Convergent mosaic brain evolution is associated with the evolution of novel electrosensory systems in teleost fishes

Larger animals tend to have larger brains and smaller animals tend to have smaller ones. However, some species do not fit the pattern that would be expected based on their body size. This variation between species can also apply to individual brain regions. This may be due to evolutionary forces shaping the brain when favouring particular behaviours. However, it is difficult to directly link changes in species behaviour and variations in brain structure.

One way to understand the impact of evolutionary adaptations is to study species that have developed new behaviours and compare them to related ones that lack such a behaviour. An opportunity to do this lies in the ability of several species of fish to produce and sense electric fields in water. While this system is not found in most fish, it has evolved multiple times independently in distantly-related lineages. Schumacher and Carlson examined whether differences in the size of brains and individual regions between species were associated with the evolution of electric field generation and sensing.

Micro-computed tomography, or μCT, scans of the brains of multiple fish species revealed that the species that can produce electricity – also known as ‘electrogenic’ species’ – have more similar brain structures to each other than to their close relatives that lack this ability. The brain regions involved in producing and detecting electrical charges were larger in these electrogenic fish. This similarity was apparent despite variations in how total brain size has evolved with body size across species.

These results demonstrate how evolutionary forces acting on particular behaviours can lead to predictable changes in brain structure. Understanding how and why brains evolve will allow researchers to better predict how species’ brains and behaviours may adapt as human activities alter their environments.

Brains are composed of multiple regions that vary widely in size across vertebrates and are associated with particular functions and behaviors (Striedter, 2005; Striedter and Northcutt, 2020). Much of the variation in brain region sizes is attributed to the allometric scaling of each region with total brain size (concerted evolution), which may result from conservation and constraint in developmental neurogenesis (Finlay and Darlington, 1995; Striedter, 2005). Seemingly disproportionately enlarged regions can have larger allometric slopes by extending the timing of neurogenesis for late developing brain regions such as the cortex and cerebellum (Finlay and Darlington, 1995). However, changes in brain region sizes independent of total brain size, or mosaic shifts, have also been observed in several taxa and are hypothesized to reflect selection on traits associated with those regions (Barton and Harvey, 2000; Striedter, 2005). Mosaic shifts in fine-scale brain regions and circuits are well accepted and have been linked to changes in behavior (Carlson et al., 2011; Vélez et al., 2017; Vélez et al., 2019; Gutiérrez-Ibáñez et al., 2014; Moore and DeVoogd, 2017; DeCasien and Higham, 2019; Krebs, 1990), but the scale at which selection can act to alter brain region sizes remains unclear. There may potentially be more flexibility for mosaic changes in nuclei or circuits dedicated to specific functions compared to major brain regions that serve multiple functions and may be subject to greater developmental and phylogenetic constraints.

Most studies looking at the scaling of major brain regions instead find evidence of concerted evolution (Finlay and Darlington, 1995; Striedter, 2005; Yopak et al., 2010). There is some evidence of mosaic evolution at these scales (Hoops et al., 2017; Sukhum et al., 2018), but the drivers and selective pressures necessary for mosaic evolution to overcome constraints, whether developmental or due to functional interconnectivity, remain unclear. Further, this is difficult to test without repeated evolution of the same phenotypes.

Mosaic brain evolution of major brain regions is hypothesized to occur more frequently at larger taxonomic scales and alongside behavioral innovations that open new niches since mosaic shifts are more likely to contribute to major differences in brain function (Striedter, 2005). In dragon lizards, mosaic brain evolution is associated with ecomorph (species similar in morphology and behavior inhabiting the same ecological niche; Hoops et al., 2017), but as many different behavioral and sensory changes occur alongside ecomorph development, it is difficult to identify specific selective pressures favoring the observed ecomorph brain structure. Weakly electric fishes are excellent for testing whether mosaic brain evolution occurs with behavioral novelty: these fishes evolved behaviorally novel active electrosensory systems with several neural innovations, which likely resulted in strong selection for electrosensory processing capabilities (Carlson and Arnegard, 2011). Further, multiple lineages independently evolved similar electrosensory systems (Crampton, 2019).

Previous studies found that African weakly electric fishes (Mormyroidea) evolved extremely large brains along with mosaic increases in the sizes of the cerebellum and hindbrain relative to other non-electric osteoglossiforms (Sukhum et al., 2018; Sukhum et al., 2016). These mosaic increases occurred alongside the evolution of an active electrosensory system (electrogenesis + electroreception), but since this is only a single lineage, it is impossible to determine whether these mosaic shifts are associated with the evolution of this electrosensory system or with other phenotypes that differentiate mormyroids from their closest living relatives. Further, one non-electric osteoglossiform, Xenomystus nigri, is electroreceptive, and there is no evidence that it has experienced mosaic brain evolution compared to other non-electric osteoglossiforms.

Electrogenesis and electroreception have evolved multiple times, but sparingly, across vertebrates—at least six times and at least two times, respectively, within ray-finned fishes (Actinopterygii; Figure 1). However, the degree to which electrosensory systems have evolved, with respect to their component parts and how they are utilized, varies across these different independent origins. For example, some lineages produce strong electrical discharges while others produce weak electrical discharges, some lineages have evolved multiple types of electroreceptors while others have evolved just one, and the usage and developmental origins of electrogenesis differ across lineages (Crampton, 2019). Here, we investigated another lineage of fishes, otophysans, which includes taxa that evolved similar electrosensory systems independent of mormyroids, but to varying degrees with respect to their components and utilization, to determine whether these mosaic shifts are found repeatedly alongside the evolution of active electrosensing. Although osteoglossiform and otophysan lineages have convergently evolved similar electrosensory systems (Crampton, 2019), there are some distinctive differences in the degree of electrosensory usage, electroreceptor type, and electrical discharge type that could indicate differential selective pressures on the brains of these species. These differences allowed us to assess how multiple aspects of electrosensory systems relate to mosaic brain evolution.

Figure 1

Download asset Open asset

To investigate how electrosensory systems relate to structural brain variation, we combined published osteoglossiform data for electrogenic mormyroids, electroreceptive Xenomystus, and non-electrosensory outgroup species (Sukhum et al., 2018) with otophysan data for two additional electrogenic lineages (Gymnotiformes and Synodontis Siluriformes), electroreceptive (but not electrogenic) siluriforms, and non-electrosensory Characiformes and Cypriniformes (outgroup otophysans). Using micro-computed tomography (μCT) scans, we measured total brain and brain region volumes for 15 electrogenic, 3 electroreceptive, and 4 non-electrosensory otophysan species. Combined with the published osteoglossiform data, this yielded a dataset of 32 species (Figure 2, Figure 2—video 1). We measured the volumes of seven distinct brain regions (olfactory bulbs, OB; telencephalon, TEL; hindbrain, HB; optic tectum, OT; torus semicircularis, TS; cerebellum, CB; and rest of brain, RoB) to determine patterns of major brain region scaling across taxa. Rest of brain includes thalamus, hypothalamus, and additional midbrain structures excluding optic tectum and torus semicircularis (see ‘Brain region delimitation’).

Figure 2 with 2 supplements see all

Download asset Open asset

Electrosensory system evolution is not associated with shifts in total brain size scaling

A recent study found a steeper brain–body allometric relationship for osteoglossiforms compared to other actinopterygians, but not for seven other focal ray-finned fish orders, which may, in part, be driven by the highly speciose mormyroids (Tsuboi, 2021). To determine if the evolution of an electrosensory system is associated with extreme encephalization or shifts in brain–body allometric relationships, we combined our data with published brain and body mass data across ray-finned fishes (Tsuboi, 2021; Tsuboi et al., 2018), which resulted in a combined dataset of 870 species across 46 orders, with phylogenetic data from a previously assembled time-calibrated phylogeny (Rabosky et al., 2018). We used Bayesian reversible-jump bivariate multiregime Ornstein–Uhlenbeck modeling (OUrjMCMC; Uyeda et al., 2017) to identify shifts in both y-intercept and slope of brain–body allometric relationships. This approach allows shifts to be identified without assuming their location a priori.

We identified eight allometric shifts across actinopterygians, three in lineages with at least three descendants: osteoglossiforms, and two shifts within percomorphs (Figure 3A). The first (hereafter referred to as percomorph grade A) included Lophiiformes, Tetraodontiforms, Acanthuriformes, some descendants from Scorpaeniformes, and some descendants from Perciformes. The second (hereafter referred to as percomorph grade B) included Gasterosteiformes, some descendants from Scorpaeniformes, and some descendants from Perciformes. Shifts were also detected for Gymnothorax meleagris, Synodontis multipunctatus, Arothron nigropunctatus, and a mormyroid lineage containing Isichthys henryi + Brienomyrus brachyistius. All other taxa best fit the ancestral allometry. However, it is worthwhile to note that additional shifts may be present within actinopterygians. There are more than 20,000 known actinopterygians while we only have data for 870 species. Further, multiple single species were identified as having increased brain–body allometries; however, it remains unclear if these single species have particularly enlarged brains compared to their closest relatives or if a more speciose lineage with enlarged brains would be identified with additional sampling of close relatives.

Figure 3

Download asset Open asset

Figure 3—source data 1

Ornstein–Uhlenbeck modeling (OUrjMCMC) and phylogenetic generalized least squares (PGLS) fitted brain–body allometries for each grade.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig3-data1-v2.xlsx

Download elife-74159-fig3-data1-v2.xlsx

Figure 3—source data 2

Results (p-values) of phylogenetically corrected pairwise post-hoc tests with a Bonferroni correction for an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of brain size against body for each identified grade.

Differences in intercept are below the diagonal and differences in slope are above the diagonal. Significant differences are shown in bold.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig3-data2-v2.xlsx

Download elife-74159-fig3-data2-v2.xlsx

Figure 3—source data 3

Model selection results for phylogenetic generalized least squares (PGLS) relationships systematically collapsing each putative grade to its ancestral grade.

The best-fit model is shown in bold.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig3-data3-v2.xlsx

Download elife-74159-fig3-data3-v2.xlsx

Figure 3—source data 4

Ornstein–Uhlenbeck modeling (OUrjMCMC) estimated marginal likelihoods and model selection results for shifts in brain–body allometries associated with electrosensory phenotypes.

