The Evolution of Arthropod Nervous Systems

Daniel Osorio, Jonathan P Bacon, Paul M Whitington. American Scientist. Volume 85, Issue 3. May/Jun 1997.

A taxonomist viewing insects or crustaceans tends to see tough, jointed exoskeletons and elaborate limbs as their salient features-so salient, in fact, that the taxonomists who named the phylum to which these animals belong called it Arthropoda, or “jointed foot.” The same biologist will also notice that the arthropod body plan is that of a modified worm, where the worm’s more-or-less uniform series of segmental units becomes well differentiated along the length of the arthropod body. This gives arthropods distinct head, middle and tail regions.

As neurobiologists considering evolution from simple to more complex animals, we are interested in the changes in the neural circuits serving locomotion and sensory coding that accompanied the transition from worm to arthropod. Such evolution is surely crucial to the emergence of advanced mobile animals. Because brains and body plans must have evolved in step from their simpler antecedents, comparative studies of neural development and of the mature nervous systems of animals promise to reveal much about arthropod evolution. In arthropods almost every nerve cell grows in a particular pattern and makes specific synaptic connections to form the elaborate neural circuits that give an animal its mobility and control; these features make each cell a recognizable identity.

Just as vertebrae and ribs are serially repeated, sets of neurons are repeated in each segment of an insect’s body, or beneath each facet of the compound eye. Using identified neurons, anatomists can compare the development and structure of nervous systems from different arthropods much as they can compare vertebrate skeletons. In addition, we can record the electrical activity from a single neuron, which allows us to compare physiological function in different lineages. So far, most work has been done on insects. But comparisons can also be made between insects and crustaceans. Such comparisons tell biologists how these arthropods, with their jointed exoskeletons and specialized limbs, evolved from simpler ancestors and how they have subsequently diversified during evolutionary history.

The Emergence of Arthropods

Among a variety of arthropod-like animals, three main classes are clearly present in the fossil record by the Middle Cambrian Period, around 520 million years ago. These include the trilobites, which are now extinct. The other two classes are the crustaceans and the chelicerates, whose living representatives include crabs and spiders, respectively two further groups of arthropods, the insects and the myriapods (millipedes and centipedes), are terrestrial, having descended from unknown marine ancestors. Myriapods first appear in rocks of the Silurian Period some 420 million years ago, and as such were among the first animals to appear on land. Insects first appear unequivocally somewhat later in the Devonian, around 375 million years ago.

As Darwin noted in the Origin of Species, the abrupt emergence of arthropods in the fossil record during the Cambrian presents a problem for evolutionary biology There are no obvious simpler or intermediate forms-either living or in the fossil record-that show convincingly how modern arthropods evolved from worm-like ancestors. Consequently there has been a wealth of speculation and contention about relationships between the arthropod lineages.

For example, some see the group Arthropoda as a true phylum descended from a unique ancestral arthropod. Others, most notably the late English zoologist Sidnie M. Manton, contend that differences in limb form, musculature and embryonic development show that insects, crustaceans and chelicerates evolved independently from soft-bodied worms. According to this view, shared characteristics, such as a jointed exoskeleton, limbs and large eyes, arose as solutions to the demands of active life-styles and were adopted independently by many groups during the Cambrian Period.

But even if the fossil record is slow to yield information about how arthropods arose from their soft-bodied ancestors, comparative studies have received a boost in recent years from work on the molecular and genetic mechanisms underlying the development of body plans in animals such as the fruit fly Drosophila and the mouse. A notable finding is the discovery of a family of genes, called homeotic (or HOM/Hox) genes, which play an important role in specifying body plans in a wide variety of animals. These genes, found in all animals and plants, code for proteins that bind to and control the expression of DNA. By comparing their DNA sequences it is possible to recognize related homeotic genes in distantly related organisms. These observations point to a hitherto unsuspected unity in the mechanisms governing the development of body plans. They also suggest that a key to the emergence of animals such as chordates (which include vertebrates) and arthropods during, or before, the Cambrian, was the evolution in a common ancestor of mechanisms that permitted the development of complex body plans.

Anatomy of the Insect Nervous System

Alteration in external appearance is but the most obvious modification wrought by evolution. As the animals changed on the outside, internal changes also took place to control and coordinate the activities of an increasingly specialized body. The wiring of the nervous system necessarily differs between animals that swim and those that walk, or between animals with and without wings. To understand the innovations that gave rise to the arthropod nervous system, neurobiologists such as ourselves study and compare nerve cells between distant relatives in modem phyla-in particular, between insects and their crustacean relatives. The details of the development, anatomy and physiology of the insect and crustacean nervous systems promise to reveal much about the evolutionary relationships between these two groups.

The basic structure of the nervous system is similar for insects and for the largest crustacean subclass, the malacostracans, which includes marine animals such as lobsters, crayfish and crabs, as well as sowbugs on land. The nervous system in these animals runs from head to tail along the belly (and not the back, as it does in vertebrates). Just as the arthropod body is segmented, so, too, is its nervous system.

Each body segment contains a mass of nerve cells called a ganglion, which serves as a local control center for that segment. In insects, each ganglion contains between 500 and 3,000 cells. The cells integrate inputs from sense organs and send out projections called axons, which transmit signals to muscles within the segment. In addition, the nerve cells in one segmental ganglion can also send messages to and receive input from the cells in neighboring ganglia. In this way, cells in the ganglia can control the appendages in their own segment, and through connections to other ganglia, they can coordinate movements between segments.

The head of insects and crustaceans is formed from a set of fused segmental units, whose precise number, perhaps seven, has proved difficult to establish. Likewise the head contains a set of segmental ganglia fused to form a brain. Because so much of the sensory equipment is contained in the animal’s front end-structures such as the eyes and antennae-head ganglia are specialized for sensory processing, which requires a relatively large number of cells. As in vertebrates, the invertebrate brain contains the majority of the nervous system. Moreover, about half of the entire complement of neurons, around 300,000 in an insect such as a blowfly, form the optic lobes that serve the compound eyes.

Development and Cellular Homology A good place to start comparisons between animal lineages is with early development. Cells that share a common developmental pathway are likely to be homologous, meaning that they originated in a common ancestor and have been inherited in the two lineages. Identification of homologous neurons and neural circuits at the level of single cells makes the arthropod nervous system a potentially valuable source of information about phylogeny. Many studies focus on the relationships between flies and grasshoppers, whose lineages diverged about 300 million years ago. We are also beginning to see close similarities between these insects and crustaceans, whose nervous systems are derived from a common ancestor that lived even earlier.

Microscopic observation of the embryonic nervous-system in the arthropods reveals that neurons in each segment are generated from a set of about 60 precursor cells called neuroblasts. Neuroblasts can be recognized by their position in the segment. They generate a set of mature neurons by a stereotyped pattern of cell divisions. Each neuron has a unique identity and place in the neural circuitry. The progression from neuroblast to mature neuron is brought about by specific patterns of gene expression that can be traced using molecular probes.

Neurons from insects of different species that have cell bodies in similar positions in the ganglion and that also share common molecular markers are generally found to have identical lineages of descent from their neuroblast precursor.

In flies and grasshoppers the pattern of neuroblasts and, where known, the expression of cell-specific genes in each embryonic segment are closely comparable. In addition, the pattern of axon growth of the central nervous system is remarkably similar in these two insects and in the wingless silverfish, an animal that is thought to resemble the flightless common ancestor of winged insects. Thus neurons that lie in equivalent positions in these three species send out axons at the same time during development, and these axons generally follow the same trajectory.

