Variations on an ancient theme — the central complex across insects

The central complex (CX) is a highly conserved region of the insect brain, and its ubiquitous occurrence suggests that its neural circuits are of fundamental importance. While its overall layout has not changed since the evolution of insect flight, substantial variations exist in the internal organization of all CX components. By changing the details of a system of repeating columns and layers, these differences affect the almost crystalline internal organization of the CX and thus the characteristic neuroarchitecture that directly links structure with function. While neuropil level changes suggest widespread differences in cellular architecture and circuits, data at these deeper levels are mostly limited to the fruit fly Drosophila . Nevertheless, interspecies neuron-level differences have begun to emerge. Whereas these differences are small compared to the astounding degree of conservation, they reveal highly evolvable aspects of the CX circuitry, providing promising starting points for future research using comparative circuit-level analysis.


Introduction
The central complex (CX) is a charismatic, midline spanning group of neuropils that is ubiquitous across insects.Evolutionarily, it predates the emergence of insects, and equivalent structures are present across the arthropods [1,2].Unpaired, midline spanning neuropils have even been found in some Polychaete annelids [3], which, if indeed homologous to the arthropod CX, would suggest that the evolutionary roots of the CX are found over 550 million years ago, close to the bilaterian ancestor and the origins of animal brains.Given these ancient roots, the neural circuits forming this midline spanning region can be expected to be of fundamental importance.Distilling the ground pattern of the CX circuitry thus promises to offer equally fundamental insights into the core functions of animal brains.While not much is known about the origin of the CX, in insects, this brain region had reached its full complexity already with the evolution of flight [1].The overall anatomical layout of the CX has thus not undergone major changes over the last 350 million years of evolution -a period more than twice as long as mammals have inhabited this planet.
This high degree of conservation of the CX makes its substructures and neural components unambiguously identifiable across insects, thus providing an opportunity to recapitulate the evolutionary trajectories of this essential brain region across the most diverse group of animals on Earth at unprecedented resolution.The insights emerging from the beginning of this endeavor are starting to hint at how neural systems evolve to enable novel behavioral abilities in the context of existing constraints, as well as illuminating how brains adjust to the varying ecologies of a species.While most recent work on the CX has focused on the fruit fly Drosophila melanogaster (reviewed in Refs.[4][5][6][7]), rich but scattered data from across the insect phylogeny has also been accumulated.In this review, I will provide an overview over recent anatomical and functional insights, focusing on cross-species variations and emerging ground patterns.

Overall layout
The principal layout of the CX neuropils has been known for almost a century [2].In all pterygote insects, the CX consists of four main compartments: the protocerebral bridge (PB), the fan-shaped body (FB), the ellipsoid body (EB), and the noduli (Figure 1a).Of those, the noduli are the only consistently paired structure, with one nodulus located on either side of the brain's midline.Each of these neuropils comprises more or less distinct anatomical subcompartments, either arranged vertically (columns) or horizontally (layers).The orthogonal intersection of these features endows the FB and EB with a neuroarchitecture of almost crystalline clarity.Importantly, any difference at the level of neuropils are direct consequences of organizational differences at the level of neurons.During development, these neurons are born from neural stem cells that express tightly controlled temporal gradients of RNA-binding proteins and transcription factors, which specify the fate of each resulting neuron [8][9][10].Many of the involved stem cells are type II neuroblasts, which give rise to neural lineages consisting of hundreds of cells [8,11].Small changes in the temporal expression profiles of regulatory proteins in these lineages can thus lead to switches in neural identity of many neurons as well as to dramatic alterations of cell numbers.Over evolutionary time scales, these processes can be expected to cause changes in neuropil structure.While the exact nature of differences at the cellular level cannot be inferred from the neuropil layout, the presence of differences between species clearly indicates that the underlying circuits have undergone changes.Detailed comparisons of the internal CX neuropil organization between species are thus crucial as long as comprehensive single-cell data are not available.

