Tuesday, April 15, 2014

Pupfish on the adaptive landscape, or: How I learned to stop worrying and love the cable tie

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

Like most evolutionary biologists, when I think of evolution I imagine rugged mountain landscapes, as Carl Zimmer eloquently introduced the concept of the adaptive landscape. Try to imagine a vast landscape connecting the phenotypes of all organisms where spatial location indicates a particular phenotype and the terrain – the height at any particular location – corresponds to the fitness of each individual phenotype (a Simpsonian landscape) or genotype (the original Wrightian landscape from 1932) along this space. The species we observe in the wild are expected to sit on the tops of fitness peaks, isolated by fitness valleys from neighboring fitness peaks corresponding to other species: alternative phenotypes capable of surviving and reproducing in the same environment. In reality, of course, this picture is even more complex: environments fluctuate, phenotype and genotype ‘surfaces’ are high-dimensional, fitness is frequency-dependent, many-to-one mapping is pervasive, and fitness may continue to increase indefinitely even on the tops of ‘peaks’; the adaptive landscape may resemble something more like the foaming surface of the ocean than a stable mountain range.

Nonetheless, the adaptive landscape is a powerful metaphor that pervades evolutionary thinking. But it is also an empirical quantity that can be measured in the wild. Since Lande and Arnold’s watershed insight that fitness landscapes can be estimated directly, there has grown an enormously successful research field measuring fitness in natural populations. There are now hundreds of studies measuring the strength and form of selection within natural populations in the wild. These have provided many novel insights about how selection acts in the short term, and have posed new paradoxes, such as why there is an abundance of disruptive selection despite our expectation that most species should sit on fitness peaks and thus experience stabilizing selection.

However, most of these studies measured selection within a population at one or two time periods. Few studies have repeatedly measured fitness within the same environment or across multiple environments. Fewer still have measured fitness landscapes across multiple species or for the intermediate phenotypes between species. Yet, this is exactly the data we need in order to understand past evolutionary trajectories and make predictions about the future. Our understanding of the larger-scale features of fitness landscapes – those peaks and valleys connecting multiple species that we all so vividly imagine – is still in its infancy.

Ideally, to increase our understanding of these large-scale features of the adaptive landscape, we need systems that are not only rapidly speciating and undergoing major morphological and ecological transitions, but that are also amenable to laboratory hybridization and transport in order to measure the fitness of intermediate hybrid phenotypes rarely observed in the wild.  A small, charismatic group of fishes named after puppy dogs (they love to beg for food and wag their tails) offer all of these features: Caribbean pupfishes! It is indeed possible to breed thousands of hybrid pupfishes in the laboratory in a month and then transport these offspring in your checked luggage back to the tropics to measure their fitness in the wild. This was my dissertation research project.


Figure 1. (a) Field enclosure on San Salvador Island. (b) Cable ties.

Within saline lakes on the tiny island of San Salvador in the Bahamas, two ecological specialists and a generalist pupfish have rapidly evolved from a generalist algae-eating common ancestor within the past 10,000 years. Importantly, these species are still in the process of speciating and adapting to novel ecological niches, so measuring the relevant environment driving adaptive radiation is still possible. Secondly, all three species live and breed within exactly the same benthic littoral habitat, which means that any differences among the species cannot be attributed to different habitats or allopatric isolation; rather, they result from the distinct niche environments that these species experience within the same habitat. Third, this young sympatric species flock has all the features of a classic adaptive radiation: (1) morphological diversification rates 50 times faster than background rates, (2) novel ecological niches, with one species specialized for tearing the scales off other pupfishes (Video 1) and another specialized for crushing hard-shelled prey with a unique nasal appendage, and (3) fitness experiments measuring hybrid growth and survival in field enclosures that demonstrate that multiple fitness peaks due to competition are driving adaptive radiation in the wild.


video

Video 1. Cyprinodon desquamator, the scale-eating pupfish, feeding on a euthanized generalist pupfish. Filmed at 2000 frames per second.  (You can also view the video at a larger size at http://adaptiveradiation.smugmug.com).

Interestingly, just a single empirical snapshot of the complex fitness landscape on this island (Figure 2) makes further predictions about speciation within the radiation. For example, I found much stronger selection against hybrids resembling the scale-eater phenotype than against against hybrids resembling the hard-shelled prey specialist phenotype. This indicates that a very large fitness valley separates the specialist niche of scale-eating from the ancestral niche of generalist algivore. In contrast, a small fitness valley separates the specialist niche of hard-shelled prey crushing from the ancestral niche of generalist algivore.


