Intraspecific brain size variation between coexisting sunfish ecotypes

Abstract

Variation in spatial complexity and foraging requirements between habitats can impose different cognitive demands on animals that may influence brain size. However, the relationship between ecologically related cognitive performance and brain size is not well established. We test whether variation in relative brain size and brain region size is associated with habitat use within a population of pumpkinseed sunfish composed of different ecotypes that inhabit either the structurally complex shoreline littoral habitat or simpler open-water pelagic habitat. Sunfish using the littoral habitat have on average 8.3% larger brains than those using the pelagic habitat. We found little difference in the proportional sizes of five brain regions between ecotypes. The results suggest that cognitive demands on sunfish may be reduced in the pelagic habitat given no habitat-specific differences in body condition. They also suggest that either a short divergence time or physiological processes may constrain changes to concerted, global modifications of brain size between sunfish ecotypes.

1. Introduction

Vertebrate brains vary in size from a few milligrams to more than 10 kg. While most variation in brain size is due to body size variation, residual variation reflects a combination of historical and more contemporary evolutionary and environmental processes [1]. Brain size variation is of interest because of its potential functional links to behavioural and cognitive performance. In addition, behavioural and cognitive diversity is thought to permit adaptive diversification by allowing populations to colonize novel habitats where they subsequently experience diversifying selection on other traits [2]. Thus, revealing the causal links between brain, behaviour and ecology could provide insights into adaptive diversification processes in vertebrates. Fish are ideal for studies of the causes and consequences of variation in brain size because, as the most diverse clade of vertebrates, they show considerable variation in brain size and occur in a wide range of freshwater and marine habitats.

A large portion of the variation in brain features across vertebrate taxa is represented by variability in whole brain size [3–7]. Because larger brains typically have more neurons and more connections between neurons, whole brain size is thought to reflect general cognitive ability [1,8,9]. However, because neural tissue is energetically costly to maintain, greater brain size should evolve and be maintained only under strong selection for cognitive ability [10–12]. This idea is supported by the evolution of smaller brains when selection on cognitive ability is relaxed [13]. Additionally, larger brains are associated with habitat spatial complexity [14], suggesting functional links between spatial complexity, cognition and brain size.

Whole brain size may reflect general cognitive ability, but it may not capture variation in the relative contributions of more specialized cognitive abilities. Fish brains vary in gross morphology. They are composed of regions that appear partially specialized [15] because the relative size of particular brain regions is often associated with environmental conditions that require particular sensory or cognitive tasks [16–21]. However, enlargement of a brain region should only occur when the sensory, cognitive or behavioural function that region serves benefit organismal performance and fitness because brain tissue is energetically costly.

If brain morphology and cognitive performance are functionally related, then fish living in habitats that require greater cognitive abilities for specific attributes of behaviour, cognition or sensory perception should have relatively larger brains and larger regions of the brain associated with those specific attributes [22]. For example, benthic shark species display more developed olfactory bulbs than pelagic species that conversely have larger cerebella [4], probably reflective of differences in movement and foraging requirements of those habitats. This and other examples of interspecific variation in brain morphology, such as in reef fish [6], cichlids [23] and gobies [24], are consistent with functional relationships between brain morphology and habitat use in fishes.

