Introduction

Cichlidae comprises four subfamilies: Etroplinae (India and Madagascar), Ptychochrominae (Madagascar), Cichlinae (Neotropical region) and Pseudocrenilabrinae (Africa and a few locations in Middle East) (Fricke et al., 2023a), forming a monophyletic group well supported by morphological (Kaufman & Liem, 1982; Stiassny, 1987) and molecular data (Zardoya et al., 1996; Friedman et al., 2013). It is the third most species-rich family in the Neotropical region (Reis et al., 2003; López-Fernández et al., 2010; 2016), with 571 valid species from a total of 1749 in the family as a whole (Fricke et al., 2023a).

Cichlids have high diversification of the color patterns, feeding habits, breeding, and other behavioral aspects (Parry et al., 2005; Carleton, 2009; Gois et al., 2015; Schneider et al., 2020), and they are a well-studied group. These traits have been the focus of investigation by evolutionary biologists, and some studies have linked morphology to behavior and function (Kotrschal et al., 1998; Arbour & López-Fernández, 2014; Edmunds et al., 2016). In this sense, the encephalon’s morphology may provide insight into how the central nervous system coordinates a variety of behavioral features. Studies on Pseudocrenilabrinae, for example, have revealed a significant difference in encephalon morphology, particularly among the three large lakes (Tanganyika, Victoria and Malawi), indicating that environmental factors (e.g. turbidity, depth) and ecological traits (e.g. eating habits) are correlated with evolutionarily neuroanatomic characteristics (van Staaden et al., 1995; Huber et al., 1997).

Besides ecological and behavioral approaches, morphology has been used to generate new phylogenetic hypotheses and for taxonomic purposes. Recently, neuroanatomy has been used characters to generate a phylogenetic hypothesis of Pseudopimelodidae providing useful information in that context, such as synapomorphies and better resolution and support of phylogenetic relationships, as well as information on the evolution of those characters (Abrahão et al., 2018) and a recent neuroanatomical analysis of a cichlid species has brought up information that could be useful in phylogenetic investigations (Oliveira & Graça, 2020). However, few studies analyze fish neuroanatomic characters within the phylogenetic context, with a predominance of osteological characters in such analyses (Datovo & Vari, 2014). Therefore, studies using encephalon morphology in fish phylogenetic systematics are lacking in Neotropical cichlids, we assume.

Allied with morphological analyses, molecular analyses are commonly used in fish phylogenetic systematics. Cichlinae was the only Neotropical subfamily of Cichlidae, now divided into seven tribes: Astronotini, Chaetobranchini, Cichlasomatini, Cichlini, Geophagini, Heroini and Retroculini (Smith et al., 2008; Ilves et al., 2018). Although, some relationships remain ambiguous even in concatenated trees, such as the divergence between Geophagus, Gymnogeophagus, and the clade “Geophagus” steindachneri, Ilves et al. (2018) confirmed the genera belonging to Geophagini.

Geophagini (sensu Ilves et al., 2018) encompasses taxa that exhibit a variety of exterior morphological traits that are linked to their feeding and behavioral patterns (Arbour & López-Fernández, 2014). As a result, they serve as a starting point and model for understanding some aspects of the relationship between the gross anatomy of the central nervous system with respect to some ecological, behavioral and evolutionary characteristics in a phylogenetic framework. The main purposes of our study were: 1) To describe the major encephalon portions and cranial nerves for all Geophagini genera, as well as some Neotropical and African cichlids; and 2) to perform a constrained analysis using neuroanatomical characters from representatives of the Geophagini tribe on the Geophagini phylogeny sensu Ilves et al. (2018), in order to reconstruct the evolution of the different parts of the encephalon and present some putative characters that are synapomorphic to the tribe and subgroups.

Material and methods

Morphological analysis

Taxonomy

Representative species of all Geophagini genera used by Ilves et al. (2018) were selected to describe the encephalon gross morphology: Acarichthys heckelii (Müller & Troschel, 1849), Apistogramma borellii (Regan, 1906), Apistogramma commbrae (Regan, 1906), Apistogramma trifasciata (Eigenmann & Kennedy 1903), Biotodoma cupido (Heckel, 1840), Biotoecus opercularis (Steindachner, 1875), Crenicara punctulata (Günther 1863), Saxatilia britskii (Kullander, 1982), Dicrossus warzeli Römer et al. (2010), “Geophagus” iporangensis Haseman, 1911, “Geophagus” steindachneri Eigenmann & Hildebrand (1922), Geophagus sveni Lucinda et al. 2010, Guianacara dacrya Arbour & López-Fernández (2011), Gymnogeophagus balzanii (Perugia, 1891), Mazarunia mazarunii Kullander 1990, Mikrogeophagus ramirezi (Myers & Harry 1948), Satanoperca acuticeps (Heckel, 1840), Satanoperca setepele Ota et al. 2022, Taeniacara candidi Myers 1935, and Teleocichla proselytus Kullander 1988. The current taxonomic status of these species follows Fricke et al. (2023b). The classification follows Ilves et al. (2018). The list of examined materials are presented in Supplementary File 1, with their respective institutions and collections.