Figure 2. Complex snapshot of natural selection on San Salvador Island, Bahamas. Fitness landscape for survival estimated from the survival of F2 hybrids of all three pupfish species placed in a 12-foot field enclosure for 3 months. For reference, photographs show the phenotypes and position of the three parental species on the landscape: generalist, durophage, and scale-eater. Although I found no evidence of a fitness peak corresponding to scale-eaters, there is still strong evidence of a large fitness valley in this region of the morphospace.

These different-sized fitness valleys on the adaptive landscape predict that gene flow between scale-eaters and generalists should be much lower than between hard-shelled prey specialists and generalists. I tested this prediction in my new paper in Molecular Ecology. In my first foray into next-gen sequencing, I used double-digest RADseq (specifically, the Elshire et al. 2011 genotyping by sequencing protocol) to genotype over 13,000 SNP’s in all three species of pupfishes on San Salvador from seven different lakes where at least two species coexist. Fortunately for me, my collaborator Laura Feinstein, a grad student in Ecology at UC Davis, already had this protocol up and running and showed me the ropes. Our study of genetic differentiation supported predictions from the empirical adaptive landscape. I’ll quickly summarize and speculate about our results below.


1. Specializing on scales drives speciation faster than specializing on snails

Across all lakes surveyed, scale-eaters were more genetically differentiated than hard-shelled prey specialists within the same habitat, despite ongoing gene flow in both species. This pattern can be understood with knowledge of the underlying fitness landscape: stronger selection against scale-eater hybrids provides a greater barrier to gene flow than moderate selection against hard-shelled prey specialist hybrids. Furthermore, this pattern of genetic differentiation was consistent across many lakes on the island and many rare dispersal events of these specialist species, suggesting that it is driven by a consistent selective regime, rather than different histories of the two species. Although speciation in these two species is far from complete, the pattern so far suggests that these two very different foraging strategies are driving different rates of speciation.


2. Not all adaptation is the same

We often talk about adaptation in general such as understanding the genetic basis of adaptation or the role of adaptation in driving reproductive isolation. Instead, our results emphasize that the niche matters: not all adaptation is the same. Adapting to a “closer” niche on the adaptive landscape results in less selection against hybrids than adapting to a “distant” niche on the adaptive landscape. This difference in the strength of postzygotic extrinsic isolating barriers – perhaps the very first isolating barrier in this system – may drive different rates of incipient speciation between diverging ecotypes, even within the same habitat. Both specialist pupfish species are rapidly adapting to new resources in a new environment, but the specific performance demands of these very different resources affect the pace of speciation.


3. The niche shapes the topography of the adaptive landscape

The difference in the strength of selection against hybrids between these two specialist niches makes sense when we consider their contrasting performance demands. Specializing on hard-shelled prey, such as snails and ostracods, simply requires modifying a simple lever system (the jaws) for higher mechanical advantage. Furthermore, these prey are not evasive and are often consumed by generalists as well. In contrast, successful scale-eating requires high-speed strikes on evasive prey (i.e., other pupfish!). Furthermore, each strike only provides a small nutritional benefit – a mouthful of skin, mucus, and scales – so that each strike must be very efficient. Indeed, all specialized scale-eating fishes are size-limited relative to their prey. This suggests that scale-eating should require a highly specific and specialized phenotype for success, whereas successful snail-eating may require fewer modifications. These different performance demands might create different topographies on the adaptive landscape – larger fitness valleys, higher-dimensional changes, and steeper fitness peaks – that all result in stronger selection against scale-eater hybrids.


4. Reinforced pre-mating isolation may result from alternative niche environments

Preliminary female mate choice trials in the lab and focal observations in the field suggest that scale-eater females show much stronger preferences for conspecific males than hard-shelled prey specialist females. This may reflect reinforced conspecific mate preferences in scale-eater females to avoid the higher costs of hybrid matings.


5. Is the rarity of a niche connected to speciation rate within that niche?

Finally, I’d like to take this opportunity to speculate wildly. Scale-eating is incredibly rare within pupfishes: to put this in context, the nearest species with convergent ecology is separated by 168 million years of evolution (my proposed index for measuring ecological novelty). I speculate that this rarity reflects an underlying distant, hard-to-reach fitness peak for scale-eating on the adaptive landscape. At the same time, a distant fitness peak surrounded by a large fitness valley results in much stronger selection against hybrids if a population is able to reach such a peak. If this association holds more generally, it suggests a positive correlation between the rarity of a niche and the rate of speciation within that niche: distant fitness peaks are hard to adapt to initially (resulting in rare use of the corresponding resource or niche), but if a population does colonize such a niche, it may evolve reproductive isolation much faster due to stronger selection against hybrids and reinforced pre-mating isolation. In other words, perhaps oddball ecological specialists speciate faster? Stay tuned!

