Brain diversification of cichlid fish


It is estimated that there are as many as 3000 different species of cichlid fish.  Lake Malawi in Malawi, Africa, is the third largest lake on the continent and is home to hundreds of distinct species of cichlid that evolved from a common ancestor over the last 500,000 years.  Given that this group of Malawi cichlids has only evolved and diversified for a relatively short span of time, it is surprising that there is such wide variety between brain structures of the different species. These species are polymorphous with nearly identical genomes, but their brain structures vary like unrelated species.  One study by Dr. Todd Streelman’s lab at Georgia Tech (Sylvester et al., 2010) attempts to understand how such diversity in brain structure develops in ecologically distinct species that are very closely related and only recently (the past 500,000 years) evolved into discrete species types.

Previous studies on brain development

Previous studies suggest that variation in brain region size comes from late development stages by changes in neurogenesis based on adaptive control. Neurogenesis is the process by which neurons in the brain mature by terminal differentiation of neuronal stem cells.  A study by Finlay et al. in 1995 concluded that variations in brain structures are caused by the timing of neurogenesis, after brain structures had formed. A 2002 study on mice (Chenn et al., 2002) showed that it was possible to increase the size of the mouse brain by altering the WNT pathway, which causes delayed neurogenesis. These studies established the viewpoint that brains evolved only through changes in late stages of development, called the “late=large theory,” which states that the later in development that neurogenesis occurs, the larger the brain is.  Sylvester et al., on the other hand, looked at the importance of early patterning events by comparing the early patterning of ecologically distinct species to determine the role that early development plays in brain diversity between recently divided lineages.

Cichlid species studied in this experiment

(Images found on Google)

This study aimed to compare the brain structures of ecologically distinct species, meaning species that live in different environments. “Mbuna” is a term used to describe any species of Lake Malawi cichlids that live in rocky areas, while “nonmbuna” describes those that live elsewhere, in this case near sand. There are hundreds of species of each of these types of fish, although they seem to have originated from a common ancestor. Observation of brains from both groups suggested that brain region size has to do with ecological environment, but there also seemed to be similarities in region proportions between species that perform similar feeding behaviors. This study used three mbuna and three nonmbuna species with different feeding habits for comparison.

Mbuna: rock-dwellers

  • Labeotropheus fulleborni (LF): scrape algae from rocks
  • Maylandia zebra (MZ): no specific feeding habits
  • Cynotilapia afra (CA): feed on plankton

Nonmbuna: sand-dwellers

  • Copadichromis borleyi (CB): feed on plankton
  • Mchenga conophorus (MC): no specific feeding habits
  • Aulonocara jacobfreibergi (AJ): find food by sensing vibrations

Major brain structures

  • Forebrain (telencephalon, diencephalon): The embryonic forebrain is made up of the telencephalon and the diencephalon, which together develop into the cerebrum. In humans, the cerebrum makes up the majority of the brain and works with the cerebellum to control voluntary physical actions. The forebrain differentiates into two major domains: rostral, which contains the telencephalon and hypothalamus, and caudal, which forms the thalamus. Cerebrotyping (Figure 1A) shows that 60-80% of mammal brains are telencephalon, while this region makes up only about 18-25% of cichlid brains.
  • Midbrain (mesencephalon, optic tectum): The embryonic mesencephalon gives rise to the midbrain, which controls sight, hearing, temperature regulation, and motor functions. The mesencephalon is part of the brainstem and bridges the forebrain and the hindbrain. Figure 1A shows that in mammals, the midbrain usually makes up less than 10% of the brain, while it is roughly 35-45% of cichlid brains. The midbrain includes the optic tectum, which receives and processes sensory input, especially from the eyes.
  • Hindbrain (cerebellum): The cerebellum is at the base of the brain structure, and is important in motor control and maintenance of equilibrium. Figure 1A shows that the cerebellum composes approximately 12-15% of brain structures in mammals, Lake Malawi cichlids, and Lake Tanganyika cichlids.

Figure 1A: Sylvester et. al., 2010

Figure 1A is a cerebrotype box plot that compares the proportion of the brain that each region composes for mammals (blue), Lake Malawi cichlids (green), and Lake Tanganyika cichlids (yellow).

Figure 1B-D: Brain structures of rock-dwelling cichlids. Sylvester et al., 2010

Figure 1B-D show these major regions in the three types of mbuna (rock-dwelling) cichlids, with the anterior region on the left.  The black arrow points to the telencephalon, the gray is the optic tectum, and the white is the cerebellum.