The best-fit model is shown in bold.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig3-data4-v2.xlsx

Download elife-74159-fig3-data4-v2.xlsx

For shifts in lineages (i.e., for each grade) containing at least three descendants, we tested the putative grades in a phylogenetic generalized least squares (PGLS) framework for shifts in both slope and y-intercept using a phylogenetically corrected analysis of covariance (ANCOVA) and phylogenetically corrected pairwise post-hoc testing with a Bonferroni correction. We found that all of the grades differ significantly in slope (p<0.05) while only osteoglossiforms relative to the ancestral grade and relative to percomorph grades A and B differ in y-intercept (p<0.05; Figure 3B, Figure 3—source data 2). However, only osteoglossiforms had an increased slope relative to the ancestral grade while the slope for percomorph grade A decreased relative to the ancestral grade with a further decrease in percomorph grade B. To confirm shifts across all putative grades, we fit the following PGLS models: all the identified shifts, each grade collapsed in turn to its ancestral grade, and a single allometric relationship across all taxa. Collapsing S. petricola to the ancestral grade was the best-fit model (ΔAIC > 3) with the model containing all putative shifts as second best (ΔAIC > 6; Figure 3—source data 3).

To explicitly address whether shifts in the allometric relationship are associated with the evolution of electrosensory phenotypes, we ran OUrjMCMC models with fixed shifts for lineages that evolved ampullary electroreceptors, lineages that evolved tuberous electroreceptors, and lineages that evolved electrogenesis in addition to the following null hypotheses: a fixed shift at the branch leading to osteoglossiforms as found previously (Tsuboi, 2021), only shifts in the allometric relationship for intercept but not slope, and a single allometric relationship across all taxa. We found that the model with eight shifts provided the best fit to the data (2ln(BF)>28; Figure 3—source data 4). In addition, when forcing a global slope across all taxa, we found no shifts in intercept with a posterior probability >0.1 despite all parameters having estimated sample sizes >1000. Taken together, these results suggest that fishes can evolve an active electrosensory system without evolving a brain as large as that of mormyroids.

Electrogenic species have similar structural brain variation

To determine how brain structure varies in association with electrosensory phenotype, we used regional measurements for all taxa and ran a phylogenetically corrected principal components analysis (pPCA). Considering the phylogenetic relationships among otophysans are still debated (Crampton, 2019; Hughes et al., 2018; Rabosky et al., 2013), our goal was only to account for relatedness to the best of our ability, not to propose a resolved phylogeny. We found that electrogenic species cluster distinctly from both electroreceptive and non-electrosensory species, which overlap considerably (Figure 4A). All electroreceptive lineages (mormyroids, Xenomystus, gymnotiforms, and siluriforms) have evolved ampullary electroreceptors, which detect relatively low-frequency electrical information, while only gymnotiforms and mormyroids have evolved additional tuberous electroreceptors that broaden the frequency range of detectable signals (Crampton, 2019). We find that electrogenic species with both electroreceptor types cluster distinctly from electrogenic fishes with only ampullary electroreceptors (Synodontis siluriforms). The first principal component (PC1) explained 92.02% of the variation in brain region volumes and is strongly correlated with total brain volume (ρ = –0.99, p<10^–16). PC2 explained 4.81% of the total variation, which is 60.3% of the variation in region volumes not explained by total brain volume. Whereas all of the brain regions loaded in the same direction for PC1, cerebellum, torus semicircularis, and hindbrain loaded negatively on PC2 while the remaining regions (telencephalon, rest of brain, optic tectum, and olfactory bulbs) loaded positively. This suggests that concerted brain evolution explains the most variation in region volumes as seen in PC1, but that mosaic brain evolution could be contributing to the observed variation in brain region volumes as seen in PC2.

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 4—source data 1

Phylogenetically corrected principal components analysis (pPCA) loadings for each non-normalized brain region.

OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig4-data1-v2.xlsx

Download elife-74159-fig4-data1-v2.xlsx

Figure 4—source data 2

Phylogenetically corrected principal components analysis (pPCA) phylogenetic generalized least squares (PGLS) model selection results for non-normalized data (N = 31).

PC1 was correlated with total brain volume, thus total brain volume was removed from all PC1 models. The best-fit model for each PC is shown in bold.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig4-data2-v2.xlsx

Download elife-74159-fig4-data2-v2.xlsx

Figure 4—source data 3

Phylogenetic flexible discriminant analysis (pFDA) results table showing the regional coefficients for each discriminant axis and the means for each electrosensory phenotype group along each axis: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig4-data3-v2.xlsx

Download elife-74159-fig4-data3-v2.xlsx

To ensure that differences in size ranges between regions were not biasing the results, we z-score-normalized the region volumes, reran the pPCA, and found nearly identical results (Figure 4—figure supplement 1). As there are multiple approaches to multivariate clustering analyses, we also ran a phylogenetic flexible discriminant analysis (pFDA) to investigate whether convergence in brain structure across electrosensory phenotypes persists irrespective of method. Again, we find distinct clustering between electrogenic species with both electroreceptor types, electrogenic species with only ampullary electroreceptors, and non-electric species, indicating three distinct electrosensory-associated cerebrotypes (Clark et al., 2001). The resulting three discriminant functions (i.e., number of electrosensory phenotype groups – 1) of the observed data accurately predicted electrosensory phenotype from the residuals of brain characters for all 32 species, further suggesting a relationship between electrosensory phenotypes and the observed brain region volume variation (Figure 4—figure supplement 2, Figure 4—source data 3). As all three clustering approaches demonstrated the same conclusions, we proceeded with the non-normalized pPCA.

To assess the relative importance of electrosensory phenotypes in explaining the axes of brain structural variation (PCs 1–4), we ran candidate models that considered body mass, total brain volume, presence or absence of electrogenesis, and electroreceptor type (tuberous and ampullary vs. only ampullary vs. none). Models that only consider allometric scaling with body mass and total brain volume would be consistent with concerted evolution while models that also consider either one or both electrosensory phenotypes would be in line with mosaic brain evolution since more than just allometric scaling explains the observed variation in brain region volumes. Since PC1 strongly correlates with brain size, we removed total brain volume as a variable from all PC1 models.

We found that the model that considers the electrogenesis phenotype better explained PC1, but is statistically indistinguishable from the concerted model (Figure 4—source data 2), which further supports the role of concerted evolution in determining the sizes of individual brain regions. The model that considers electroreceptor type better explained PC2 (Figure 4—source data 2), which supports our hypothesis that the electrosensory system is related to mosaic evolutionary changes in brain region scaling. PC3 explains 1.68% of the total variation (21.1% of the variation not explained by total brain volume) and largely reflects the variation between olfactory bulbs and optic tectum with no separation between electrosensory phenotypes (Figure 4B). The model that considers electroreceptor type better predicts PC3 but is statistically indistinguishable from the concerted model, suggesting that this axis of brain variation likely evolved concertedly with brain size (Figure 4—source data 2). PC4 explains 0.54% of the total variation (6.8% of the variation not explained by total brain volume), largely reflects the variation between torus semicircularis and cerebellum, and is better predicted by the model that considers both electrosensory phenotypes (Figure 4B, Figure 4—source data 2). Taken together, these results highlight the importance of both concerted and mosaic brain evolution in producing the observed variation in brain region volumes.

Mosaic shifts in electrogenic species relative to non-electric species

To directly test for mosaic shifts associated with electrosensory phenotypes, we fit PGLS regressions for each brain region against total brain size for species with each electrosensory phenotype and used an ANCOVA to test for significant differences in both slope and y-intercept of the PGLS relationships for each brain region. Considering the debate on how best to assess patterns of brain region scaling (Finlay and Darlington, 1995; Yopak et al., 2010), we also determined PGLS relationships for each brain region against total brain volume minus the focal brain region (remaining brain volume).

Since electroreceptive-only species and non-electrosensory species overlapped considerably in the pPCA, we combined all non-electric taxa and performed a phylogenetically corrected ANCOVA with phylogenetically corrected pairwise post-hoc testing to compare electrogenic taxa with both tuberous and ampullary electroreceptors (mormyroids and gymnotiforms), electrogenic taxa with only ampullary electroreceptors (Synodontis siluriforms), and non-electric taxa (Figure 5, Figure 5—figure supplement 1, Figure 5—source data 1, Figure 5—source data 2, Figure 5—source data 3, Figure 5—source data 4, Figure 5—source data 5, Figure 5—source data 6). We found a significant increase in y-intercept in cerebellum for all electrogenic species relative to non-electric species (p<0.05). For torus semicircularis, we found a significant increase in y-intercept in electrogenic + ampullary-only taxa relative to non-electric taxa (p<0.05) and a further increase in y-intercept in electrogenic + ampullary + tuberous taxa (p<10^–4). We found a significant increase in y-intercept in hindbrain for electrogenic + ampullary + tuberous taxa relative to non-electric taxa (p<0.01) with electrogenic + ampullary-only taxa being intermediate (p>0.05). These results were the same for both regressions against total brain volume (Figure 5, Figure 5—source data 1, Figure 5—source data 2) and regressions against remaining brain volume (Figure 5—figure supplement 1, Figure 5—source data 4, Figure 5—source data 5, Figure 5—source data 6).

Figure 5 with 2 supplements see all

Download asset Open asset

Figure 5—source data 1

Results of an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain volume for each electrosensory phenotype: slope p, intercept p, and Pagel’s lambda (λ).

Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig5-data1-v2.xlsx

Download elife-74159-fig5-data1-v2.xlsx

Figure 5—source data 2

Results (p-values) of pairwise post-hoc tests with a Bonferroni correction for an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain volume for each electrosensory phenotype: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

Differences in intercept are below the diagonal, and differences in slope are above the diagonal for each brain region. Significant differences are shown in bold. Post-hoc tests were not performed when the ANCOVA revealed no significant differences (indicated by ‘n.s.’). OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig5-data2-v2.xlsx

Download elife-74159-fig5-data2-v2.xlsx

Figure 5—source data 3

Estimated effect sizes (Cohen’s d) of each contrast for an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain volume for each electrosensory phenotype: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

Post-hoc tests were not performed when the ANCOVA revealed no significant differences (indicated by ‘n.s.’). OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig5-data3-v2.xlsx

Download elife-74159-fig5-data3-v2.xlsx

Figure 5—source data 4

Results of an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain–region volume for each electrosensory phenotype: slope p, intercept p, and Pagel’s lambda (λ).

Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig5-data4-v2.xlsx

Download elife-74159-fig5-data4-v2.xlsx

Figure 5—source data 5

Results (p-values) of pairwise post-hoc tests with a Bonferroni correction for an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain–region volume for each electrosensory phenotype: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

Differences in intercept are below the diagonal, and differences in slope are above the diagonal for each brain region. Significant differences are shown in bold. Post-hoc tests were not performed when the ANCOVA revealed no significant differences (indicated by ‘n.s.’). OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig5-data5-v2.xlsx

Download elife-74159-fig5-data5-v2.xlsx

Figure 5—source data 6

Estimated effect sizes (Cohen’s d) of each contrast for an analysis of covariance (ANCOVA) comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain–region volume for each electrosensory phenotype: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

Post-hoc tests were not performed when the ANCOVA revealed no significant differences (indicated by ‘n.s.’). OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig5-data6-v2.xlsx

Download elife-74159-fig5-data6-v2.xlsx

There were significant decreases in y-intercept in olfactory bulbs and rest of brain for electrogenic + ampullary + tuberous taxa relative to both electrogenic + ampullary-only and non-electric taxa when regressed against total brain volume (p_OB<0.05, p_RoB<0.05). When regressed against remaining brain volume, we found similar results for olfactory bulbs (p<0.05), but we only found a significant decrease in y-intercept for electrogenic + ampullary + tuberous taxa relative to non-electric taxa in rest of brain (p<10^–4). For optic tectum, we found a significant decrease in y-intercept in electrogenic + ampullary + tuberous taxa relative to non-electric taxa (p<10^–4) with electrogenic + ampullary-only taxa as intermediate (p>0.05) when regressed against total brain volume. When regressed against remaining brain volume, we found a significant decrease in optic tectum y-intercept for electrogenic + ampullary + tuberous taxa relative to both electrogenic + ampullary-only and non-electric taxa (p<0.05). There were no significant differences in y-intercept for telencephalon when regressed against either total brain volume or remaining brain volume (p>0.05).

These results show that there are similar mosaic shifts in lineages that independently evolved electrogenesis regardless of analysis method. We also found a significant increase in slope in cerebellum and a significant decrease in slope in optic tectum for electrogenic + ampullary-only species relative to non-electric species when regressed against total brain volume (p_CB<0.05, p_OT<0.05), which may be related to the reduced species sampling and brain size distribution of electrogenic + ampullary-only species (N = 4) relative to non-electric (N=10) and electrogenic + ampullary + tuberous species (N = 18). When regressed against remaining brain volume, we found a significant decrease in slope in cerebellum for non-electric taxa relative to both electrogenic + ampullary + tuberous taxa and electrogenic + ampullary-only taxa, which may be related to the obvious non-electric outlier Pantodon buchholzi. We did not find a significant difference in optic tectum slope when regressed against remaining brain volume (p>0.1). All other slope comparisons were nonsignificant for both analyses (p>0.1).

To assess how different brain regions covary with respect to electrosensory phenotype, we fit PGLS regressions for each region-by-region comparison and performed phylogenetically corrected ANCOVAs with phylogenetically corrected pairwise post-hoc testing for the same electrosensory phenotype groups (Figure 6). For electrogenic + ampullary + tuberous taxa relative to non-electric taxa, we found significant differences in y-intercept for olfactory bulbs against telencephalon (p<0.05); rest of brain against cerebellum (p<10^–3); telencephalon against olfactory bulbs (p<0.01), hindbrain (p<0.01), and cerebellum (p<0.01); hindbrain against optic tectum (p<10^–4) and telencephalon (p<10^–3); torus against hindbrain (p<0.01); and cerebellum against olfactory bulbs (p<10^–3), optic tectum (p<10^–4), and rest of brain (p<10^–4). For electrogenic + ampullary + tuberous taxa relative to both electrogenic + ampullary-only and non-electric taxa, we found significant differences in y-intercept for olfactory bulbs against hindbrain (p<0.01), torus (p<0.01), and cerebellum (p<0.05); optic tectum against hindbrain (p<0.01), torus (p<10^–3), and cerebellum (p<0.05); rest of brain against hindbrain (p<0.01) and torus (p<10^–4); telencephalon against torus (p<0.01); hindbrain against olfactory bulbs (p<0.05) and rest of brain (p<0.01); and torus against olfactory bulbs (p<0.01), optic tectum (p<10^–3), rest of brain (p<10^–4), and telencephalon (p<0.05). For both electrogenic + ampullary + tuberous taxa and electrogenic + ampullary-only relative to non-electric taxa, we found significant differences in y-intercept for cerebellum against telencephalon (p<0.05) and in slope for rest of brain against optic tectum (p<0.05); telencephalon against optic tectum (p<0.05); hindbrain against optic tectum (p<0.01); torus against optic tectum (p<0.01); and cerebellum against optic tectum (p<0.01) and telencephalon (p<0.05). All remaining comparisons for y-intercept and slope were nonsignificant (p>0.05).

Figure 6

Download asset Open asset

Figure 6—source data 1

Matrix of region-by-region analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships for each electrosensory phenotype: slope p (S), intercept p (I), and Pagel’s lambda (L).

Columns indicate the region on the y-axis, and rows indicate the region on the x-axis. Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig6-data1-v2.xlsx

Download elife-74159-fig6-data1-v2.xlsx

Figure 6—source data 2

Matrix of region-by-region analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships for each electrosensory phenotype: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

Columns indicate the region on the y-axis, and rows indicate the region on the x-axis. Numbers in each matrix cell are the results (p-values) of pairwise post-hoc tests with a Bonferroni correction. Differences in intercept are below the diagonal, and differences in slope are above the diagonal for each inset. Significant differences are shown in bold. Post-hoc tests were not performed when the ANCOVA revealed no significant differences (indicated by ‘n.s.’). OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig6-data2-v2.xlsx

Download elife-74159-fig6-data2-v2.xlsx

Figure 6—source data 3

Matrix of region-by-region analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships for each electrosensory phenotype: electrogenic + tuberous and ampullary electroreceptors (E + A + T), electrogenic + only ampullary electroreceptors (E + A), and non-electric (Not E).

Columns indicate the region on the y-axis, and rows indicate the region on the x-axis. Numbers in each matrix cell are the estimated effect sizes (Cohen’s d) of each contrast. Post-hoc tests were not performed when the ANCOVA revealed no significant differences (indicated by ‘n.s.’). OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig6-data3-v2.xlsx

Download elife-74159-fig6-data3-v2.xlsx

Taken together, these results suggest that there are two covarying brain structure groupings, with olfactory bulbs, optic tectum, telencephalon, and rest of brain falling into one group and hindbrain, torus, and cerebellum falling into the other group as all comparisons of group 1 regions against group 2 regions have significant mosaic shifts between electrosensory phenotypes. However, this pattern is strongest for tuberous receptor taxa relative to non-electric taxa. These findings mirror the pPCA results, in which these taxa were separated primarily by PC2, for which olfactory bulbs, optic tectum, telencephalon, and rest of brain loaded positively, whereas hindbrain, torus, and cerebellum all loaded negatively (Figure 4). Differences in slope were only found in regions where there are obvious outliers in the non-electric fishes (cerebellum: P. buchholzi; optic tectum: Microglanis iheringi), suggesting that these species might have mosaic shifts relative to other non-electric osteoglossiforms and otophysans, respectively, but additional studies are needed to investigate this possibility further. We did not find any slope differences in the olfactory bulb comparisons, which also have an obvious outlier (M. iheringi); however, there is more variability in olfactory bulbs compared to the other brain regions, which may decrease the influence of outliers.

Lineage-specific mosaic shifts within electrosensory phenotypes

To determine if there are lineage-specific differences within electrosensory phenotypes, we performed a phylogenetically corrected ANCOVA of each brain region between mormyroids and gymnotiforms (Figure 7A, Figure 7—figure supplement 1A, Figure 7—source data 1, Figure 7—source data 2) and between non-electric osteoglossiforms and otophysans (Figure 7, Figure 7—figure supplement 1B, Figure 7—source data 1, Figure 7—source data 2). When regressed against total brain volume, we found a shallower slope and smaller y-intercept for mormyroids in olfactory bulbs (p_slope<10^–4, p_intercept<0.05) and rest of brain (p_slope<10^–2, p_intercept<0.05), shallower slope for mormyroids in torus semicircularis (p<10^–2), and steeper slope for mormyroids in cerebellum (p<0.05). We found similar results when regressed against remaining brain volume, but we did not find a significant difference in cerebellum slope between mormyroids and gymnotiforms (p>0.05). All remaining comparisons for y-intercept and slope were nonsignificant for both analyses (p>0.05). For the region-by-region comparisons, we did not find the same groupings as across electrosensory phenotypes (Figure 7—figure supplement 2, Figure 7—source data 3), further suggesting that the observed differences in brain structure are related to evolving electrosensory systems. We did find significant differences in y-intercept and/or slope in most olfactory bulb comparisons (p<0.05), except olfactory bulbs and torus comparisons and optic tectum against olfactory bulbs (p>0.05). We also found significant differences in y-intercept for the telencephalon and torus versus rest of brain comparisons (p<0.05) and in slope for some comparisons between optic tectum, hindbrain, torus, and cerebellum (p<0.05). These differences are likely contributing to the secondary clustering between mormyroids and gymnotiforms in the pPCA (Figure 4A) and suggest that there are more nuanced distinctions in brain structure between these two lineages. For non-electric fishes, we found that osteoglossiforms have a significantly larger y-intercept for telencephalon (p_TEL<10^–2), but no differences in either slope or y-intercept in the other brain regions (p>0.1) when regressed against both total brain volume (Figure 7B, Figure 7—source data 1) and remaining brain volume (Figure 7—figure supplement 1B, Figure 7—source data 2). We found that osteoglossiforms also have significantly larger y-intercepts for telencephalon against all other regions (p<0.05) and for torus against cerebellum (p<0.05), but no differences in either slope or y-intercept in the other comparisons (p>0.05; Figure 7—figure supplement 3, Figure 7—source data 4).

Figure 7 with 4 supplements see all

Download asset Open asset

Figure 7—source data 1

Lineage analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain volume for each lineage: slope p, intercept p, Pagel’s lambda (λ), Cohen’s d.

Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig7-data1-v2.xlsx

Download elife-74159-fig7-data1-v2.xlsx

Figure 7—source data 2

Lineage analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships of region volume against total brain–region volume for each lineage: slope p, intercept p, Pagel’s lambda (λ), Cohen’s d.

Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig7-data2-v2.xlsx

Download elife-74159-fig7-data2-v2.xlsx

Figure 7—source data 3

Matrix of region-by-region analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships for mormyroids vs. gymnotiforms: slope p (S), intercept p (I), Pagel’s lambda (L), and Cohen’s d (D).

Columns indicate the region on the y-axis, and rows indicate the region on the x-axis. Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig7-data3-v2.xlsx

Download elife-74159-fig7-data3-v2.xlsx

Figure 7—source data 4

Matrix of region-by-region analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships for non-electric osteoglossiforms vs. non-electric otophysans: slope p (S), intercept p (I), Pagel’s lambda (L), and Cohen’s d (D).

Columns indicate the region on the y-axis, and rows indicate the region on the x-axis. Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig7-data4-v2.xlsx

Download elife-74159-fig7-data4-v2.xlsx

Figure 7—source data 5

Matrix of region-by-region analysis of covariance (ANCOVA) results comparing phylogenetic generalized least squares (PGLS) relationships for wave vs. pulse gymnotiforms: slope p (S), intercept p (I), Pagel’s lambda (L), and Cohen’s d (D).

Columns indicate the region on the y-axis, and rows indicate the region on the x-axis. Significant differences are shown in bold. OB, olfactory bulbs; TEL, telencephalon; HB, hindbrain; OT, optic tectum; TS, torus semicircularis; CB, cerebellum; RoB, rest of brain.

https://cdn.elifesciences.org/articles/74159/elife-74159-fig7-data5-v2.xlsx

Download elife-74159-fig7-data5-v2.xlsx

Some gymnotiforms produce continuous electrical discharges with variable frequency (wave-type), while others produce discrete electric discharges separated by variable periods of stasis (pulse-type). To assess whether there are mosaic shifts associated with the evolution of electrical discharge type, we performed a phylogenetically corrected ANCOVA between wave-type and pulse-type gymnotiforms (Figure 7C, Figure 7—figure supplement 1C, Figure 7—source data 1, Figure 7—source data 2). We found significant increases in slope for wave gymnotiforms in olfactory bulbs against hindbrain and cerebellum and in telencephalon against hindbrain and cerebellum (p<0.05; Figure 7—figure supplement 4, Figure 7—source data 5). However, we found no differences in either y-intercept or slope of any region against total brain volume between wave and pulse gymnotiforms and only a significant increase in slope in telencephalon against remaining brain volume for wave gymnotiforms (p<0.05), suggesting that the transition between discharge types did not relate to brain structure at this scale. It was not possible to directly test this in mormyroids since there is only one wave-type mormyroid species (Gymnarchus niloticus) that is sister to the family of all pulse-type mormyroids. However, the electrosensory system of the wave-type mormyroid is similar to that of gymnotiforms (Bell and Maler, 2005), and the wave mormyroid tends to be more similar to gymnotiforms in both brain region residuals (Figure 7A) and the pPCA (Figure 4). After excluding the wave mormyroid, we found the same regional differences associated with all mormyroids, along with a significant increase in the cerebellum y-intercept in pulse mormyroids relative to gymnotiforms (p<0.05), suggesting that the extraordinarily enlarged cerebellum of some mormyroid species is the result of both a steeper allometric relationship and a mosaic shift in the ancestor of pulse mormyroids.

We used osteoglossiform and otophysan fishes to test whether mosaic shifts in brain region volumes are associated with the convergent evolution of behaviorally novel active electrosensory systems. Although the mosaic shifts previously found in mormyroids were hypothesized to be related to the evolution of electrogenesis, it remained unknown if these patterns would be found in other electrogenic lineages. The brain scaling patterns of electrogenic versus non-electric osteoglossiforms and otophysans are strikingly similar despite the considerable phylogenetic distance between them, revealing distinct electrosensory associated cerebrotypes (Figure 8). Further, these electrosensory cerebrotypes converged independent of the variable brain–body allometric relationships of osteoglossiforms and otophysans (Figure 3).

Figure 8

Download asset Open asset

Gymnotiforms have the most similar electrosensory system to mormyroids in terms of electrosensory structures, neural processing, and behavioral usage (Hopkins, 1995). In both of these lineages, we found mosaic increases in cerebellum, hindbrain, and torus semicircularis compared to their non-electric relatives, which suggests that these evolutionary shifts in brain structure largely reflect their coevolution with electrogenesis and tuberous electroreceptor phenotypes. First-order electrosensory processing takes place in the electrosensory lateral line lobe (ELL) of the hindbrain, which projects to the torus for further processing of electrocommunication and electrolocation signals (Baker et al., 2013; Bell and Maler, 2005; Metzen and Chacron, 2021). Electrosensory information projects from the torus both directly and indirectly to areas of the cerebellum with large, reciprocal connections between cerebellum, torus, and the ELL of both mormyroids and gymnotiforms. Multiple areas of the mormyroid cerebellum show responses to electrosensory stimuli (Russell and Bell, 1978). More generally, cerebellum is also known to be involved in predicting the sensory consequences of motor movements and subsequent error detection, nonmotor functions, and learning (Hull, 2020; Popa and Ebner, 2019; Strick et al., 2009). The overwhelming evidence of feedback circuits between initial electrosensory processing regions and cerebellum in both mormyroids and gymnotiforms suggests that the cerebellum may also be involved in processing electrosensory information in addition to electromotor control (Bell and Maler, 2005; Paulin, 1993). The hindbrain is also involved in generating electromotor output (Caputi et al., 2005), which suggests that both electrosensory processing and electromotor control are related to evolutionary changes in relative region sizes.

Electrogenic Synodontis, which only have ampullary electroreceptors, have significant mosaic increases in cerebellum and torus semicircularis relative to non-electric fishes. Synodontis electrical discharges are likely detectable by their ampullary electroreceptors (Hagedorn et al., 1990; Zupanc and Bullock, 2005) and involved in electrocommunication (Albert and Crampton, 2006; Boyle et al., 2014), which could relate to enlargement of these regions relative to electroreceptive but non-electric species. We found that Synodontis were intermediate between electrogenic fishes with tuberous receptors and non-electric fishes in torus and hindbrain, although the difference in hindbrain was not significant despite the hindbrain’s role in generating electromotor output in synodontids (Hagedorn et al., 1990; Kéver et al., 2020). The hindbrain also contains the facial and vagal lobes, which are enlarged in siluriforms and cypriniforms (Striedter, 2005), potentially obscuring a relationship with electrogenesis, and further highlighting the complexity of gross-scale brain region evolution.

The addition of tuberous electroreceptors increases the range of detectable signals and total electrosensory input to the brain relative to only ampullary electroreceptors (Crampton, 2019), and more subregions of the torus and hindbrain are devoted to electrosensory processing in tuberous electroreceptor species (Bell and Maler, 2005). The enlarged torus in Synodontis may also relate to acoustic communication. Some Synodontis species produce swim bladder sounds (Boyle et al., 2014), and the torus is involved in auditory processing (Fay and Edds-Walton, 2008). This is also true for mormyroids, as they are known to have specialized hearing, and some species produce acoustic signals (Ladich and Winkler, 2017). However, gymnotiforms are not known to produce acoustic signals, and all otophysans possess accessory structures that improve hearing (Ladich and Winkler, 2017). Additionally, M. iheringi and several Corydoras species are also known to utilize acoustic communication (Kaatz et al., 2010), but we did not find an enlarged torus in these taxa.

Interestingly, all electrogenic fishes have a significant mosaic increase in cerebellum regardless of electroreceptor type. Relative brain region sizes of electroreceptive only species are largely consistent with non-electrosensory species, which suggests that the evolution of electrogenesis strongly relates to structural brain composition. However, two chondrichthyan lineages have independently evolved electrogenesis, Torpediniformes and Rajidae (Bennett, 1971), with no evidence of mosaic shifts in cerebellar size in these taxa (Mull et al., 2020; Yopak et al., 2010; Figure 5—figure supplement 2). It is possible this is because chondrichthyan cerebellums are already massively enlarged compared to their closest relatives, agnathans, which arguably lack a proper cerebellum (Striedter and Northcutt, 2020). Unfortunately, the brain structure of other independently evolved electrogenic lineages, such as Malapterurus catfishes, Astroscopus stargazers, and Uranoscopus stargazers, remain unknown. The addition of these lineages could better elucidate the relationship between the evolution of electrogenesis and structural brain composition, especially as stargazers are the only electrogenic fishes that lack electroreceptors of any type.

Alternatively, it is possible the enlarged cerebellum in Synodontis is unrelated to the evolution of electrogenesis. Like Synodontis, Rajidae produce sporadic discharges likely used for electrocommunication, while mormyroids and gymnotiforms produce near continuous discharges (Bennett, 1971; Crampton, 2019). Considering we find a further enlargement in the cerebellum of pulse mormyroids which produce electrical discharges at varying and complex timing intervals, it is possible that the specific usage and complexities of electrical discharges may relate to the degree of cerebellar enlargement. We do not find any differences in cerebellar volume of wave and pulse gymnotiforms, but pulse gymnotiforms produce discharges at regular intervals like wave-type fishes and unlike the irregularly discharging pulse mormyroids (Caputi et al., 2005). Further research on the usage of sporadic electrical discharges and the related electrosensory pathways are needed to better elucidate the relationship between electrogenesis and cerebellar enlargement. In particular, further research on Synodontis species is greatly warranted as electrogenesis in these species appears evolutionarily labile (i.e., some species produce continuous discharges, some produce sporadic discharges, and one species produced no discharges under experimental conditions), yet only 13 of 200+ species have been investigated for electrogenesis, and very little is known about electrosensory processing in these species (Baron et al., 1994; Baron, 2002; Boyle et al., 2014; Hagedorn et al., 1990). We also found a shift in the brain–body allometry of S. multipunctatus; however, we are unable to determine whether this is associated with any specific electrosensory phenotypes without additional research into more synodontid species. Our results only further highlight the value of these species in understanding the relationship between electrogenesis and brain evolution.