Monday, April 7, 2014

Peaks and Valleys in the Genome

(This post is by Marius – I am just putting it up. Andrew.)

Driven by methodological advances, evolutionary biology is currently much concerned with understanding the way selection shapes the genome. In the search for such signatures of selection – and ultimately the loci associated with them – we often pursue a similar strategy: we compare populations at thousands of genetic markers with the hope of finding genomic regions of particularly high or low differentiation relative to the genome-wide baseline. We then believe that such regions can be directly linked to distinct selective processes. On the one hand, genomic regions of high divergence are thought to be the result of selection acting in opposite ways (divergent selection) between populations. Low divergence regions, on the other hand, are commonly taken as evidence for balancing selection. The results of our recent paper published in Molecular Ecology, however, challenge these common assumptions.

Figure 1. Parallel adaptation to similar derived habitats (blue) from a common source population inhabiting an ecologically distinct habitat (gray).
For our paper, we first implemented theoretical models in which we considered several populations deriving from a common source population into selectively new and similar habitats – that is, parallel adaptation (Figure 1). We demonstrate that among derived populations, this process drives a region of particularly low divergence around a selected locus. How come? Due to common ancestry, the derived populations do not only share the actual variant being selected, but also the genomic background linked to that variant. Thus, the same variant together with this background are driven to fixation in the derived populations. Consequently, when we compare such populations, we find a genomic region of low divergence surrounding a locus involved in parallel adaptation (Figure 2). Admittedly, this explanation for low divergence within parts of a genome is intuitive. Nevertheless, it is normally overlooked when interpreting low divergence regions in genome scans. From now on, let us call such a region a ‘divergence valley’.

Figure 2. The peak-valley-peak divergence signature of parallel adaptation from shared genetic variation. Important to note here is that the selected locus is actually located at the bottom of the divergence valley, and not at any of the flanking peaks!
 To either side of the divergence valley, our models reveal exaggerated divergence (Figure 2). This is because the selected variant and its linked genomic background are associated with different haplotypes in the different derived populations. Such haplotypes then become, to some extent, selected along with the actual variant and its immediately linked background under selection (this process is called ‘genetic hitchhiking’). As a consequence, different haplotypes increase in the different populations to different frequencies. It is important to note here that this phase of genetic hitchhiking only initiates the high-divergence regions flanking the divergence valley! These regions of high divergence – I will refer to them as ‘twin peaks’ from now on – grow higher and become sharper over time, even after selection has fixed the favorable variant at the locus under selection in all derived populations (Figure 2). The reason for this is that the locus under divergent selection between the source and derived populations and under parallel selection among the derived populations acts as a barrier to ongoing gene flow from the source to the derived populations. The divergence valley will thus, to a great extent, be sheltered from such genetic introgression and remain a low-divergence region among derived populations within the genome. Next to it, however, some genetic variation will introgress from migrants stemming from the source population. Because this happens only occasionally and randomly, the through hitchhiking initiated divergence twin peaks grow higher. Also, the twin peaks become sharper over time because even further away from the selected locus, genetic variation can flow almost unconstrained between the source and derived populations. This gene flow homogenizes the genome of the derived to baseline divergence. A detailed explanation and a graphical illustration of these different – yet together acting – processes can be found in our paper. Also, you will find there a thorough dissection of many factors influencing these divergence patterns (recombination rate, strength of selection, time, migration, number of initial colonizers, and multiple interacting selected loci). What I can tell you already is that the above-explained patterns emerged very consistently in all our simulations!

Picture 1: Top: The Sayward River estuary, where one of the marine stickleback samples was taken for our empirical analysis. Bottom: A typical marine stickleback. Note its characteristic armor plating all along the body axis, which is absent in most freshwater stickleback (Picture: M. Roesti).
Agreed, up to now, this blog post has been quite theoretical. Luckily, we have a great model system at hand to take these theoretical predictions out into the wild. That model system is the threespine stickleback fish. Stickleback have repeatedly colonized and adapted to freshwater (parallel adaptation) from a common marine source population since the last glaciation period. This corresponds exactly to our modeled situation above. In a second part of our paper, we thus predicted to find a divergence valley flanked by twin peaks (together, we can refer to them as ‘peak-valley-peak’; Figure 2) around three particular genes. These genes are great candidates for being under strong divergent marine-freshwater selection, and thus seemed ideal to test whether we would find the peak-valley-peak divergence signature of parallel adaptation to freshwater. We included a total of eight freshwater populations from Vancouver Island (BC, Canada) and two marine samples from the coast of that island in our empirical analyses (Picture 1 and 2). As expected, marine and freshwater stickleback proved strongly differentiated at all three genes. To calculate differentiation, we used haplotype information taken from targeted sequencing as well as the classic divergence measure FST calculated at thousands of polymorphisms along the genome (RAD sequencing data). We further applied an alternative approach to calculate differentiation, for which we looked at the separation of marine and freshwater stickleback within many phylogenetic trees along the genome. Now, our main interest was in divergence among the derived populations adapted in parallel to freshwater. Excitingly, comparing these freshwater populations among each other indeed revealed the predicted peak-valley-peak divergence signature around all three genes! As this worked out so well, we then searched the entire stickleback genome for further such signatures and found many more of them. This allowed us to propose new genes that have been important for replicate freshwater adaptation. Interestingly, we also found that those chromosomes harboring many of these signatures of selection exhibited the strongest overall divergence between marine and freshwater stickleback. This indicates that divergently selected loci can drive heterogeneity in genomic divergence on a chromosome-wide scale.

