Tuesday, July 29, 2014

It’s all about the variance: science and life at N > 2

Variation is the grist for, and the flour from, the evolutionary mill. Without variation, no evolution occurs. With variation, evolution can generate even more variation by causing organisms in different environments to evolve different traits. We all know this, and we proceed accordingly in our research; but perhaps we too often take it for granted. Only sometimes are we smacked in the face by variation in such a way that it makes us pause and re-evaluate the way we view the world. Well, variation smacked me upside the head a few weeks ago during a trip into the field. In so doing, it made me reflect on how we estimate and interpret variance – and how this flavors the way we view our research and our daily experiences.

Threespine stickleback (Gasterosteus aculeatus)
Threespine stickleback could be the world’s most variable vertebrate. In some populations, average size at maturity is less than 30 mm – in others it is greater than 85 mm. In some populations, the pelvis is a huge structure – in others it is completely lacking. In some populations, the side of the fish is almost completely covered with bony plates – in others plates are entirely absent.  In some populations, mature males are almost entirely black – in others they have massive amounts of red – and in others black and red can be minimal. In some populations, the head is huge and the mouth massive – in others they are very small. In some populations, mean egg size (dry mass) is less than 0.047 mg – in others it is greater than 0.089 mg. This is just a small set of examples: stickleback, even just in freshwater, vary dramatically both within and among populations in almost any trait one cares to measure. This is why they are such a spectacular model system for studying adaptation.

A representation of stickleback diversity. The marine ancestor is surrounded by various freshwater forms.
Tom Reimchen, a professor at the University of Victoria, has long maintained that variation in stickleback on Haida Gwaii, a modest-sized archipelago off the coast of northern British Columbia, Canada, are as variable as are stickleback across the rest of their massive range in the northern hemisphere. Ever the skeptic, I have – when reviewing or editing Tom’s papers – pointed out that this assertion isn’t strictly true as (slightly) smaller stickleback are found in North Uist, Scotland. I am sure my nit was annoying to Tom as it was just a technicality and it required him inserting some rather pointless qualifiers into a few of his papers.

Several weeks ago, I had the opportunity to visit Tom in the field to see Haida Gwaii stickleback for myself. The first lake we visited was Drizzle, where Tom lived for 15 years and worked for 40. Drizzle is a modest-sized (112 ha) and heavily-stained (tannic, the color of very strong tea) lake with large and dark stickleback. A highlight here (besides camping and having a breakfast of bannock beside the lake) was walking the shoreline on Tom’s annual survey of loon-induced stickleback mortality. Several species of loon, particularly common and red-throated loons, congregate on Haida Gwaii lakes like Drizzle in numbers I had not thought possible, despite visiting countless lakes in my life. On Drizzle, dozens of loons would cruise nearby checking us out during our survey. And they would capture stickleback as if on cue – probably dozens were dispatched as we watched. Not surprisingly, many of the stickleback we found on the shore had been captured and killed, but not eaten, by loons. (Of course, many others are eaten - but we obviously can't find those on the shore.) Tom has an effective strategy for motivating search efforts. The person in front gets one point for every dead stickleback found. The following person gets two points. The third person gets three points. Tom was first, then Hannah, then me. Although it was like following two vacuum cleaners – I named one Hoover and the other Roomba – I held my own as tail-end Charlie (on points anyway).

Tom's cabin at Drizzle Lake (the lab is the wing at left).
A common loon with a Drizzle Lake stickleback.
The next lake we visited was Mayer, where Ric Moodie had – before I was born – discovered and described what is probably the world’s largest freshwater stickleback. This lake is larger (627 ha) than Drizzle, also quite stained, and even more overrun by loons. I had the good fortune, the day before meeting Tom, to happen by Mayer Lake just as it had stopped raining, in perfect time to cook my breakfast while watching 33 loons swim back and forth in front of me. Our next planned stop was Boulton Lake, in which more than half of the stickleback completely lack a pelvis, but this plan was derailed by happenstance. It seems that some delinquent and potentially dangerous kids had run off into the woods around Boulton Lake, and police parked along the highway nearby strongly discouraged us from going in.

So we instead hiked into Rouge Lake. This lake is a very shallow and small (1.5 ha) lake in the middle of a bog near the northern end of Graham Island (Drizzle and Mayer are on this same island). Rouge Lake stickleback are exceptional in several respects, especially their frequent lack of one of their dorsal spines, their (until recently) extreme red colour, their occasional possession of two dorsal fins, and the complete fixation of an otherwise locally-rare genetic variant (the Japanese clade of mitochondrial DNA). It was on the way back from this lake, tramping my way through bog behind Tom and three students, that variance smacked me upside the head. Just walking to these few lakes and hearing about (and seeing some of) their stickleback had finally brought home Tom’s assertions about the exceptional variation on Haida Gwaii and, more generally, the exceptional variation that organisms can achieve on very small spatial scales.

The Abbey Road of stickleback biology – the Rouge Lake trip. (Note: the picture is inverted for a reason that should be obvious.)
Along with this abrupt realization came a more fundamental epiphany: why had I been really impressed by the variance only after the third lake (not counting our aborted attempt to visit Boulton)? All of a sudden, I was struck by the parallel that, in statistics, we require a minimum sample size of three to get our first proper (albeit still weak) estimate of variance. The reason is that we need at least N = 2 to estimate a mean, and estimating a variance requires first estimating the mean and then needing at least one more data point. This makes sense statistically, of course, but – walking back from Rouge Lake – I began to wonder if our brains work the same way. That is, we really have to experience three things before we begin to get some mental perspective on how much they vary – because we need to consider the possibility of outliers. That is, with N = 3, we can see if any of the points stick out particularly far with respect to the mean – something that is impossible with N = 2 because in that case each point is equally distant from the mean. Stated another way, a sense of how variable things are first requires us to get a sense of the “average” or “typical” value and then a distribution of values around this average, which requires at least N = 3. Perhaps statistical principles match our mental processing machinery.

Now I can hear you saying: “Sheesh, N = 3 is way too low for a proper variance estimate.” You are, of course, correct. My point is simply that an assessment of variation, both statistically and mentally, can only begin at N = 3. Getting this third data point (visiting that third site) is the first moment when one has the potential to be impressed by that variation. Following that, much more data needs to be collected (many more sites experienced) to get a real estimate/understanding of the variance, but N = 3 is the first time you might be inspired by experience to try further. Hmmm, in writing this, I am reminded that I have only two kids. “Sweetheart, I’ve been thinking …”


Some other cool Haida Gwaii experiences:

More Haida Gwaii photos: https://www.flickr.com/photos/andrew_hendry/sets/72157645239128218/

Eagles were everywhere.
Bleeding tooth fungus - how cool is that?
Sandhill crane in the rain.
Drizzle Lake
Native tree frog.
Masset, Haida Gwaii.
River otters in a tiny rainforest stream.
Sandpiper squadron.

Thursday, July 24, 2014

Evolutionary costs, ecological currency, and baby fish

[ This post is by Darren Johnson; I am just putting it up. –B. ]

I’m just going to say it – I like cute, baby fish. As a longtime SCUBA diver, I’ve spent countless hours on reefs throughout the world, and one of my true delights is noting the arrival of baby fish. Yes, they are often adorable, but one of the most fascinating things is that sometimes there are giant schools of baby fish, and other times there are few, if any, to be seen.  The future population of adults depends on these babies, yet the replenishment of populations by new babies (a process referred to as recruitment) is notoriously variable.  In fact, it is so variable that this phenomenon has a name.