Columns -the structural pillars of the CX
The division of the CX into repeating vertical columns is highly conserved and transcends all regions of the CX except the noduli.Functionally, these columns comprise a representation of azimuthal space around the insect, with each column representing an equal fraction of the 360° of surrounding space [6].Developmentally, this organization emerges from a series of eight type II neuroblasts, each of which gives rise to several hundred neurons that innervate single columns of the PB and project to distinct FB or EB columns (developmental lineages) [11,12].These columnar neurons form the backbone of the computational architecture of the CX and the details of the projection patterns between the PB and FB/EB directly define the implemented computational algorithms [13][14][15][16][17].This means that the anatomical structure of the CX is highly predictive of functional properties, providing an exceptional model system for exploring how neural structure underlies brain function.
Originally, a division of the PB into eight columns was described in the CX of locusts [18] and flies [19], a pattern that was assumed to be valid across insects.With the arrival of transgenic lines in Drosophila, which allowed to drive fluorescent markers in entire populations of cells of the same type, it was shown that the Drosophila PB in fact consists of nine columns [20,21].Rather than being exceptional, volume electron microscopical work in bumblebees recently confirmed a fly-like arrangement in bees [22], and careful immunohistochemical studies combined with single-cell dye injections also added a ninth column to the PB of the locust [23] and cockroach [24].The ground pattern of a nine-column architecture has thus already evolved in hemimetabolous insects, possibly as early as in dragonflies [25].As the size of the ninth column ranges from vestigial to fully developed, the columnar organization of the CX in general appears to be more flexible than previously acknowledged (Figure 1b), and neither an eight-or nine-column organization can be taken for granted in any species.While the functional relevance of varying the number of columns is still unknown, it likely impacts the circuit mechanisms of how insects represent azimuthal space.
The columns of the PB are directly matched by nine corresponding columns in the EB [13].In Drosophila, the two ends of the EB are fused to produce a ring-like structure.This fusion merges the lateral-most EB columns into a single compartment, generating a regular ring of eight equal EB segments (Figure 1d).The resulting eightfold radial symmetry directly reflects the underlying circuit topology and is also implicit in the linear default layout of this neuropil in other insects.In many species, the outermost columns are only half as wide as the remaining ones, effectively comprising two hemicolumns that complement seven regular columns [18,22,26].However, elaborations of this simple structure that cannot be easily derived from the above outline exist as well, for example, in cockroaches, where different EB layers possess different numbers of columns in a complex, interlocking arrangement (Figure 1d) [24].This variation prevents conclusions about an ancestral pattern in the EB until more data become available.
In the FB, the columnar architecture is not as obvious as in the other regions.This is because different columnar cell types follow different projection patterns, which divide the width of the FB into either 8, 9, 12, 16, or 18 columns [13,22,24].The large lateral overlap generated by these neurons thus blurs the column boundaries present in any individual cell type.

Layers -expanding complexity
Across all insects, horizontally oriented layers are present in the FB, EB, and the noduli.They are formed by CX tangential neurons, each of which laterally innervates one or several layers within CX neuropils and which provide connections with many brain regions outside the CX.While in Drosophila layers are firmly based on the connectome, descriptions of layers in other species are based on immunohistochemistry or collections of singlecell morphologies, naturally providing a coarser and less complete data basis.
In the EB of Drosophila, horizontal domains are defined by the projection fields of 11 ring neuron supertypes (EB ring neuron [ER] cells) [11,13].These GABAergic input neurons are a hallmark of the CX across insects and define the structure of the EB [27].Across species, they yield a similar organization as in Drosophila, yet with numbers of identified layers ranging from two to six and differences in their detailed spatial organization (Figure 1e) [22,26,28,29].Interestingly, in all species, the reported number of EB layers is smaller than in Drosophila.
To what extent this is due to the less comprehensive cellular description or a genuinely simpler EB anatomy remains to be shown.As different ER neuron types carry parallel inputs from the diverse sensory modalities that contribute to head direction coding [5,13,[30][31][32][33][34][35][36], the complexity of the EB layers could correspond to the richness of sensory information used by an insect for orientation and navigation.
The FB shows the largest variability in its organization across species (Figure 1c).In Drosophila, nine linearly stacked layers span the FB from anterior to posterior [13].While some species, for instance, the Monarch butterfly or dung beetles, show a similar linear stacking [28,37], other insects, such as locust and cockroaches, possess layers that wrap around each other in more complex ways [24,29] (Figure1c).In some cases, repeating domains of distinct neurochemical identity are disrupting the FB layers and add significant complexity, for example, in stick insects, crickets, or cockroaches [38].Irrespective of individual complexities, in most species, three to four main layers have been identified using immunohistochemical labeling of neurotransmitters.Interestingly, using the same methods, expression patterns of GABA and dopamine showed no obvious layering in the FB of the basal, flightless firebrats (order: Zygentoma), while serotonin expression in the same species revealed two layers (Figure 1c) [27,38], suggesting that a simple, possibly bilayered FB might represent an ancestral state.Although complex layering of the FB has since evolved in all examined insect orders, due to the differences in both shape and number, any homology of FB layers across species is difficult to assess.In line with recent evidence that the FB is involved in encoding and selecting navigational goals [39-42], the variable layout of this region could be interpreted as a correlate of species-specific behavioral repertoires (figure 2) The noduli have emerged with the evolution of flight and are thus only present in pterygote insects [1].They consistently have two principal compartments (Figure 1f), one linked to the EB (NOs) and one linked to the FB (NOm) [21,24,28,37,43,44].While the NOs is uniform with no further divisions, the NOm is larger and more complex in organization.In all insects except hymenopteran species it contains at least three layers, each linked in parallel to a layer of the FB.The NOs and each layer of the NOm receive input from dedicated tangential cells of the lateral accessory lobes, relaying different types of self-motion information to specific sets of FB or EB columnar cells, essential for many aspects of navigation and orientation [13,[15][16][17]45].Despite their most complex navigation behaviors, in hymenopteran insects, the NOm has a simpler organization, with only one compartment and no obvious layering (Figure 1f) [22].However, hymenopterans possess one more nodulus subunit, the cap region (NOc).This region lacks the input from the tangential cells of the lateral accessory lobes that defines the NOm and NOs, but instead exclusively receives input from a set of FB tangential cells.Interestingly, the NOc appears to meet many predictions from computational models of circuits required for complex vector navigation [22,46].These data suggest that the NOs and NOm perform key functions and contain the core nodulus circuits across insects, but that this circuit can be amended to add complexity that serves the behavioral needs of particular species.