Electrosensory information also projects to the telencephalon (Bell and Maler, 2005), but we do not find a mosaic increase in any electrosensory taxa relative to non-electric taxa. This is likely because the telencephalon is involved in higher-order sensory integration across many different sensory modalities (Striedter and Northcutt, 2020) while sensory systems mostly remain segregated in the lower-order processing of hindbrain and torus (Meek and Nieuwenhuys, 1998). Surprisingly, we find that non-electric osteoglossiforms have a mosaic increase in telencephalon relative to non-electric otophysans. The telencephalon of osteoglossiforms is highly differentiated, more so than other teleosts (Meek and Nieuwenhuys, 1998), which suggests that osteoglossiforms evolved enlarged telencephalons relative to Elopomorpha, followed by a relative decrease in mormyroids alongside the evolution of electrogenesis and mosaic increases in other brain regions. Further research is needed to determine why osteoglossiforms have enlarged telencephalons.

Optic tectum is also involved in sensorimotor integration, particularly with respect to the electrosensory, lateral line, and visual systems, in addition to being the primary target of visual input to the brain (Meek and Nieuwenhuys, 1998). Yet we find that tuberous receptor fishes have a mosaic decrease in optic tectum relative to non-electric fishes while electrogenic + ampullary-only fishes are intermediate. Additionally, tuberous receptor fishes have a mosaic decrease in olfactory bulbs relative to fishes lacking tuberous receptors, which together could indicate decreased reliance on visual and olfactory systems. Gymnotiforms are thought to have poor vision (Takiyama et al., 2015), and different mormyroid lineages specialize to varying degrees in visual versus electrocommunication systems (Stevens et al., 2013). We do find a mosaic decrease in rest of brain in electrogenic taxa with tuberous electroreceptors relative to fishes lacking tuberous receptors and a mosaic decrease in rest of brain in mormyroids relative to gymnotiforms that could reflect a trade-off in one or more of the subregions that comprise the rest of brain, but since we combined these subregions, we are unable to speculate about their evolution.

Here, we assume that increased brain region volume corresponds to increased neuron number, which increases processing power and reflects behavioral changes that natural selection can act upon in one lineage with respect to another. However, increases in absolute regional volume can result from increased neuron number, neuron size, glia number, or any combination thereof (Herculano-Houzel, 2012; Marhounová et al., 2019). Previous studies found that neuron number and size tend to scale with brain size, but this scaling varies across both lineages and brain regions, with some regions having more neurons than expected given total brain size (Barton, 2012; Herculano-Houzel et al., 2014; Kverková, 2022; Marhounová et al., 2019). These findings suggest that volume measures could over- or underrepresent neuron number, and future studies should investigate the neuronal composition of these regions to better investigate how absolute region volumes and processing capabilities have changed.

Changes in relative regional volumes can result from any of the aforementioned mechanisms in the focal region, but they can also result from changes in other regions that cause a shift in the relative proportion of any given region. In particular, we want to emphasize that all of our identified mosaic shifts are relative to total brain size and to other lineages. An increase in the size of one region necessitates a decrease in one or more of the other brain regions since all regional measurements are relative to total brain size. For example, an increase in the number of neurons in the cerebellum of mormyroids would lead to an increase in absolute cerebellar volume. Even if the neuron number, size, and glial content of the telencephalon remained constant in all osteoglossiforms and thus no changes in absolute telencephalon volume occurred, mormyroid telencephalon size would have necessarily decreased relative to total brain size since total brain size has increased with the addition of more cerebellar neurons. Thus, the relative proportion of telencephalon may have decreased in mormyroids relative to other osteoglossiforms due solely to increases in the absolute sizes of cerebellum, hindbrain, and torus. However, it is quite difficult to distinguish this scenario from the possibility that absolute telencephalon size has also changed in some manner, especially when other factors such as body size differ as well. Further, the patterns of regional scaling in one lineage are relative to others, and may reflect differences in relative investment across lineages.

Due to the complex evolutionary histories of different brain regions, subregions, and total brain size, we are unable to make any claims regarding the evolution of absolute region sizes at this phylogenetic scale. Even commonly used ‘reference’ brain regions such as our ‘rest of brain’ have subregions like the lateral hypothalamic regions and preglomerular complex that are known to vary tremendously in size across teleosts, and the preglomerular complex is extensive in mormyroids in particular (Wullimann, 2020; Wullimann and Northcutt, 1990). Regardless of the specific differences in absolute region volumes, we still find overwhelming evidence of relative differences in the volume of individual regions that, although the mechanism is currently unknown, are likely still reflecting biologically meaningful differences in relative neural processing investment across these fishes.

Evidence of mosaic evolution at smaller subregional, nuclei, and circuit levels is readily available (Carlson et al., 2011; Vélez et al., 2017; Gutiérrez-Ibáñez et al., 2014; Moore and DeVoogd, 2017; DeCasien and Higham, 2019; Krebs, 1990), but rarely have mosaic shifts been observed at the level of major brain regions and even less so alongside convergence in behavioral evolution. Our findings support the hypothesis that mosaic brain evolution occurs more readily under substantial selective pressure that favors a greater expansion of a particular brain region than allometric scaling can accommodate without incurring a substantial energetic cost (Striedter and Northcutt, 2020). Although not necessarily exclusive to electrosensing or sensory systems in general, we suspect the evolution of this novel sensory system provided such a strong selective pressure. Indeed, our clearest example of gross-scale mosaic evolution occurs alongside evolutionary changes in both the sensory and motor system, and we suggest looking towards other instances of behavioral novelty for additional potential examples of gross-scale mosaic brain evolution.

However, the evolutionary pressures on brain structure are multifaceted. As tasks require further integration of different areas of the brain, selective pressure favoring one trait could instead lead to coordinated selection favoring the expansion of all regions (Avin et al., 2021; Striedter and Northcutt, 2020). Indeed, we find that multiple regions covary with electrosensory phenotypes (Figure 6). Not only are there evolutionary forces acting on brain region scaling to consider, but also the evolutionary forces acting upon total brain size. After constraint on total brain size was added in a bare-bones model of brain region evolutionary dynamics, the probability of mosaic evolution increased under most tested scenarios (Avin et al., 2021). The decoupling of brain–body allometries has been reported in birds and mammals, whereas actinopterygians are consistently found to have more constrained allometric relationships, whether through explicit constraints or strong stabilizing selection (Tsuboi, 2021; Tsuboi et al., 2018). Thus, it is entirely possible that the interactions of both region size and total brain size scaling may increase the likelihood of gross-scale mosaic brain evolution in fishes relative to birds and mammals. This may reflect differences in evolutionary strategies to enlarge brain size in conjunction with indeterminant and determinant growth, respectively. However, more research is needed on the evolution of brain–body and region–brain allometries in other lineages with indeterminant growth.

Our results also highlight the importance of considering differences in allometric slope and relaxing the assumption of shared allometric relationships for major taxonomic groups. With brain–body allometries, average slopes across major vertebrate taxonomic levels (class to genus) are relatively constant (Tsuboi et al., 2018); however, when not a priori defining grades based on strict taxonomic-level distinctions, significant differences in the allometric relationships of various groups at different taxonomic levels are readily detected (Ksepka et al., 2020; Smaers et al., 2021; Figure 3). Indeed, when we allow only intercept to vary between grades while assuming parallel slopes, we no longer detect any reliable grade shifts within actinopterygians. Even when allowing slope to vary, two of our identified shifts were undetectable when assuming the brain–body allometry is constant within orders despite analyzing much of the same data (Tsuboi, 2021).

Like previous work in other lineages (Finlay and Darlington, 1995; Yopak et al., 2010), we also find differences in slope across region–brain allometries at varying taxonomic levels, indicating that the dramatic differences in observed region volumes can result from both mosaic shifts between lineages and evolutionary changes in total brain size within a lineage (Figure 5—source data 1, Figure 7—source data 1). As interspecific (evolutionary) allometries are an emergent property of developmental (within individuals) and static (within species) allometries (Pélabon et al., 2014; Tsuboi, 2021), it is difficult to assess the mechanism leading to these differences in slope across species without further research into evolutionary changes in developmental and static allometries. Previous work considering the effect of static allometries on evolutionary allometries found that evolutionary changes in both the static slope and intercept are contributing to the steeper evolutionary slope found across osteoglossiforms (Tsuboi, 2021). Steeper static slopes may indicate a higher rate of brain growth to body growth in adult stages while larger static intercepts might reflect increased brain mass at the transition between embryonic and juvenile growth phases in fishes (Oikawa et al., 1992; Oikawa and Itazawa, 1984; Tsuboi, 2021). Previous work in marsupials did not find any mechanistic links between regional neurogenesis timing or growth rate and static or evolutionary region–brain allometric differences in either slope or intercept despite finding extensive heterochronic differences between species (Carlisle et al., 2017). This finding suggests that the intraspecific mechanisms resulting in these interspecific scaling differences may differ across species, but additional research is needed to determine whether there are shared intraspecific mechanisms resulting in interspecific differences in slope versus intercept in other lineages, especially in those with extensive adult neurogenesis.

Different lineages can independently evolve the same phenotype via the same mechanism (parallel evolution) or different mechanisms (convergent evolution). Given the phylogenetic distance between osteoglossiforms and otophysans, it would be more remarkable to find that the different electrosensory systems and mosaic shifts in brain region volumes evolved in parallel rather than by convergence. Given that the mechanism of these regional increases remains unknown, we argue for a more conservative assumption of convergent evolution for electrosensory cerebrotypes. This is supported by the fact that the cerebellar subregion that has expanded the most in mormyroids is the valvula cerebelli while in gymnotiforms, it is the corpus cerebelli (Meek and Nieuwenhuys, 1998). Additionally, the torus is laminar in gymnotiforms while there are distinct nuclei in the non-laminar torus of mormyroids (Bell and Maler, 2005), and we find evidence of a steeper brain–body allometric relationship for osteoglossiforms that appears unrelated to the evolution of an electrosensory system in mormyroids. Interestingly, we found a subsequent decrease in the brain–body allometry of two small-brained sister mormyroids (I. henryi and B. brachyistius), which further suggests the steeper brain–body allometry of osteoglossiforms is unrelated to electrosensory capabilities. Further research is needed into osteoglossiforms as a whole to determine why this lineage has an increased brain–body allometry. Regardless of whether the mechanism of relative regional scaling evolution is convergent or parallel, we provide evidence of repeated, independent mosaic evolution of major brain regions in association with a convergent behavioral novelty. These findings demonstrate that evolutionary changes in gross-scale brain structure are surprisingly predictable alongside the evolution of active electrosensory systems, even when the underlying brain–body allometry differs. More broadly, these findings suggest that mosaic brain evolution may occur alongside the evolution of behavioral novelty and could reflect a degree of predictability in brain evolution with behavioral evolution, especially when constraint in brain–body allometries are strong.