Encephalon nomenclature and preparation

Encephalons were extracted following Datovo & Vari (2014) with modifications proposed by Oliveira & Graça (2020) for Neotropical cichlids. Neuroanatomic nomenclature and abbreviations of encephalon morphological regions followed Meek & Nieuwenhuys (1998). Photographs were taken with a camera coupled with a stereomicroscope. The encephalon was immersed entirely in 70% ethanol (at a depth of ~ 1 mm over the surface tissue) to avoid possible refractive problems, according to White & Brown (2015). An ellipsoid model was used to determine the volume of each encephalic region (i.e., dorsal medulla (gustative lobes), corpus cerebelli, tectum mesencephali plus torus semicircularis, hypothalamus, hypophysis, and telencephalon). This method assumes that each region has an idealized elliptical shape (Staaden et al., 1995; Huber et al., 1997; Wagner, 2003; Lisney & Collin, 2006; Pollen et al., 2007; Ullmann et al., 2010; White & Brown 2015; Abrahão et al. 2018). Linear measurements were made based on standardized images of dorsal, lateral and ventral views, using the Opticam Microscopy OPTHD 3.7.8718 software (Opticam, 2003–2017). Measurements of length, width and height of each lobe, including even hemisphere lobes, followed Abrahão et al. (2018). Linear measurement values were converted to volume measurements (V) using the following formula: V =  \({1 \mathord{\left/ {\vphantom {1 6}} \right. \kern-0pt} 6}\pi {\text{lwh}}\) (where l = length, w = width and h = height) to calculate each lobe’s volume and total encephalon volume, according to Abrahão et al. (2018). Colored illustrations were made using the computer program GIMP, based on photographs and direct stereomicroscopic observations of selected specimens (Table 1).

Table 1 Morphometry of encephalon of Geophagini species and comparative material examined. Total volume of encephalon (EV) is in cubic millimeters; other data are expressed as percentages of EV
Full size table

Data analysis

Phylogenetic analysis

A total of 23 characters were used in the phylogenetic hypothesis of Ilves et al. (2018), and are presented in two groups, continuous and discrete (Table 2), in order to facilitate localization they are separated by encephalon structure. The character matrix is presented in Supplementary File 2.

Table 2 Continuous and discrete characters used for phylogenetic analysis, separated by encephalon structure to facilitate localization
Full size table

Two constrained phylogenetic analyses were performed using two weighting schemes: equal weighting and implied weighting with K = 3,0000. These analyses were conducted using TNT 1.5 software (Goloboff et al., 2008) from a matrix of neuroanatomic characters on the phylogenetic hypothesis proposed by Ilves et al. (2018). Constrained analysis has the purpose of visualizing the distribution of the character states in the phylogenetic tree, reconstructing the putative ancestral states at each node.

Maximum Parsimony (MP) was the optimization criterion. The phylogenetic analysis was performed using “Traditional search” on TNT. The “Random addition sequence” (RAS) algorithm was selected, with 100 replications. The swapping method used was “Tree bisection and reconnection” (TBR). The RAS and TBR technique searches for consistent analysis of up to 100 taxa. Character states were unordered. For resampling data, we used the Bootstrap index, with default replications (100). As the search was performed with 1,000 replications, the total replication number of replications was were 100 times 1000.

Results

Geophagini encephalon gross morphology

Cichlid encephalon is divided into four great divisions: Rhombencephalon, Diencephalon, Mesencephalon, and Telencephalon (Fig. 1). In all species analyzed, the encephalon is positioned above parasphenoid, prootic and basioccipital and below supraoccipital. Tectum mesencephali occupies the main volume of the encephalon (Table 1). All species have different encephalon shapes and sizes. Through subgroups within Geophagini (Fig. 2), encephalon morphology varies more in crenicichlines (Fig. 3a–d) and apistogrammines (Fig. 3e–j), presenting the most visible difference in size and shape between homologous structures within each group. Less intra-group variation can be seen among mikrogeophagines (Fig. 4a–c) and guianacarines (Fig. 4d–e) structures. In turn, geophagines (Fig. 4f–i) have the most similar encephalon. In dwarfed species, crenicaratines (Fig. 5a–b), Mikrogeophagus (Fig. 4c) and Apistogramma (Fig. 3f–g), the encephalon almost fills the skull cavity. Among Neotropical (Figs. 5e–c, 6a–b) and African cichlids (Fig. 6c–f), the encephalon also has high variation. Except by Retroculus, gustative lobe is less developed in those Neotropical and African species, as discussed below.

Fig. 1
figure 1

Illustration of encephalon gross morphology of Geophagus sveni NUP 18979, 120.6 mm standard length, in dorsal (A), lateral (B) and ventral (C) views. Scale bar = 1 mm

Full size image
Fig. 2
figure 2

Synapomorphies derived from encephalic morphology, as recovered by constrained analysis based on the topology by Ilves et al. (2018). Colored clades correspond to subgroups within Geophagini. Green, upward-directed arrowheads represent increases in continuous characters; red, downward-directed arrowheads, a decrease; blue lozenges, synapomorphic discrete character states with parallel acquisitions in other clades; blue rectangles, homoplasy-free synapomorphic discrete character states

Full size image
Fig. 3
figure 3

A, B encephalon gross morphology in dorsal, lateral and ventral view of Crenicichlines: A, Acarichythys heckelii NUP 4892, 83.85 mm standard length (SL); and B, Biotecus opercularis UFRO-I 6070, 22.46 mm SL. C, Saxatilia britiskii NUP 7953, 93.62 mm standard length (SL); and D, Teleocichla proselytus MZUSP 22017, 16.85 mm SL. E-G-C encephalon gross morphology in dorsal, lateral and ventral view of Apistogrammines: E, Apistogramma borellii NUP 4267, 31.99 mm standard length (SL); F, Apistogramma trifasciata NUP 16240, 29.5 mm SL; and G, Apistogramma commbrae NUP 16467, 26.29 mm SL. H–J, encephalon gross morphology in dorsal, lateral and ventral view of Apistogrammines: H, Satanoperca acuticeps NUP 4885, 58.88 mm standard length (SL); I, Satanoperca setepele, NUP 22313, 97.83 mm SL; and J, Taeniacara candidi UFRO-I 20710, 27.74 mm SL. 1, telencephalon; 3, lateral preglomerular nucleus; 4, lobus inferior hypothalami; 5, hypophysis; 6, saccus vasculosus; 7, tectum mesencephali; 8, corpus cerebelli; 10, lobus facialis; 11, lobus vagi; 12, medulla oblongata; 13, medulla spinalis. Scale bars = 1 mm