Neural circuits -linking structure and function
The neural circuits of the CX are formed by the interplay of tangential input neurons and the structured and spatially confined projection patterns of the columnar neurons.These cell types are supplemented by multicolumnar neurons, that is, neurons that connect several columns within either the PB (∆7 neurons) and the FB (pontine cells).
The high degree of conservation of the PB and EB is also reflected by the neurons that form the core circuits in these areas.While ER neurons provide the input to the EB [13,30,36], two sets of columnar cell types interconnect columns of the EB and the PB in a stereotypical manner.They form two recurrent loops that are the foundation for head direction coding in the insect CX [13, [47][48][49][50].The first loop comprises a direct recurrence between the EB and PB (stabilizing loop, formed by EPG and PEG neurons), while the second loop comprises a recurrence that is offset by one column between the EB and PB (shifting loop, formed by EPG and PEN neurons).Together with the ∆7 cells of the PB, these cell types form a topological ring at the core of the head direction circuit, which produces a sinusoidal pattern of activity, the peak of which encodes the current heading of the insect.Functionally, this circuit has been comprehensively characterized only in Drosophila (reviewed in Refs.[4][5][6][7]), but electrophysiological recordings from individual components of this circuit from a wide range of insects confirm that the anatomical resemblance is indeed matched by identical functions [15,34,[51][52][53][54][55][56].
While highly conserved in principle, variability in the details of cellular architecture and in aspects of projection patterns that are dictated by the species-specific Current Opinion in Behavioral Sciences The central complex across species Heinze 5 layout of the EB and PB generates slightly different versions of the head direction circuits in each species.Differences are especially pronounced at the point of the circuit that closes the topological ring [22,57].Consistent with the regions showing the largest structural variability, the involved neurons are located at the edges of the EB and in PB columns 1 and 9 [22].That such small changes in the circuit layout can indeed lead to pronounced differences in the circuit behavior was demonstrated by computational modeling [57].This suggests that evolution can modify highly conserved circuits at key points to meet shifting demands imposed by a species ecology.
Similar variations in the projection patterns of existing circuits are found in the CX output neurons (PFL neurons).These patterns dictate how the head direction signal encoded in the PB is compared to goal encoding signals from the FB.The anatomical offset, or phase shift, between the innervation points in these two structures determines how the PFL neurons encode steering movements in response to mismatches between the animal's goal and its current heading [14,15,40,41].The magnitude of these offsets appears to be species specific [22], likely representing adaptations to each species movement patterns, agility, and habitat constraints.
Besides modifying projection patterns, interspecies differences can be also achieved at more fundamental levels, giving rise to entirely new CX circuits that do exist in some branches of insects but not in others.One example are the neural circuits associated with the noduli.Recapitulating the structural differences of the noduli, the neurons at the core of these circuits (PFN neurons) are twice as numerous in bees compared to flies [22], despite all other described columnar cells occurring in identical numbers.The majority of the added PFN cells in bees appear to generate distinct, parallel subcircuits, occupying only the NOc region and differing in their neural input.As PFN neurons are among the latest born neurons in the columnar axon bundles connecting the PB with the central body (in Drosophila, born 36-48 hours after hatching; [8]), it appears likely that these new cells have resulted from an expansion toward the end of their developmental lineages.Although the identity of the new cells is similar to the classic PFN cells and they largely duplicate existing circuits, they show distinct differences indicating that they have been recruited for novel functions, specific to bees [22].In general, adding neurons to the end of lineages could be a key mechanism for how the CX circuits can be adapted to meet new functional demands in different species without disrupting the existing ancestral functions needed by all insects.