  • B: LF fish are algal scrapers, which were determined to have relatively large telencephala, but small optic lobes
  • C: MZ fish use many sources of food.
  • D: CA fish feed on plankton, and had relatively large optic lobes.
  • Not pictured: AJ “sonar hunters” had large telencephala and cerebella.

Brain development

Regions of the brain develop based on anterior and posterior signaling molecules, which initiate this polarity soon after gastrulation. Wnt1 is the major signaling protein necessary for the development of posterior structures of the brain, and is secreted by the midbrain-hindbrain boundary (MHB). This signal is suppressed in the anterior region by wnt antagonists released from the anterior neural ridge (ANR), such as six3 and tlc. The zona limitans intrathalamica (ZLI) is another signaling center formed after the MHB and ANR, which released sonic hedgehog (shh), important for development of the forebrain and midbrain. Functional neurons arise from neurogenesis as these three signaling centers direct patterning and morphogenesis of individual brain compartments.

The major genes examined in this study work together to form a regulatory circuit to direct development of the telencephalon and thalamus.

  • wnt1: functions for development of the posterior structures of the brain
  • shh: sonic hedgehog, released by the ZLI for development of the forebrain
  • fezf2: transcription factor that regulates development of parts of the forebrain
  • irx1b: involved in the development and positioning of the ZLI and thalamus, mediates WNT signaling
  • six3: transcription factor antagonist of WNT secreted from the ANR, plays a role in ZLI formation and telencephalon development

The three brain compartments develop individually based on signals released from the three major signaling centers.  Four hypotheses were suggested to explain the diversification of brain structures.

  1. The MHB and ANR release their respective signals at different times or in different amounts.
  2. The early patterning boundaries (MHB and ANR) form at different locations, leading to variation in region size.
  3. Neurogenesis occurs at different times, at different rates, or in different amounts during the development process.
  4. Multiple of the above are working together to cause differences in structure between closely related cichlids.

This study by Sylvester et al. looks at the second of these hypotheses using in situ hybridization to determine the location of the patterning boundaries early in the development process. The third hypothesis was supported by previous studies mentioned above, so it was thought that a combination of multiple causes work together to regulate brain region size.

The Experiment

The WNT pathway is important in signaling the development of brain regions and regulating Shh expression. The WNT pathway is active all over the body to signal for cell proliferation and differentiation.

The WNT pathway with GSK3 inhibition (1)

This pathway controls the transcription factor beta-catenin, which is active once it is transferred from the cytoplasm into the nucleus of the cell. Beta-catenin is down-regulated by glycogen synthase kinase (GSK3), which binds to the beta-catenin and targets it for degradation by proteasomes. Lithium chloride inhibits GSK3, therefore stabilizing beta-catenin and causing it to be more active. DMSO, or dimethyl sulfoxide, is a solvent used to help LiCl reach cells, so control fish are treated with DMSO to ensure that this does not alter results.

Like most species, cichlid development is described by stages, each of which represents a time period in development. It was determined that stage 16 was the earliest at which the regions of the brain could be discerned and measured reliably.
This table shows that the embryonic mbuna had relatively larger telencephala and smaller thalami, while the embryonic nonmbuna had smaller telencephala and larger thalami. This clear variation in brain region size is based on ecological habitat. It was proposed that the mbuna have larger telencephala due to their social nature and need to navigate rocky habitats that are more complex than sand.

Sylvester et al. then wanted to look at differences in the patterning of the forebrain by manipulating the Wnt pathway, to see how this plays a role in physical variation between brain regions.  The genes previously described were known to initiate AP polarization as well as to determine the ZLI location within the forebrain region.  wnt1 establishes the posterior regions, and knockout mice had been shown to develop larger telencephala, smaller thalami, and to have the ZLI regions shifted towards the posterior. fefz2 and irx1 also position the AP axis and the ZLI location.  irx1 knockdown in zebrafish had been shown to shorten the forebrain, and extend the ZLI towards the posterior, thereby shortening the thalamus [Scholpp et al., 2006].  From this previous knowledge, it was hypothesized that changes in anterior-posterior forebrain patterning cause the variation in forebrain size between mbuna and nonmbuna, which could be observed by the expression of this gene network.