Patterns of gene expression can secure homologies that are based on purely morphological grounds, rather than the more rigorous criterion of developmental lineage. A good example is the gene even-skipped (eve), which is a determinant of morphology in some identified neurons of the fruit fly Drosophila. The homologous eve gene in the grasshopper has now been cloned and sequenced. The grasshopper’s eve gene is expressed in the same identified neurons, called RP2, aCC and pCC, as in Drosophila. For aCC and pCC this confirms homology of the Drosophila and grasshopper neurons established by lineage, but eve expression provides a useful signature for comparative studies where detailed developmental study is impractical. The silverfish (a wingless insect) and also the crayfish (a crustacean!) have putative homologues of these three cells, based on cell-body position and axon trajectory, and it will be interesting to see whether they too express the silverfish and crayfish eve genes.

In both grasshopper and Drosophila another neural precursor, MP2, divides only once to form two dissimilar cells. One of these is the dMP2 neuron, which sends its axons posteriorly, to the animal’s rear end. The other cell is vMP2, which sends its axons anteriorly, toward the head. But not so in the silverfish. Here we have found that the axons of both its putative dMP2 and vMP2 neurons run toward the head.

Such observations can lead to insights about the developmental, and maybe evolutionary, differences between winged and wingless insects. A clue to these differences is suggested by the Drosophila gene called numb, which produces a protein in MP2 that segregates entirely into dMP2 (and not to its sister cell vMP2) and determines its identity. Should the numb gene be nonfunctional in Drosophila, the effects become apparent via a transformation of dMP2. In these numb mutants, the axon of the cell, which usually grows toward the posterior end of the animal, now grows anteriorly toward the head, as in silverfish. This observation raises the intriguing possibility that an evolutionary change in axon morphology may be the result of a change in the expression of this gene. In crustaceans, as in silverfish, the neurons in the equivalent positions to the insect dMP2 and vMP2 neurons both send their axons anteriorly, and so resemble numb-mutant Drosophila. We intend to look at evolutionary relationships between the neurons by examining numb expression–or the lack of itin silverfish and crustaceans.

Given the similarities in development, it is not surprising that the adult nervous systems of diverse insects often share much the same neural circuitry. We have examined one of the sensory interneurons, the tritocerebral commissure giant (TCG) neuron of the grasshopper and its putative homologues in cricket and praying mantis. We say “putative” because we have not established the TCG developmental lineage in any of these insects. Nevertheless the TCG is easily recognized from its overall structure. It has a cell body and dendrites in the brain, where it receives inputs from sense organs on the head. Its axon crosses the midline in the tritocerebral commissure and runs to the thoracic ganglia, which are the main centers of flight control; it is this distinctive form that provides the main argument for homology. In the grasshopper the TCG provides wind information to the flight motor, and it may have a similar role in the cricket and the praying mantis. But the TCG puts a different weighting on sensory input from head hairs, antennae and compound eyes in each species according to the relative importance of the sensory structures on the particular animal’s head. Thus, for example, the cricket is bald in comparison to the grasshopper, but its antennae are longer. These differences in sensory integration are accomplished with only minor changes in the general form of the TCG neuron. This conservation of form seems to be widespread in cells such as the TCG, with substantial evolutionary changes in external organs and behavior reflected by only minor modifications in the central nervous system.


Vision is the chief sense in almost all mobile animals that live where there is enough light to see. Large compound eyes with a facet lens for each “pixel” in the optical image are a distinctive feature of insects and many crustaceans. Beneath these eyes much of the brain is devoted to vision. By comparison, simpler and sedentary animals generally have small eyes and a rudimentary visual system.

Although Darwin said that contemplating the perfection of the human eye gave him a “cold shudder,” it is not difficult to evolve the superb optics of our eye. In an elegant piece of modeling, our colleagues Dan Nilsson and Susan Pelger of Lund University in Sweden have recently shown that an optically perfect eye with a single lens, known as a simple eye, can be reached over a short evolutionary time by incremental improvements from an unfocused eyespot. If evolution takes a different path, the aggregation of many simple units, each with its own lens and a small number of receptors, will give a compound eye. Given the importance of vision, it is no surprise that good eyes, both simple and compound, have evolved independently on many occasions. Vertebrates such as people and mollusks such as squid have simple eyes. A wide range of invertebrates have compound eyes. The presence of compound eyes in various arthropods is not in itself evidence that they share a common evolutionary origin.