Extrapolating from the noduli, the even larger structural variability in the FB predicts large-scale differences in its neural composition across insects.Besides the differences imposed on the FB by the axonal branches of the added PFN neurons in bees, the large and diverse set of tangential input cells of the FB are the most likely cells to differ between species.Unfortunately, data on these cells are sparse outside Drosophila [11,13].While the diversity of the reported FB tangential cells from across insects is indeed large, with only few cell types clearly identifiable as fly homologs, there is no quantitative data on neurons or cell types [29, [58][59][60].As these cells likely relay information on internal states, contextual cues, and output from the memory circuits of the mushroom body in a species-specific way [13,39,61], the extent in which they vary can be expected to reveal insights in how species establish their specific behavioral repertoire.
An additional set of neurons for which to expect variability are the intrinsic neurons of the FB, the pontine cells.They provide structured links between specific sets of layers and columns and are involved in remapping population activity that encodes various types of vectors across the width of the FB [13,16,17,39].These vectors can represent goal angles for directed behaviors, and the transformations that generate them are key processes toward producing the neural representations of a selection of possible target actions of the animal [40,41].[60].FBt, fan-shaped body tangential cells.(b) Main CX cell types are highly conserved across insects.Illustrated by PFL neurons from six different species from five insect orders.(c) Wiring principles of the head direction circuit are conserved from flies to bees.External compass cues are integrated with rotational velocity signals to generate a head-direction population code in the PB.The anatomical phase shift between PEN and EPG neurons allows rotational velocity signals to move the head direction activity peak to a new position in line with ongoing body rotations.Note that the entire circuit is repeated in a symmetrical manner on the other hemisphere.The asterisk indicates assumed functions in the bee.(d) While conserved in principle, the head direction circuit shows species-specific variations at the edges of the circuit.Note that EPG neurons in the innermost PB column project to the ipsilateral EB, while they project contralaterally in the fly.Other differences are present in the PEN projections originating in PB column nine.Bumblebee data from Ref. [22].Fly data from Ref. [13].(e) Novel circuits originate in parallel to existing circuits in the noduli (NO).Left: Schematic layout of the connections of the Drosophila NO [11,13].Note the matched layering of the NO and FB.In bees, no layering is obvious in the main NO compartment or in the corresponding part of the FB.Yet, all cell types homologous to the fly cells also exist in bumblebees.Beyond those conserved components, there are unique cell types forming distinct neuropil subregions that are suited to operate in parallel.Bold font: cell type names.Pink highlights: Novel circuit elements in bees.Cell-type acronyms are used according to Drosophila nomenclature.For alternative names, see Table 1 in Ref. [4].Neuron morphologies in B obtained from https://neuprint.janelia.org/(Drosophila, data from Ref [13]) and www.insectbraindb.org(data from Refs.
While the principle types of pontine cells are conserved in a wide range of insects [24,43,[58][59][60]62], most details about these cells remain elusive.Together with the FB tangential cells, the pontine cells are emerging as one of the most promising starting points to explore behaviorally relevant variations of how computational algorithms are implemented in the regular neural matrix of the FB.