Brain measurement data is located in Supplementary File 1. Brain mass data is located in Supplementary File 2. All analysis code and phylogenetic trees are available in Dryad. The raw micro-computed tomography scans are too large to post (multiple TBs), but are available upon request. To request raw otophysan and/or osteoglossiform scans, contact the corresponding author. We ask that those who want access to the scan data send us an external hard drive, which we will upload all the data to and then return.

1. 1. Schumacher EL
   2. Carlson B

   (2022) Dryad Digital Repository

   Data from: Convergent mosaic brain evolution is associated with the evolution of novel electrosensory systems in teleost fishes.

   https://doi.org/10.5061/dryad.7d7wm37w5

1. 1. Tsuboi M
   2. van der Bijl W
   3. Kopperud BT
   4. Erritzoe J
   5. Voje KL
   6. Kotrschal A

   (2018) figshare

   Data from: Brain mass and body mass datasets and phylogenies linked to brain-body allometry and the encephalization of birds and mammals.

   https://doi.org/10.6084/m9.figshare.6803276.v1

1. 1. Abrahão VP
   2. Pupo FM
   3. Shibatta OA

   (2018) Comparative brain gross morphology of the Neotropical catfish family Pseudopimelodidae (Osteichthyes, Ostariophysi, Siluriformes), with phylogenetic implications

   Zoological Journal of the Linnean Society 184:750–772.

   https://doi.org/10.1093/zoolinnean/zly011

   • Google Scholar
2. Book

   1. Albert JS
   2. Crampton WGR

   (2006)

   Electroreception and Electrogenesis

   In: Evans DH, Clairborne JB, editors. The Physiology of Fishes. Florida, United States: CRC Press. pp. 431–472.

   • Google Scholar
3. 1. Avin S
   2. Currie A
   3. Montgomery SH

   (2021) An agent-based model clarifies the importance of functional and developmental integration in shaping brain evolution

   BMC Biology 19:97.

   https://doi.org/10.1186/s12915-021-01024-1

   • PubMed
   • Google Scholar
4. 1. Baker CA
   2. Kohashi T
   3. Lyons-Warren AM
   4. Ma X
   5. Carlson BA

   (2013) Multiplexed temporal coding of electric communication signals in mormyrid fishes

   The Journal of Experimental Biology 216:2365–2379.

   https://doi.org/10.1242/jeb.082289

   • PubMed
   • Google Scholar
5. 1. Baron VD
   2. Morshnev KS
   3. Olshansky VM
   4. Orlov AA

   (1994) Electric organ discharges of two species of African catfish (Synodontis) during social behaviour

   Animal Behaviour 48:1472–1475.

   https://doi.org/10.1006/anbe.1994.1387

   • Google Scholar
6. 1. Baron VV

   (2002)

   Triggering of electric discharges in catfish synodontis serratus and clarias gariepinus

   J Ichthyol 42:S223–S230.

   • Google Scholar
7. 1. Barton RA
   2. Harvey PH

   (2000) Mosaic evolution of brain structure in mammals

   Nature 405:1055–1058.

   https://doi.org/10.1038/35016580

   • PubMed
   • Google Scholar
8. 1. Barton RA

   (2012) Embodied cognitive evolution and the cerebellum

   Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 367:2097–2107.

   https://doi.org/10.1098/rstb.2012.0112

   • PubMed
   • Google Scholar
9. Software

   1. Bartoń K

   (2013)

   MuMIn: Multi-model inference. R package version, version 1.10.0

   Multi-Model Inference.
10. Book

    1. Bell CC
    2. Maler L

    (2005)

    Central neuroanatomy of electrosensory systems in fish

    In: Bullock TH, Hopkins CD, Popper AN, Fay RD, editors. Electroreception. New York: Springer. pp. 68–111.

    • Google Scholar
11. Book

    1. Bennett MVL

    (1971)

    Electric OrgansFish Physiology

    New York: Academic Press.

    • Google Scholar
12. 1. Boyle KS
    2. Colleye O
    3. Parmentier E

    (2014) Sound production to electric discharge: sonic muscle evolution in progress in Synodontis spp: catfishes (Mochokidae)

    Proceedings. Biological Sciences 281:20141197.

    https://doi.org/10.1098/rspb.2014.1197

    • PubMed
    • Google Scholar
13. Book

    1. Burnham K
    2. Anderson D.

    (2002)

    Model selection and multi- mode inference: a practical information theoretic approach

    New York: Springer.

    • Google Scholar
14. Book

    1. Caputi AA
    2. Carlson BA
    3. Macadar O

    (2005)

    Electric organs and their control

    In: Bullock TH, Hopkins CD, Popper AN, Fay RR, editors. Electroreception. New York: Springer. pp. 410–451.

    • Google Scholar
15. 1. Carlisle A
    2. Selwood L
    3. Hinds LA
    4. Saunders N
    5. Habgood M
    6. Mardon K
    7. Weisbecker V

    (2017) Testing hypotheses of developmental constraints on mammalian brain partition evolution, using marsupials

    Scientific Reports 7:4241.

    https://doi.org/10.1038/s41598-017-02726-9

    • PubMed
    • Google Scholar
16. 1. Carlson BA
    2. Arnegard ME

    (2011) Neural innovations and the diversification of African weakly electric fishes

    Communicative & Integrative Biology 4:720–725.

    https://doi.org/10.4161/cib.17483

    • PubMed
    • Google Scholar
17. 1. Carlson BA
    2. Hasan SM
    3. Hollmann M
    4. Miller DB
    5. Harmon LJ
    6. Arnegard ME

    (2011) Brain evolution triggers increased diversification of electric fishes

    Science 332:583–586.

    https://doi.org/10.1126/science.1201524

    • PubMed
    • Google Scholar
18. 1. Clark DA
    2. Mitra PP
    3. Wang SSH

    (2001) Scalable architecture in mammalian brains

    Nature 411:189–193.

    https://doi.org/10.1038/35075564

    • PubMed
    • Google Scholar
19. 1. Crampton WGR

    (2019) Electroreception, electrogenesis and electric signal evolution

    Journal of Fish Biology 95:92–134.

    https://doi.org/10.1111/jfb.13922

    • PubMed
    • Google Scholar
20. 1. Darriba D
    2. Taboada GL
    3. Doallo R
    4. Posada D

    (2012) jModelTest 2: more models, new heuristics and parallel computing

    Nature Methods 9:772.

    https://doi.org/10.1038/nmeth.2109

    • PubMed
    • Google Scholar
21. 1. Day JJ
    2. Peart CR
    3. Brown KJ
    4. Friel JP
    5. Bills R
    6. Moritz T

    (2013) Continental diversification of an African catfish radiation (Mochokidae: Synodontis

    Systematic Biology 62:351–365.

    https://doi.org/10.1093/sysbio/syt001

    • PubMed
    • Google Scholar
22. 1. DeCasien AR
    2. Higham JP

    (2019) Primate mosaic brain evolution reflects selection on sensory and cognitive specialization

    Nature Ecology & Evolution 3:1483–1493.

    https://doi.org/10.1038/s41559-019-0969-0

    • PubMed
    • Google Scholar
23. Book

    1. Dubois E

    (1913)

    On the Relation between the Quantity of Brain and the Size of the Body in Vertebrates

    Amsterdam: Koninklijke Akademie van Wetenschappen te Amsterdam.

    • Google Scholar
24. Book

    1. Fay RR
    2. Edds-Walton PL

    (2008)

    Structures and Functions of the Auditory Nervous System of Fishes In

    In: Webb JF, Popper AN, Fay RR, editors. Fish Bioacoustics. New York, NY: Springer Handbook of Auditory Research. pp. 49–98.

    • Google Scholar
25. 1. Finlay BL
    2. Darlington RB

    (1995) Linked regularities in the development and evolution of mammalian brains

    Science 268:1578–1584.

    https://doi.org/10.1126/science.7777856

    • PubMed
    • Google Scholar
26. 1. Guindon S
    2. Gascuel O

    (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood

    Systematic Biology 52:696–704.

    https://doi.org/10.1080/10635150390235520

    • PubMed
    • Google Scholar
27. 1. Gutiérrez-Ibáñez C
    2. Iwaniuk AN
    3. Moore BA
    4. Fernández-Juricic E
    5. Corfield JR
    6. Krilow JM
    7. Kolominsky J
    8. Wylie DR

    (2014) Mosaic and concerted evolution in the visual system of birds

    PLOS ONE 9:e90102.

    https://doi.org/10.1371/journal.pone.0090102

    • PubMed
    • Google Scholar
28. 1. Hagedorn M
    2. Womble M
    3. Finger TE

    (1990) Synodontid catfish: a new group of weakly electric fish Behavior and anatomy

    Brain, Behavior and Evolution 35:268–277.

    https://doi.org/10.1159/000115873

    • PubMed
    • Google Scholar
29. 1. Hastie T
    2. Tibshirani R
    3. Buja A

    (1994) Flexible discriminant analysis by optimal scoring

    Journal of the American Statistical Association 89:1255–1270.

    https://doi.org/10.1080/01621459.1994.10476866

    • Google Scholar
30. Software

    1. Hastie T
    2. Tibshirani R
    3. Leisch F
    4. Hornik K
    5. Ripley BD

    (2020)

    Mixture and Flexible Discriminant Analysis, version 0.5-3

    Mda.
31. 1. Herculano-Houzel S

    (2012) The remarkable, yet not extraordinary, human brain as a scaled-up primate brain and its associated cost

    PNAS 109 Suppl 1:10661–10668.

    https://doi.org/10.1073/pnas.1201895109

    • PubMed
    • Google Scholar
32. 1. Herculano-Houzel S
    2. Manger PR
    3. Kaas JH

    (2014) Brain scaling in mammalian evolution as a consequence of concerted and mosaic changes in numbers of neurons and average neuronal cell size

    Frontiers in Neuroanatomy 8:77.