Most studies exploring the relationship between brain morphology and habitat use invoke interspecific comparisons that can confound current with historic functional utility [6,23–26]. Species comparisons have the advantage of substantial and consistent ecological divergence rarely observed within species, but intraspecific comparisons are complementary by isolating current effects of habitat use on brain morphology. Here, we explore the potential for adaptive diversification in brain size and morphology between ecotypes diversifying into littoral and pelagic lake habitats. Various studies suggest that the shallow inshore littoral and deeper offshore pelagic habitats of aquatic environments require different sensory, cognitive and behavioural abilities in fishes because they differ in many biotic (e.g. species interactions) and abiotic (e.g. physical structure and visual complexity) conditions [4,6,24]. For example, bluegill sunfish (Lepomis macrochirus) use a more thorough searching style during foraging under littoral compared with pelagic conditions when searching for large but cryptic littoral macroinvertebrates compared with small pelagic zooplankton prey [27]. Ecologically related alternative phenotypes (hereafter ‘ecotypes’) occur in littoral and pelagic lake habitats, and are widespread in fishes in northern post-glacial lakes [28–30]. Because of their young age since the last glacial retreat (less than 12 000 years bp) and replication within and among species, these ecotypes provide opportunities to explore relationships between brain features, cognitive performance and early ecological diversification. Specifically, we hypothesize that the littoral ecotype requires larger brains than the pelagic ecotype after accounting for the effects of body size because of the greater structural complexity of littoral compared to pelagic habitat. Similarly, if each habitat has specific cognitive requirements, then the proportional size of brain regions relative to the whole brain will also differ between fish that occupy each habitat, assuming that brain regions are unconstrained in their ability to change in size.

2. Material and methods

(a). Study organism

Pumpkinseed sunfish (Lepomis gibbosus) are a freshwater ray-finned fish native to northeastern North America that can exhibit divergent ecotypes within a single lake population. Pumpkinseeds typically occur in an inshore shallow-water or littoral habitat and have evolved adaptations including specialized pharyngeal jaws for consuming large armoured invertebrate prey like snails, and tolerance for low oxygen tension in the warm, shallow and structurally complex littoral habitat [31]. However, in larger nutrient-poor oligotrophic post-glacial lakes, adults are frequently found offshore in the deep pelagic habitat, particularly in proximity to submerged rocky shoals where they can spawn and find refuge from predators. Pelagic pumpkinseeds feed extensively on zooplankton throughout the summer growing season, whereas littoral pumpkinseeds do not [31,32]. Ecotypes have evolved in sympatry in numerous geographically separated post-glacial lakes from a local common ancestor, although local lake conditions [33] and colonization history [34] shape each local divergence. Variation in body form and behaviour generally follow functional expectations of the different foraging requirements in the different habitats. For example, pelagic fish tend to have shorter heads, more densely packed gill rakers and smaller jaws appropriate for a planktonic diet [32,34,35], while the littoral fish display the opposite characteristics, better suited for larger benthic invertebrate prey. Body form and behavioural variation are related to feeding performance in laboratory [36] and in field studies [37]. Common garden studies indicate that phenotypic variation is influenced by both genetic and environmental components [36,38]. The combination of ecological, phenotypic and performance variation along with genetic differences among ecotypes suggests recent adaptive diversification between ancestral littoral and derived pelagic ecotypes.

(b). Study population

Samples of pumpkinseed sunfish were collected from Ashby Lake, Ontario (45.092 N, 77.351 W), which contains considerable habitat diversity that limits fish movement. Ashby Lake is a mid-sized Canadian Shield lake (surface area 2.59 km², maximum length 2.6 km, width 1.8 km, depth 36.6 m). Three lake basins are dominated by large volumes of pelagic habitat (3–37 m) occasionally punctuated by submerged rocky shoals without vegetation. Littoral habitat occurs in shallow bays that contain soft sediments supporting macrophyte vegetation separated by long expanses of rocky shoreline. Littoral and pelagic shoal habitats are segregated by large expanses of open water where pumpkinseeds are never observed. Divergent pumpkinseed sunfish ecotypes persistently inhabit both habitats over the summer growing season [32,34] and have similar population sizes [39].

(c). Sampling

Adult fish (71–126 mm) were sampled by angling from four pelagic and four littoral sites in August 2016 (n = 50) and 2017 (n = 81) after spawning had ceased in late July to minimize any potential effects of reproduction on brain form [40]. Spatial replication here is in response to the highly spatially structured nature of this population. Pumpkinseed sunfish ecotypes display high site fidelity. Mark–recapture studies show that the probability of an individual being recaptured at its site of origin during the summer season is greater than 97% [41], and is 50% among pelagic sites and 90% among littoral sites between years [39]. Furthermore, natural movement between habitats is also less than 7% within a summer season or between years [39,41]. Thus, replication here refers to independent effects of local conditions on individual traits in fish spending significant periods of their lives at the same site. Captured fish were euthanized with an overdose of clove oil (100 ppm) and preserved in 10% buffered formaldehyde. Sampling procedures were approved by the University of Guelph animal care committee under the guidelines of the Canadian Council on Animal Care.