Conclusion
Based on functional data and a complete connectome of the Drosophila CX and supported by data from other species, the insect CX plays an essential role in context-dependent selection of navigation behavior.It establishes a reliable representation of the animal's current heading, generates goal representations by combining contextual cues, memory, and internal states, and compares both angles to select an action, for example, a steering movement.All these processes are tightly linked to the anatomical details of the neural organization of the CX, with anatomical principles being highly predictive of the implemented computational algorithms.The conservation of these features across insects suggests that the general function of the almost crystalline regularity of the CX is to serve the generation of sine wave-shaped population codes that represent vectors, to transform these signals, and to select an action to adaptively change ongoing behavior.As the gross anatomy of the CX reveals numerous differences across insects, comparing the details of the underlying CX circuits provides a promising roadmap toward illuminating how the functional principles discovered in Drosophila can drive an astonishing range of species-specific behaviors, carried out by insects with vastly different body shapes, and inhabiting fundamentally different environments.

•
Kandimalla P, Omoto JJ, Hong EJ, Hartenstein V: Lineages to circuits: the developmental and evolutionary architecture of information channels into the central complex.J Comp Physiol 2023,679-720, https://doi.org/10.1007/s00359-023-01616-y.A deep analysis of the Drosophila connectome data focusing on tangential neurons of the CX.The paper organized these cells according to their developmental origin from defined neural lineages and thus provides an immensely helpful view onto this diverse and understudied set of neurons.Possibly the most important paper on the CX of the last decades.For the first time, a complete connectome of all neurons of this region in any species is presented.The detailed account of the Drosophila CX covers not only over 300 pages of synaptic resolution circuit descriptions but offers highly accessible, functional interpretations at many different levels of the circuits, generating hypotheses for years to come.[17], this paper reveals how the intricate circuits of the FB and the noduli can translate optic flow information perceived by the fly's eyes into a vector that indicates the movement direction of the fly, irrespective of whether the fly's body angle is aligned with its movement direction.These results are an outstanding example of how connectomics data, functional results, and computational models can be combined to achieve insights into computations realized in biological circuits.

•
Lyu C, Abbott LF, Maimon G: Building an allocentric travelling direction signal via vector computation.Nature 2021, 601:92-97.Together with Ref. [16], this paper reveals how the intricate circuits of the FB and the noduli can translate optic flow information perceived by The central complex across species Heinze 7 the fly's eyes into a vector that indicates the movement direction of the fly, irrespective of whether the fly's body angle is aligned with its movement direction.These results are an outstanding example of how connectomics data, functional results, and computational models can be combined to achieve insights into computations realized in biological circuits.

••
Sayre ME, Templin R, Chavez J, Kempenaers J, Heinze S: A projectome of the bumblebee central complex.Elife 2021, 10:e68911.The first attempt of volume electron microcopy -based analysis of the CX outside of Drosophila.Despite limited in resolution, the paper finds astounding degrees of conservation in the cellular composition of the CX between flies and bees but also highlights key differences in several circuits, hypothesized to be important in bee specific behaviors.

•
Timm J, Scherner M, Matschke J, Kern M, Homberg U: Tyrosine hydroxylase immunostaining in the central complex of dicondylian insects.J Comp Neurol 2021, 529:3131-3154.Taking the reader on a journey across the insect phylogeny, the paper describes the evolution of dopaminergic neurons of the CX.The results beautifully highlight the conserved nature of the core principles of this brain region but also showcase the variations present in each insect order.

••
Matheson AMM, Lanz AJ, Medina AM, Licata AM, Currier TA, Syed MH, Nagel KI: A neural circuit for wind-guided olfactory navigation.Nat Commun 2022, 13:4613.This remarkable paper demonstrates how defined circuits of the CX can drive opposite behavioral choices in response to the same stimulus, dependent on the olfactory context.The authors beautifully link functional studies, insights from connectomics and computational modeling to address how attractive odors can cause a switch from upwind to downwind orientation in flies.

•
Westeinde EA, Kellogg E, Dawson PM, Lu J, Hamburg L, Midler B, Druckmann S, Wilson RI: Transforming a head direction signal into a goal-oriented steering command.Nature 2024, 819-826, https://doi.org/10.1038/s41586-024-07039-2.Combining functional studies and computational modeling, this paper illuminates the activity patterns of the main output neurons of the Drosophila CX during goal-directed behaviors.It provides a biologically constrained model of how the neural circuits in the CX can drive downstream descending pathways to achieve effective steering movements depending on the current deviations from the goal direction, postulating a clear functional role for a previously elusive cell type.

