Differential demands on cognitive ability may be expected to result in the evolution of cognition and associated changes in underlying neural mechanisms. While most comparative studies of cognition have focused on volumetric brain measurements, it remains unclear whether neuron morphology, which appears to be directly linked to cognitive functions, may be responsive to differential selection on cognitive ability.
Food-caching birds rely on caches to survive winter and use spatial memory to recover previously stored food. Birds in more harsh winter climates have been hypothesized to be more dependent on cached food, and therefore, their winter survival may be expected to be more memory-dependent relative to their conspecifics from the milder winter climates. Here, we show that neuron soma size in the hippocampus, a brain region involved in memory function, exhibits significant population variation associated with different environmental pressures on spatial memory related to differences in winter climate harshness in two species of food-caching chickadees. Comparing ten populations of black-capped chickadees and three populations of mountain chickadees along a gradient of winter climate harshness, we found that birds from harsher environments had significantly larger hippocampal neuron soma sizes.
Using chickadees from the two most divergent populations reared in a laboratory environment, we showed that these differences appear to be at least partly heritable as significant differences between these populations remained in birds sharing the same laboratory environment. At the same time, laboratory-reared birds had significantly smaller neuron soma size compared with the wild-sampled birds, suggesting that at least some variation in neuron soma size may be due to environment-related plasticity.
Our data suggest that environment-related selection on memory may generate differences in neuron morphology, which appear to be controlled by some heritable mechanisms and likely underlie population differences in spatial memory.
Cognition and variation in cognitive traits are likely to play an important role in evolutionary processes (Dukas 2004). Environmental differences that generate differential selection pressures on cognitive traits might be expected to result in the evolution of cognitive differences along with differences in the underlying neural mechanisms of these traits (Sherry 2006). At the same time, enhanced cognition can allow animals to successfully invade new environments, as well as to survive perturbations associated with changing environments (Sol et al. 2005). As cognition is a product of neuronal processes (Dukas 2004), selection on cognitive traits is expected to produce some changes in the neural structures supporting those cognitive traits. Much of the research on the evolution of cognition, and especially comparative studies of cognition, relies heavily on volumetric/size measurements of the brain or some specific regions of the brain to understand the evolution of enhanced cognition (Healy & Rowe 2007; Roth et al. 2010). Yet, it is frequently unknown which specific neurobiological mechanisms can produce these volumetric changes (Healy & Rowe 2007; Roth et al. 2010).
Recent studies argue that volumetric variation of the brain does not scale well with variation in cognitive abilities and that the total number of neurons appears to be a much more accurate predictor of cognitive ability (Herculano-Houzel 2011, 2012). Indeed, cognitive functions such as learning and memory appear to be based on the synaptic activity of specific neurons (Kandel & Schwartz 1982; Reijmers et al. 2007; Chen et al. 2012), yet evolutionary comparative studies rarely investigate whether and how specific selection pressures might be associated with differences in neuron morphology.
Food-caching species have been used extensively to investigate how environmental pressures related to food caching are associated with evolutionary changes in spatial learning abilities and the hippocampus, a brain region known to be involved in spatial learning (Krebs et al. 1989; Sherry et al. 1989). While initial studies compared the relative hippocampal volume of food-caching and non-caching species (Krebs et al. 1989; Sherry et al. 1989), more recent approaches involved within-species population comparisons (Pravosudov & Clayton 2002; Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011; Freas et al. 2012; Roth et al. 2012).
Many food-caching species exist across large gradients of environmental conditions, which were hypothesized to be associated with differential dependence on food caches and thus on spatial memory used for cache retrieval (Pravosudov & Clayton 2002). Most food-caching species are non-migratory and winter in temperate environments, which exhibit large differences in winter climate severity on both small and large geographical scales. Populations wintering in more severe environments with colder and longer periods of winter conditions are hypothesized to be more highly dependent on cached food than populations wintering in more moderate climates, especially in small bodied species such as chickadees (Pravosudov & Clayton 2002; Roth & Pravosudov 2009). Winter periods are characterized by lower ambient temperatures, which require more energy intake to meet potentially increased metabolic demands, yet naturally available food is unpredictable and usually in short supply. Longer winters and lower ambient temperatures increase metabolic requirements, which should result in a need for more caches and more successful cache retrieval for survival. Cache retrieval is spatial memory-dependent, and therefore, harsher environments likely produce stronger selection pressures on spatial memory, which is hippocampus-dependent. Therefore, it can be predicted that environments with more severe winter climates should favour the evolution of enhanced spatial memory, which could be achieved via neurobiological changes in the hippocampus.