(d). Sample processing

Fish were blotted and weighed, and their standard length and external oral jaw width (maximum distance between maxillaries) measured using digital callipers. Following external measurements, fish were sexed via gonad examination, and heads were removed. A small incision was made in the brain case, and heads were replaced in buffered formaldehyde for an additional 24 h to allow post-fixation of brain tissue. Whole brains were then removed by dissection. Sixteen individuals from 2016 were excluded from analysis due to improper dissection, leaving a sample of 115 fish.

Digital images of complete brains were made from dorsal, ventral and lateral orientations using a Canon Powershot G10 camera connected to an Olympus SZ61 dissection microscope. Consistent lateral orientation was achieved by ensuring that both sides of the large bilaterally symmetrical lobes (optic tectum and telencephalon) were vertically aligned. The brains were trimmed of excess cranial nerves and the spinal cord was cut at the level of the obex. Brains were blotted to remove excess formaldehyde and weighed using an Accu-124D scale (Fisher Scientific) at a resolution of 0.0001 g.

The linear length, width and depth of each region (cerebellum, optic tectum, telencephalon, olfactory bulb and hypothalamus) were measured from digital images using the Neurolucida quick line measurement function at a resolution of 0.0001 mm (MBF Bioscience, Williston, VT). Only the left side of the brain was photographed, so the depth of both lobes of bilaterally symmetrical brain regions was assumed to be the same. The volume of each region was estimated using the ellipsoid formula V = (L × W × H)π/6, following reliability analyses by White & Brown [42]. Two individuals collected in 2016 were excluded from analyses of the olfactory bulb and PCA due to damage during dissection to that region (n = 113). Observer bias was avoided during brain preparation and measurement by using labels containing no information on capture location.

(e). Statistical methods

Relationships between brain mass or brain region volume and explanatory variables were evaluated with linear mixed-effects models. Body size was estimated from standard length (SL) because it is unaffected by body condition that can influence mass. Brain mass, brain region volumes and standard length were all log-transformed to meet the assumption of residual normality. Habitat use was assessed at two scales with separate variables. A factorial sample source habitat variable (littoral versus pelagic) indicated capture habitat. Because individuals may also specialize somewhat on benthic or pelagic resources within habitats, microhabitat use was quantified by residual variation in a continuous resource use trait, oral gape width scaled to body length [43,44]. Fish that tend to feed more benthically have larger jaws relative to body size, allowing them to consume larger macroinvertebrates [45].

The final whole brain mass model included standard length, capture habitat, sex, residual gape width and collection year as fixed effects, and capture site (within habitat) as a random effect. An interaction between standard length and habitat was included in an initial analysis to test for differences in brain allometry between the habitats, but was not included in the final model as it was not a significant predictor, nor did it increase model fit. Other interactions between fixed effects were not included in the mixed effects model because specific predictions relating to brain size were not available, and because exploratory analysis indicated that the best model based on Akaike's information criterion excluded these interactions. To evaluate possible differences in energy intake between the habitats and resulting differences in body condition that may affect brain size, we also fit a mixed effects model relating body mass to standard length, capture habitat, residual gape width and sex as fixed effects, treating capture site as a random effect.