For decades, fisheries scientists used the term “recruitment problem” to describe both the challenges of understanding why recruitment varies, and the difficulty of predicting adult numbers from the abundance of earlier stages such as eggs or larvae. Relationships between the numbers of young fish and the numbers of adults that those young fish eventually become have been studied for many, many species. Of course, abundance of young matters, but a common conclusion reached by such studies is that we need to know more than just numbers to understand recruitment variability with any degree of accuracy.

You’re prettier when you’re younger… Dascyllus albisella on small patches of coral in Hawai’i. Photo credit: D. Johnson.

One reason it’s so difficult to predict recruitment is that most fish species produce huge numbers of offspring (egg numbers in the millions are not uncommon), and only a few survive to adulthood. So despite the big-eyed optimism of those cute baby fish, the average outlook is grim. As if that weren’t enough, life for baby fish is exceedingly unfair. It is well known that individual fish with certain phenotypes fare better than others. For example, large babies seem to survive much better than their smaller counterparts. Similarly, faster-growing fish often survive better too. Even when such advantages are small, they are important. For populations that typically start out in the millions, even slight variations in survival rates can result in order-of-magnitude differences in the population of adults.

Times of plenty: a school of blackfin chromis (Chromis vanderbilti) hover above a Hawaiian reef.  There’s also a yellow tang (Zebrasoma flavescens) in there. Photo credit: D. Johnson.

Although unfair for baby fish, these links between phenotypes and relative survival might offer some insight into recruitment variability. With this in mind, our recent study examined recruitment from an eco-evolutionary perspective. We wanted to know the extent to which phenotype-mediated differences in individual survival probabilities added up to affect the dynamics of whole populations. In other words, there appears to be an evolutionary cost associated with individuals having an inferior phenotype. We wanted to take these evolutionary costs (measures of selection) and convert them to ecological currency (estimates of average survival within populations). To address this question, we gathered all the studies that we could find that repeatedly measured relationships between phenotypes and relative survival of fish. We then analyzed these selection measurements in combination with observed variation in the distributions of phenotypes.

We found that most of the mortality experienced by populations of larval and juvenile fishes is selective mortality. That is, most mortality is related to variation in phenotypes such as body size, growth, etc.  In addition, the amount of selective mortality varied widely among different cohorts of the same species (e.g., different groups of fish that arrived to the reef at different times). Together, these results suggest that variation in selective mortality, rather than non-selective mortality, is the biggest source of recruitment variability.  Taking these results a step forward, it suggests that if the relationship between phenotype and survival is relatively consistent, then understanding how phenotypic variation interacts with selection might hold the key to understanding recruitment variability.

Tagging fish to measure how phenotype affects relative survival.  A juvenile bicolor damselfish (Stegastes partitus) in the Bahamas gets a tattoo.  Photo credit: Nikita Schiel-Rolle.

Toward this goal, we provide a conceptual and mathematical framework for analyzing fitness surfaces – functions that relate phenotypic value to relative survival across a broad range of phenotypes. The framework can accommodate cases in which fitness depends on multiple traits, and cases in which fitness depends on population density. We illustrate that fitness surfaces can be relatively constant, and that interactions between phenotypic variability and fitness surfaces can vastly increase our ability to explain recruitment variability.

The relationship between selection gradients and mean phenotypes can be used to reconstruct the fitness surface (solid curve in lower panel). Groups of fish whose phenotype distributions show little overlap with the fitness surface (e.g., group 1) have low rates of overall survival (and more intense selection), whereas groups with greater overlap (group 8) have greater survival (and less intense selection).

Beyond fish and recruitment, our study suggests that in many ecological scenarios (though certainly not all of them), fitness surfaces might be reasonably constant. Our study also suggests that fitness surfaces are often nonlinear, which might result in complex relationships among phenotype distributions, selection, and average fitness. For example, in some cases variation in phenotypes has a larger effect on average survival than mean phenotype does.  Understanding (and properly estimating) fitness surfaces will be critical to understanding how variation in phenotypes ultimately drives variation in the dynamics of populations.


Johnson, D.W., Grorud-Colvert, K., Sponaugle, S. and Semmens, B.X. (2014). Phenotypic variation and selective mortality as major drivers of recruitment variability in fishes. Ecology Letters 17(6), 743–755. DOI: 10.1111/ele.12273

Friday, July 18, 2014

Passport required to reproduce: Local adaptation persists despite frequent dispersal

[ This post is by Daniel Peterson; I am just putting it up.  –B. ]

Alaska contains roughly half of the wilderness in the United States. That’s over 230,000 square kilometers of pristine habitat – a place where ecosystem processes disrupted almost everywhere else can be observed in a natural state. Of course, that also means a large proportion of the state isn’t accessible by road, a fact I could barely comprehend as a brand-new grad student stepping off a plane from the East Coast. I soon found myself getting used to travelling exclusively by boat, keeping an eye peeled for bears and moose, and tying a decent bowline (that last trick learned only after the shameful loss of a Secchi disk to the deep). What I’ll never get used to is the thing that brought me there in the first place: gravelly streams full of bright red, desperately spawning salmon.

Little Togiak Lake, Wood-Tikchik State Park, Alaska.

The Alaska Salmon Program at the University of Washington has been doing research in the Bristol Bay region since 1946, well before Alaska became a state. One major focus of our work has been understanding the ecological effects of habitat diversity on Alaskan salmon stocks, now one of the world’s best examples of a productive and sustainable fishery. Hilborn et al. 2003 showed that the Bristol Bay sockeye salmon (Oncorhynchus nerka) are relatively stable in abundance and resilient to shifts in climate conditions because they are not one gigantic, panmictic population, but instead a metapopulation of independent breeding aggregations adapted to unique spawning habitat types (streams, rivers, and lake beaches). But how do these populations that are adapted to different habitats interact with each other on an evolutionary time scale? Are they on the road to speciation or does gene flow limit their divergence and possibly their adaptation to local conditions?

Despite the famous ability of pacific salmon to home to their natal spawning grounds after years in the ocean, dispersal rates among nearby populations have been measured to be 2–10%, potentially high enough to swamp the effects of ecologically divergent selection. However, even in the presence of considerable dispersal, gene flow may be limited if dispersers have low reproductive success in already-occupied habitats. Where local adaptation has arisen, dispersers between populations occupying distinct habitat types might be maladapted to their new habitat compared to philopatric (non-dispersing) individuals and dispersers between similar habitats, reducing gene flow and reinforcing local adaptation.

We set out to empirically assess the effect of local adaptation on the reproductive success of dispersers between beach- and stream-adapted populations. Beach-spawning fish, especially males, tend to be large and deep-bodied, while stream-spawning fish are more slender. Differences between these ecotypes have also been observed in other ecologically important traits such as egg size and migration timing. In order to isolate the fitness effects of local adaptation from those of dispersal itself, we compared the reproductive success of dispersers between populations that shared the same spawning habitat type with dispersers between ecologically distinct spawning habitat types.