Full size image
Fig. 4
figure 4

A–C, encephalon gross morphology in dorsal, lateral and ventral view of Mikrogeophagines: A, Biotodoma cupido NUP 13014, 57.68 mm standard length (SL); B, “Geophagus” iporangensis NUP 3717, 80.52 mm SL; and C, Mikrogeophagus ramizeri MZUSP 96547, 24.23 mm SL. D, E, encephalon gross morphology in dorsal, lateral and ventral view of Guianacarines: D, Guianacara dacrya ROM 96095, 67.21 mm standard length (SL); and E, Mazarunia mazarunii ROM 89586, 55.37 mm SL. F–I, encephalon gross morphology in dorsal, lateral and ventral view of Geophagines: F, “Geophagus” steindachneri LBP 18635, 76.73 mm standard length (SL); and G, Geophagus sveni NUP 18976, 83.93 mm SL. H, Gymnogeophagus balzanii NUP 3035, 84.34 mm standard length (SL); and I, Gymnogeophagus meridionalis NUP 18037, 67.29 mm SL. 1, telencephalon; 3, lateral preglomerular nucleus; 4, lobus inferior hypothalami; 5, hypophysis; 6, saccus vasculosus; 7, tectum mesencephali; 8, corpus cerebelli; 9, eminentia granularis; 10, lobus facialis; 11, lobus vagi; 12, medulla oblongata; 13, medulla spinalis. Scale bars = 1 mm

Full size image
Fig. 5
figure 5

A, B, encephalon gross morphology in dorsal, lateral and ventral view of Crenicaratines: A, Crenicara punctulatum UFRO-I 12763, 27.87 mm standard length (SL); and B, Dicrossus warzelii MZUSP 25423, 37.53 mm SL. C, D, encephalon gross morphology in dorsal, lateral and ventral view of Cichlasomatini: C, Aequidens plagiozonatus NUP 194, 76.65 mm standard length (SL); and D, Cichlasoma paranaense NUP 1936, 73.41 mm SL. E, F, encephalon gross morphology in dorsal, lateral and ventral view of Cichlasomatini: A, Acaronia nassa NUP 17654, 55.23 mm standard length (SL); and B, Bujurquina vittata NUP 179, 63.03 mm SL. G–I, encephalon gross morphology in dorsal, lateral and ventral view of Chaetobranchini (A) and Heroini (B, C): A, Chaetobranchus flavescens NUP 19495, 150.25 mm standard length (SL); B, Amphilophus citrinellus NUP 14730, 82.09 mm SL; and C, Parachromis managuensis NUP 22314, 111.62 mm SL. 1, telencephalon; 3, lateral preglomerular nucleus; 4, lobus inferior hypothalami; 5, hypophysis; 6, saccus vasculosus; 7, tectum mesencephali; 8, corpus cerebelli; 9, eminentia granularis; 10, lobus facialis; 11, lobus vagi; 12, medulla oblongata; 13, medulla spinalis. Scale bars = 1 mm

Full size image
Fig. 6
figure 6

A, B, encephalon gross morphology in dorsal, lateral and ventral view of Cichlini (A) and Retroculini (B): A, Cichla kelberi NUP 2014, 74.44 mm standard length (SL); and B, Retroculus acherontos NUP 22315, 122.59 mm SL. C, D, encephalon gross morphology in dorsal, lateral and ventral view of Pseudocrenilabrinae: A, Coptodon rendalli NUP 2000, 78.0 mm standard length (SL); and B, Cynotilapia afra NUP 14722, 49.59 mm SL. E, F, encephalon gross morphology in dorsal, lateral and ventral view of Pseudocrenilabrinae: A, Hemichromis bimaculatus NUP 14724, 75.3 mm standard length (SL); and B, Oreochromis niloticus NUP2840, 113.99 mm SL. 1, telencephalon; 3, lateral preglomerular nucleus; 4, lobus inferior hypothalami; 5, hypophysis; 6, saccus vasculosus; 7, tectum mesencephali; 8, corpus cerebelli; 10, lobus facialis; 11, lobus vagi; 12, medulla oblongata; 13, medulla spinalis. Scale bars = 1 mm

Full size image

Rhombencephalon

The rhombencephalon (Fig. 1) is the most posterior portion, located just anterior to the spinal cord, posterior to the tectum mesencephali. The medulla spinalis is tubular and passes through the vertebral canal. Anteriorly lies the medulla oblongata, which is an intumescent area, larger in its anterior part, tapering posteriorly. There is no visible division between the medulla spinalis and the medulla oblongata in any of the species studied. They are located above the basioccipital. The medulla oblongata, in its anterior portion, lies posterolateral to lobus vagi. The latter is located below the posterior part of the supraoccipital process.

In all species analyzed, the gustative lobes (lobus vagi and lobus facialis) are located in the intermediodorsal rhombencephalic region. The lobus vagi is divided into two parts that are symmetrically positioned in the laterodorsal view. The two halves vary among cichlid species, with some having grooves on the dorsal surface, as in Satanoperca (Fig. 3h–i), others having a smooth surface. In the dorsal view, the two halves vary among species, forming in some a tubular slot as in Gymnogeophagus (Fig. 4h–i), in others a straight one (Figs. 3g, i, 4f–g). The lobus facialis is located anterior to the lobus vagi, ventral to the corpus cerebelli in those species that have a greater caudal prominence, as in Saxatilia and Teleocichla (Fig. 3c–d). It is not as visible in some species (Fig. 5a–b), but is discernible in others, sometimes being more prominent than the lobus vagi (Fig. 5f). Like the lobus vagi, the lobus facialis is composed of two halves on either side of the intermediodorsal zone of the rhombencephalon.

