this question could also address the suggestion of Dick Fay and others that otolithic inputs might be combined centrally and thus underly sound localization.

I made two decisions after a summer working out the methodology I would use. The first was to focus on the dorsal DO because it is the largest population of medullary neurons known or implicated to be auditory in all fish. The second was to do the study on the goldfish, even though its otophysic peripheral auditory system is specialized relative to the ancestral condition for actinopterygian (ray-finned) fishes. I reasoned that if otolith end organ inputs converged onto common auditory neurons in an otophysic species, that such convergence might reflect the ancestral level of organization. In addition, pilot studies indicated that large numbers of experiments would be required; thus, I needed a readily available and economical test species.

Confocal analysis could overcome a limitation of the tract-tracing methods used in previous studies of otolithic organ projections. Counterstains used in these studies revealed somata (Fig. 7a), but at best, only the most proximal parts of their dendrites. Significantly, the bulk of lagenar and utricular terminals are located in a cell-poor, but dendrite-rich, area lateral to the somata in the dorsal DO, and many of these dendrites emanate from these somata (Larsell, 1967). Thus, inputs to laterally extending dendrites of such neurons could not be visualized. This limitation was overcome by making multiple injections of different tracers in a given fish. This resulted in more complete labeling of dorsal DO somata and their dendrites and allowed the inputs of two otolith end organs per fish to be differentiated [Figs. 7(b) and 7(c)].

The results of this study were striking in two ways (McCormick and Wallace, 2012). First, they highlighted a range of dorsal DO neuronal subtypes and their dendritic arbors, some of which had not been previously described [Fig. 7(d); also see Fig. 5 from Yamamoto and Ito (2005)]. All of these subtypes, which are grouped into medial and lateral populations, project to neurons in the torus semicircularis. Second, they showed unequivocally that most dorsal descending neurons receive input from at least two end organs (Fig. 8). Although the characteristics of the confocal microscope available to me did not allow me to visualize simultaneously three end organ inputs along with the DO in a given fish, the combinations of nerves labeled across different specimens makes it conceivable that three end organs could supply a given neuron.

There is, however, one neuronal subtype in the goldfish DO that seems to be exclusively supplied by the saccule, the dorsomedial spherical neurons [Md in Figs. 7(a) and 7(d)]. Goldfish, like other otophysic species, display Pattern C, in which the dorsal DO is noticeably larger than in Pattern B (Fig. 5). The dorsomedial spherical neurons [Fig. 7(e)] constitute a significant part of the hypertrophied dorsal area. Because the goldfish saccule is known to process sound pressure, one can speculate that the dorsomedial spherical neurons process this auditory input and that the other cell types in dorsal DO process particle motion. Since at least some of these putative particle motion sensitive neurons receive input from at least two otolith end organs, they are in line with Dick's suggestion that convergence of end organ inputs might underlie sound location (Fay, 1984).

In addition to describing some key anatomical features of the dorsal DO, which along with the anterior octaval nucleus constitutes the first central components of the auditory system in jawed fishes, I have tried to emphasize the importance of comparative studies to our understanding of piscine hearing and its evolution among fishes. In terms of central anatomy, defining the ancestral organization of any system is essential because this organization constitutes the fundamental scaffolding upon which further evolution takes place.

If we are going to fully understand how fish localize sound, defining the anatomical substrate for directional processing is essential. Although it can be hypothesized that inputs to individual dorsal DO neurons from more than one otolith end organ form part of this anatomical substrate, this is by no means proven. It is even possible that some of these inputs provide vestibular rather than auditory information to dorsal DO neurons, although no theories of sound localization in fish have proposed that positional/tilt information contributes to computations underlying this behavior.

In the oyster toadfish, a homotopic commissural tract connecting the right and left DO and extracellular recordings indicate binaural processing occurs in this first-order population (Edds-Walton, 1998; Edds-Walton and Fay, 1998, 2008) in contrast to initial binaural processing that occurs at the second-order level in amniotes. A small number of fibers from the sound pressure-sensitive end organ in goldfish (saccule) and gizzard shad (utricle) traverse into the contralateral dorsal DO (McCormick 1997; McCormick and Wallace, 2012), which likewise implies binaural processing at the first-order level. Thus, future physiological and further comparative anatomical studies are needed to uncover the details of how the dorsal DO encodes sound direction.

More studies of the contribution of the mechanosensory lateral line to hearing are also needed. Many anatomical studies have noted that while the vast majority of lateral line nerve fibers provide input to the first-order mechanosensory nucleus medialis, a small number of fibers course into the dorsal DO (reviewed in McCormick , 2016). In the goldfish, these fibers contact specific cell populations within the laterally located cell portion of the dorsal DO (McCormick , 2016). The lateral line system can be stimulated by low frequency sounds (<200 Hz) at close range (reviewed in Coombs and Montgomery, 1999; Braun and Sand, 2014; Higgs and Radford, 2016). Various hypotheses about how the lateral line system might complement audition have been proposed (reviewed in McCormick , 2016), including a role in sound localization (Zeddies , 2010; Radford and Messinger, 2014). Higgs and Radford (2013) reported that in the goldfish the auditory evoked potential at 100 and 200 Hz is a combination of the activity of both the lateral canal neuromasts and the otic auditory organs, whereas at higher frequencies it results only from the inner ear. They suggested that at such low frequencies, the auditory evoked potential of other fish species might be re-visited to see whether this is a common characteristic.

Differing cell morphologies imply functional differences. The neurons in the dorsal DO that project to the acoustic midbrain in goldfish form three medial and four lateral populations, and each population is different in morphology and inputs from the inner ear and the lateral line (McCormick and Wallace, 2012; McCormick , 2016). The notably large size of the somata and dendritic terminals of some of these populations provides an opportunity to begin an exploration of how convergent inputs from the otolith and lateral line end organs shape the output of these populations to higher-order levels and potentially to other dorsal DO neurons.

Fibers from the saccule supply the most medial neurons in the dorsal DO in all fishes. In otophysans such as the goldfish, in which the saccule is a specialized sound pressure receptor by virtue of its connection to the swimbladder, the saccule has an exclusive projection to the most dorsal and medial of these neurons, Md. However, in the gizzard shad (a clupeid), the physical relationship between the utricle and the auditory bullae imparts sound pressure sensitivity to this end organ, and it is the utricle that projects to neurons corresponding to the goldfish Md! What developmental signals underlie this difference in connectivity?

I thank Art Popper and two anonymous reviewers for providing me with useful comments and help with various aspects of the manuscript. I appreciate the Oberlin college undergraduates who carried out research in my lab relevant to this contribution, whether they were my co-authors or not. For many years, Sandra Grellinger Ronan created a welcoming and supportive atmosphere in the lab that, along with her technical expertise, benefitted us all. Finally, I am grateful for the insightful suggestions of Mark Braford, my steadfast partner both personally and professionally. My research has been supported by the Oberlin College Department of Biology and grants from Oberlin College and the National Science Foundation.

The author has no conflicts to disclose.

Data sharing is not applicable to this article as no new data were created or analyzed in this study.