Male and female sockeye salmon from the stream-spawning ecotype (above) and the beach-spawning ecotype (below). Photos: Tom Quinn.

To get direct measurements of individual dispersal and reproduction, we conducted exhaustive sampling of adults in two stream-spawning populations (A and C Creeks) every year from 2004 through 2010. We walked the full length of both streams every day during the spawning season (late July through late August), tagging any newly observed fish with unique IDs and noting the location of each previously tagged fish. We observed and fin-clipped a total of 4473 individuals in A Creek and C Creek in 2004 and 2005 (the parent years) and 2008, 2009, and 2010 (the years their offspring returned), plus 166 individuals that settled on the beach habitat in 2004 and 2005 (as a genetic baseline with which to identify dispersers from the beach to the streams). In the two parent years, 12% of sampled fish were immigrants (fish genetically assigned to a population other than the one in which they were sampled). C Creek had more immigrants from the other stream as well as from the beach than A Creek, but there was no clear sex bias in dispersing individuals. The number of individuals immigrating to the streams from the beach-spawning populations (N=108) was greater than the number of dispersers between stream-spawning populations (N=85).

Surveying C Creek. Photo: Jocelyn Lin.

Using pedigree reconstruction to calculate the number of returning adult offspring produced by each individual in the parental generation, we compared the lifetime reproductive success of all philopatric fish, dispersers between streams, and dispersers from adjacent lake beaches to the streams. We found that the reproductive success of dispersers between the two stream-adapted populations did not differ significantly from that of philopatric individuals, but immigrants from the beach population had significantly lower mean reproductive success than both philopatric fish and immigrants from the other stream. On average, beach-to-stream dispersers produced about one fewer offspring than between-stream dispersers, a reduction in fitness equivalent to almost half of the average reproductive success.

We don’t know the mechanistic reasons why dispersers from the beach produced fewer offspring. Morphological maladaptation to the stream environment could have limited the reproductive success of beach-adapted immigrants by reducing adult lifespan during the spawning period through selective bear predation or stranding in shallow water. Previous studies have shown that the abundant brown and black bears preferentially kill larger salmon (thereby selecting against them), but we found that dispersers from the beach were less likely to be found dead after being killed by bears and more likely to disappear without a trace. Recent PIT tagging work by Bentley et al. has shown that stream-spawning sockeye salmon show a wide variety of movement strategies, with many fish moving between stream and lake on a daily basis. It may be that the mere presence of bears indirectly affects the reproductive success of large-bodied dispersers by eliciting predator avoidance behavior and thereby limiting reproductive opportunity. Alternatively, reduced physical access to shallower areas of the stream could limit access to mates and spawning sites, encouraging the departure of larger individuals. Either way, adaptive behavioral differences between ecotypes may influence the conversion of dispersal into gene flow.

Spawning in the stream. Photo: Allan Hicks.

We usually expect that genetic differentiation, adaptive or otherwise, will only develop when diverging populations are insulated from gene flow by barriers to dispersal (either intrinsic or extrinsic). In our study, beach-to-stream dispersers were more prevalent than between-stream dispersers, suggesting that barriers to dispersal between habitat types are not strong in this system. The high fitness cost to dispersers that move between habitats might therefore be crucial to the maintenance of these morphologically and genetically recognizable stream- and beach-spawning ecotypes.

In the long term, we might expect that when dispersers have low reproductive success, selection will drive the evolution of intrinsic barriers to dispersal. However, additional factors might select against such barriers. For example, in dynamic metapopulations, rare subpopulation recolonization events might substantially bolster the long-term fitness of dispersal alleles even if dispersers have limited reproductive success in occupied subpopulations. Moreover, flexible behavior patterns in systems that allow for reversal of dispersal decisions could minimize the fitness cost of dispersal in unfavorable conditions. Thus, in many metapopulations, reduced immigrant reproductive success might be more important than barriers to dispersal for the maintenance of intraspecific biodiversity.


Peterson DA, Hilborn R, & Hauser L (2014). Local adaptation limits lifetime reproductive success of dispersers in a wild salmon metapopulation. Nature Communications, 5.  DOI:10.1038/ncomms4696

Gordon Research Conference on Speciation: March 15–20, 2015

Hi everybody.  An upcoming conference on speciation has just been announced:

Gordon Research Conference on Speciation
Modes of Diversification, Ecological Mechanisms, and Genomic Signatures

March 15–20, 2015
Ventura, California, USA

A preliminary program is up with a great list of speakers.  More information is available at the conference’s home page.  Save the date!

Thursday, July 10, 2014

A small mammal with an outsized impact

[ This post is by Craig W. Benkman; I am just putting it up.  –B. ]

When we think of species having large and disproportionate impacts on communities, animals like sea otters come to mind. By eating and depleting sea urchins, sea otters prevent urchins from eating and depleting kelp. The huge difference between having kelp forests and their diverse community of fishes, sea lions, and eagles, versus largely kelp-less barrens arises simply from contemporary ecological processes; otters directly eating urchins and indirectly facilitating the increase in kelp. Such cascading effects are thought to be widespread both on land and in sea where you have strongly interacting species like mammals. Remove the mammals and the world is different. Unfortunately, we humans are very good at that.

Although less appreciated, strongly interacting predators and herbivores potentially have strong evolutionary effects on their prey. And, in the case in which the prey dominate a landscape, such evolutionary effects could be just as profound as those found for the more commonly studied trophic cascades of otters and their like. Our research indicates that Rocky Mountain lodgepole pine (Pinus contorta ssp. latifolia), which dominates tens of millions of hectares across the northern Rocky Mountains, is one such dominant species. The evolutionary effect is the result of differential seed predation by the aptly named pine squirrel (Tamiasciurus hudsonicus, also referred to as the American red squirrel; Talluto and Benkman 2014 Proceedings of the National Academy of Sciences USA 111:9543-9548; see also our 2013 paper in Ecology 94:1307-1316).

Map of lodgepole pine from Wikipedia. Rocky Mountain lodgepole pine occurs in and dominates much of the area highlighted by light green.

Driving across Yellowstone and other Rocky Mountain locales, one is struck by the variation in lodgepole pine seedling and sapling densities in areas recovering from fire. This variation has obvious consequences for the structure of plant communities varying from sparse pine seedlings and domination by various grasses and forbs to dense growth of pines with little else. Other components of the communities such as the various animal communities (e.g., birds, mammals, and invertebrate pollinators) must also vary accordingly.

A dense carpet of lodgepole pine saplings several years after a fire. (Photo from Wikipedia)

The key question is what causes the variation in the initial pine seedling density following fire. Ecosystem and landscape ecologists Monica Turner, Dan Tinker, and their colleagues found that the best predictor of seedling density after the 1988 Yellowstone fires was the frequency of serotiny in the pre-fire forest (Turner et al. 2003, Frontiers in Ecology and the Environment 1:351-358). Serotiny occurs when woody plants in fire-prone habitat encase their seeds for multiple years in woody structures (hard woody cones in lodgepole pine) creating an arboreal seed bank that is released soon after a stand-replacing fire. When the frequency of serotiny among the pines is high, large numbers of seeds are released from serotinous cones after fire sweeps through, resulting in large numbers of seedlings. Lower frequencies of serotiny result in fewer seeds available after the fire and many fewer seedlings. Turner, Tinker and colleagues found that the density of seedlings per hectare ranged from 600 when less than one percent of the trees were serotinous to 211,000 when 65% of the trees were serotinous. These remarkable differences resulted in dramatic differences in nutrient flows and how the communities developed after a fire.