Our previous studies supported this prediction by showing that more severe winter conditions are associated with enhanced spatial memory, an enlarged hippocampus, higher total number of hippocampal neurons and higher hippocampal neurogenesis rates on a large continental scale in black-capped chickadees, Poecile atricapillus (Pravosudov & Clayton 2002; Roth & Pravosudov 2009; Chancellor et al. 2011; Roth, LaDage & Pravosudov 2011; Roth et al. 2012), and on a small spatial scale along an elevation gradient of winter climate severity in mountain chickadees, P. gambelli, in the Sierra Nevada (Freas et al. 2012). Furthermore, these population differences in black-capped chickadees appear to be dependent on some heritable mechanisms, at least in some populations as shown in a ‘common garden’ experiment (Roth et al. 2012).
The total number of neurons in the brain appears to be a more accurate predictor of cognitive abilities than the volume of the brain (Herculano-Houzel 2011, 2012) suggesting that enhanced cognition in populations experiencing increased selection pressure on cognitive traits could be achieved via the larger number of neurons. However, it is equally important to understand whether evolutionary processes affecting cognition might also generate differences in neuron morphology (for example via affecting its developmental programme), yet little is known about population/species variation in neuron morphology associated with specific selection pressures.
Neurons are essential for learning, and dendritic and synaptic structures of neurons are known to be directly involved in learning and memory function (e.g. Kandel & Schwartz 1982; Kolb & Whishaw 1998). Therefore, it is plausible to hypothesize that selection on neuronal functions, such as learning and memory, could indirectly affect neuron properties involved in these functions, assuming these properties have some heritable mechanism(s).
In this study, we focused on neuron soma size, which is suggested to be associated with cognitive ability (e.g. De Voogd & Nottebohm 1981; Oberlander et al. 2004). In songbirds, seasonal changes in volume of song control brain nuclei involved in song learning are associated with changes in neuron soma area (e.g. Tramontin, Hartman & Brenowitz 2000; Thompson & Brenowitz 2005). Montagnese et al. (1993) reported that two food-caching bird species had larger calbindin-immunoreactive cells in the hippocampus compared with two non-caching species. Schizophrenia and major depressive disorder, which, among many other things, are also known to be related to reduced cognitive ability, have been associated with reduced neuron soma size in the hippocampus and in the anterior cingulate cortex in humans (Benes, Sorensen & Bird 1991; Jonsson et al. 1999; Cotter et al. 2001; Stockmeier et al. 2004). In zebra finches (Taenopygia guttata), neural aromatization resulted in enhanced spatial memory acquisition and in larger neuron soma size in the hippocampus (Oberlander et al. 2004). Larger neuron soma could contain larger cellular and metabolic machinery needed to serve a larger neuron dendritic tree, more synaptic connections and higher neuron activity (e.g. De Voogd & Nottebohm 1981; Kolb & Whishaw 1998; Oberlander et al. 2004). Larger memory ability is associated with more dendritic arborization and a larger dendritic tree (e.g. Kolb & Whishaw 1998); therefore, larger soma size may be associated with memory via supporting such increased dendritic arborization and neuron activity. Higher neuron activity is dependent on more mitochondria, which appear to originate in neuron soma (Kann & Kovacs 2007). Therefore, it might be hypothesized that neuron soma size could be reflective of memory and learning ability, with larger neuron soma supporting enhanced memory and learning ability.