Picture 2: One of many breathtaking watersheds on Vancouver Island (BC, Canada) inhabited by freshwater stickleback (Picture: M. Roesti).
So what does this all mean? Our results show that parallel adaptation – the very process involving similar selection pressures – can drive high population divergence within parts of a genome. These high-divergence regions, however, are not holding the actual targets of selection themselves; instead, these targets are located in particularly low-divergence regions when the same genetic variation has been re-used for adaptation. Our results are certainly relevant to many organisms for which we have evidence or a strong feeling that parallel adaptation from shared variation has happened. Also, the case where similar selection pressures act in different populations on parts of the genome may be more common than what appears ‘ecologically intuitive’ to us. Threespine stickleback fish provide a particularly neat model system because we can here draw on many independent and parallel adaptation events to freshwater. Also, we can sample marine stickleback, contemporary representatives of the genetic source underlying this parallelism.

Overall, our findings should be taken into consideration when reasoning on divergence signatures within a genome. Finally, our insights can be used as explicit tools in the hunt for selection signatures, and ultimately, adaptation genes. I hope you will enjoy reading our paper!

Full story:

Roesti M, Gavrilets S, Hendry AP, Salzburger W, Berner D (2014). The genomic signature of parallel adaptation from shared genetic variation. Molecular Ecology (From the Cover). 
http://onlinelibrary.wiley.com/doi/10.1111/mec.12720/full

Friday, April 4, 2014

The Coelacanth has had its day

Who hasn’t wanted to bring an extinct species back into existence? Sure, there are risks, both physical (T. rex and pathogens) and ethical (Neanderthals), and sure, we’re better off without some species (smallpox and mososaurs), but how about the gastric brooding frog and the thylacine and the dodo and so on? Surely the world would be a better – or at least not worse – place if we hadn’t lost them. Enter the de-extinction movement, which seeks to bring extinct critters back to life. It hasn’t happened yet, of course, and it might never happen given not only the risks but the costs and difficulties. Even better than de-extinction – and without any of the ethical baggage – is when things thought to be extinct are found not to be (unextinction?).

Being a fish guy, one of the most inspiring unextinction stories is the discovery of the coelacanth – though to be extinct for more than 60 million years. Found on December 22, 1938, in the bottom of a pile of fish on a trawling ship by the young curator (Marjorie Courtenay-Latimer) of a tiny museum in East London, South Africa, this first specimen was bundled into a cab with the help of a very reluctant cab driver, sketched iconically and the sketch mailed to Professor J.L.B. Smith at nearby Rhodes University. The discovery rocked the scientific world but was less than ideal given the lack of preservative available in East London. Then came the search for another specimen, found only 14 years later in the Comoros. Here the story only gets better, with midnight calls to the Prime Minister of South Africa, a clandestine military evacuation, and an indignant French establishment. And then, in 1998, a second coelacanth species was found in Indonesia in a fish market by a couple on their honeymoon. This one was quickly named by French scientists who hadn’t seen it, in apparent retaliation for the loss of the Comoros specimen. (If these tidbits intrigue you, read the full coelocanth story in A Fish Caught in Time.)

Marjorie Courtenay Latimer and the famous sketch that started it all.
De-extinction the way it should be – unextinction! This story had always been one of my favorites and so I had long been excited to see a coelacanth – if only in a big vat of preservative. However, few coelacanths are in North American museums, partly because of the monopoly the French exerted for years after the Comoros scoop by the Brits. So I didn’t get to see one until I went to France, where every Podunk museum and aquarium in every tiny town seems to have one – I saw mine in Biarritz. My daughter even got to see it, although I am not sure at the age of one she had much appreciation for its scientific significance. Truly an inspiring moment – although some day I would love to see one in the wild (probably harder to see than nearly anything else).