The models for brain region volumes were the same as that for whole brain mass, except that whole brain mass was used as a size covariate instead of standard length. We also evaluated the influence of explanatory variables on the covariation among brain region sizes by extracting principal components of the five brain region sizes. A correlation matrix-based PCA revealed a first principal component accounting for 78% of total variation in brain region size with all five regions making almost equal contributions. PC1 was also highly correlated with brain mass (r = 0.95). This indicates that PC1 reflects variation in region size due to whole brain size, which we do not pursue because this is redundant with our analysis of whole brain mass. PC2 accounted for an additional 10.5% of total variation dominated by variation in the volumes of olfactory bulb and cerebellum, and PC3 accounted for 5.3% of total variation dominated by cerebellum and hypothalamus volumes (all other principal components accounted for less than 5% of total variation; electronic supplementary material, appendix 1). The PC2 and PC3 scores of individuals were evaluated separately in the same mixed-effects model as described for individual brain regions above.

3. Results

(a). Brain mass

Brain mass was greater in sunfish caught from the littoral habitat compared with fish caught from the pelagic habitat after accounting for the effects of other variables in the model (figure 1a; source habitat estimate ± s.e. = −0.081 ± 0.016, t₆ = −5.19, p = 0.002). The mean brain mass was, on average, 8.3% greater in littoral fish after accounting for other model effects (adjusted littoral mass = 0.118 g, pelagic = 0.109 g). Furthermore, brain mass increased with residual gape width after accounting for an effect of habitat (0.031 ± 0.012, t₁₀₃ = 2.63, p = 0.01), indicating that fish with larger mouths for their body size within each habitat also possessed larger brains (figure 1b). Brain mass was not influenced by any other model variable (table 1). There was no evidence that body mass, adjusted for standard length, differed between the habitats (1.59 ± 1.73, t₇₀ = −0.92, p = 0.39), suggesting that energy allocation was broadly similar in fish from both habitats.

Figure 1.

Whole brain mass variation after adjusting for other factors in the mixed-effects model. Red symbols indicate littoral habitat and blue symbols indicate pelagic habitat. (a) The relationship between adjusted brain mass and standard length (both log-transformed), along with the linear fits for each habitat. (b) The relationship between brain mass (log-transformed) and gape width residuals, along with linear fits for each habitat. In both analyses, linear fits were parallel (no interaction between habitat and standard length in (a): 0.13 ± 0.084, t₁₀₂ = 1.58, p = 0.12; and no interaction between habitat and gape width residuals in (b): 0.0028 ± 0.025, t₁₀₂ = 0.11, p = 0.91).

Table 1.

Mixed-effects model results predicting brain size, brain region volume or principal component brain region volume scores based on models that included body size as a covariate (SL, standard length; or brain mass), source habitat (littoral versus pelagic), sex, oral jaw size that reflects resource use (residual gape width (GW) after adjusting for standard length) and year (2016–2017). A positive effect for habitat indicates an increased value of the response variable in the pelagic habitat; for sex, an increased value in males; for GW residual, an increased value with larger gape width; and for date, an increased value in 2017. Significant p-values are in italic type.