The goal of this study was to investigate whether environmental differences associated with differential demands for memory-dependent food caches are associated with differences in neuron morphology in hippocampal neurons assessed by estimating neuron soma area. We used brain tissue from the birds collected for previously published studies and compared 10 populations of food-caching black-capped chickadees and 3 populations of mountain chickadees previously shown to be significantly different in spatial memory, hippocampal volume, total number of hippocampal neurons and hippocampal neurogenesis rates (Roth & Pravosudov 2009; Chancellor et al. 2011; Roth, LaDage & Pravosudov 2011; Freas et al. 2012; Roth et al. 2012). Additionally, to investigate whether any potential population differences in hippocampal neuron soma area might be due to plasticity related to local environmental differences and/or whether these differences might be controlled by some heritable mechanisms, we analysed previously collected brain tissue and compared the two most diverse populations of black-capped chickadees that were reared and maintained in the same laboratory environment (‘common garden’) and compared them with those from the wild (Roth et al. 2012). We specifically predicted that birds from more harsh environments should have a larger hippocampal neuron soma size that potentially supports their enhanced spatial memory ability and that an increased soma size in these populations is a result of increased selection on spatial memory and/or a plastic response to environmental differences.
Materials and methods
Neuron soma area measurements were newly conducted on brain tissue collected for our previous studies of black-capped and mountain chickadees (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011; Freas et al. 2012; Roth et al. 2012).
Black-capped chickadees: wild-sampled populations
We used brain tissue from 120 black-capped chickadees that were collected from 10 populations on a large continental scale along both longitudinal and latitudinal gradients of winter climate severity for our previous studies of hippocampal morphology (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011). These locations have been ranked on winter climate severity based on average air temperature over winter (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011): Alaska-Fairbanks (AKF) < Alaska-Anchorage (AKA) < Maine (ME) < British Columbia (BC) < Minnesota (MN) < Montana (MT) < Iowa (IA) < Colorado (CO) < Kansas (KS) < Washington (WA). These locations can also be roughly separated into two groups, one having harsh winter environments (AKF, AKA, ME, BC, MN) and the other exhibiting mild winter environments (IA, CO, KS, IA, WA) with MT falling in the middle as we have done in our previous analyses of population genetics, winter climate and hippocampal morphology (Pravosudov et al. 2012).
At each location, we trapped 12 birds using mist nets during late September-early October and perfused all birds in the field to avoid any potential captivity effects on the brain (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011). Brains were extracted immediately following perfusion and shipped to the laboratory at the University of Nevada, Reno where they were processed, frozen and then sectioned at 40 μm. Every 4th section was mounted on slides and Nissl-stained (for all detailed procedures see Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011).
Mountain chickadees: wild-sampled populations
We used brain tissue from twenty-four mountain chickadees that were collected at three elevations (8 birds per elevation, 2400 m, 1800 m and 1200 m) along an elevation gradient of winter climate severity in the Sierra Nevada in northern California for our previous study on the relationship between winter climate and hippocampus morphology (Freas et al. 2012). At each elevation, wild chickadees were captured with mist nets during late September-early October and sacrificed on the same day to avoid any captivity effects on the brain. Birds were perfused and their brains were extracted, processed, frozen and sectioned at 40 μm. Every 4th section was mounted on slides and Nissl-stained (Freas et al. 2012).
Black-capped chickadees: common garden experiment
To investigate whether any potential differences in hippocampal neuron soma area were due to plastic responses to immediate environmental differences and/or whether these differences were due to some potentially heritable mechanisms, we used brain tissue from 24 black-capped chickadees that were previously collected from the two extremely different populations (around Anchorage, Alaska – 12 birds and around Manhattan, Kansas – 12 birds) and reared and maintained under the same laboratory ‘common garden’ environment (Roth et al. 2012). All chickadees were collected at approximately 10 days of age before their eyes were open while the chicks were in a dark nest cavity (Roth et al. 2012). From the time of collection, chicks from both populations were hand-reared and maintained in the same common environment for their entire life until they were sacrificed for the brain analyses at approximately 8·5 months of age. Each of the 24 birds used in this study came from a different nest, and all details about hand-rearing and maintenance have been published previously (Roth et al. 2012).