Serotinous lodgepole pine cones remain closed with seeds secured inside for years until the heat from a fire causes the scales to open and shed their seeds. These cones are over 5 years old. (C. Benkman photo)

Non-serotinous cones open and seeds are shed several weeks after the seeds mature in early autumn. (C. Benkman photo)

Traditionally, forest ecologists have focused on variation in the frequency of fire to account for variation in the occurrence of serotiny. More frequent fires favor higher frequencies of serotiny, and observations, including those contrasting low and high elevations in Yellowstone, support such a relationship. However, the variation in serotiny mentioned above occurs in the lower-elevation lodgepole pine forests of Yellowstone where fire frequency is rather high and where one might expect the frequency of serotiny to be uniformly high if fire alone was the main selective agent driving the occurrence of serotiny.

This is where pine squirrels come in. Theoretically, a seed predator could select against serotiny by preferentially preying on seeds in the serotinous cones and thereby depleting the arboreal seed bank. Pine squirrels are such a predator. Both females and males defend separate territories year-round where they harvest, cache, and defend thousands of closed cones with seeds secured within. Because serotinous cones remain closed and hold their seeds for years if not decades, seeds in serotinous cones remain vulnerable to squirrels for years. In contrast, squirrels harvest non-serotinous cones almost exclusively during the several weeks between seed maturation and scale opening and seed shedding in autumn. Consequently, the overall probability of a squirrel harvesting a given serotinous cone is about 100 times higher than that for a non-serotinous cone. Depending on the density of squirrels, this can result in extremely strong selection against serotiny. Indeed, our models taking into account the life histories of the pines, differential seed predation, and fire frequency show that when squirrel densities exceed one and a half squirrels per hectare, selection by squirrels against serotiny overwhelms countering selection by fire favoring serotiny. At lower squirrel densities the balance between selection by squirrels and fire should set the frequency of serotiny.

Pine squirrel biting off the overlapping scales of a lodgepole pine cone to reach the underlying seeds. (C. Benkman photo)

Our model accounts quite well for the patterns of variation in the frequency of serotiny across Yellowstone and elsewhere. Both at both low elevations where fire frequency is high and at high elevations where fire frequency is lower, the frequency of serotiny is inversely related to the density of squirrels; as predicted by the models and observed, the average frequency of serotiny is higher at lower elevations where fire frequencies are higher. Our model also explains why the frequency of serotiny in ranges east and west of the Rockies, where squirrels were unable to colonize, is uniformly around 90 percent and exceeds the frequency found in areas with squirrels. Indeed, following a fire in the Cypress Hills, where there were no squirrels and the frequency of serotiny averages 92 percent, the density of seedlings was 2,500,000 per hectare!

Much of the variation in the frequency of serotiny, therefore, represents the outcome of two conflicting selection agents, one abiotic, fire, and one biotic, differential seed predation by pine squirrels. Importantly, at high enough squirrel densities the selection they exert overwhelms selection from even high frequencies of fire favoring serotiny. The balance of these conflicting selective agents in turn influences the density of seedlings after a fire with huge community and ecosystem consequences. Because serotiny is a genetically based trait with 11 “loci” accounting for over 50 percent of the variation in the trait (Parchman et al. 2012 Molecular Ecology 21:2991-3005), we can begin to link “genes” to ecosystems over vast expanses of the Rocky Mountains where lodgepole pine dominates the landscape, all mediated by the two main conflicting selection agents. In short, a small squirrel has an evolutionary impact with outsized ecological consequences.

Monday, July 7, 2014

Sometimes it’s better to be lucky than good

[ This post is by Devon Pearse; I am just putting it up. –B. ]

Identifying the genomic basis of complex, ecologically important phenotypes has become a major obsession among many evolutionary geneticists. Recently, some of the best successes in finding them have been in systems in which strong divergent selection has driven genetic changes at a few significant loci. These situations in the wild are analogous to the strong artificial selection imposed in plant and animal breeding. Certain traits selected for during the domestication process in dogs, for example, are determined by just a few genes of major effect rather than having the more polygenic architecture previously assumed. Similarly, examples of complex phenotypes with simple genetic architectures have been found in studies of wild organisms undergoing recent strong selection, such as those driven by anthropogenic impacts.

One such trait is smoltification, the process through which anadromous fishes mature from their freshwater juvenile rearing habitats into an ocean-going phase to grow to adulthood. This transformation is the epitome of a complex phenotype, involving changes in physiological, morphological, and behavioral traits preceding downstream migration to the ocean. And in some species, including the enigmatic salmonid species Oncorhynchus mykiss, undergoing this transformation is optional—individuals that smolt become ocean-going steelhead, while other individuals remain in freshwater to mature as rainbow trout. Not surprisingly, efforts to use genomic techniques such as GWAS and QTL mapping have identified numerous significant associations, but different studies have produced highly variable results, often identifying completely different chromosomal regions as important in determining individual expression of anadromy.

Photo: Anadromous and resident O. mykiss. Credit: Morgan Bond. 

This brings us to the subject of this post, recent studies in which my colleagues and I took a different approach (Martinez et al. 2011; Pearse et al. 2014). Taking advantage of pairs of wild populations above and below barriers to migration, we reasoned that, from the perspective of the above-barrier populations, selection would be extremely strong to stay above the falls—fish that migrate downstream can’t come back, so any gene that increases the probability of smoltification would rapidly remove itself from the above-barrier populations, never to return. Using a simple Fst-outlier approach, we detected parallel adaptive changes in the allele frequencies at multiple loci in multiple populations, many of which had been isolated above-barrier for just a few decades following introduction or dam construction. These patterns demonstrated that a large genomic region on chromosome Omy5 plays a key role in the evolution and expression of smoltification in nature, and that repeated evolution of the resident freshwater phenotype above-barriers has resulted in a parallel evolutionary genetic response in this region.

Allele frequencies of 55 SNP loci located on Omy5 in populations above and below barriers to migration. Orange and blue indicate frequencies of alleles associated with resident and anadromous populations, respectively. Loci are ordered left to right based on the strength of linkage disequilibrium between them. (from Pearse et al. 2014).

Given the hard selection in these above-barrier populations against alleles associated with downstream migration, how strong of an effect is required to create these population differences? A simple simulation study illustrates this, and shows that given reasonable population parameters, a fitness differential as small as 5-10% could explain the observed patterns of parallel evolution. Thus, the population-level allele frequency changes we detected, even though they occurred over extremely rapid evolutionary timescales, could be the result of fairly small shifts in individual probability of smoltification.

Simulations using PopGen (v3.3, 2008) show that a small reduction in relative fitness for AA individuals (RR=1.0, AR=0.95, AA=0.9) can reduce the frequency of the A allele over ~30 generations, even with a small population size (N=100).