response variable	variable	estimate	t-value	d.f.	p-value
brain size
brain mass	SL habitat sex GW residual year	1.26 ± 0.043 −0.081 ± 0.016 −0.0035 ± 0.013 0.031 ± 0.012 0.0094 ± 0.014	29.4 −5.19 −0.27 2.63 0.69	103 6 103 103 103	<0.0001 0.002 0.78 0.001 0.49
univariate brain region size
cerebellum volume	brain mass habitat sex GW residual year	1.1 ± 0.074 0.020 ± 0.046 −0.020 ± 0.029 −0.021 ± 0.027 −0.047 ± 0.031	14.7 0.41 −0.66 −0.77 −1.51	103 6 103 103 103	<0.0001 0.69 0.51 0.44 0.13
optic tectum volume	brain mass habitat sex GW residual year	0.95 ± 0.035 −0.016 ± 0.017 −0.013 ± 0.014 −0.0095 ± 0.013 −0.010 ± 0.015	26.8 −0.98 −0.92 −0.72 −0.52	103 6 103 103 103	<0.0001 0.37 0.36 0.47 0.52
telencephalon volume	brain mass habitat sex GW residual year	1.1 ± 0.043 −0.023 ± 0.021 0.014 ± 0.017 −0.040 ± 0.016 −0.024 ± 0.018	25.9 −1.1 0.80 −2.5 −1.33	103 6 103 103 103	<0.0001 0.32 0.42 0.014 0.18
olfactory bulb volume	brain mass habitat sex GW residual year	0.75 ± 0.089 0.023 ± 0.049 −0.015 ± 0.036 0.015 ± 0.033 −0.13 ± 0.038	8.4 0.46 −0.41 0.44 −3.30	101 6 101 101 101	<0.0001 0.66 0.68 0.66 0.0013
hypothalamus volume	brain mass habitat sex GW residual year	1.03 ± 0.064 0.047 ± 0.030 0.024 ± 0.025 0.023 ± 0.024 −0.065 ± 0.027	16.2 1.57 1.00 1.00 −2.41	103 6 103 103 103	<0.0001 0.22 0.43 0.33 0.046
multivariate brain region size
PC2	BM habitat sex GW residual year	−0.54 ± 0.35 0.1 ± 0.21 0.084 ± 0.14 0.12 ± 0.13 −0.47 ± 0.15	−1.55 0.46 0.61 0.9 −3.21	101 6 101 101 101	0.12 0.66 0.54 0.37 0.0018
PC3	BM habitat sex GW residual year	−0.34 ± 0.25 −0.049 ± 0.15 −0.13 ± 0.1 −0.052 ± 0.093 −0.081 ± 0.11	−1.34 −0.32 −1.25 −0.55 −0.75	101 6 101 101 101	0.18 0.76 0.21 0.58 0.45

(b). Brain region volumes

None of the five region volumes were related to capture habitat after statistically accounting for the effects of the other variables in the model (table 1). Furthermore, only telencephalon volume was negatively related to gape width residuals (−0.40 ± 0.016, t₁₀₃ = −2.5, p = 0.014), meaning a larger relative telencephalon volume was associated with a smaller gape width regardless of habitat. The volumes of the olfactory bulb and hypothalamus and the PC2 scores were all greater in 2016 than 2017 (table 1), possibly due to less precise brain dissections in 2016 that subsequent inspection revealed had damaged the edges of the hypothalamus and olfactory bulbs. Variation in brain region PC2 and PC3 were also unrelated to capture habitat or residual gape width (table 1), indicating that covariation of brain region volumes was unrelated to capture habitat or microhabitat use. Additionally, sex had no effect on brain region size or any of the principal components. An effect of collection year on PC2 was also probably due to the dissection effects in 2016 noted above.

4. Discussion

We found that during the summer growing season, relative brain size was greater in pumpkinseed sunfish that make use of the littoral habitat than those that use the pelagic habitat, but that the proportional size of individual brain regions did not consistently differ between fish from these lake habitats. These data suggest that fish living in the littoral habitat may face increased cognitive requirements that require a larger brain. A lack of brain region differences between littoral and pelagic sunfish either suggests that specific cognitive requirements do not differ between these habitats, or that habitat-related differentiation of brain region size is constrained in these sunfish. Our observational study cannot discriminate whether these differences in brain size reflect plastic responses, constitutive genetic differences or their combined effects.

(a). Brain size variation

Three explanations may account for differences in brain size between sunfish from these lake habitats: differences in energetics, in head size or in cognitive requirements. Brains are energetically costly organs [10–12], so fish that consume less food may have smaller brains. We considered and rejected the idea that the smaller size of brains in fish from the pelagic habitat could be a consequence of less available energy because body condition (mass adjusted for standard length) did not differ between adults from these habitats. Furthermore, a prior study indicates no mean difference in adult ecotype condition between habitats [37], and young of year juveniles have better growth over their first summer season in the pelagic compared to the littoral habitat (B.W.R. 2015, unpublished results), indicating potentially greater energy availability in the pelagic habitat, at least for juveniles. Hence, the smaller brain size of the pelagic sunfish ecotype seems unlikely to reflect less available energy.