The most anterior part of rhombencephalon is the cerebellum, an unpaired lobe with an upwardly directed bulging area called the corpus cerebelli (Fig. 1). In some cichlid species, a small bulged area, the eminentia granularis, emerges at each side of the cerebellum peduncle. The shape of the corpus cerebelli varies greatly among the species studied. In dorsal view, it can have a smooth, anteroposteriorly ovate shape (Fig. 5c), or a transversely ovate shape (Fig. 4b, f, h–i), or a rounded shape (Figs. 4a, 6c–f), or irregular borders of different shapes (Fig. 3a–d). The distal part of the corpus cerebelli usually has a posterior prominence of varying degree (Figs. 3c–d, 4d–e, h), as if the distal partof the corpus cerebelli was bent posteriorly. Furthermore, the cerebellum varies in its height in lateral view, in most species its height determines the upper encephalic margin, being higher than height of the tectum mesencephali.

Eight pairs of nerves emerge from the rhombencephalon (Fig. 1): nervus trigeminus (V), nervus abducens (VI), nervus facialis (VII), nervus octavus (VIII), nervus glossopharyngeus (IX), nervus vagus (X), nervus linea lateralis anterior (Nlla) and nervus linea lateralis posterior (Nllp). Nervi V, VII and VIII emerge together from a common stem that is divided into the three nerves. They are located at the anterior mid-lateral part of the rhombencephalon, ventral to the cerebellum. Nlla fibers lie anteriorly, passing between the lobus inferior hypothalami and the lateral preglomerular nucleus, in the same way as nervi V, VII and VIII, making it sometimes difficult to separate and identify these four nerves. Nllp and nervus IX arise in sequence, ventrolateral to lobus vagi, the former being anterodorsal to the latter. Nervus X is the most porterior encephalic nerve, also arising ventrally to the lobus vagi. Unlike nervi V, VII and VIII, nervus X is composed of fibers that rise separately and unite to form a common stem. Nervus X exits the neurocranium through a foramen situated in the exoccipital. Finally, in the ventral view of the rhombencephalon, nervus VI rises, a pair of slender nerves that are difficult to see in most species and are therefore most often damaged during dissection.

Mesencephalon

The tectum mesencephali, the greater part of mesencephalon, is a paired, oval-shaped lobe in lateral view in all species, always with a smooth surface. In some species, it has irregular borders in dorsal view (Fig. 3j). It lies dorsal to the diencephalic structures and is connected to this area by a mass of nervous tissue, the encephalic truncus, and the torus semicircularis. Nervus II (ophthalmic nerve) (Fig. 1), arises from the anteroventral part of tectum mesencephali. This nerve has a chiasma opticum, which is the region where the contralateral ophthalmic nerves cross, one of which runs ventrally to the other. Nervus III (nervus oculomotoris) (Fig. 1) is a slender stem rising from the upper part of the torus semicircularis, just ventral to the tectum mesencephali. Nervus IV (Fig. 1) (nervus trochlearis) is thin and rises from the mesencephalon posterodorsally to nervus III.

Diencephalon

The diencephalon (Fig. 1) is a portion located on the ventral side of the encephalon, consisting mainly of the hypothalamus and the pituitary gland or hypophysis. In ventral view, the paired lobus inferior hypothalami is easily reconizable. In some species it is smooth (Fig. 3c), while in others it forms many bulges and depressions on the posteroventral surface (Fig. 3f, h, j), fitting the sagitta. Viewed from the side, it has a deep groove between the lateral preglomerular nucleus and the lobus inferior hypothalami (Fig. 1), where nerves V, VII, VIII and Nlla, coming from the rhombencephalon, lie together with nerves III and IV, coming from the mesencephalon. The pituitary gland is located ventrally to the hypothalamus (Fig. 1) and it varies in shape and size among different taxa. In some species, it is triangle-shaped in lateral view (Fig. 4h), while in others, it is flattened (Fig. 3a) or rounded (Fig. 3c). Posterior to the pituitary gland in ventral view, lies the saccus vasculosus, sometimes smaller than the pituitary gland (Fig. 3a), sometimes larger (Fig. 5g, i).

Telencephalon

The telencephalon is the most anterior portion of the encephalon, constituting a paired lobe (Fig. 1) anterior to the tectum mesencephali, and varies widely in shape and size both inter- and intraspecifically (Supplementary File 3). In some species it is grooved and triangle-shaped (Fig. 3b, d), while in others it is slender and elongated anteroposteriorly ((Figs. 3a, h, 4b, 5b). From its anterior border, the nervus I (nervus olfactorius) arises. In all cichlid species studied, the proximal part of this nerve forms the bulbus olfactorius (Fig. 1). In dwarf species, most of the encephalon was in direct contact with a thin surrounding neurocranial bone layer. Conversely, in larger species, there is a large space between the encephalon and the skull. Thus, in dwarf species, part of the encephalon case adheres to the encephalon, mostly to the tectum mesencephali and telencephalon, which were difficult to extract without damage.

Phylogenetic constrained analysis

The minimum score required to obtain the predetermined topology was 97.610 for the unweighted analysis and 8.0960 for the weighted analysis (Fig. 7). See Supplementary File 4 for more details on the changes in each character in the analyzed species.