Regardless of the apparent strength of the association between the Omy5 region and life-history variation in O. mykiss, it is important to remember the fundamental truth of complex phenotypes—they are complex. To use the analogy of playing cards, the hand that is dealt may represent an individual’s genetic makeup, but how it is played has a huge influence on the outcome (phenotype). And, unlike card games, the genetic hand that each individual human, fish, or other organism is dealt is spread across tens of chromosomes, with potentially 100s or 1000s of genes and segregating genomic regions, so it is hardly surprising that such variable associations between phenotype and genotype have been described. Some cards may be more influential than others (an ace might greatly increase an individual’s probability of ‘winning’), while the influence of other cards may depend on the combination of cards dealt—a three of hearts could be very important in a hand with two other 3s or a run of hearts! Finally, evolution may link important genes together through chromosomal inversions or other mechanisms; skewing the odds like a set of aces that are always dealt together, these ‘supergenes’ may combine to exert much greater influence on the phenotype. The long-awaited publication of a genome sequence for O. mykiss (Berthelot et al. 2014, Nature Communications) will greatly facilitate further exploration into the genomic architecture that maintains the linkage block on Omy5.

Photo: Ben Alfred (Flickr)

Moving forward, the challenge is now to determine the extent to which variation in genes associated with phenotypes at the population level influence individual phenotypic expression in nature. Even for single, large-effect regions, such as the one on Omy5, phenotypic plasticity and ecological interactions will have a strong influence on trait expression, much like human genetic information that provides a ‘risk factor’ for an individual patient’s likelihood of developing a given disease. And, as in humans, there may be huge differences in genetic architecture among populations across the range of diverse species in nature. Understanding this variability will be critical to improving both our understanding of the evolutionary genetic basis of complex phenotypes and to efforts to conserve this variation.


Martínez, A., Garza, J.C. & Pearse, D.E. 2011 A microsatellite genome screen identifies chromosomal regions under differential selection in steelhead and rainbow trout. Transactions of the American Fisheries Society 140, 829-842.

Pearse, D. E., Miller, M. R., Abadía-Cardoso, A., & Garza, J. C. 2014. Rapid parallel evolution of standing variation in a single, complex, genomic region is associated with life history in steelhead/rainbow trout. Proc. Roy. Soc. B. 281: 20140012. http://dx.doi.org/10.1098/rspb.2014.0012

(And a tip o’ the hat to Robin Waples for an insightful chat a few years back that greatly encouraged me to pursue this work)

Friday, July 4, 2014

Carnival of Evolution #73: World Cup Edition

The 73rd Carnival of Evolution has been hosted by Pleiotropy, in a tour de force “World Cup Edition” of the Carnival.  The good news: we were a top-seeded team, tied for second place, based on the all-around evolutionary excellence of our nominated post, Katie Piechel’s “Add It Up: The Genetic Basis of Ecological Speciation”.  Go check out the Carnival to see how we did.

In honor of the World Cup, we have two additional items for your delectation:

The evolution of World Cup soccer balls through time.  Interesting, no?  I didn't realize the truncated icosahedron shape was such a modern thing.  Photo credit: Jens Heilmann.

And this is, of course, from Jerry Coyne’s website; I hope he doesn’t mind us propagating this meme, for it is a high-fitness meme indeed.

Monday, June 30, 2014

A polymorphism persists in panmixia despite unfit heterozygotes

Discovering the genes that underlie phenotypic variation within species is increasingly common, but figuring out how they respond to natural selection is a major challenge. Stickleback researchers have been particularly successful at identifying genes or genomic regions of major phenotypic effect. The Pitx1 gene, for example, is known to influence pelvic girdle and spine development. QTL studies have revealed genomic regions underlying variation in phenotypes ranging from juvenile growth rate, gill rakers, pigmentation, schooling, and sensory systems. The Ectodysplasin gene (Eda) is particularly interesting, because of its strong effect on the number of bony lateral plates in stickleback (Colosimo et al. 2004). Multiple freshwater stickleback populations have a reduced number and/or size of lateral plates relative to their marine ancestors, suggesting that natural selection has repeatedly acted on Eda following the colonization of freshwaters by stickleback.

Stickleback plate morphs found in Kennedy Lake (Marchinko and Schluter 2007), where the partial morph is surprisingly rare (Marchinko et al. 2014).

Over a decade has passed since the discovery of Eda, and numerous speculative hypotheses have emerged about what selective forces affect Eda to control armor plate evolution in wild stickleback populations. In 2008, Barrett and colleagues did a selection experiment with the aim of observing how colonization of freshwater could drive phenotypic evolution of stickleback armor (Barrett et al. 2008). They introduced marine stickleback, heterozygous at the Eda locus, into freshwater ponds and found strong positive selection for the low-plated Eda allele. As predicted, the low-plated phenotypes had a fitness advantage once lateral plates had developed, possibly due to enhanced growth in freshwater environments (Marchinko and Schluter 2007). However, in early life stages (prior to plate development) selection favored the high-armor (“complete”) allele, suggesting that the strength and direction of selection varied among life stages. They concluded that Eda was potentially having either direct or epistatic effects on other phenotypic traits affecting fitness. As simple as this experiment was, it revealed considerable complexities about how natural selection drives phenotypic evolution, even when the phenotype has a relatively simple genetic basis.

Prior to this experiment, Marchinko and colleagues had begun to investigate a polymorphism in body armor of a population of stickleback in Kennedy Lake on Vancouver Island. Plate polymorphisms are not unique to Kennedy Lake, but in the Pacific basin it is relatively uncommon to have a high frequency of both completely plated and low-plated individuals. Typically, freshwater populations are either low-plated or fully plated. In some cases, a polymorphism is present transiently, as the frequency of plates changes over time in response to changing selection pressures (Kitano et al. 2008). In Kennedy Lake, however, this plate polymorphism has been stable since at least 1965.

Typically, polymorphisms are thought to be maintained either by heterozygote advantage (when the fitness of Aa is higher than that of both AA and aa genotypes), or more commonly by spatially varying selection (whereby each of AA and aa have highest fitness in different parts of the population). In contrast, population genetic theory indicates that a polymorphism is highly unlikely to persist when heterozygotes have a disadvantage (when the fitness of Aa is lower than AA or aa). This is because in the absence of assortative mating, a higher fraction of copies of whichever allele is rarest (A or a) end up in low-fitness heterozygotes. Selection is expected to drive the common allele to a frequency of p = 1, eliminating the polymorphism. With the discovery of Eda’s effects on lateral plates, the Kennedy population became an attractive system to test the role of natural selection in maintaining a polymorphism.

The first hint that something interesting was going on in the Kennedy Lake stickleback population came from our first sampling campaign in 2004. We found that the vast majority of individuals were low-plated or completely plated, whereas intermediates were scarce. Low-plated and completely plated morphs differed both in their morphology (head shape) and stable isotopes (reflecting diet/habitat choice). The results were difficult to interpret, as we had no strong a priori prediction about how diet or morphology might differ between the morphs.

Sampling Kennedy Lake

The story became much more interesting when we started genotyping. Over multiple years of sampling the adults in the population, we repeatedly found a deficit of Eda heterozygotes that was not associated with population structure. This indicated that the polymorphism was unlikely to result from either heterozygote advantage or gene flow between locally adapted populations. Maybe assortative mating could explain the heterozygote deficit? Not so. We raided egg clutches from stickleback nests and found they were in Hardy-Weinberg equilibrium.

Really?!? No assortative mating?