A second explanation for brain size variation among ecotypes is that head size variation constrains brain size [46,47]. We considered and rejected the idea that adult pelagic sunfish brains are smaller due to smaller head size (correlated with smaller jaws). First, there is no evidence that braincase size differs between sunfish ecotypes; the differences observed in head size are limited to the buccal region (see [31,33]). Additionally, and consistent with previous observations in fish [22], a significant amount of the adult sunfish braincase of both ecotypes is not occupied by the brain, but filled with lipid tissue

Lastly, differences in brain size between habitats could reflect differences in the cognitive requirements of living in each habitat. Two observations support this explanation. First, brain size was consistently associated with habitat use both at between- and within-habitat scales. Individuals inhabiting the inshore littoral habitat and individuals that make greater use of littoral–benthic resources within each habitat (as indicated by oral jaw size) tended to have larger brains than individuals living in the pelagic habitat or more likely to use pelagic prey. Second, larger whole brain size was associated with the more spatially complex littoral habitat, consistent with functional predictions relating cognitive requirements to habitat complexity in vertebrates [14,48].

Furthermore, these ecotype differences are consistent with previous work that suggests differences between littoral and pelagic habitats can affect brain size. The 8.3% difference in relative brain size between pumpkinseed sunfish ecotypes is consistent with or slightly greater than differences found between fish species that occupy these habitats [6,24]. Variation in brain size is also comparable to evolved responses observed under artificial selection for large and small brains in guppies (Poecilia reticulata) [49], as well as developmental responses in brain size to acute oxygen manipulation in African cichlids (Pseudocrenilabrus multicolor) [50]. This suggests that habitat-specific differences in cognitive requirements or environmental conditions that influence brain size can have strong effects among individuals within a species.

Studies of intraspecific variation in relative brain size as a response to habitat variation are limited, highlighting the novelty of our results. Moran et al. [51] and Ahmed et al. [52] found no consistent relationship between habitat and relative brain size in Mexican tetra (Astyanax mexicanus) and threespine stickleback (Gasterosteus aculeatus), respectively. Similarly, Evans et al. [53] found no difference in brain size within species pairs of lake whitefish (Coregonus clupeaformis). Intraspecific differences in brain size between spatially separated populations of fish have been reported in other studies [54–56]. Gonda et al. [57] found that freshwater pond-dwelling nine-spine sticklebacks (Pungitius pungitius) had larger brains (adjusted for body size) than marine sticklebacks, and they emphasized that multiple proximate mechanisms could generate this diversity (discussed below). As far as we know, our study is the first to find a relationship between habitat and relative brain size variation between ecotypes that coexist in a single lake population of fish.

It is important to recognize that we cannot distinguish between three proximate mechanisms that could generate a functional association between habitat and brain size among ecotypes. First, individuals could select habitats based on functional variation in brain size or other correlated traits. Second, diversifying selection acting on variation in brain size among individuals in each habitat could favour different optimum brain sizes in each habitat. If heritable variation in brain size were available, then differences in brain size could evolutionarily diverge between ecotypes. Third, brain size may be phenotypically plastic [56,58–60], raising the possibility that individual sunfish may develop brain characteristics that improve the functional match to local conditions over their lifetimes. Common garden, plasticity and selection studies could distinguish among these proximate mechanisms, although it is possible that all three mechanisms can contribute to phenotypic variation in nature [36]. We also do not know whether brain size variation influences ecologically relevant cognitive ability in this system (e.g. [49]). Nevertheless, the uncertainty around the relative importance of the proximate mechanisms shaping brain size variation and their cognitive consequences does not detract from our expectation that habitat-specific cognitive requirements probably contribute to differences in brain size between sunfish ecotypes in this population.