Fig. 7
figure 7

Trees resulting from the constrained unweighted (A) and weighted (B) analysis. Numbers at the base of branches represent bootstrap values. One tree retained, steps = 97.610. Consistence index = 0.237; Retention index = 0.346

Full size image

The continuous character states 2[0.220–0.224] and 4[0.001–0.002] were synapomorphic to Geophagini (Fig. 2; Supplementary File 4, Figs. 3 and 5; Supplementary File 5). Character 2, tectum mesencephali volume, tended to increase in this clade, whereas character 4, volume proportion of hypophysis, tended to decrease in Geophagini. Regarding the subgroups within Geophagini, character state 2[0.238–0.241] were synapomorphic to crenicichlines, consisting of Saxatilia plus Teleocichla as the sister group to Acarichthys plus Biotoecus (Fig. 2; Supplementary File 4, Fig. 3; Supplementary File 5), and tended to increase in this group.Character state 8[0.663–0.697] was synapomorphic for apistogrammines, consisting of Satanoperca as sister of Apistogramma plus Taeniacara (Fig. 2; Supplementary File 4, Fig. 9; Supplementary File 5), as the telencephalon width/length ratio increased in this clade. The character states 1 [0.055], 3 [0.062] and 4 [0.000] were synapomorphic to guianacarines, composed of Mazarunia and Guianacara (Fig. 2; Supplementary File 4, Figs. 2, 4 and 5; Supplementary File 5). Character 1, cerebellum volume, increased in this clade, being larger than in all other Geophagini. Character 3, lobus inferior hypothalami volume, tended to decrease in this clade. Character 4, hypophysis volume, was smaller in guianacarines than in other Geophagini.

Here, we will deal first with the characters uniting mikrogeophagines plus geophagines, because it is a monophyletic group indeed. Character state 7[0.903], which deals with the size of the tectum mesencephali in the lateral view, was synapomorphic for this clade (Fig. 2; Supplementary File 4, Fig. 8; Supplementary File 5). However, this character changed several times within the same clade. Considering only the geophagines, character state 2 [0.027] was considered a synapomorphic for the clade, tending to show lower volumes of the tectum mesencephali (Fig. 2; Supplementary File 4, Fig. 3; Supplementary File 5).

The character states 0[0.006], 2[0.277], 3[0.086], 5[0.093], 7[0.783], 8[0.615], 15[1] and 16[1] were synapomorphic to Crenicaratines, composed of Dicrossus and Crenicara (Fig. 2; Supplementary File 4, Figs. 1, 3, 4, 6, 8, 9, 16 and 17; Supplementary File 5). All continuous characters showed varied changes along the cladogram. Character 0, gustative lobes volume, decreased in this clade. Character 2, tectum mesencephali volume, increased for this clade. Character 3, volume of the lobus inferior hypothalami, also increased in this clade. Character 5, telencephalon volume, also increased, but presented variation within the clade. Character 7, tectum mesencephali height/length ratio, decreased in this clade, being smaller in Crenicara. Character 8, telencephalon width/length ratio, decreases in this clade with variation, being even smaller in Dicrossus. Character 15, cerebellum height relative to tectum mesencephali, has a small size 15[1] in Biotoecus, Teleocichla and Apistogrammines. Character 16, distinctiveness of the eminentia granularis, was parallel 16 [1] in Mikrogeophagus, Biotoecus and the apistogrammines, because it is not visibly distinct.

The continuous character 0 (Fig. 2; Supplementary File 4, Fig. 1; see states for each group/species in Supplementary File 5), gustative lobes volume, underwent several changes along the cladogram. The clades that presented higher volumes of the gustative lobes within Geophagini were mikrogeophagines (except Mikrogeophagus ramirezi) plus geophagines, Satanoperca (even greater in Satanoperca sp.), Biotoecus and Apistogramma borellii. We observed convergent values in the outgroup to Coptodon, Retroculus and Chaetobranchus. However, low gustative lobes volume was observed in crenicichlines (even lower in Saxatilia), and convergence was observed in the outgroup to Acaronia plus Bujurquina, Parachromis, Cichla and Cynotilapia.

Discussion

Encephalon gross morphology

Cichlid encephalon examined herein presents a typical division of a Teleost encephalon, designated from the posterior to anterior margin in rhombencephalon, diencephalon, mesencephalon and telencephalon (see Meek & Nieuwenhuys, 1998). Sutures between neurocranial bones are hardly distinguishable in most cichlids studied herein. Nonetheless, some differences in the relative positions of encephalon parts and skeletal components could be detected when compared to other fish families that had their encephalon gross morphology studied, such as Pseudopimelodidae (Abrahão & Pupo, 2014; Abrahão et al., 2018) and Bathydraconidae (Eastman & Lannoo, 2003). Differently from Pseudopimelodidae, our findings demonstrate that in cichlids the medulla spinalis and the medulla oblongata are located dorsally to the basioccipital instead of the parasphenoid. Rhombencephalon is divided into four main regions, which innervate and receive inputs from viscera: ventral (somatomotor), intermedioventral (visceromotor), intermediodorsal (viscerosensory) and dorsal zone (somatosensory) (Meek & Nieuwenhuys, 1998). Although Staaden et al. (1995) & Huber et al. (1997) used the term "dorsal medulla" to describe African cichlid rhombencephalon dorsal structures, which is not incorrect due to its location, this portion has specializations such as facial and vagal lobes in the intermediodorsal zone (Meek & Nieuwenhuys, 1998).

Our data indicates that lobus vagi is well developed in some cichlid species (Table 1; also see Figs. 1–2 in Oliveira & Graça, 2020), but it was shown to be larger in cyprinids, due to the specialized pharyngeal palatal organ in these fishes (Meek & Nieuwenhuys, 1998), a chemosensitive and muscular structure used to select food particles among gravel (Sibbing, 1984; Finger, 1988). Our findings imply that in most of the cichlids, lobus vagi is paired, also covering a great portion in the dorsal rhombencephalon as Meek & Nieuwenhuys, 1998 show for cyprinids. Lobus facialis receives sensory input from the facial cranial nerve (VII) and is connected to taste buds in the mouth cavity and external taste buds localized on the lips and body surface. In cichlids, although the lobus facialis is discernible in most of the species, it is less developed than in other teleosts, such as ictalurids, because they have an elaborated taste system (Meek & Nieuwenhuys, 1998).

The cerebellum is the most anterior portion of the rhombencephalon, with a role in processing somatosensory input of lateral line afferent fibers in its posterior part. Our research shows that in cichlids, it is as large as found in most of the other teleosts (Meek & Nieuwenhuys, 1998), but in others, it is greater than in cichlids studied herein, like in Pseudopimelodidae species (Abrahão et al., 2018). According to Meek & Nieuwenhuys (1998) the cerebellum comprises three major divisions, a vestibulolateral zone posteriorly, the corpus cerebelli in surface and a valvula cerebelli. In cichlid species, the eminentia granularis and corpus cerebelli are easily visible. The vestibulolateral zone is composed of the caudal lobe and eminentia granularis. This one is a visible mass of granular cells, involved in lateral line sensory reception. In contrast to cichlids examined in our study, catfish have two sections (Tong & Finger, 1983). The corpus cerebelli is a tubular lobe (Meek & Nieuwenhuys, 1998) which may be directed rostrally, as seen in Pseudopimelodidae (Abrahão et al., 2018), or caudally, as seen in Bathydraconidae species (Eastman & Lannoo, 2003), and in cichlids in previously (van Staaden et al., 1995) but in the cichlids analyzed herein, it sometimes appears caudally, sometimes upwardly. Meek (1992) found no functional significance in shape disparity. It is important to know that the afferent centers were found in anterior and posterior to the peduncle of the cerebellum (Meek & Nieuwenhuys, 1998) and efferent centers project to several areas of the encephalon (Wullimann & Northcutt, 1988). Finally, the valvula cerebelli is a portion located anteriorly to the cerebellum (Meek & Nieuwenhuys, 1998).

In cichlids studied herein, nerves are placed in the ventral rhombencephalic, intermedioventral rhomboencephalic, and intermediodorsal rhomboencephalic zones, as in other teleosteans. Meek & Nieuwenhuys (1998) found the abducens motor nucleus (VI), whose fibers innervated the rectus externus extra-ocular muscle in the ventral rhombencephalic zone. In the intermedioventral rhomboencephalic zone, visceromotor nuclei of nerves V, VII, IX and X have been found innervating striated peripharyngeal muscles of branchial arches (Meijer, 1975). Sensory nuclei of nerves VII, IX and X are distributed in the intermediodorsal rhomboencephalic zone (Meek & Nieuwenhuys, 1998). The system of somatosensory region in the dorsal zone is composed of trigeminal sensory nuclei and processes general information of the head, such as touch, temperature and proprioception (Meek & Nieuwenhuys, 1998). According to them, there are acoustic, vestibular, mechanical and electrosensory receptors with cells called “hair cells” in this region, which codify environmental information and are innervated by nervi VIII, Nlla and Nllp. Although lateral line receptors are utilized to detect water movements in cichlids, as in other teleosts, there are some groups with electroreception specialized lateral line systems, such as the Gymnotidae and others (Meek & Nieuwenhuys, 1998).

Mesencephalon is involved in the motor (ventromedially located tegumentum) and sensory (torus semicircularis) functions, although there is no rigorous separation of motor and sensory functions, as there is in rhombencephalon (Meek & Nieuwenhuys, 1998). The ophthalmic nerve (II) innervates the eyeball in its medial face. Some authors have observed in other teleosts that the oculomotor nucleus has cholinergic neurons (Rhodes et al., 1986; Ekstrom, 1987; Brantley & Bass, 1988). The nervi III and IV, along with abducens (VI, from rhombencephalon), also innervate other extraocular muscles (Meek & Nieuwenhuys, 1998). Motoneurons of trochlear nerve (IV) supply the contralateral obliquus superior eye muscle (Luiten & Dijkstra-de Vlieger, 1978; Graf & McGurk, 1985; Szabo et al., 1987). The Tectum mesencephali, commonly known as the optic tectum, is paired and occupies a considerable portion of the mesencephalon in cichlids analyzed herein. The intertectal commissure connects their halves (Meek & Nieuwenhuys, 1998). Despite its name, the optic tectum does not process only visual inputs, but also integrates visual signs with other sensory information to provide coordination of goal-directed movements (Meek & Nieuwenhuys, 1998). Hence, it is the primary sensorimotor integration center of the teleost central nervous system (Meek & Nieuwenhuys, 1998). In cichlids, the tectum mesencephali was shown to be the most developed region in the encephalon (Kotrschal et al. 1998).

The diencephalon is commonly divided into the epithalamus, dorsal thalamus, ventral thalamus and hypothalamus zone in cichlids examined herein and the most visible structure is hypothalamus, situated beneath the thalamus, being the ventral diencephalic area. A pituitary stalk connects it to the hypophysis and contains nerve fibers that have a part in neuroendocrine regulation processes (Meek & Nieuwenhuys, 1998). The lobus inferior hypothalami is composed by two bulbs visible in ventral view (Meek & Nieuwenhuys, 1998) and saccus vasculosus is positioned beneath, posteriorly to the pituitary gland. Jansen (1973) showed that the saccus vasculosus is vascularized and contains cerebrospinal fluid in contact with bipolar neurons and cells called coronet.

The telencephalon in teleosts is likewise a paired lobe, located anteriorly to other encephalon sections. Compared to other vertebrate groups, the teleost telencephalon is everted rather than evaginated (Nieuwenhuys, 1962, 1963), so as in cichlids examined herein. This part includes the olfactory bulb which may be sessile, as in cichlids (herein analyzed and see Kotrschal et al., 1998) and Perciformes (Bathydraconidae in Eastman & Lannoo, 2003), or stalked as in Pseudopimelodidae (Abrahão & Shibatta, 2015) which is connected to the telencephalon by secondary fibers through the tractus olfactorius (Meek & Nieuwenhuys, 1998). Nonetheless, the forebrain (telencephalon) presented the main interspecific variation in size and shape in African cichlids (van Staaden et al., 1995), as well as in our study.