All birds were perfused and their brains extracted, processed, frozen and sectioned at 40 μm following the exact procedures used to process brains from wild-sampled chickadees (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011). Every 4th section was mounted on slides and Nissl-stained (Roth, LaDage & Pravosudov 2011; Roth et al. 2012).
In addition to comparing brain tissue from these ‘common garden’ birds from two populations, we also compared them with the brain tissue from chickadees that were previously wild-caught in the same populations (Roth et al. 2012).
Estimating hippocampal neuron soma area
Neurons were identified in Nissl-stained tissue as cells that exhibited one or two darkly stained nucleoli, a clear nucleoplasm, and lightly coloured extensions of dendritic processes, (Fig. 1) using the same procedure as in our previous studies (Pravosudov & Clayton 2002; Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011; Roth et al. 2012), following widely accepted neuron identification guidelines (e.g. Barnea & Nottebohm 1994; Sherwood et al. 2006).
Neuron soma area measurements were taken from every 9th mounted section (every 36th section) starting with the second section used previously to measure hippocampal volume (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011; Freas et al. 2012; Roth et al. 2012). Overall, we measured neuron soma area on a minimum of 4 and a maximum of 6 hippocampal sections. On each section, we divided hippocampal tissue into three approximately equally sized areas and sampled one randomly chosen location within each area on both left and right hemispheres (Fig. 1). Such sampling ensured random coverage of the entire hippocampal area, but we did not focus on any specific regions of the hippocampus. Once the location was chosen, we used a frame (75 × 110 μm) to insure unbiased sampling of all neurons and measured neurons independent of their size by selecting and measuring the first five neurons within the frame by moving left to right, beginning at the left side of the frame. The criterion for selecting the first neuron was that it should either cross the left line of the frame or be located to the right of the left line. If there were fewer than five neurons within the frame, we moved the frame to the right for as long as it took to identify the five required neurons.
For each selected neuron, we traced the cell area using StereoInvestigator software (MBF Biosciences, Williston, VT, USA), which estimated neuron soma area based on the tracing. It is important to note that cells are likely not positioned in the same way within the sections, but, assuming random orientation of cells within the hippocampus, measuring the soma area across multiple randomly selected cells should accurately reflect the average area of the neuron soma. Our sampling scheme resulted in measurements of 15 neurons for each hemisphere for each measured tissue section for a total of 30 neurons for each tissue section. Overall, we estimated neuron soma area for a minimum of 120 and a maximum of 180 hippocampal neurons per bird. All of these measurements were averaged to produce one mean neuron soma estimate per bird, which was used in the statistical analyses. While there may be several neuron types in the avian hippocampus, our study did not differentiate between them. We measured all cells as long as they met our criteria for neuronal phenotype and our sampling covered the entire hippocampal formation; our measurements reflect an average neuron soma area across all neuron types and all hippocampal regions. All measurements were carried out blind to the identity of each bird by CAF.
We used previously published estimates of telencephalon volume in the same individuals (Roth & Pravosudov 2009; Roth, LaDage & Pravosudov 2011; Freas et al. 2012; Roth et al. 2012) to test for relative differences in hippocampal neuron soma area.
We also measured neuron soma in two other telencephalon regions – the mesopallium (M), across the ventricular zone from the hippocampus and the hyperpallium apicale (HA), adjacent to the hippocampus (Fig. 1) in the two populations (Washington and Alaska-Anchorage) that showed the largest differences in hippocampal neuron soma area. We applied the same methods by sampling three randomly chosen locations and sampling 5 neurons at each location. Overall, we estimated HA neuron soma area for a minimum of 90 and a maximum of 150 neurons per bird and M neuron soma area for a minimum 120 and a maximum of 180 neurons per bird.
There were no significant effects of sex in any of the comparisons, and therefore, sexes were pooled.
We used a general linear model (GLM) to test for differences between populations and treatment groups. For all comparisons, we tested for potential differences by adding telencephalon volume (to control for potential scaling of neuron soma area with the brain size) as a covariate, as well as using raw neuron area with no covariates. Variation in hippocampal volume is generally associated with variation in telencephalon volume (e.g. Pravosudov & Clayton 2002) and using telencephalon as a covariate would allow disassociating any existing population effects on hippocampal neuron soma area from any potentially confounding effects associated with overall brain size. To test for the predicted relationship between winter climate severity and hippocampal neuron soma area, we used an ordered heterogeneity test (Gaines & Rice 1990) that specifically tests for ordered associations as we have done in our previous studies (Roth & Pravosudov 2009; Freas et al. 2012).
Black-capped chickadees: wild populations along longitudinal and latitudinal gradients of winter climate severity
There was significant variation in hippocampal neuron soma area relative to the telencephalon size among ten black-capped chickadee populations over a large continental scale (F9,109 = 11·53, P <0·001; Telencephalon as a covariate: F1,109 = 21·3, P <0·001). There was also a significant relationship between winter climate severity and relative neuron soma area (Figs 2a,b and 3a, rs = 0·915, Ordered Heterogeneity Test, rsPc = 0·915, k =10, P <0·001), with birds from populations with more severe winter climates having larger mean neuron soma size relative to telencephalon volume. This relationship remained highly significant (F9,110 = 13·68, P < 0·001; rsPc = 0·96, P < 0·001) when telencephalon volume was not used as a covariate.
When we separated all ten populations into two groups containing five independent populations and sharing either a relatively harsh (Alaska-Anchorage, Alaska-Fairbanks, British Columbia, Maine, and Minnesota) or a relatively mild winter environment (Kansas, Washington state, Colorado, Iowa and Montana), there were also significant differences in the hippocampal neuron soma area (t-test, t8 = 3·89, P =0·005).
Finally, we tested whether population differences in hippocampal neuron soma size were independent of any potential differences in neuron soma size in the rest of the telencephalon by comparing the two most divergent populations (AKA and WA) and using mean neuron soma size in HA and M as covariates. Population differences in hippocampal neuron soma size remained highly significant with HA neuron soma size as a covariate (F1,21 = 41·5, P <0·001, HA neuron soma area – F1,21 = 15·09, P <0·001) and with M neuron soma size as a covariate (F1,21 = 43·42, P <0·001, M neuron soma area – F1, 21 = 4·17, P =0·054). At the same time, both HA (F1,21 = 8·19, P < 0·01) and M (F1,21 = 7·53, P = 0·012) neuron soma areas were also significantly larger in Alaska birds compared with birds sampled from milder climates in Washington state.
Mountain chickadees: wild populations along an elevation gradient of winter climate severity in the mountains
The significant relationship between more severe winter climate and larger neuron soma area was also present in mountain chickadees along the elevation gradient; larger hippocampal neuron soma area relative to telencephalon size was associated with higher elevations (Figs 2c,d and 3b, General Linear Model, F2,20 = 5·06, P =0·02, Telencephalon as a covariate, F1,20 = 5·1, P =0·03; Ordered Heterogeneity Test, rs = 1, Pc = 0·98, k = 3, P <0·01). Differences among the three elevations remained significant when telencephalon was not used as a covariate (F2,21 = 3·83, P = 0·038).
Black-capped chickadees: common garden experiment
We compared the neuron soma area of chickadees reared and maintained in a controlled laboratory environment and chickadees from the same populations (Kansas and Alaska-Anchorage) that were trapped directly from the wild (Roth, LaDage & Pravosudov 2011; Roth et al. 2012). There were significant differences between the two populations (F1,43 = 30·89, P <0·001), as well as between chickadees sampled directly from the wild and birds reared and maintained in the same laboratory conditions (F1,43 = 6·21, P = 0·016). The interaction between population (Kansas vs. Alaska) and condition (wild vs. laboratory-reared) was not significant (F1,43 = 2·06, P = 0·16), but telencephalon size was positively related to neuron soma area (F1,43 = 7·18, P = 0·01). Chickadees from Alaska had relatively larger hippocampal neuron soma areas than birds from Kansas independently of their condition (wild vs. common garden), but laboratory-reared and maintained chickadees also had smaller relative hippocampal neuron soma area than their conspecifics sampled directly from the wild (Fig. 4). These results remained significant (P <0·05) even if we did not use telencephalon as a covariate.