Without evidence for assortative mating, the most likely explanation was that every year disruptive natural selection regenerated the heterozygote deficiency in the adult population. We estimated that heterozygotes were 20–30% less fit than individuals with the fully plated homozygous genotype, and up to 80% less fit than the low-plated homozygous genotype. The low fitness of the heterozygotes has been observed only once before, in a previously published experiment by Zeller et al. (2012). The superior fitness of the low-plated homozygous genotype was consistent with an observed growth advantage of this genotype in a previous study (Marchinko and Schluter 2007).

At this point we had some nice patterns from a single stickleback population, but why is this interesting to non-stickleback folk? First, evidence for stable genetic polymorphisms with heterozygote disadvantage are extremely rare, even though theory suggests that they might arise under a wide range of ecological conditions. Second, the polymorphism appears to be maintained under a seemingly worst-case scenario. As we described above, maintenance of a polymorphism with heterozygote disadvantage is theoretically difficult. However, the solution might be frequency-dependent selection: each allele might have an advantage overall as it becomes rare. Third, such a polymorphism can lead to a wide range of evolutionary outcomes, including the evolution of dominance, assortative mating (and possibly sympatric speciation), or the loss of the polymorphism. Although the specific outcome is difficult to predict, Kennedy Lake sticklebacks are well suited for studying how frequency-dependent selection can maintain genetic and phenotypic diversity within a population.

Kennedy Lake is the largest lake on Vancouver Island, and definitely worth a visit.

Our study also adds another chapter to the Eda story. The mere existence of the stable polymorphism means that fully plated and low-plated populations are not the only options. However, we still know very little about the agents of selection that affect Eda in natural populations, and so we can only speculate about the forces driving disruptive selection (and rare-allele advantage, if it exists) in the Kennedy Lake population. Hopefully our study has exposed our ignorance and will inspire further experimental tests and new comparative studies of polymorphic stickleback populations.

Cited work

Barrett, R. D. H., Rogers, S. M., & Schluter, D. (2008). Natural selection on a major armor gene in threespine stickleback. Science, 322(5899), 255–257. doi:10.1126/science.1159978

Colosimo, P. F., Peichel, C. L., Nereng, K., Blackman, B. K., Shapiro, M. D., Schluter, D., & Kingsley, D. M. (2004). The Genetic Architecture of Parallel Armor Plate Reduction in Threespine Sticklebacks. PLoS Biology, 2(5), e109. doi:10.1371/journal.pbio.0020109

Kitano, J., Bolnick, D. I., Beauchamp, D. A., Mazur, M. M., Mori, S., Nakano, T., & Peichel, C. L. (2008). Reverse Evolution of Armor Plates in the Threespine Stickleback. Current Biology, 18(10), 769–774. doi:10.1016/j.cub.2008.04.027

Marchinko, K. B., & Schluter, D. (2007). Parallel evolution by correlated response: lateral plate reduction in threespine stickleback. Evolution, 61(5), 1084–1090. doi:10.1111/j.1558-5646.2007.00103.x

Marchinko, K. B., Matthews, B., Arnegard, M. E., Rogers, S. M., & Schluter, D. (2014). Maintenance of a Genetic Polymorphism with Disruptive Natural Selection in Stickleback. Current Biology, 1–4. doi:10.1016/j.cub.2014.04.026

Zeller, M., Lucek, K., Haesler, M., Seehausen, O., & Sivasundar, A. (2012). Little evidence for a selective advantage of armour-reduced threespined stickleback individuals in an invertebrate predation experiment. Evolutionary Ecology, 26(6), 1293–1309. doi:10.1007/s10682-012-9566-2

Tuesday, June 24, 2014

Add it up: the genetic basis of ecological adaptation

[This post is by Katie Peichel. I am just putting it up.]

What are the genetic and genomic changes that underlie adaptation to divergent environments? How do these changes lead to the formation of new species? These two questions have driven my research for the past 15 years - and it’s been an exciting time to be asking them! The field of evolutionary genetics has begun to make great progress in identifying specific genetic changes that underlie phenotypes thought to be important for adaptation to particular environments. Yet there usually isn’t just one "magic" phenotype important for adaptation to a complex environment, and many different phenotypes need to work together for an organism to be successful in a new environment. Importantly, we still don’t have a clear idea of how the genetic changes that underlie specific phenotypes contribute to whole-organism performance or fitness in a particular environment. Also, we know very little about how these genetic changes lead to the evolution of reproductive isolation between species that have adapted to divergent environments.

To address these questions, I have had the good fortune to work with Matt Arnegard and Dolph Schluter on the benthic-limnetic stickleback species pairs – the poster children for the study of ecological speciation. These two species have evolved within the past 12,000 years as a consequence of adaptation to divergent trophic habitats. First discovered and characterized by Don McPhail (University of British Columbia - UBC), the benthic-limnetic species pairs have independently evolved in several lakes in British Columbia, including Paxton Lake on Texada Island. A body of work by Dolph and his group has led to our understanding of the ecological factors that have driven divergence between the species and of the isolating barriers that maintain them in the face of gene flow.

Dolph Schluter explaining his quest to understand the origin of species using the benthic-limnetic species pairs at the Texada Island Stickleback meeting in 2010.

Matt, Dolph and I decided to investigate the genetic basis of the traits responsible for adaptation to the two trophic niches (benthic and limnetic) found in Paxton Lake, as well as the traits that contribute to reproductive isolation. Many of these traits, such as foraging naturally on different food resources, female mate preference, or male nest site preference are either difficult or impossible to measure in the laboratory. Thus, we took advantage of Dolph’s amazing new pond facility at UBC. These ponds were established in 2008, and we conducted the first experiments in this new facility. Matt worked hard to establish the environments in these ponds using material from Paxton Lake, and multiple lines of evidence suggest that they approximate the two habitats found within the lake.

Photo of experimental pond 4 at UBC in fall 2008, when we collected the F2 juveniles used in our study. The pond has both a benthic habitat (shallow littoral zone) and a limnetic habitat (deep open-water zone), approximating the two environments present within Paxton Lake. (Figure from Arnegard et al. 2014)

We designed a novel approach that would allow us to uncover the genetic basis of whole-organism feeding performance in the contrasting habitats. To do so, we first made F1 hybrids between Paxton benthics and limnetics in the lab. In the spring of 2008, we put 40 of these F1s into one pond (pond 4) and allowed them to mate freely. The F2 intercross population that resulted thus spent their entire lives in the pond, where they were free to choose their habitat and diet. We collected 633 F2s as juveniles in the fall of 2008, when we measured the consequences of these choices on the performance of the fish, measured as body size. In addition, we measured the diet of the F2s using both stable isotope analyses and gut contents, with help from Blake Matthews (EAWAG). We also measured the morphological traits that are expected to contribute to performance in the two trophic habitats, including functional morphological traits that Matt McGee (UC Davis) has shown to be important for prey capture and retention in the benthics and limnetics.

Matt Arnegard and Katie Peichel looking for the Paxton benthic-limnetic F1 hybrids in spring 2008, just after 40 F1s were released into pond 4. These F1s mated freely to create the large F2 population used for genetic mapping in our study.