Forebrain patterning visualized using in situ hybridization Sylvester et al. 2010

This image shows in situ hybridization of various genes involved in the forebrain patterning. The diagram at the top shows a summary of expression locations and interactions. A, B, and C show the expression of shh, fefz2, six3, and irx1b in nonmbuna species at stage 10. The ZLI is initiated in stage 10, and is distinguishable by expression of shh (shh shown in blue in part A, and ZLI outlined with the white dotted line in part B).  Part D compares the expression of wnt1 and six3 in nonmbuna AJ (left) versus mbuna MZ (middle), with a schematic showing expression locations for clarity.  This shows that nonmbuna had extended forebrain regions with wnt1 expression beginning 46% of the way through the brain, while for mbuna it began at 36%, closer to the anterior. In both mbuna and nonmbuna, six3 is expressed in the anterior-most area, but with less expression in the nonmbuna.

Part E shows the expressions of irx1 (red) compared between MC nonmbuna (left), LF mbuna (middle), and LF mbuna treated with LiCl.  The LF had a shorter anterior region compared to MC, with wnt1 expression beginning at 47% of the brain for MC but 38% for LF.  When treated with LiCl, the nonmbuna species showed an elongated anterior region, with wnt1 expression beginning at 55% of the brain structure.

Part F shows stage 11 shh expression between MC nonmbuna and MZ mbuna, both of which are generalist feeders. This expression marks the ZLI, which is outlined in black in the figures. The angle of the ZLI was determined, and it was found that nonmbuna have a larger ZLI angle (average 129 degrees) than mbuna (average 108), but with LiCl treatment the mbuna angle becomes larger, more similar to that of the nonmbuna (133 degrees).

Summary of results

  • Mbuna: reduced extent of wnt1-irx1b, small thalami, smaller ZLI angles, larger telencephalon
  • Nonmbuna: extension of wnt1-irx1b expression, larger ZLI angles, smaller telencephala, larger thalami

wnt1 was altered using agonist LiCl, and then sacrificed the treated fish at stages 11-17 to see how the ZLI angle changed as development continued after treatment.  Increased WNT pathway signaling was hypothesized to cause mbuna brains to become more similar to nonmbuna.  Results of the ZLI angle of LiCl treated fish against controls confirmed that the manipulation of the Wnt pathway at early stages in brain development was sufficient to cause the variation between mbuna and nonmbuna.


  • Changing the patterning of pathways responsible for brain development early in the process (before structures even arise) can ultimately affect the size of adult brain structures
  • This is in contrast to the previous “late = large” theory, which suggested that differences in brain size are the result of only neurogenesis late in brain development.
  • This study supported the fourth hypothesis of brain development mentioned above, that multiple causes contributed to the diversification of cichlid brain structures. With this recent research, it is believed that this process is due to a combination of neurogenesis timing and early patterning events.

Critiques of the paper

Overall, this paper provides important insight into the link between brain development and evolution. This research ends when the fish hatch, so a significant amount of development still occurs after the time span of this paper. These later stages of development, particularly of neurogenesis, could have been included in order to see the effect of the entire neurogenesis process on the changes in patterning that were induced. A longer time span of experimentation to include post-hatching neurogenesis would give a more complete picture. Also, there are many developmental pathways involved in neurogenesis, including Notch, Bmp, Hedgehog, and Fgf, so it is possible that some of these experiments altered pathways other than just WNT. In future studies, the actions of other pathways should be examined for further understanding, and to ensure that data was not the result of manipulation of multiple pathways involved in more than just neurogenesis. If up-regulation of the WNT pathway increases brain size, a WNT antagonist should be used to see if this decreases brain region size.


Chenn, A. and Walsh, C. (2002). Regulation of cerebral cortical size by control of cell cycle exit in neural precursors. Science. Vol. 297, No. 5580, pp 365-369.

Finlay, B., Darlington, R. (1995). Linked regularities in the development and evolution of mammalian brains Science. Vol. 297, pp 1578-1584.

Kimmel, C., Ballard, W., Kimmel, S., Ullmann, B., Shilling, T. (1995). Stages of embryonic development of the zebrafish. Developmental Dynamics. Vol. 203, pp 253-310.

Scholpp, S., Wolf, O., Brand, M., Lumsden, A. (2006). Hedgehog signaling from the zona limitans intrathalamica orchestrates patterning of the zebrafish diencephalon. Development. Vol. 133. pp 855-864.

Sylvester, J., Rich, C., Lon, Y., Staaden, M., Fraser, G., Streelman, J. T. (2010). Brain diversity evolves via differences in patterning. PNAS. Vol. 107, No. 21, pp 9718-9723.


[1] Wnt pathway: