Evolution of spatial memory and the hippocampus in food caching birds
Animal species present a great diversity of brain sizes and cognitive skills and it is likely that both have been shaped by natural selection. The question of cognition and brain evolution has attracted much attention and behavioral ecologists employed comparative analyses trying to unravel how different cognitive skills and their underlying neural substrates might have evolved. This approach, which is often called neuroecology, implies that using a comparative method it is possible to investigate the evolution of functions and their underlying mechanisms and it should also allow understanding of functions of specific areas of the brain. In particular, a strong emphasis of many studies has been on the relationship between spatial memory and the hippocampus, an area of the brain involved in spatial memory processing and food-caching birds were at the center of these investigations.
The first two groundbreaking studies by Krebs et al. (1989) and by Sherry et al. (1989) suggested that food-caching birds have relatively larger hippocampal volume compared to non-caching birds. These studies formed an adaptive specialization hypothesis that states that increased selection pressure on spatial memory needed to retrieve numerous food caches have resulted in enhanced memory and an enlarged hippocampus in food-caching birds. In a broader sense, the adaptive specialization hypothesis suggests that any ecological condition providing high demands for certain cognitive skills should result in the evolution of these skills along with their underlying neural mechanisms.
I have been pursuing two lines of investigation pertaining to the general question of evolution of spatial memory and the hippocampus: (1) intraspecific comparisons and (2) interspecific comparisons of species from Eurasia and North America.
If this adaptive specialization hypothesis is correct, we should be able to show that animals which have evolved under higher demands for better spatial memory (e.g. food-caching species) should have a better spatial memory and a larger hippocampal formation compared to species which do not experience such high demands for spatial memory. The available data, however, are quite controversial. Some studies do provide results that support the adaptive specialization hypothesis (e.g. Balda and Kamil, 1989; Kamil et al., 1994; Olson et al., 1995; Biegler et al., 2001), but overall the results have been mixed (reviewed by Shettleworth, 1995, Bolhuis & MacPhail, 2001; Brodin & Lundborg 2003). Recent comparative analyses and reviews have argued that there is no convincing evidence that the adaptive specialization hypothesis is true because there is no clear pattern relating food-caching behavior to better spatial memory and an enlarged hippocampus (Bolhuis & MacPhail 2001; MacPhail & Bolhuis 2001; Brodin & Lundborg 2003). The later comparisons, however, again claimed that there is a relationship between food caching and the hippocampus volume (Lucas et al. 2004; Garamszegi & Eens 2004), but concluded that there is a significant difference between the continents and that North American birds have a relatively smaller hippocampus and smaller brains (Lucas et al. 2004).
It appears that some problems with comparative studies concern ecological differences that are unrelated to food caching behavior. Reliance on stored food is confounded with many other factors such as sociality, territory size and diet, which differ among the species compared. For example, in contrast to specialist feeders that only cache seeds, species such as the western scrub jay that cache both perishable and non-perishable foods may need to remember the relative time since caching, and the contents of their caches (Clayton & Dickinson, 1998, 1999). When comparing species with different diets and ecologies, reliance on stored food per se may not provide an accurate reflection of the memory demands imposed by food caching as memory is also used for many other purposes.
An alternative analysis that could be effective in testing the adaptive specialization hypothesis and whether memory and the hippocampus respond to environmental conditions that place high demands for spatial memory should concentrate on intra-specific comparisons of reproductively isolated populations of food-caching species living under different ecological demands (Pravosudov & Clayton 2002). Many food-caching species are non-migratory and occupy extremely broad geographic ranges and comparisons can be made within several species, where northern and southern populations may have different degrees of reliance on cached food due to differences in environmental conditions, such as day length available to forage, ambient temperature, snow cover, length of winter season, availability of naturally available food, etc. Significance of cached food can be drastically influenced by environmental conditions. Energetically demanding northern conditions place higher demands on cached food and on efficiency of memory-dependent cache retrieval. Thus, better spatial memory should have a stronger impact on survival probability of animals living in northern, more ecologically demanding environments compared to animals living in milder southern environments (Pravosudov and Clayton 2002). Showing that differences in memory exist within different populations of the same species would provide support to the hypothesis that heavier reliance on food caching is related to better memory and relatively larger hippocampus and it could indicate possible evolutionary path for changes in memory and the hippocampus.
In my preliminary study, I compared northern and southern populations of the same species, the black-capped chickadee (Poecile attricapilla), in which the main difference between the populations lay in significance of cached food based on the difference in the local environment (Pravosudov & Clayton 2002). Reliance on stored food should be much greater for individuals living in harsher northern environments where failure to recover food caches in the winter is more likely to result in death from starvation (Pravosudov & Grubb, 1997; Pravosudov & Lucas, 2001). I found that in identical laboratory conditions wild-caught black-capped chickadees from Alaska had significantly larger hippocampi with more neurons compared to Colorado conspecifics (Pravosudov & Clayton 2002). Alaska chickadees also cached more food, showed more efficient cache recovery and performed better on a spatial memory tests compared to chickadees from Colorado. In contrast, chickadees from Alaska and Colorado performed similarly in a color version of an associative learning task which is hippocampus-independent. However, just as with the inter-specific analyses of the relationship between the hippocampus, spatial memory, and the degree of reliance on cached food, it is crucial to document that the same pattern exists between populations experiencing different ecological conditions within other food-caching species that might differ in other aspects of their biology but should be similar in their reliance on food caches.
CONTINENTAL INTERSPECIFIC COMPARISONS:
Lucas et al. (2004) re-analyzed the data in Brodin & Lundborg (2003) and discovered that the relationship between food-caching specialization and hippocampal volume becomes significant if a continent effect was added as an independent variable. Lucas et al. (2004) concluded that North American corvids and parids have significantly smaller relative hippocampal and telencephalon volume compared to Eurasian corvids and parids. Garamszegi & Eens (2004) made another large-scale comparison by adding more non-caching species to the database and reported that food-caching species have larger hippocampi and larger brains than non-caching species without using a continent as a variable. Later, Garamszegi & Lucas (2005) reported that continental differences exist even in the larger data set and, even more surprisingly, they reported that such continental differences are not limited to food-caching birds but exist among non-caching birds as well. Garamszegi & Lucas (2005) concluded that there might be large ecological differences between North American and Eurasian species that might have caused evolution of different brain sizes on these two continents. This would be fascinating and groundbreaking news if it proved to be true. However, before we assume that such an important conclusion is true we must validate the data on it is based.
Careful analyses of the data reveals that (1) with a few exceptions, most species in all comparative studies are represented by one or two data points, (2) data for many species comes from different laboratories and (3) only a handful of species were analyzed by the same laboratory. Thus it appears possible that these analyses might produce spurious results because of numerous and significant differences in brain collection, processing and analyses.
Garamszegi & Eens (2004) and Garamszegi & Lucas (2005) claimed that they ruled out the possibility that the differences between the laboratories in brain tissue preparation and processing provided a serious bias in their results. There are, however, many steps in brain processing that might cause significant variation in brain measurements between the studies. (1) The time of brain collection. None of the large-scale comparative studies controlled for the time of brain collection yet it is possible that both hippocampal and telencephalon volume undergo significant seasonal variation (e.g. Smulders et al. 1995, 2000). (2) The type of fixative. To fix the brain tissue some labs use formalin whereas other labs use paraformaldehyde. The type of fixative may greatly affect tissue shrinkage and might be a source of significant variation. (3) Timing of perfusion and post-fixation period. Different labs use different time for perfusions and for post-fixation. The length of time in fixative may also greatly influence tissue shrinkage. (4) Tissue processing. Some labs embedded brain tissue in paraffin while others simply use frozen tissue. Again, this may have a tremendous effect on tissue shrinkage. (5) Analyses. The boundaries of the hippocampal formation are not defined too clearly and are subject to an observer bias. It is critically important to do inter-observer reliability analyses to insure that the measurements are consistent across the studies. None of these sources of variation have been controlled for in the large-scale comparative studies, but Brodin & Lundberg (2003) did discuss some of these potential sources of variation. However, these differences between the labs might easily result in significantly different results.
Strikingly, only a few species were repeatedly measured in several different labs to provide sufficient data for consistency in measurements and these measurements provide fairly large variance. For example Garamszegi & Eens (2004) used mean hippocampal volume of 26 mm3 for European willow tits (Parus montanus) and 12.8 mm3 American black-capped chickadees (Poecile atricapillus). However, Smulders et al. (1995) for instance, reported that in October the mean hippocampal volume of adult black-capped chickadees was 21.97 mm3, which is much closer to the mean hippocampal volume of willow tits (26.0 mm3). Looking at data in Cristol (1996), it seems that hippocampal volume of willow tits in his study ranged from 12.2 to 21 mm3 (recalculated from Log transformed data), which is much smaller than the mean value using in comparative analyses (26.0 mm3). Data on black-capped chickadees in MacDougall-Shackleton et al. (2003) also suggests larger hippocampi (range 12-19 mm3) than reported in Garamszegi & Eens (2004). Finally, Barnea & Nottebohm (1996) reported drastically different estimates for the hippocampal volume in black-capped chickadees – around 4.5 mm3. There is also a fairly large variance in telencephalon measurements. Very limited data on neuron numbers in the hippocampus also contradict the conclusion that European willow tits and North American black-capped chickadees are strikingly different. The total number of hippocampal neurons presented in Cristol (1996) for willow tits (appr. 1,600,000) is quite similar to the number of hippocampal neurons in black-capped chickadees from Alaska (appr. 1,400,000) in Pravosudov & Clayton (2002) or from New York (appr. 1,400,000) in Smulders et al.(2000).
Most dramatic example of between-the-lab-differences is the study of North American corvids by Basil et al. (1996). All data on North American corvids included in published comparative analyses on Eurasian and North American species comes from a single study by Basil et al. (1996). Newly available data on western scrub-jays, Aphelocoma californica (Pravosudov & de Kort 2006), on the other hand, showed that hippocampal and telencephalon volume in these birds was at least twice as large compared to those reported in Basil et al. (1996) (Fig. 1). Furthermore, the relationship between the hippocampus and the telencephalon volumes was also significantly different in the new analysis compared to what was reported in Basil et al. (1996). Basil et al. (1996) found that relative hippocampal volume of western scrub-jays was smaller than that of Clark’s nutcrackers (Nucifraga columbiana) whereas Pravosudov & de Kort (2006) showed a significantly opposite result. This example clearly demonstrates that the differences between the labs in brain processing and measurements could significantly change the patterns of the hippocampus-telencephalon, hippocampus-body mass and telencephalon-body mass relationships. Data in Pravosudov & de Kort (2006) show that western scrub-jays have one of the largest relative hippocampal volumes and largest total brain volumes among all studied corvids including Eurasian species (Fig. 3). The differences between the data in Basil et al. (1996) and in Pravosudov & de Kort (2006) most likely originate from the differences in tissue preparation. Basil et al. (1996) embedded brain tissue in paraffin whereas Pravosudov & de Kort (2006), as well as many other labs, used frozen tissue. Embedding tissue in paraffin results in severe tissue shrinkage and it seems that the estimates of paraffin-embedded tissue are incompatible with estimates made with frozen tissue. Pravosudov & de Kort (2006) suggested that such shrinkage is not proportional and thus data on all corvids in Basil et al. (1996) should not be used in comparative studies using other species. Data on North American corvids from Basil et al. (1996) were a significant part of all large-scale comparisons leading to a conclusion that North American birds have relatively smaller hippocampus compared to Eurasian birds. Removing data of Basil et al. (1996) from these analyses might invalidate some of the conclusions made by Lucas et al. 2004 and Garamszegi & Lucas 2005. Data in Pravosudov & de Kort (2006) also show that differences between the labs could potentially lead to large differences not only in hippocampal measurements but also in measurements of the telencephalon.
Another critical issue is that hippocampal volume might not be the best predictor of spatial memory abilities (Roth & Dicke 2005a,b). Roth & Dicke (2005a,b) argued that neuronal features including the total number of neurons would be much better predictors of cognitive traits. Whereas hippocampal volume and neuron numbers usually correlate (e.g. Healy et al. 1996; Pravosudov & Clayton 2002) there are studies that detected differences between the groups only in neuronal density but not hippocampal volume (Cristol et al. 2004). Thus measuring both volume and the total number of neurons should be the standard for all comparative studies (also see Jacobs et al. 2005). So far, there have been absolutely no large-scale comparative studies using the total number of neurons and these data are largely absent for most of the bird species (Reader et al. 2005) leaving a large gap in our understanding of evolution of neural substrates for cognitive abilities including spatial memory.
Thus it is important to measure both hippocampal volume and the total neuron number in several key food-caching and non-caching species from Europe and North America in the same lab to provide an unbiased test of the two important hypotheses: (1) do food-caching birds have relatively larger hippocampal volume with more neurons compared to non-caching birds and (2) do North American birds have relatively smaller hippocampal volume with fewer neurons compared to Eurasian birds.