Amazingly, we found strong evidence for a performance landscape within the pond! When we mapped body size of the F2s onto their position in stable isotope space, we found two peaks of F2s with large body size, separated by a saddle of F2s with smaller body size, as well as a valley of F2s with very small body size. One group of large F2s (group B) was eating mostly benthic resources and had more benthic-like functional morphology, whereas the other group of large F2s (group L) was eating mostly limnetic resources and had more limnetic-like functional morphology. The F2s with smaller body size were intermediate in both diet and morphology. The really small F2s (group A) were eating an alternative food resource not usually consumed by benthics or limnetics in the wild (we think it blew into the pond) and had a strong phenotypic mismatch in two oral jaw traits previously shown by Matt McGee to be key for feeding on zooplankton, a major food resource of limnetics. Importantly, these results showed that there is not one “magic” trait that allows sticklebacks to feed on these alternative food resources and that whole-organism performance is driven by integrated suites of phenotypic traits.

Performance landscape in pond 4. The positions of individual F2s are plotted based on their stable isotope values, overlain with contours of loess-smoothed body size. Individuals in the three groups with the most extreme body sizes are shown in black symbols (group B, downward triangles; group L, upward triangles; group A, squares), and the remaining individuals are shown in gray circles. PC1 represents the major axis of isotope variation in the F2s and is consistent with the major axis of niche divergence between the benthics and limnetics in the wild. (Figure from Arnegard et al. 2014)

We then determined the genetic basis of performance in this trophic landscape. First, we genotyped the F2s with a panel of single nucleotide polymorphism (SNP) markers developed by David Kingsley, Felicity Jones and Frank Chan (Stanford). We then performed genetic linkage mapping of all the morphological traits that we measured, with a particular focus on the “component” traits that contributed to variation in performance. We found that over half of the 21 stickleback chromosomes had at least one genetic locus that contributed to variation in these component traits. We then asked how the genetic loci that underlie individual component traits combine to determine the performance of an F2 in niche space. Figuring out how to robustly conduct these genetic analyses was probably one of the most challenging parts of the study. Matt, Dolph and I had many discussions about how to do best do these analyses; I think our final breakthrough occurred during a break from skiing during the Peichel-Schluter Lab retreat at Mt Baker! Matt persevered through many iterations of the analyses, and in the end, we demonstrated that the genetic basis of niche divergence is largely polygenic and additive. That is, the addition of a benthic allele at any of the loci for individual morphological traits moves a fish in isotope space by the same amount as at any other locus. We did find evidence that epistatic interactions between loci have an effect on niche divergence, but the core genetic architecture is largely additive.

I personally was quite surprised by our findings that such a complex and distributed genetic architecture has arisen in such a short evolutionary time, especially in the face of gene flow! Perhaps naively, I had originally expected that we would find a few clusters of loci with relatively strong effects on performance in the divergent habitats. The genetic architecture we instead uncovered could be a consequence of strong and multifarious selection on multiple traits. It is also possible that this complex genetic architecture did not arise de novo in the past 12,000 years but is the reassembly of ancestral variation that was segregating in the marine ancestors of the benthic and limnetic species. I would love to test these ideas in the future.

The work also has implications for the genetic basis of speciation in this system. Multiple lines of evidence indicate that hybrids do not perform well in either parental environment in Paxton Lake and that ecological selection against hybrids contributes to reproductive isolation. Clearly, some F2 hybrids in our study performed better than others and some performed particularly poorly (the group A fish). If the slower growth of intermediate hybrids in the pond would lead to lower fitness (and in sticklebacks, growth has been associated with fitness), then our results suggest that the genetic architecture of extrinsic hybrid inviability might be largely additive. This is a very different genetic architecture than the epistatic genetic interactions that underlie intrinsic hybrid incompatibilities. Nonetheless, the group A F2 hybrids have combinations of traits from the parent species that are badly mismatched. Thus, our data suggest a parallel between the genetic mismatch found in intrinsic incompatibilities under “mutation-order” speciation and the phenotypic mismatch found in extrinsic incompatibilities under ecological speciation.

Our genetic mapping results are consistent with recent genome scans in sticklebacks and several other systems showing that many regions distributed across the genome can show high levels of divergence between even young species. As Patrik Nosil’s group nicely demonstrated using experimental population genomics in stick insects, at least some of these genomic regions of high divergence can result from divergent selection. Both of these studies further highlight the need to perform experimental genetic and genomic studies in natural or semi-natural conditions in order to identify loci associated with fitness and performance in the wild. In the future, it will be important to combine the results of our phenotype-driven genetic mapping studies with population genomic studies in order to obtain an integrated view of the genotypes and phenotypes that contribute to ecological adaptation and speciation in the wild.

Arnegard ME, McGee MD, Matthews BW, Marchinko KB, Conte GL, Kabir S, Bedford N, Bergek S, Chan YF, Jones FC, Kingsley DM, Peichel CL*, and Schluter D* (2014) Genetics of ecological divergence during speciation. Nature. doi:10.1038/nature13301

Saturday, June 14, 2014

Evolution Bucket List - alpha 1

Humans have a penchant for making “best of” (Top 10, Top 5, anything from Buzzfeed, ...) lists. David Letterman’s Top 10 lists are, of course, a long-standing classic. Many magazines and websites have Best Of lists for the year: best photos, best movies, etc. And now we have “best of” YouTube collections for pretty much anything that might interest you. It seems to be human nature to make such lists – although the characters in the movie High Fidelity argue it is a tendency more common or more enhanced in men than women.

The ultimate best-of lists are Bucket Lists – the things you need to see/do before you die. Presumably these are the 100 (or whatever number of) things that would enrich your life experience more than any other. The things that – if you didn’t do them – would have you looking back from the grave and thinking “Damn, I never went cave diving” or “Why didn’t I ever see a Broadway Play?” or whatever. Sometimes we see global bucket lists presumably applicable to all humanity (or perhaps the “average” human), but other times we see more targeted Bucket Lists for things such as Sporting Events, Fishing, Skiing, Reading, Dining, Sex, and so on. This got me to thinking: what would be an Evolution Bucket List? A Google search suggests no such list exists and so I figure we should develop one.

I will start the ball rolling with some initial ideas. I hope folks will suggest additional options in the comments. Perhaps with a few versions of the list batted around, we can develop the optimal set. What should be on such a list? Several categories stand out. (1) Locations where organisms are particularly special, such as Galápagos. (2) Particular organisms of evolutionary significance or novelty, such as the platypus. (3) Locations of historical/contemporary importance to the study of evolution, such as Darwin’s home. (4) Amazing interactions between organisms, such honey guides and honey badgers or the parasitic isopod that replaces the tongue of some marine fish. (5) Specific dramatic or important fossils or fossil sites, such as the Burgess Shale. So I will start our group effort by suggesting some items for the Evolution Bucket List: some I have already experienced, some I expect to experience, and some that would be great to experience but that I probably never will. 