    https://doi.org/10.3389/fnana.2014.00077

    • PubMed
    • Google Scholar
33. 1. Hoops D
    2. Vidal-García M
    3. Ullmann JFP
    4. Janke AL
    5. Stait-Gardner T
    6. Duchêne DA
    7. Price WS
    8. Whiting MJ
    9. Keogh JS

    (2017) Evidence for concerted and mosaic brain evolution in dragon lizards

    Brain, Behavior and Evolution 90:211–223.

    https://doi.org/10.1159/000478738

    • PubMed
    • Google Scholar
34. 1. Hopkins CD

    (1995) Convergent designs for electrogenesis and electroreception

    Current Opinion in Neurobiology 5:769–777.

    https://doi.org/10.1016/0959-4388(95)80105-7

    • PubMed
    • Google Scholar
35. 1. Hughes LC
    2. Ortí G
    3. Huang Y
    4. Sun Y
    5. Baldwin CC
    6. Thompson AW
    7. Arcila D
    8. Betancur-R R
    9. Li C
    10. Becker L
    11. Bellora N
    12. Zhao X
    13. Li X
    14. Wang M
    15. Fang C
    16. Xie B
    17. Zhou Z
    18. Huang H
    19. Chen S
    20. Venkatesh B
    21. Shi Q

    (2018) Comprehensive phylogeny of ray-finned fishes (Actinopterygii) based on transcriptomic and genomic data

    PNAS 115:6249–6254.

    https://doi.org/10.1073/pnas.1719358115

    • PubMed
    • Google Scholar
36. 1. Hull C

    (2020) Prediction signals in the cerebellum: beyond supervised motor learning

    eLife 9:e54073.

    https://doi.org/10.7554/eLife.54073

    • PubMed
    • Google Scholar
37. 1. Kaatz IM
    2. Stewart DJ
    3. Rice AN
    4. Lobel PS

    (2010) Differences in pectoral fin spine morphology between vocal and silent clades of catfishes (Order Siluriformes): Ecomorphological implications

    Current Zoology 56:73–89.

    https://doi.org/10.1093/czoolo/56.1.73

    • Google Scholar
38. 1. Kéver L
    2. Bass AH
    3. Parmentier E
    4. Chagnaud BP

    (2020) Neuroanatomical and neurophysiological mechanisms of acoustic and weakly electric signaling in synodontid catfish

    The Journal of Comparative Neurology 528:2602–2619.

    https://doi.org/10.1002/cne.24920

    • PubMed
    • Google Scholar
39. Book

    1. Klingenberg CP

    (1996) Multivariate Allometry In: Marcus LF

    In: Corti M, Loy A, Naylor GJP, Slice DE, editors. Advances in Morphometrics. Boston, MA: Springer US. pp. 23–49.

    https://doi.org/10.1007/978-1-4757-9083-2_3

    • Google Scholar
40. 1. Krebs JR

    (1990) Food-storing birds: adaptive specialization in brain and behaviour?

    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 329:153–160.

    https://doi.org/10.1098/rstb.1990.0160

    • PubMed
    • Google Scholar
41. 1. Ksepka DT
    2. Balanoff AM
    3. Smith NA
    4. Bever GS
    5. Bhullar B-AS
    6. Bourdon E
    7. Braun EL
    8. Burleigh JG
    9. Clarke JA
    10. Colbert MW
    11. Corfield JR
    12. Degrange FJ
    13. De Pietri VL
    14. Early CM
    15. Field DJ
    16. Gignac PM
    17. Gold MEL
    18. Kimball RT
    19. Kawabe S
    20. Lefebvre L
    21. Marugán-Lobón J
    22. Mongle CS
    23. Morhardt A
    24. Norell MA
    25. Ridgely RC
    26. Rothman RS
    27. Scofield RP
    28. Tambussi CP
    29. Torres CR
    30. van Tuinen M
    31. Walsh SA
    32. Watanabe A
    33. Witmer LM
    34. Wright AK
    35. Zanno LE
    36. Jarvis ED
    37. Smaers JB

    (2020) Tempo and pattern of avian brain size evolution

    Current Biology 30:2026–2036.

    https://doi.org/10.1016/j.cub.2020.03.060

    • PubMed
    • Google Scholar
42. 1. Kverková K

    (2022) The evolution of brain neuron numbers in amniotes

    PNAS 119::e2121624119.

    https://doi.org/10.1073/pnas.2121624119

    • Google Scholar
43. 1. Ladich F
    2. Winkler H

    (2017) Acoustic communication in terrestrial and aquatic vertebrates

    The Journal of Experimental Biology 220:2306–2317.

    https://doi.org/10.1242/jeb.132944

    • PubMed
    • Google Scholar
44. Software

    1. Lenth R

    (2020)

    Estimated Marginal Means, aka Least-Squares Means, version 1.7.5

    Emmeans.
45. 1. Loomis C
    2. Peuß R
    3. Jaggard JB
    4. Wang Y
    5. McKinney SA
    6. Raftopoulos SC
    7. Raftopoulos A
    8. Whu D
    9. Green M
    10. McGaugh SE
    11. Rohner N
    12. Keene AC
    13. Duboue ER

    (2019) An adult brain atlas reveals broad neuroanatomical changes in independently evolved populations of mexican cavefish

    Frontiers in Neuroanatomy 13:88.

    https://doi.org/10.3389/fnana.2019.00088

    • PubMed
    • Google Scholar
46. 1. Maler L
    2. Sas E
    3. Johnston S
    4. Ellis W

    (1991) An atlas of the brain of the electric fish Apteronotus leptorhynchus

    Journal of Chemical Neuroanatomy 4:1–38.

    https://doi.org/10.1016/0891-0618(91)90030-g

    • PubMed
    • Google Scholar
47. 1. Marhounová L
    2. Kotrschal A
    3. Kverková K
    4. Kolm N
    5. Němec P

    (2019) Artificial selection on brain size leads to matching changes in overall number of neurons

    Evolution; International Journal of Organic Evolution 73:2003–2012.

    https://doi.org/10.1111/evo.13805

    • PubMed
    • Google Scholar
48. Book

    1. Meek J
    2. Nieuwenhuys R

    (1998)

    Holosteans and Teleosts In

    In: Nieuwenhuys R, Ten Donkelaar H, Nicholson C, editors. The Central Nervous Sytem of Vertebrates. Germany: Springer. pp. 759–938.

    • Google Scholar
49. Thesis

    1. Merzin M

    (2008)

    Applying stereological method in radiology. Volume measurement (Bachelor’s thesis

    University of Tartu, Estonia.

    • Google Scholar
50. 1. Metzen MG
    2. Chacron MJ

    (2021) Population coding of natural electrosensory stimuli by midbrain neurons

    The Journal of Neuroscience 41:3822–3841.

    https://doi.org/10.1523/JNEUROSCI.2232-20.2021

    • PubMed
    • Google Scholar
51. 1. Moore JM
    2. DeVoogd TJ

    (2017) Concerted and mosaic evolution of functional modules in songbird brains

    Proceedings. Biological Sciences 284:20170469.

    https://doi.org/10.1098/rspb.2017.0469

    • PubMed
    • Google Scholar
52. 1. Motani R
    2. Schmitz L

    (2011) Phylogenetic versus functional signals in the evolution of form-function relationships in terrestrial vision

    International Journal of Organic Evolution 65:2245–2257.

    https://doi.org/10.1111/j.1558-5646.2011.01271.x

    • PubMed
    • Google Scholar
53. 1. Mull CG
    2. Yopak KE
    3. Dulvy NK

    (2020) Maternal investment, ecological lifestyle, and brain evolution in sharks and rays

    The American Naturalist 195:1056–1069.

    https://doi.org/10.1086/708531

    • PubMed
    • Google Scholar
54. 1. Oikawa S
    2. Itazawa Y

    (1984) Relative growth of organs and parts of the carp, cyprinus carpio, with special reference to the metabolism-size relationship

    Copeia 1984:800.

    https://doi.org/10.2307/1445176

    • Google Scholar
55. 1. Oikawa S
    2. Takemori M
    3. Itazawa Y

    (1992) Relative growth of organs and parts of a marine teleost, the porgy,Pagrus major, with special reference to metabolism-size relationships

    Japanese Journal of Ichthyology 39:243.

    https://doi.org/10.1007/BF02905482

    • Google Scholar
56. 1. Pagel M

    (1999) Inferring the historical patterns of biological evolution

    Nature 401:877–884.

    https://doi.org/10.1038/44766

    • PubMed
    • Google Scholar
57. 1. Paradis E
    2. Claude J
    3. Strimmer K

    (2004) APE: analyses of phylogenetics and evolution in R language

    Bioinformatics 20:289–290.

    https://doi.org/10.1093/bioinformatics/btg412

    • PubMed
    • Google Scholar
58. 1. Paulin MG

    (1993) The role of the cerebellum in motor control and perception

    Brain, Behavior and Evolution 41:39–50.

    https://doi.org/10.1159/000113822

    • PubMed
    • Google Scholar
59. 1. Pélabon C
    2. Firmat C
    3. Bolstad GH
    4. Voje KL
    5. Houle D
    6. Cassara J
    7. Rouzic AL
    8. Hansen TF

    (2014) Evolution of morphological allometry: The evolvability of allometry

    Annals of the New York Academy of Sciences 1320:58–75.

    https://doi.org/10.1111/nyas.12470

    • PubMed
    • Google Scholar
60. Website

    1. Pinheiro J
    2. Bates D
    3. DebRoy S
    4. Sarkar D
    5. Team RC

    (2019) nlme: Linear and Nonlinear Mixed Effects Models

    Accessed November 11, 2015.

    https://svn.r-project.org/R-packages/trunk/nlme/
61. 1. Pinton A
    2. Agnèse JF
    3. Paugy D
    4. Otero O

    (2013) A large-scale phylogeny of Synodontis (Mochokidae, Siluriformes) reveals the influence of geological events on continental diversity during the Cenozoic

    Molecular Phylogenetics and Evolution 66:1027–1040.

    https://doi.org/10.1016/j.ympev.2012.12.009

    • PubMed
    • Google Scholar
62. 1. Popa LS
    2. Ebner TJ

    (2019) Cerebellum, predictions and errors

    Frontiers in Cellular Neuroscience 12:524.

    https://doi.org/10.3389/fncel.2018.00524

    • PubMed
    • Google Scholar
63. Book

    1. R Core Team

    (2019)

    R: A Language and Environment for Statistical Computing

    Vienna, Austria: R Foundation for Statistical Computing.