••
Mussells-Pires P, Zhang L, Parache V, Abbott LF, Maimon G: Converting an allocentric goal into an egocentric steering signal.Nature 2024, 808-818, https://doi.org/10.1038/s41586-023-07006-3.This paper demonstrates the existence of neurons in the CX that encode navigational goals.These cells generate a population code of neural activity that is compared to the similarly encoded head direction signal to drive the activity of steering cells.Combining a range of functional, anatomical, and computational methods, this paper experimentally shows for the first time how the CX can be used to achieve goal orientated behavior.

•
Beetz MJ, Kraus C, el Jundi B: Neural representation of goal direction in the monarch butterfly brain.Nat Commun 2023, 14:5859.Using extracellular tetrode recordings from Monarch butterflies flying in a virtual reality arena, the authors present neurons located in the CX whose activity is correlated with behavioral goals, while other neurons are correlated with head direction and steering movements.These results establish that goal-related neural activity is a conserved feature of the CX.

Figure 1 Current
Figure 1

Figure 2
Figure 2 10. Hamid A, Gattuso H, Caglar AN, Pillai M, Steele T, Gonzalez A, Nagel K, Syed MH: The conserved RNA-binding protein Imp is required for the specification and function of olfactory navigation circuitry in Drosophila.Curr Biol 2024, 34:473-488.e6.
12. Williams L, Boyan GS: Building the central complex of the grasshopper Schistocerca gregaria: axons pioneering the w, x, y, z tracts project onto the primary commissural fascicle of the brain.Arthropod Struct Dev 2008, 37:129-140.Haberkern H, Franconville R, Turner-Evans DB, Takemura S, Wolff T, Noorman M, Dreher M, Dan C, Parekh R, et al.: A connectome of the Drosophila central complex reveals network motifs suitable for flexible navigation and contextdependent action selection.Elife 2021, 10:e66039.
18. Williams L: Anatomical studies of the insect central nervous system: a ground-plan of the midbrain and an introduction to the central complex in the locust, Schistocerca gregaria (Orthoptera).J Zool 1975, 176:67-86.19.Strausfeld NJ: Atlas of an Insect Brain.Springer; 1976.20.Wolff T, Iyer NA, Rubin GM: Neuroarchitecture and neuroanatomy of the Drosophila central complex: a GAL4based dissection of protocerebral bridge neurons and circuits.J Comp Neurol 2015, 523:997-1037.21.Wolff T, Rubin GM: Neuroarchitecture of the Drosophila central complex: a catalog of nodulus and asymmetrical body neurons and a revision of the protocerebral bridge catalog.J Comp Neurol 2018, 526:2585-2611.
23. Hensgen R, Dippel S, Hümmert S, Jahn S, Seyfarth J, Homberg U: Myoinhibitory peptides in the central complex of the locust Schistocerca gregaria and colocalization with locustatachykinin-related peptides.J Comp Neurol 2022, Althaus V, Heckmann J, Janning M, Seip A, Takahashi N, Grigoriev C, Kolano J, Homberg U: Neuroarchitecture of the central complex in the Madeira cockroach Rhyparobia maderae: Pontine and columnar neuronal cell types.J Comp Neurol 2023,1689-1714, https://doi.org/10.1002/cne.25535.An outstanding example of classic neuroanatomical work that uses single-cell dye injections and immunohistochemistry to describe the neural composition of the CX of a cockroach.This paper provides an essential comparison to the rich data available in flies and locusts, confirming high degrees of overall conservation of neuronal cell types but also highlighting surprising variations.25.Homberg U, Kirchner M, Kowalewski K, Pitz V, Kinoshita M, Kern M, Seyfarth J: Comparative morphology of serotoninimmunoreactive neurons innervating the central complex in the brain of dicondylian insects.J Comp Neurol 2023, 531:1482-1508.

three-dimensional atlas of the honeybee central complex, associated neuropils and peptidergic layers of the central body
This paper provides the first thorough description of a hymenopteran CX.Via detailed neurochemical mapping of a wide range of neuropeptides, it generates an average shape CX that highlights the variability of the central body across insects.

Neuroarchitecture of the central complex of the desert locust
This paper provides the first large data set on tangential neurons of the CX outside of the fruit fly, establishing a crucial stepping stone toward illuminating the interspecies variability of these most versatile but essential class of CX neurons.30.Fisher YE, Lu J, D'Alessandro I, Wilson RI: : tangential neurons.J Comp Neurol 2020, 528:906-934.