Things already in my bucket

1. Galápagos Islands. For me, Galápagos is the single most iconic and inspirational location for someone interested in evolution. Not only did it inspire Darwin and many evolutionary biologists since, but it is home to some of the world’s most unique and interesting organisms: marine iguanas, Darwin’s finches, flightless cormorants, tropical penguins, and many others. Of course, we could easily list each of these organisms as separate items on the bucket list but – given that one can knock all of them off in a relatively short time in a relatively small area – I think they are better encompassed as a location. I have been fortunate enough to visit Galápagos many times, although I still haven’t seen a flightless cormorant. BLOG POST

Darwin called them "imps of darkness."
2. Down, EnglandNo location was more instrumental in the development and exposition of Darwin’s theory. It was here that Darwin did nearly all of the work that led us to our modern understanding of life on earth and how it came to be. The most obvious thing to see is Down House (especially his study) and its grounds (especially the Sandwalk). However, what was even more exciting for me was Darwin’s Pub. BLOG POST

Having a pint in Darwin's Pub.
3. ArchaeopteryxDiscovered soon after Darwin published his magnum opus, Archaeopteryx made flesh the intermediate forms between extant groups (birds and reptiles) and was thus a crucial contributor to acceptance of Darwin’s theory. Simply put, Archaeopteryx is the most famous fossil in the world: it is even in MS Word’s spell-check dictionary which, I have just discovered, knows how to spell it better than me. It is also the world’s most beautiful fossil – and spectacular versions can be seen in many museums. I have seen them in the Berlin Natural History Museum and the British Natural History Museum.

The Berlin Archaeopteryx
4. OilbirdsOne of the most exciting things for an evolutionary biologist is evolutionary novelties – species that just stick out in ways that set them apart from even the most closely related species. That is, they are the sorts of species that might not be imaginable if they didn’t actually exist. Oilbirds are my current favorite. They nest in caves and feed on fruit at night, and they echolocate! They are also huge outliers on the evolutionary tree of birds. And they live in spectacular settings in tropical forests. I have seen them in Trinidad both in caves and clicking their way along river corridors at night. BLOG POST

Me shooting pictures of oilbirds. (Photo by Felipe Perez-Jvostov)
5. Carnivorous plantsI was sorely tempted to list another cool animal (mudskippers or leafcutter ants) but I suppose plants might also be important and interesting – at least to some. Being an animal guy, the plants that are most fascinating to me are the ones that seem almost like animals – the predatory plants. Most iconically, this category (actually several independent evolutionary lineages) jncludes pitcher plants, sundews, and Venus flytraps. It is one of life’s guilty pleasure, at least for kids, to catch and “feed” insects to these plants. And, of course, it helps that Darwin was a big fan, even writing a book about them. Indeed, various carnivorous plants are featured to this day in the greenhouse at Down House (PHOTO).

Sundew in British Columbia.

Things reasonably likely to end up in my bucket

6. MadagascarAnother evolutionary marvel. A place, like Galapagos or New Zealand, long cut off from the rest of the world and so able to embark on an independent evolutionary trajectory. In the case of Madagascar, what I would most like to see are lots of lemurs: ring-tailed lemurs, mouse lemurs, and – most importantly – aye-ayes. Surely the aye-aye is one of the most bizarre and amazing mammals in existence – my kids think so anyway. Check out this awesome video True Facts about the Aye-Aye (3.5 million views and counting).

7. Burgess ShaleOne of the most important fossil finds was in the Canadian Rockies where, 500 million years ago, a large mudslide covered an almost intact fauna from the Cambrian Explosion, a period during which animal life exploded in diversity. This site has told us most of what we know about this time, half a billion years age. In fact, I am shocked that I have not already been to the Burgess Shale, given that I grew up only a few hours away (What the hell were my parents thinking not taking me there?) and I still drive through the area all the time. You can even get a guided tour. What could be simpler? Surely, I will soon check it off.

8. Chimpanzees – in the wildOur closest relative, and popularly consider 95+% genetically similar to us – to me! I am captivated watching them in any of the countless BBC videos. It is like a window into the past, in the sense that our common ancestor probably looked a lot more like a chimp than a human. I am pretty confident I will soon knock this one off too, given that I have collaborators working in Kibale National Park, Uganda, where chimpanzees are abundant and easily seen. Yet another reason to do some work there.

9. Platypus – in the wildOK, I suppose some other monotreme, like the echidna, would do in a pinch but the platypus just seems so much more bizarre. In fact, it bedeviled scientists for almost a century, as playfully described in the book titled, wait for it… Platypus. Of course, I envision my encounter will be like the one in David Attenborough’s Life of Mammals – me sitting peacefully on the edge of a pool while a platypus swims playfully (or at least indifferently) about my feet in crystal-clear water. I have already been to Australia but didn’t seek out a platypus at the time, dammit. Yet I will surely be back to Australia soon. In the meantime, I will have to be satisfied with the dusty old male platypus moldering away in the basement of the Redpath Museum. Not quite the same, but inspirational nonetheless.

10. Flying herpsThey don’t really fly, of course: flying has evolved only four times (bats, birds, insects, pterosaurs), but these gliders can be just as amazing. I grew up watching flying squirrels, which were cool enough, but what about flying snakes (VIDEO), flying lizards (VIDEO), or flying frogs (VIDEO)? The snakes can dramatically change direction in mid-air. The lizards extend their ribs until they look like a kite. The frogs were described by Alfred Russell Wallace – the co-discover of natural selection.

Things with varying degrees of unlikeness.

11. Bolas spider in actionLooks like a bird turd. Gives off a moth pheromone. Spins a ball of sticky web on a string and swings it around to catch sexed-up moths in flight (VIDEO). This sight should be the most accessible of these relatively inaccessible bucket list options, given that bolas spiders are widespread, seemingly even near my house. I just don’t know where or how to start looking. (OK. so maybe this one actually is achievable and I should have picked some rarer but equally cool spider.)

Many spiders could make the list. 
How cool is this crab spider waiting for dinner to fly up?
12. Antarctic breeding coloniesAlbatrosses. Elephant seals. Penguins. By the millions. South Georgia Island. The Kerguelen Islands. One of life’s great spectacles. Ironically, I once passed up a chance to go to the Kerguelen Islands. I probably won’t get another.

Penguin (check - Galapagos), Elephant Seal (check - California). Millions of each on a beach in (or near) Antarctica (pending).
13. Killer whales swimming onto the beach to get at sea lionsI must have watched videos of this behavior hundreds of times – truly amazing stuff. Here is one of those VIDEOS – 12 million views. In reality, this spectacle should be reasonably accessible (at least in relation to those that follow) but I am betting the chances are still remote because it happens in only a few places (Península Valdés, Argentina) for only part of the year (Feb–Apr – and episodically even then) and access seems to be strictly controlled.

Just need to get this orca and this sea lion on the beach in the same picture!
14. Hydrothermal vent communitiesEntire communities that thrive independently of energy from the sun. First seen in 1979, these communities of tube worms, crabs, fish, and other critters were too bizarre and unpredicted to be believed – except that we could actually see them on VIDEO. For this I just need a submarine.

15. TylacineIf one is to see a cool marsupial, one couldn’t do better than a Thylacine – the Tasmanian Tiger, in colloquial terms. OK, I know you’re saying well, duh, it’s extinct. But not everyone is convincedand a few of these critters might still be hanging out in the woods of Tasmania. To give a hint of what you might see, check out this chilling VIDEO of the last one alive in a zoo – viewed more than 2 million times. For this I might just need a time machine.

I would trade several thousand kangaroos for one (OK, maybe two) thylacines.

So, there’s my first attempt at an alpha version of an Evolution Bucket List. Please send me some new ideas so we can make a definitive list. If you have actually knocked off the item, then a tiny description like those above would be great – with a video or picture, ideally one you took yourself.

Have at ’er.