Our study showed that there is a significant association between winter climate severity and hippocampal neuron soma area in both black-capped chickadees on a large continental scale and in mountain chickadees on a small spatial scale along a montane gradient. Because variation in hippocampal neuron soma size mirrors predicted population differences in reliance on spatial memory, our findings suggest that larger neuron soma may subserve enhanced spatial memory. Secondly, our study showed that population differences in hippocampal neuron soma size appear to be controlled at least partially by some heritable mechanisms, because differences between black-capped chickadees from the two most divergent populations remained significant even when birds were reared and maintained in the same laboratory conditions. In addition to having some heritable basis, neuron soma size also appears plastic in response to environmental conditions, most likely as a result of potential GxE interactions, as laboratory-reared birds had smaller neuron soma size than birds sampled directly from the wild.
Compared with chickadees from Washington state, neuron soma areas in HA and M were significantly larger in Alaska birds, these locations representing the two most divergent populations showing the largest differences in hippocampal neuron soma size, suggesting that other brain regions might also be associated with differences in winter environmental conditions. However, the relationship between neuron soma size and winter climate appears to be significantly stronger specifically in the hippocampus, as the differences in the hippocampal neuron soma size between these populations were highly significant relative to the mean neuron soma area in at least two other telencephalon regions (HA and M). Alaska chickadees' hippocampal raw neuron soma area was 51% larger than that of Washington chickadees, while such difference was only 23% for HA neuron soma size and 12% for M neuron soma size.
Hippocampal neuron soma area was significantly and positively associated with the telencephalon volume suggesting that individuals with larger brains have larger hippocampal neurons. While this association is noteworthy, it was not critical for our main conclusions as the relationship between winter climate and hippocampal neuron soma area was highly significant even when raw neuron soma area was used in the analyses.
Our results provide support for the hypothesis that a harsh winter environment is associated with increased selection pressure on spatial memory needed for successful cache retrieval. It appears that such pressure likely resulted in the observed significant differences in spatial memory, hippocampal volume, total number of hippocampal neurons and adult hippocampal neurogenesis rates among different populations of two species of food-caching chickadees inhabiting different environments. This study shows that these differences extend to the hippocampal neuron morphology, with birds in harsher environments tending to have significantly larger neurons relative to the neuron soma size in the rest of the telencephalon. This is a significant finding that indirectly suggests that enhanced spatial memory in chickadees from populations under increased selection on memory-based cache retrieval might be, at least partially, subserved by neurons with larger soma size that likely provide the increased cellular metabolic machinery needed to subserve a larger neuron dendritic tree and/or more neuron activity (e.g. Kolb & Whishaw 1998).
It is interesting that larger hippocampal neuron soma size in birds from environments with a more severe winter climate is also associated with significantly higher total number of hippocampal neurons as well as with significantly higher adult hippocampal neurogenesis rates (Roth & Pravosudov 2009; Chancellor et al. 2011; Roth, LaDage & Pravosudov 2011; Freas et al. 2012; Roth et al. 2012). These data suggest that it is unlikely that larger neurons in birds from harsher environments may simply be a reflection of their relatively older age, as birds with larger neurons also appear to have higher neuron turnover rates. While we could not differentiate potentially different neuron types in the hippocampus, we specifically measured all cells that exhibited neuronal properties irrespective of their size, so our estimates should be an accurate representation of an average neuron soma size in the entire hippocampal formation.
The observed large population variation in hippocampal neuron soma size (or at least part of it) could be potentially explained by environmentally induced plasticity. All wild-caught birds in this study were sampled during early fall when environmental conditions were relatively mild for all of the compared populations, and therefore, it is unlikely that population differences in neuron soma size were simply due to immediate differences in severity of winter conditions when memory may be especially important. Nonetheless, it is possible that at least some of the population differences in neuron soma size may be due to differences in food-caching intensity, which is usually highest during the autumn months (Pravosudov 2006) and which likely involves heavy spatial memory use to encode memory for numerous cache locations (Male & Smulders 2007). This possibility is also reinforced by the data showing that birds reared and maintained in a laboratory environment had significantly smaller neuron soma area compared with the wild-sampled conspecifics from the same populations.