The Biarritz coelacanth - a treat for me and for Aspen!
Of course, coelacanths remain very rare, perhaps forever at risk of re-extinction, something we should surely seek to prevent. Indeed, the goal of preserving truly unique species is gaining steam in conservation biology. The basic idea is that many species are likely at risk of extinction and we need some sort of criterion for deciding which to preserve. One criterion that has been put forth is phylogenetic distinctiveness. That is, the species that warrant the most protection are those that represent the last remaining bits of long isolated branches of the evolutionary tree – lose that last species and forever lose a big chunk of the history of life. A recent incarnation of this idea is EDGE (evolutionarily distinct globally endangered), which seeks to prioritize species conservation based on a joint consideration of phylogenetic distinctiveness and degree of endangerment. So far so good; finally conservation biologists are fully using evolutionary criteria for species conservation! Something all evolutionary biologists can get behind – or is it?

A few weeks ago, I was in College Station, at Texas A&M University, as one of the invited speakers for the Ecological Integration Symposium organized by graduate students (my host was Emily Rose). My talk was on ecological speciation, David Reznick spoke on eco-evolutionary dynamics, Brian Bowen discussed marine speciation, and Tom Lovejoy gave an overview of global climate change effects. At the end of our plenary session, we had a panel discussion. At first, the four of us thought it might be awkward as our talks had been on very different things – but we quickly converged on a topic to which we could all provide perspectives: What, precisely, should we be conserving? A large part of the discussion focused on the importance of preserving not only species but also intra-specific variation, but we also discussed which species should be preserved. At one point, I went through the above rationale about phylogenetic distinctiveness being an important criterion and then Brian grabbed the mic out of my hand.

“THE COELACANTH HAS HAD ITS DAY” was the first sentence out of his mouth. He then went on to describe how species come and go all the time and those old rare relicts just hanging on (coelacanths, tuataras) are probably not long for this world (extinctive?) regardless of human influence – so perhaps we shouldn’t bother. Instead, we should focus our efforts on groups that are rapidly diversifying – African cichlids was his prime example – as they are the future of biodiversity. Ok, sure, I like cichlids as much as (probably more than) the average person, but let the coelacanth go? Heresy. Fear. Fire. Foes. Right then and there I excommunicated him from the pantheon of evolutionary biologists, revoked his citizenship in a compassionate humanity, unfriended him on Facebook, and started a smear campaign to discredit him.

Me and Brian Bowen, the most dangerous man alive - for coelacanths anyway. (Photo by Melissa Giresi.)
On sober (actually just the opposite) subsequent refection, however, I began to question my reaction. Try this thought experiment. How many cichlid species is the coelacanth worth? I think we can surely say at least one. Taking inspiration from Phil Pister’s “species in a bucket”, if I had all of the world’s Pseudocrenilabrus multicolor in a bucket in my left hand and all the world’s coelacanths in a bucket in my right hand, I would probably saw through my left wrist before dropping the bucket in my right. (I might hesitate longer if the hands were reversed.) I would probably decide the same for 10 cichlids or maybe a hundred but what about a thousand or ten thousand – what if it was the entire cichlid fauna of Lake Tanganika or Malawi or Victoria? By the EDGE perspective, I expect that I would save the coelacanth. By the de-extinction perspective, I would probably do the same (it would be much harder to re-evolve the coelacanth than start a new radiation of cichlids – after all, they do it all over the place). But by the Bowen perspective, I would clearly drop the coelacanth without a second thought. And at some level, I see the point. Coelacanths aren’t going to give us anything new. At best, they will still be around a million years in the future looking pretty much the same. The cichlids, however, will likely produce many new species in that time. The future of biodiversity is perhaps better off with the cichlids than coelacanths.

Fortunately, I am not a manager and don’t have to make such decisions, because the truth is I want both cichlids and coelacanths! But perhaps I could do without ticks and chiggers and dengue and AIDS and TB and definitely poison ivy.

An amazing giant isopod in the Texas A and M invertebrate collection. (Photo by Melissa Giresi.)

Tuesday, April 1, 2014

Carnival #70 is up!

Carnival of Evolution #70 is now up!  We had an embarrassment of riches this month, with guest posts from Jacques Labonne about the new Basque fish blog, Fish&Bits, Dan Hasselman on how Anthropogenic habitat disturbance can impact species integrity, Katja Räsänen on the mosaic of reproductive isolation… not to mention posts from Andrew on the meaning of “extinctive” and his visit to Darwin’s pub.

In the end, though, we nominated a fascinating post from Jonathan Richardson about microgeographic adaptation and the spatial scale of evolution.  If you've ever wondered just how “local” local adaptation is, this post will give you some food for thought.