(b). Brain region size variation

A key observation here was of little evidence for an effect of habitat on the proportional size of brain regions. There are three possible explanations for this. First, brain region size differences are much more subtle than we are able to detect. However, there is no reason to expect that a finer resolution of region sizes would detect differences when we observe such large-scale differences in whole brain size. Second, the specific cognitive requirements of the different brain regions examined may not differ between these two lake habitats. Instead, there could be characteristics of the different habitats that select for greater overall cognitive abilities, such as structural complexity, which would result in larger relative brain size without favouring specific cognitive abilities rooted in different brain regions. This seems unlikely given that previous work has found evidence for different specific cognitive requirements for fish living in different aquatic habitats [6,22,24]. Third, independent changes in fish brain region sizes could be constrained, a possibility that has been tested, but with no consistent result [1,61,62]. This constraint could be caused either by neuro-developmental processes such as limited ability to target growth hormones to particular brain regions or by evolutionary processes such as limited standing genetic variation in or genetic correlations among brain region volumes that could constrain rapid evolutionary divergence. In both constraint scenarios, brain region differences functionally affect sensory processing, cognition or behaviour, but increasing the size of a particular brain region generates a correlated increase in whole brain size because of the constraint. We cannot yet distinguish between these explanations for the absence of brain region differences between ecotypes.

A notable exception to the lack of effects of habitat on brain region volumes here was the negative relationship between relative telencephalon volume and gape width, suggesting that the telencephalon may be important for feeding in the pelagic habitat. It is possible that this reflects a type 1 error, because we performed a large number of non-independent tests on five related brain regions. However, prior work suggests that the telencephalon may be involved with environmental space use in fishes [23,61,63–66]. Our results are consistent with Wilson & McLaughlin [67], Park & Bell [65] and Gonda et al. [57], who all found larger telencephalon in fish more likely to forage for prey in the water column. A larger telencephalon may be advantageous when feeding on zooplankton prey in the water column due to the three-dimensional nature of the pelagic habitat, although we found no evidence of larger telencephala in pelagic sunfish overall. Interestingly, Edmunds et al. [68] found smaller telencephala in fish species using more pelagic compared to littoral resources in the same lake. The inconsistency of these results warrants further studies of telencephalon function in fishes.

(c). Brain, behaviour and adaptive divergence

A long-standing evolutionary hypothesis proposes that behavioural diversification is critical to ecological diversification when behaviour allows organisms to colonize and persist in novel environments where diversifying selection can subsequently operate on other ecologically important traits [2,69,70]. Under this scenario, variation in brain size and morphology that is causally linked to ecologically important behavioural variation should diversify simultaneously with ecological divergence, as observed here. However, a novel feature revealed here is that colonization of the pelagic habitat from the ancestral littoral habitat by sunfish may be facilitated by reducing the cognitive and energetic costs of ecologically important novel behaviours.

5. Conclusion

Evaluating the functional links between variation in brain size and habitat use between individuals at an early stage of ecological divergence within a species is uncommon. Our observations from a recent divergence of pumpkinseed sunfish indicate that the lake habitat in which an individual reside as well as foraging preferences of individuals within that habitat are associated with variation in brain size. This suggests that variation in brain size may be an adaptive evolutionary or functional plastic response to conditions that distinguish littoral from pelagic habitats. Specifically, the littoral habitat was associated with larger brains than the pelagic habitat, suggesting that littoral–benthic conditions are likely to be a more cognitively challenging environment [27]. Individual brain region sizes were not related to habitat, possibly indicating constraints that limit the ability of regions to change in size independent of other regions, at least over the relatively short 12 000-year post-glacial time scale over which these sunfish ecotypes have ecologically diverged. The intraspecific differences in total brain size described here echo those between some fish species and so strongly encourage further tests of functional links between brain size, cognition, ecology and adaptive divergence in lake fishes.

Ethics

Sampling procedures were approved by the University of Guelph animal care committee under the guidelines of the Canadian Council on Animal Care.

Data accessibility

Data are available on the Dryad database: http://dx.doi.org/10.5061/dryad.709ch3g [71].

Authors' contributions

C.J.A. participated in all aspects of study design, field collecting, sample processing, statistical analyses and drafted the manuscript. F.L. and B.W.R. contributed to study design, data analysis and manuscript preparation.

Competing interests

We declare no competing interests.

Funding

This research was supported by separate Discovery grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) to B.W.R. and F.L.