But what of the large compound eyes of today’s insects and crustaceans? Are these products of convergent evolution occasioned by these two groups following similar life-styles, or are their eyes and visual systems homologous structures that have been inherited from a common ancestor?

Ideally, resolution of this question should rest on identifying cellular homologies of the kind we have seen in the segmental ganglia, but so far, developmental studies of compound eyes do not permit identification of homologues by lineage. Nevertheless, close similarities in the cellular architecture of the basic unit of the compound eye, or ommatidium, of insects and crustaceans have been recognized since the 19th century. Ommatidia of insects and crustaceans are in fact virtually indistinguishable at a cellular level; the optical elements comprise two corneal cells and four cone cells that together make a lens that focuses light onto a set of eight or so photoreceptors. These receptors transduce light and send electrical signals to the optic lobe. Interestingly, compound eyes of arthropods such as the horseshoe crab Limulus (a chelicerate) and the centipede (a myriapod) differ from the plan common to insects and crustaceans. Even though they have been proposed as comparatively close relatives of crustaceans and insects respectively, the eyes of chelicerates and myriapods seem more likely to have separate origins.

The similarities between eye structures of insects and crustaceans mirror those in their segmental nervous systems. Both seem likely to have been inherited more or less unchanged from a common ancestor, but in neither case can we be sure that this ancestor was itself a sophisticated arthropod. In particular, an isolated ommatidial unit could evolve from the eyespot of a simple worm. Many worms have such eyes, and compound eyes could have evolved in parallel by multiplication of the basic module. However, animals with isolated eyelets tend to have simple visual behaviorsfor example, moving toward or away from light sources or withdrawing after sudden changes in light intensity-and these do not require a complex nervous system. For this reason it is worth paying particular attention to the neural circuits serving more sophisticated visual behaviors of which insects and crustaceans are capable; homologies here would strongly indicate that their common ancestor itself had good vision.

Examples of such behaviors include the recognition of landmarks for navigation and the use of optical cues to guide flight. As an animal moves about, it receives a stream of optical signals that need to be analyzed. The animal must interpret the SD layout of surrounding objects to determine where it is heading and the location of mates or food. Vision demands a heavy investment in neural machinery; about half of the primate cerebral cortex is taken up by visual areas, and likewise in insects about half of all the nerve cells are involved in vision.

The first stages of vision involve point-by-point transformation and coding of the image on the retina. This is done by neurons in the optic lobe, whose columnar architecture of almost crystalline regularity beautifully reflects their function. Studies of the arrangements and connections of neurons in the optic lobe and electrophysiological recording of their responses to visual stimuli allow us to build up a picture of how vision is implemented.

The neural architecture of the optic lobe is much alike in insects and malacostracan crustaceans. Signals from the eye pass successively through three ganglia: the lamina, the medulla and the lobula. In the outer two gangliathe lamina and the medulla-the neurons are arranged to form columnar modules, with a set of cells repeated beneath each ommatidium. This architecture indicates that the same set of operations is performed in parallel on each point, or pixel, in the optical image. At these peripheral stages of vision there are some 60 individual neurons for each facet on the eye, so a set of some 60 “neural images” is derived from the single optical image on the retina. These neural images, roughly speaking, filter and transform the stimulus to encode information about local image features such as color, patterns of motion and object borders. In later stages of the visual pathway, these different features of the visual scene are combined by neurons with larger receptive fields, first in the lobula, the third ganglion in the optic lobe, where there is a coarse columnar organization with a set of neurons for every four facets in the eye, and then in the brain where neurons have wide visual fields.

Lamina Anatomy and Physiology

The first and smallest ganglion in the optic lobe is the lamina; it receives direct inputs from the retina, as well as feedback from the medulla, and has only about 10 cells in each column. The lamina’s anatomical simplicity and its position early in the visual system make it an attractive place for analyzing neural circuitry and physiology, as well as for making comparisons between different groups of animals.

The work of determining the lamina circuitry was started early this century by the Spanish neuroanatomist Santiago Ramon y Cajal and his colleague Domingo Sanchez. Cajal was a master of the Golgi staining technique, which labels a very small proportion of the total population of cells in the nervous system and reveals the entire structure of individual neurons within a clear mass of unstained cells. Cajal spent many years looking at vertebrate brains, where millions of comparatively undifferentiated cells form highly interconnected networks. He was therefore enchanted by the precise construction of the insect optic lobe, which he memorably compared to a well-built pocket watch, in contrast to the crude “wall clock” of the vertebrate retina. Cajal noted that the photoreceptor cells send outputs from the retina. There are eight photoreceptor cells in each ommatidium of the fly, but nine in the honeybee. Six of these have short axons that terminate in the lamina, whereas two or three have long axons that pass through the lamina and terminate in the medulla.

By painstaking inspection of various insect laminas Cajal found that each columnar module, or lamina cartridge, is built from six main types of neurons. The 10 or so neurons of the lamina were named by Cajal according to their anatomical form and the positions of their cell bodies. Five monopolar cells have their cell bodies between the retina and the lamina and send signals from the lamina to the medulla, as does a single T cell whose cell body lies between the lamina and medulla. In addition, Cajal surmised that one to three C (for “centrifugal”) cells, whose cell bodies lie much deeper in the optic lobe between the medulla and lobula, send feedback from the medulla to the lamina. Finally, amacrine cells are restricted to the lamina and make lateral connections between neighboring columns. Using only the Golgi technique, Cajal predicted the direction of information flow in each neuron. Remarkably, 80 years later physiologists have been able to confirm these predictions.

The orderly structure and the small number of cells in the lamina cartridge have allowed contemporary workers, especially Ian Meinertzhagen, Steve Shaw and their colleagues at Dalhousie University in Halifax, Nova Scotia, to extend analysis of lamina anatomy with electron-microscopic studies of the neural circuitry. By cutting the lamina into a series of thin sections, it is possible to follow individual neurons and see where and how synaptic connections are made between the cells. Elsewhere, neurophysiologists have been making intracellular recordings from lamina neurons with dye-filled micropipettes. This allows them to establish the responses of the neurons to optical stimuli and then to label cells with dye. Once the cells are labeled, the neurophysiologists can identify the cell and so fit together physiology and the neural circuits described by anatomists.

Physiologically, the most accessible lamina neurons are a subclass of monopolar cells called the large monopolar cells (LMCs), so called because of the large diameter of their axons-about 3-5 micrometers. Neurobiologists have recorded from the LMCs of many insects as well as from crayfish. It turns out that LMCs are not only morphologically similar, but in each animal studied, the cells also share physiological characteristics, as well.

LMCs are the sole lamina neurons to receive direct input from the receptors and to project to the medulla, and as such, they form the main pathway from the eye to the visual-processing centers of the brain. Both in insects and in crustaceans such as the crayfish, LMC responses are easily recognized because they do not produce the action potentials typical of most neurons. Instead, they propagate signals by graded potentials. LMCs do not signal absolute intensity, but instead signal fluctuations about a mean background level, or contrast. They respond to a step increase in illumination of the principal receptor inputs by becoming hyperpolarized. Unlike receptors, the responses of LMCs are not sustained, and they drop back to their resting level after 100 milliseconds or so. If the light stimulus is outside the visual field of the principal input receptors, responses are much smaller and inverted, so that an increase in intensity causes a depolarization.