By the way, it’s great to be getting so many guest posts, and we’d like to encourage more.  If you’re a researcher in evolutionary ecology or eco-evolutionary dynamics and you’d like to contribute, please feel free to send an email to Andrew Hendry!

The theme of the new Carnival, at Synthetic Daisies, is evolutionary games.  Enjoy.

Fig. 1.  Evolutionary game theory.  I used to own a stand-up Millipede console.  Now that was a great video game; I wish I still owned it.  Galaga, Defender, Millipede, Donkey Kong, Tempest, Q*bert, Marble Madness… those were the days.

Sunday, March 23, 2014

Darwin’s Pub

Surely the greatest contribution that England has made to the world (apart from deep-fried Mars bars) is Charles Darwin. Certainly, then, the most important tourist destinations in England should be sites associated with Darwin. At least, that has always been my opinion. This post is about my failures and successes in attempting to visit Darwin’s haunts – and a few unexpected and uncommon discoveries along the way.

Would Chuck D have partaken?
On my first visit to London a number of years ago, I had half a day to spare and so sought out Darwin’s grave at Westminster Abbey. I showed up at the door, all aquiver with anticipation, only to be told that it was the one day of the year when tourists were not allowed – a special day instead for worship only. Damn. The next time I visited England, I had a whole day to spare (owing to that annoying policy of airlines charging almost double if you don’t stay over a Saturday night) and so I set my sights on a pilgrimage to Darwin’s home, Down House. Seeing his study and walking his Sandwalk, his “thinking path,” would surely be a great inspiration – and it must certainly be on the bucket list of every evolutionary biologist. After arriving in London on that trip, I looked Down House up on the internet and discovered that it was closed for renovations. Double damn. Instead, I visited the British Natural History Museum, where I could at least see the statue of Darwin. This statue figured prominently in a David Attenborough video for Darwin’s 200th birthday that explained how the statue of Richard Owen, who was instrumental in the museum’s history but a vocal critic of evolution, had recently been removed and replaced by this monument to his archrival Darwin.

Westminster Abbey
I visited London again last week, and I promised myself that I would visit both Darwin’s grave and his home. I even checked the opening times of Down House before booking my flight – Saturdays and Sundays only. So, on the Friday after our bioGENESIS meeting (see this post), I set out for Westminster Abbey. After waiting in line for nearly an hour, I finally made it inside. It was crowded and I was awash in hundreds of graves and monuments all over the floor and walls. Where was Darwin? The audio guide didn’t mention him – as I had been certain it would – so I had to ask. It turned out to be a plain white marble slab on the ground. I had expected something more dramatic, maybe with finch beaks engraved on it, but it was still fun to see the grave and compose pictures of it with the backdrop of an institution that – initially at least – felt so threatened by his ideas. After leaving Darwin’s grave, I tried to take a photo of the “grave of the unknown soldier” (definitely on the audio guide) and was promptly informed that photos were not allowed in the Abbey. Oops. I guess no one cares enough about Darwin’s grave to guard against photography. Even so, it was great to see the founder of evolutionary biology buried in the most important religious institution in England. (Writing this, I wonder if Bishop “Soapy Sam” Wilberforce is also in the church, perhaps with a perpetual frown in Darwin’s direction. Or maybe he is in some lesser church, with an even bigger frown.)

Westminster Abbey
The next day I was off for Down House, which proved to be quite a commute from the hotel, as befit Darwin’s desire to escape the city. I was even forced to wait about an hour for the bus from South Bromley to Downe Village (the “e” was added after Darwin’s time to distinguish it from another Down elsewhere). Fortunately, a Starbuck’s was right beside the bus stop, and so I could sip a non-fat no-whip hot chocolate (tastes the same the world over) and edit a paper. Eventually the bus came and about 20 minutes later we stopped at St. Mary’s Church in Downe. From there it was a 10 minute walk along a narrow lane between some fields and I had the great fun of seeing a pheasant prancing about – did Darwin shoot at its ancestors and miss? Down House was amazing, of course, particularly Darwin’s study and his thinking path, where I made a video to ask the pressing question: How did Darwin walk his sandwalk?


I could well have written an entire post about the wonders of Down House: Darwin loved billiards and would play every day with his butler, Darwin would leave his office dozens of times a day just to get a pinch of snuff from the hallway outside, Darwin rode horses until he fell and gave up, and so on. However, what happened after I left proved to be even more surprising and inspiring and so I will turn to that story.