Phylogenetic implications in Geophagini

The constrained analysis allowed us to understand the neuroanatomic characters in the phylogeny proposed by Ilves et al. (2018), demonstrating possible adaptive convergences and how these characters are arranged in the evolutionary history of Geophagini. There are two synapomorphies for Geophagini clade by Ilves et al. (2018) characters 2 (tectum mesencephali volume percentage) and 4 (hypophysis volume percentage). Previous works have also proposed a phylogenetic hypothesis on the cichlid intrarelationships with morphological and molecular data. However, topologies vary among studies. For instance, Kullander (1998) recovered monophyly of Geophaginae (essentially equivalent to Geophagini of Ilves et al., 2018) using morphological data, but differently from a subsequent study by López-Fernández et al. (2005), Crenicichla (we adopt Crenicichla britskii belonging to Saxatilia Varella et al. 2023) was placed as the sister group of Cichla. In the current study, Cichla and Saxatilia present a convergence in the character 0, represented by a decrease in the volume of gustative lobes, showing that morphological (osteological and neuroanatomical) characters tend to cluster the two genera.

Within Geophagini, character state 2 [0.238–0.241] also is synapomorphic to crenicichlines sensu Ilves et al. (2018), a clade not recovered as monophyletic in Kullander (1998) & López-Fernández et al. (2005). In the total evidence analysis shown in the latter, Biotoecus and Saxatilia were sister groups, while Acarichthys was the sister to Guianacara.

Character state 8 [0.663–0.697] are synapomorphic to the apistogrammines clade sensu Ilves et al. (2018). In Kullander, these genera were placed within Geophagini (mikrogeophagines and geophagines sensu Ilves et al., 2018), but not as a sister group. Apistogrammines were recovered as monophyletic within the Satanoperca clade by López-Fernández’s (2005) total evidence analysis. When the authors analyzed only morphological data, they discovered Taeniacara and Apistogramma with all other small-bodied taxa (Crenicara, Dicrossus, Mikrogeophagus and Biotoecus), and Satanoperca was found in a polytomy formed by Geophagus, Gymnogeophagus and a clade that includes Biotodoma as the sister of Acarichthys plus Guianacara. Guianacarines monophyly sensu Ilves et al. (2018) has three synapomorphies, characters 1, 3, and 4. Previous morphological studies (Kullander, 1998; López-Fernández et al., 2005), did not analyze Mazarunia, and both recovered Guianacara in Acarichthyini, as the sister of Acarichthys.

Character state 7 [0.903], height/length ratio of the tectum mesencephali are synapomorphic to mikrogeophagines and geophagines sensu Ilves et al. (2018), while character state 2 [0.027] is synapomorphic for geophagines. Kullander’s (1998) tribe Geophagini included both mikrogeophagines and geophagines sensu Ilves et al. (2018), as well as Satanoperca and Apistogramma. Mikrogeophagines, composed of Mikrogeophagus, “Geophagus” iporangensis and Biotodoma, do not form a monophyletic clade since they lack the geophagines species Gymnogeophagus, “Geophagus” steindachneri and Geophagus sveni (Fig. 2 and see Ilves et al., 2018). Therefore, both did not have their relationships satisfactorily resolved by Ilves et al. (2018).

Many characters state (0[0.006], 2[0.277], 3[0.086], 5[0.093], 7[0.783], 8[0.615], 15[1] and 16[1]) are synapomorphic to the clade formed by crenicaratines sensu Ilves et al. (2018). Crenicara and Dicrossus were recovered as sister genera in previous phylogenies (Kullander, 1998; López-Fernández et al., 2005). When just morphological data was analyzed in López-Fernández et al (2005), these genera were recovered within a clade composed of the small-bodied species, as previously explained.

Character 0, gustative lobes, was important among Geophagini. Gustative lobes tend to increase in Gymnogeophagus balzanii, Gymnogeophagus meridionalis, “Geophagus” iporangensis, Geophagus sveni, “Geophagus” steindachneri, Satanoperca sp. and Satanoperca acuticeps. It is possible to observe convergences in this structure among these taxa and Retroculus acherontos. Although Geophagus, Gymnogeophagus, Satanoperca and Retroculus do not form a monophyletic group (Ilves et al., 2018), the first three genera were placed together in other morphological phylogenies (Kullander 1998; López-Fernández et al. 2005), and in other studies, Retroculus was included in Geophagini (Cichocki, 1976; Landim, 2007).

Convergence in Geophagini: an ecological approach

Using the characters proposed herein in the constrained analysis onto the topology of Ilves et al. (2018) is enlightening within an ecological and evolutionary perspective. It is essential to understand the major functions of the encephalon in this context and link them to any behavioral and ecological characteristics of cichlid species.

The primary function of the nervous system is to coordinate the interactions between the organism and the environment via efferent (motor) and afferent (sensorial) systems, of which the latter appears to be more adaptable (Meek & Nieuwenhuys, 1998). Fishes have a variety of sensory organs, namely olfactory, gustatory, visual, acoustic, vestibular and somatosensory, which receive signals from the environment and are connected with the central nervous system by spinal and cranial nerves (Meek & Nieuwenhuys, 1998). According to the authors, a higher development of a specific region of the encephalon seems to correlate with the presence of specialized organs. For example, groups that are capable of detecting electric current in the environment, such as Mormyridae and Gymnotidae, present a better-developed lateral line system specialized for electroreception than fish that only use the lateral line as a detector of water mechanosensory stimulus (Meek & Nieuwenhuys, 1998), such as cichlids that lack a well-developed lateral-line lobe.

However, even in the absence of specialized organs, highly developed encephalon parts imply that the senses they process are crucial to an individual species’ life habits. For example, the rainbow trout Oncorhynchus mykiss (Walbaum, 1792) has a heightened sense of vision and consequently a high development of the tectum mesencephali (Meek & Nieuwenhuys, 1998). In the current study, the vision center has the largest volumes among encephalon parts in all species studied, and the high volume of the tectum mesencephali is synapomorphic to Geophagini clade. This agrees with Kotrschal et al. (1998), once the vision center, commanded by tectum mesencephali, is the most well-developed structure in cichlid encephalon in general.