In attempting to discern evolutionary relationships, it is of crucial importance to establish to what degree similarities such as those shared by the LMCs of insects and crustaceans arise from a common inheritance and to what degree they reflect convergent evolution resulting from their having a common role in visual coding. For this reason it is important to compare other cell types in the lamina as well, but intracellular recording from the smaller lamina neurons is more difficult than from the LMCs. Working with Andrew James in the Centre for Visual Sciences at the Australian National University, we have described all the main lamina-cell types in the grasshopper. Raymon M. Glantz and his colleagues at Rice University in Houston have done much the same in crayfish. We see marked similarities both in anatomy and physiology. For example, both the insect and the crustacean T cell response resembles that of the LMCs, but it is more sluggish and has a less sharply defined receptive field.

Similarly, one of the small monopolar cells in grasshoppers, crayfish and flies does not receive direct receptor input and responds with the opposite polarity to the LMCs-it is depolarized by light. This monopolar cell is strongly inhibited by off-axis illumination. Another small monopolar cell in the grasshopper and the fly is excited by light over a much wider area and is briefly excited by any intensity change, whether dimming or brightening. This latter cell seems to be absent from crayfish. Finally, a single C cell recorded from the locust gives delayed responses to light with a very complex wave form, not unexpected for a feedback neuron. There is anatomical evidence for C cells in crayfish, but these have not been recorded physiologically.

The picture that emerges from comparative studies of lamina in insects and crayfish is one of remarkable uniformity from animals whose life-styles and evolutionary lineages should set them far apart. Not only are they probably separated by over 500 million years of independent evolution, but insects also are fast moving aerial animals, whereas crayfish are comparatively sluggish, aquatic and often nocturnal. Even so, we find the same set of neurons making up a lamina cartridge, with the same physiological responses. Given that a worm-like life-style does not require much of a “visual brain,” it is probable that the elaborate neural circuitry such as that in arthropod optic lobes would evolve de novo in a mobile and maneuverable animal. If, as we expect, it turns out that the circuitry of the laminas in animals as diverse as grasshoppers, flies and crayfish is indeed homologous, this would provide strong evidence that the ancestor common to all three animals was itself a comparatively sophisticated arthropod.

The Emergence of Complexity

A fascinating aspect of evolution is the emergence of complex biological systems from simpler antecedents. Eukaryotic cells evolved from prokaryotic bacteria, multicellular animals from single-celled ancestors, and modern arthropods and vertebrates from wormlike ancestors. But complexity in animal form and behavior is hard to define; in a recent book John Maynard Smith of Sussex University and Edrs Szathmary from Budapest review the major transitions in evolution from the first appearance of life in prebiotic chemical systems to human language. They see changes in organization as increases in complexity but offer no measure of complexity other than the total amount of DNA in the genome. Others have suggested the number of histologically recognized cell types as a measure of complexity in multicellular organisms. Yet in comparing a worm to a jellyfish or a sponge, a crab to a worm, or a human being to a lamprey, the numbers of genes or cell types are at best correlates of differences in organization and behavior. These differences are themselves much better indexes of our intuitive understanding of biological complexity.

Over recent years developmental genetics has shown remarkable and previously unsuspected unity in the mechanisms and controls regulating the formation of body plans in animals as diverse as mice and flies, above all by the HOM/Hox genes. The message is that just as the genetic code appeared only once, so the emergence of mechanisms for building the complex animals we know today was a unique event, one that perhaps took place late in the Precambrian era. But, as we have observed, bodies and behaviors must evolve in tandem, and the complexity of the nervous systems of animals, both in differentiation of cell types and the intricacy of their connections, far exceeds that of other organs and the overall body plan. Thus what applies to body plans is true a fortiori for the nervous system. Similarities of the development and function of neural circuits in distantly related and diverse arthropods may well reflect a common heritage from the Cambrian. Since that time, circuits devised for early vision or control of locomotion seem to have been retained and modified only minimally to serve the diverse habits and habitats of contemporary insects and crustaceans.