However, our comparison of the ‘common garden’ birds that showed significant population differences in neuron soma size despite sharing the same laboratory environment from 10 days of age suggests that neuron soma size also appears to be controlled by some heritable mechanisms involved in neuron development, which may have a genetic basis. Combined with such potential heritable mechanisms, differences between wild-sampled and laboratory-reared birds, as well as population differences in hippocampal neuron soma size, suggest that hippocampal neuron soma size is both plastic to some degree and also determined by some heritable, possibly genetic, mechanisms so that the overall variation in soma size is likely a product of some GxE interactions.
It remains plausible that not all of the reported population differences can be explained by heritability, as we performed our ‘common garden’ experiment with only the two most divergent populations and only in black-capped chickadees. As a result, it is possible that differences in neuron soma size between populations of both mountain and black-capped chickadees over small spatial scales were entirely due to environment-induced plasticity. We also could not rule out potential maternal and/or epigenetic effects because our ‘common garden’ birds were collected at 10 days of age. Nonetheless, differences between populations shown in a ‘common garden’ experiment lend more support to the hypothesis that population differences in hippocampal neuron soma size could be a result of differential selection pressures on memory associated with environment-dependent reliance on memory-based food caching (Pravosudov et al. 2012; Roth et al. 2012). Assuming that natural selection can act on memory ability, such selection can indirectly produce significant changes in the neural structures underlying memory as long as these structures are controlled by some heritable mechanisms.
It is interesting that birds reared and maintained in the same laboratory environment had significantly smaller hippocampal volume compared with the wild-sampled birds from the same populations, yet there were no significant differences in the total number of hippocampal neurons (Roth et al. 2012). The present study, on the other hand, showed significant reduction in hippocampal neuron soma size in the laboratory-reared birds, which suggests that changes in hippocampal volume were at least partially due to changes in neuron soma size and not the total number of neurons.
Overall, our results showed that increased winter climate severity is associated with an enlarged neuron soma area in the hippocampus in two species of food-caching chickadees and that population differences in neuron soma size appear to be based on some heritable mechanisms. While our data do not explain the exact mechanism of how increased neuron soma area may be related to spatial memory, it points at neuron soma as a potential mechanism involved in memory function. Our findings that population differences in neuron soma size are associated with differences in spatial memory indirectly suggest that enhanced spatial memory in birds from harsher winter environments may be subserved by hippocampal neurons with larger soma size. Because larger soma size may provide increased cellular and metabolic machinery needed to maintain larger dendritic arborization and higher neuron activity, chickadees from harsher environments may potentially be able to store more memories associated with the larger number of caches. The data reported here on hippocampal neuron soma size add significantly to the mounting evidence that environmental conditions in food-caching birds are associated with differential selection pressure on memory-dependent cache retrieval, resulting in significant modifications of the neural mechanisms, including hippocampal neuron soma size, which appears to support enhanced spatial memory. While environment and experience-induced plasticity appear to explain at least some of the observed population variation in neuron soma size, population genetics (Pravosudov et al. 2012) and ‘common garden’ experiments (Roth et al. 2012, this study) suggest that such variation is consistent with a history of natural selection. Nonetheless, future studies are needed to establish whether the observed population differences in neuron soma size have a genetic basis or whether they are caused by either maternal or epigenetic effects.
This research was partially supported by grants from the National Science Foundation (IOB-0615021 to VVP and IOB-0918268 to LDL and VVP). The research reported in this study was performed on brain tissue collected for previously published studies, which all adhered to the ASAB/ABS guidelines for the use of animals in research and all legal and institutional requirements. We thank Matt Forister for providing critical comments on the manuscript. We also thank Anders Brodin and an anonymous reviewer for providing valuable comments that significantly improved the manuscript. All necessary permits and Institutional Animal Care and Use Committee protocols were listed in these previously published studies. No new birds were collected or sacrificed for this study.