Submitting a paper at Down House.
After about four hours at Down House, I walked back to the church in Downe to catch the bus. I had a few minutes to spare and so I walked around the church (and saw a plaque saying the sundial was in Darwin’s honor) and in the church (where written material explained how Darwin and his butler, Mr. Parslow, were an integral part of the community). As the bus was arriving, I saw a pub across the street from the church – the George & Dragon. Hmmm, I thought, how could I not have a drink in the bar in Darwin’s home town? So I let the bus go by, committing myself to at least an hour in Downe, and walked across the street to have a pint of Guinness. On my way there, I started to wonder. Could Darwin have gone to this pub? It looked quite old – perhaps he stopped in for a beer or two. Or maybe he spent the whole church service there after his beloved daughter Annie died and his faith was thus permanently shattered.

Emma’s church.
I entered the pub and was reinforced in my romantic hope as it looked really old, down to the low ceiling with rough-hewn and sagging support beams. But it still seemed a silly hope, so I started by asking the bartender some leading questions. “How old is Guinness?” – “Oh, hundreds of years.”  “Cool – and how old is this pub.” – “Oh, considerably older than Guinness.”  “Really,” I say, my excitement mounting. “Could Darwin have come in here for a pint.” – “Oh, yes, certainly. In fact, he stayed upstairs while visiting Downe and looking at the house.”  “Awesome. Perhaps he had a pint of Guinness here – just like I am doing.” – “Oh, that seems likely as he did some business here – see the photo and inscription on the wall.”

Darwin’s pub – the George and Dragon
Guinness in hand, I walk over to a framed document, which included a picture of the pub in the old days – originally called the George Inn – accompanied by an excerpt from the Bromley Record, July 1, 1867.

On Tuesday, 11th June, the Downe Friendly Benefit Society held their 17th anniversary at the GEORGE INN where a most excellent dinner was provided by Mr. and Mrs. Uzzell. The chair was taken by Mr. Snow and the vice-chair by Mr. Parslow. After the cloth was removed and the usual loyal toasts and healths of the treasurer C. R. Darwin Esquire and others, had been given …

Be still my beating heart.

Over the next few hours, I sat in a big comfy chair beside a fireplace that might have warmed Darwin (but not me, owing to fire regulations) and drank several pints while bus after bus went by without me. I edited a paper about the evolution of resistance to parasites. I edited the video asking How did Darwin walk his sandwalk? And I generally absorbed the ambiance and reveled in the thought that I might be sitting in the place where Darwin first scribbled his “I think” diagram – perhaps on a bar napkin.

Darwin’s thinking chair?
OK, I realize I am being overly romantic here. Guinness was probably not on tap in 1860. And, if it was, it was probably not available in the George Inn. And, if it was, Darwin’s delicate stomach probably made him gravitate toward easier fare. And bar napkins probably didn’t exist. And, if they did, Darwin probably didn’t bring his quill to the bar. And, if he did, he probably wasn’t thinking about evolution while drinking. And, of course, he probably scribbled his I think diagram somewhere else (indeed, he did so before buying Down House). But the experience was nevertheless inspiring and the scenario at least plausible in that Darwin might have had some eureka moments in the same physical location I was occupying. Certainly, most of my good ideas have come in bars over a pint of beer or a glass of whisky – at least most of my good blog ideas anyway.

Or maybe Darwin would have preferred this sherry - photo by Mike Hendry
So, the next time you’re in England, by all means visit Darwin’s grave and Down House. Marvel at his writing chair. Be inspired on the sandwalk. But – most of all – don’t forget to visit Darwin’s pub. Bring your computer – do some science. Darwin would want you to.

Wednesday, March 19, 2014

From a conundrum to the mosaic of reproductive isolation

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

In a paper published in a special issue on ecological speciation in 2012 (Räsänen et al. 2012, J. Int. Ecol), we showed that lake and stream threespine stickleback (Gasterosteus aculeatus) from the Misty system, Vancouver island, Canada, do not mate assortatively by ecotype. This apparent lack of mating barriers between highly distinct ecotypes contrasts strongly with the findings from the benthic-limnetic stickleback (now a central model for ecological speciation) and posed us with an intriguing conundrum.

For all of you who struggle with the word “conundrum”: according to Wikipedia it is “a riddle whose answer is or involves a pun or unexpected twist” and/or “a logical postulation that evades resolution, an intricate and difficult problem”. The lake and (inlet) stream fish from the Misty system are phenotypically very divergent (even when reared in the lab over multiple generations) and genetically distinct, yet there are no strong geographic barriers to movement between the lake and its inlet or outlet streams. To us this indeed is a riddle – but does the answer involve a pun or an unexpected twist? That is yet to be seen. It certainly is logical to postulate that we would expect there to be some form of strong reproductive barrier between clearly genetically distinct populations – when there are no strong barriers to movement of individuals. So, if assortative mate choice (e.g. via the “mate with your own type” rule) is not the reproductive barrier, what then keeps these ecotypes distinct?