    • Google Scholar
64. 1. Rabosky DL
    2. Santini F
    3. Eastman J
    4. Smith SA
    5. Sidlauskas B
    6. Chang J
    7. Alfaro ME

    (2013) Rates of speciation and morphological evolution are correlated across the largest vertebrate radiation

    Nature Communications 4:1958.

    https://doi.org/10.1038/ncomms2958

    • PubMed
    • Google Scholar
65. 1. Rabosky DL
    2. Chang J
    3. Title PO
    4. Cowman PF
    5. Sallan L
    6. Friedman M
    7. Kaschner K
    8. Garilao C
    9. Near TJ
    10. Coll M
    11. Alfaro ME

    (2018) An inverse latitudinal gradient in speciation rate for marine fishes

    Nature 559:392–395.

    https://doi.org/10.1038/s41586-018-0273-1

    • PubMed
    • Google Scholar
66. 1. Rambaut A
    2. Drummond AJ
    3. Xie D
    4. Baele G
    5. Suchard MA

    (2018) Posterior Summarization in Bayesian Phylogenetics Using Tracer 1.7

    Systematic Biology 67:901–904.

    https://doi.org/10.1093/sysbio/syy032

    • PubMed
    • Google Scholar
67. 1. Revell LJ

    (2012) phytools: an R package for phylogenetic comparative biology (and other things

    Methods in Ecology and Evolution 3:217–223.

    https://doi.org/10.1111/j.2041-210X.2011.00169.x

    • Google Scholar
68. 1. Russell CJ
    2. Bell CC

    (1978) Neuronal responses to electrosensory input in mormyrid valvula cerebelli

    Journal of Neurophysiology 41:1495–1510.

    https://doi.org/10.1152/jn.1978.41.6.1495

    • PubMed
    • Google Scholar
69. 1. Schindelin J
    2. Arganda-Carreras I
    3. Frise E
    4. Kaynig V
    5. Longair M
    6. Pietzsch T
    7. Preibisch S
    8. Rueden C
    9. Saalfeld S
    10. Schmid B
    11. Tinevez JY
    12. White DJ
    13. Hartenstein V
    14. Eliceiri K
    15. Tomancak P
    16. Cardona A

    (2012) Fiji: an open-source platform for biological-image analysis

    Nature Methods 9:676–682.

    https://doi.org/10.1038/nmeth.2019

    • PubMed
    • Google Scholar
70. 1. Schneider CA
    2. Rasband WS
    3. Eliceiri KW

    (2012) NIH Image to ImageJ: 25 years of image analysis

    Nature Methods 9:671–675.

    https://doi.org/10.1038/nmeth.2089

    • PubMed
    • Google Scholar
71. Software

    1. Schumacher EL
    2. Carlson BA

    (2022) Convergent mosaic brain evolution is associated with the evolution of novel electrosensory systems in teleost fishes

    Dryad.

    https://doi.org/10.5061/dryad.7d7wm37w5
72. 1. Smaers JB
    2. Rothman RS
    3. Hudson DR
    4. Balanoff AM
    5. Beatty B
    6. Dechmann DKN
    7. de Vries D
    8. Dunn JC
    9. Fleagle JG
    10. Gilbert CC
    11. Goswami A
    12. Iwaniuk AN
    13. Jungers WL
    14. Kerney M
    15. Ksepka DT
    16. Manger PR
    17. Mongle CS
    18. Rohlf FJ
    19. Smith NA
    20. Soligo C
    21. Weisbecker V
    22. Safi K

    (2021) The evolution of mammalian brain size

    Science Advances 7:eabe2101.

    https://doi.org/10.1126/sciadv.abe2101

    • PubMed
    • Google Scholar
73. 1. Stevens JA
    2. Sukhum KV
    3. Carlson BA

    (2013) Independent evolution of visual and electrosensory specializations in different lineages of mormyrid electric fishes

    Brain, Behavior and Evolution 82:185–198.

    https://doi.org/10.1159/000355369

    • PubMed
    • Google Scholar
74. 1. Strick PL
    2. Dum RP
    3. Fiez JA

    (2009) Cerebellum and nonmotor function

    Annual Review of Neuroscience 32:413–434.

    https://doi.org/10.1146/annurev.neuro.31.060407.125606

    • PubMed
    • Google Scholar
75. Book

    1. Striedter G.

    (2005)

    Principles of Brain Evolution

    Sunderland, Massachusetts: Sinauer Associates.

    • Google Scholar
76. Book

    1. Striedter GF
    2. Northcutt RG

    (2020) Brains Through Time

    New York, NY: Oxford University Press.

    https://doi.org/10.1093/oso/9780195125689.001.0001

    • Google Scholar
77. 1. Suchard MA
    2. Lemey P
    3. Baele G
    4. Ayres DL
    5. Drummond AJ
    6. Rambaut A

    (2018) Bayesian phylogenetic and phylodynamic data integration using BEAST 1.10

    Virus Evolution 4:vey016.

    https://doi.org/10.1093/ve/vey016

    • PubMed
    • Google Scholar
78. 1. Sukhum KV
    2. Freiler MK
    3. Wang R
    4. Carlson BA

    (2016) The costs of a big brain: extreme encephalization results in higher energetic demand and reduced hypoxia tolerance in weakly electric African fishes

    Proceedings. Biological Sciences 283:20162157.

    https://doi.org/10.1098/rspb.2016.2157

    • PubMed
    • Google Scholar
79. 1. Sukhum KV
    2. Shen J
    3. Carlson BA

    (2018) Extreme enlargement of the cerebellum in a clade of teleost fishes that evolved a novel active sensory system

    Current Biology 28:3857–3863.

    https://doi.org/10.1016/j.cub.2018.10.038

    • PubMed
    • Google Scholar
80. 1. Takiyama T
    2. Luna da Silva V
    3. Moura Silva D
    4. Hamasaki S
    5. Yoshida M

    (2015) Visual capability of the weakly electric fish apteronotus albifrons as revealed by a modified retinal flat-mount method

    Brain, Behavior and Evolution 86:122–130.

    https://doi.org/10.1159/000438448

    • PubMed
    • Google Scholar
81. 1. Tsuboi M
    2. van der Bijl W
    3. Kopperud BT
    4. Erritzøe J
    5. Voje KL
    6. Kotrschal A
    7. Yopak KE
    8. Collin SP
    9. Iwaniuk AN
    10. Kolm N

    (2018) Breakdown of brain-body allometry and the encephalization of birds and mammals

    Nature Ecology & Evolution 2:1492–1500.

    https://doi.org/10.1038/s41559-018-0632-1

    • PubMed
    • Google Scholar
82. 1. Tsuboi M

    (2021) Exceptionally steep brain-body evolutionary allometry underlies the unique encephalization of osteoglossiformes

    Brain, Behavior and Evolution 96:49–63.

    https://doi.org/10.1159/000519067

    • PubMed
    • Google Scholar
83. 1. Ullmann JFP
    2. Cowin G
    3. Kurniawan ND
    4. Collin SP

    (2010) A three-dimensional digital atlas of the zebrafish brain

    NeuroImage 51:76–82.

    https://doi.org/10.1016/j.neuroimage.2010.01.086

    • PubMed
    • Google Scholar
84. 1. Uyeda JC
    2. Pennell MW
    3. Miller ET
    4. Maia R
    5. McClain CR

    (2017) The evolution of energetic scaling across the vertebrate tree of life

    The American Naturalist 190:185–199.

    https://doi.org/10.1086/692326

    • PubMed
    • Google Scholar
85. Software

    1. Uyeda JC
    2. Eastman J
    3. Harmon L

    (2020)

    Bayesian fitting of ornstein-uhlenbeck models to phylogenies, version 2.1.1

    Bayou.
86. 1. Vélez A
    2. Kohashi T
    3. Lu A
    4. Carlson BA

    (2017) The cellular and circuit basis for evolutionary change in sensory perception in mormyrid fishes

    Scientific Reports 7:3783.

    https://doi.org/10.1038/s41598-017-03951-y

    • PubMed
    • Google Scholar
87. 1. Vélez A
    2. Ryoo DY
    3. Carlson BA

    (2019) Sensory specializations of mormyrid fish are associated with species differences in electric signal localization behavior

    Brain, Behavior and Evolution 92:125–141.

    https://doi.org/10.1159/000496493

    • Google Scholar
88. 1. Wullimann MF
    2. Northcutt RG

    (1990) Visual and electrosensory circuits of the diencephalon in mormyrids: an evolutionary perspective

    The Journal of Comparative Neurology 297:537–552.

    https://doi.org/10.1002/cne.902970407

    • PubMed
    • Google Scholar
89. 1. Wullimann MF

    (2020) Neural origins of basal diencephalon in teleost fishes: Radial versus tangential migration

    Journal of Morphology 281:1133–1141.

    https://doi.org/10.1002/jmor.21237

    • PubMed
    • Google Scholar
90. 1. Xie W
    2. Lewis PO
    3. Fan Y
    4. Kuo L
    5. Chen MH

    (2011) Improving marginal likelihood estimation for Bayesian phylogenetic model selection

    Systematic Biology 60:150–160.

    https://doi.org/10.1093/sysbio/syq085

    • PubMed
    • Google Scholar
91. 1. Yopak KE
    2. Lisney TJ
    3. Darlington RB
    4. Collin SP
    5. Montgomery JC
    6. Finlay BL

    (2010) A conserved pattern of brain scaling from sharks to primates

    PNAS 107:12946–12951.

    https://doi.org/10.1073/pnas.1002195107

    • PubMed
    • Google Scholar
92. Book

    1. Zupanc GKH
    2. Bullock TH

    (2005) From Electrogenesis to Electroreception: An Overview In

    In: Bullock TH, Hopkins CD, Popper AN, Fay RR, editors. Electroreception. New York: Springer. pp. 5–46.

    https://doi.org/10.1007/0-387-28275-0_2

    • Google Scholar

Author details

1. Erika L Schumacher

   Department of Biology, Washington University, St Louis, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2492-1117

2. Bruce A Carlson

   Department of Biology, Washington University, St Louis, United States

   Contribution

   Conceptualization, Writing - review and editing

   For correspondence

   carlson.bruce@wustl.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2151-0443

• 1,411

  views

• 306

  downloads

• 13

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.