Once our findings of Cichlid species have established a well-developed vision center, it is critical to comprehend the significance of this center to cichlids in general. First, the encephalon processes vision in all vertebrates via light absorption in the photoreceptors of the retina (cones and rods) (Meek & Nieuwenhuys, 1998). In cichlids, there are seven different genes of cone opsins and one of rod opsin (Carleton et al., 2016), which are expressed in different sets across species (Carleton et al., 2016; Schneider et al., 2020) and along ontogeny (Carleton, 2009), allowing for the absorption of a wide range of light spectra (Carleton et al., 2016; Schneider et al., 2020). In this sense, the evolution of visual sensitivity in cichlids appears to have occurred through altering the expression of these genes, and appears to have been selected mostly for the properties of light incident on the habitat, which vary depending on the water turbidity and depth (Schneider et al., 2020). Thus, this complex vision system and a well-developed mesencephalon in Cichlid species found herein, could be linked to visual orientation of cichlids in relation to recognition of conspecific coloration (Schneider et al., 2020), brood guarding behavior and territory defense (Gois et al., 2015), foraging (Parry et al., 2005) and habitat selection (Carleton, 2009). On the other hand, cichlids are considered microsmatic, i.e., they have a poorly developed olfactory system, therefore their bulbus olfactorius is less developed than in other groups, which is consistent with their small olfactory organs (Ridet & Bauchot, 1990).

About feeding habits, some Neotropical cichlid species were discovered to be mostly as benthivorous or piscivorous fishes (López-Fernández et al., 2012). It is difficult to extrapolate feeding habits for all Geophagini genera, because, in that study, there were Crenicichla/Saxatilia species which may consume both prey types. Despite this, most of Geophagini have a benthic-feeding behavior, with the exception of Crenicichla/Saxatilia, using their protractible jaws to sift the substrate. Geophagini sifters have modifications of the pharyngeal apparatus (weak pharyngeal jaws and the presence of epibranchial 1 lobe), indicating a link between morphology and this feeding behavior (López-Fernández et al., 2012).

Thus, occasionally, it is possible to identify a correlation between encephalon shape and feeding adaptations (Oliveira & Graça, 2020). Our study detected proportionally greater gustative lobes (lobus facialis and lobus vagi) in sediment-sifting cichlids, such as most mikrogeophagines and geophagines, Satanoperca and Retroculus, suggesting that taste is important in sorting edible from non-edible particles during winnowing (Fig. 2; Supplementary File 4, Fig. 1; although Acarichthys, Biotodoma, Guianacara and Mikrogeophagus present no such increase). Well-developed lobus vagi were also found in a previous study conducted with Geophagus sveni, which showed a possible relation with adaptations to winnowing (Oliveira & Graça, 2020), and in Cyprinids, which have a specialized pharyngeal palatal organ (Meek & Nieuwenhuys, 1998), and in species that select particles among gravel (Sibbing, 1984; Finger, 1988).

However, in piscivores (Acaronia, Cichla, Saxatilia and Parachromis) and in some dwarf species (crenicaratines and Taeniacara) the volume of gustative lobes decreased. Perhaps taste is a less important sense in visual predators, which is corroborated by the fact that piscivores have a larger tectum mesencephali (van Staaden et al., 1995; Huber et al., 1997). Bujurquina and Guianacara, which are neither piscivorous nor dwarf species but have a small gustative lobe volume, and Biotoecus, a dwarf species with a large gustative lobe volume, are exceptions to this rule.

Encephalon gross morphology varied among Geophagini and other cichlid species. Encephalon variation may be the reflection of both phylogenetic distance and environmental conditions experienced by fishes throughout time (van Staaden et al., 1995; Huber et al., 1997; Abrahão et al., 2018). Thus, feeding specialization itself does not entirely explain interspecific variation in encephalon morphology, once microhabitat use is also associated with increasing or decreasing in some cichlid encephalon structures between species (van Staaden et al., 1995; Huber et al., 1997). A study conducted in the African lakes, Victoria, Tanganyika and Malawi, revealed that turbidity, depth, and substrate complexity could predict variability, but not causality, of differences in encephalon structures in different cichlids species (Huber et al. 1997). For example, shallow rock environments were associated with small gustatory lobes in the encephalon (lobus vagi and facialis).

Conclusion

Geophagini encephalon gross morphology varies interspecifically and the differences provided[ putative characters. In contrast, using neuroanatomic characters on a previous phylogenetic hypothesis we were able to uncover probable adaptive convergences. Four genera analyzed herein (Acaronia, Cichla, Saxatilia and Parachromis) that converged to piscivorous habits developed independently smaller gustative lobes in comparison with other taxa investigated, with the exception of some dwarf species (crenicaratines and Taeniacara). Conversely, most of the specialized winnowers, namely geophagines, “Geophagus” iporangensis, Retroculus and Satanoperca developed large gustative lobes, apparently to facilitate sorting of edible and non-edible particles during sifting. Although it is not quite certain if the ancestor of all Geophagini was a specialized sediment-sifter or not, within the tribe this behavior is correlated to large gustative lobes, with few exceptions, and may represent adaptive convergence. Concurrently, Retroculus certainly has developed sifting habits independently from Geophagini, and also possesses well-developed gustative lobes. Furthermore, the tectum mesencephali, which is the largest structure in proportion to the encephalon in all cichlids analyzed herein, is even larger in Geophagini. An increase in its volume is one of the characters that supports the tribe in the constrained analysis. Using encephalon gross morphology to access information under the phylogenetic constrained approach reveals unique characteristics that emphasize the evolution of this structure in Geophagini. In this context, neuroanatomic characters are informative in phylogenetic and ecological studies and they could be used allied to a larger set of other morphological structures, despite their homoplastic nature.