Reproductive isolation (RI) can arise via many different ecologically and non-ecologically mediated pre- and post-mating barriers. (See Nosil et al. 2005, Evolution 59: 705-719 for a nice overview). For the Misty stickleback, we have evidence that assortative mate choice is weak or non-existent and that genetic incompatibilities are unlikely (hybrid crosses are easy to perform and hybrids are viable). So in a recently published paper (Räsänen and Hendry 2014, Ecology and Evolution), we report on the next step in our hunt for reproductive barriers in this system. Here we tested for selection against migrants (SAM), whereby individuals from populations adapted to a given environment (for example, a lake) are selected against when migrating to other environments (for example, a stream).

What makes the Misty system particularly interesting in this context is that two streams are connected to the lake: the inlet stream, where water flows into the lake, and the outlet stream, where water flows out of the lake. (Why the direction of flow matters, I will return to below). Previous work has repeatedly demonstrated that Inlet stickleback are phenotypically and genetically strongly divergent from the Lake fish, whereas Outlet stickleback are phenotypically and genetically pretty similar to the Lake fish (especially close to the lake).


IMAGE 1.  Representative individuals of wild caught threespine stickleback from the Misty system. Typically, Inlet fish are smaller, deeper bodied and more drab in colouration, whereas Lake and Outlet fish have shallower and more streamlined bodies and darker colouration. These differences are genetically based. © A.P. Hendry.

This difference in divergence has been suggested to reflect the balance between adaptive divergence and gene flow: gene flow from the lake to the outlet is higher at least in part due to the ease of dispersing in the direction of the water flow, combined with the large population size of lake stickleback, whereas gene flow from the lake to the inlet is apparently constrained at least in part because stickleback, unlike salmonids, do not like to swim upstream. However, if the strong phenotypic and genetic divergence between the Lake and Inlet stickleback does reflect adaptive divergence, we would expect any individuals moving to the “wrong” environment to be selected against due to their reduced performance: selection against migrants (SAM), as described earlier.

To test for SAM, we conducted a reciprocal transplant experiment in situ using enclosures. For this, we used fish that were collected from the wild – and hence express the full phenotype within a given environment (keeping in mind here that it is the composite phenotype, including direct genetic, epigenetic, parental and environmental effects, that is under selection in the wild). We erected enclosures in the lake, outlet, and inlet, and then transferred a randomly selected subset of mature-size fish from each site to each enclosure.


IMAGE 2. I am releasing fish to the Inlet enclosure.

We made sure the fish were in good shape prior to transfer, tagged them with coded wire tags, weighed them (to get a grasp of size differences) and photographed them, and then released the fish for three weeks to do whatever fish like to do in a lake or in a stream environment. (Presumably mostly eating the local food and “socializing” – positively or negatively – with their neighbours…)

We made two key predictions. First, since Lake and Inlet fish are strongly phenotypically different (Inlet fish are smaller, deeper bodied and have fewer and shorter gill rakers than Lake fish), whereas Lake and Outlet fish are phenotypically similar (pretty big buggers, with dark blue nuptial coloration, and more limnetic-type feeding morphology), we predicted that Lake and Inlet fish should perform differently, whereas Lake and Outlet fish should perform similarly in all environments. Second, if each population is locally adapted to its native environment, we would expect that each ecotype (Lake, Inlet or Outlet) should perform best in its native environment (Lake in lake, Inlet in inlet…and so on). (Of course, given that our Outlet fish are phenotypically similar to the Lake fish, they should perform well in the lake too…)

It is this second prediction that is the key prediction for SAM: if SAM contributes to RI, we would expect that Inlet fish perform poorly when transferred to the Lake (= Inlet fish would be selected against in the lake), whereas Lake fish perform poorly when transferred to the inlet (= Lake fish would be selected against in the inlet). Nice and easy!

But what did we find?

For the first prediction – yes indeed, phenotypically different Inlet fish did perform differently from Lake (and Outlet) fish in all environments, whereas the phenotypically similar Outlet and Lake fish performed similarly in all environments. (The measures of performance here being change in body mass and survival). This confirmed that phenotype does make a difference for fitness. All good. However, for the second prediction the results were not quite what we would have expected. Basically, although Inlet fish indeed did perform poorly in the lake (they lost a lot weight and had much lower survival), Lake fish seemed to be quite content in the inlet (they had high survival and did not lose much weight).  (For the result details, do check the paper…)

This finding suggests that SAM in the Misty system is asymmetric: it works in one direction (from the inlet to the lake), but not in the other (from lake to the inlet). (I now ignore the Outlet fish as they don’t do much that is unexpected). Ok…so what does this then mean for the conundrum? Does it help to solve the riddle?

Well, yes – partially. It does suggest that SAM indeed makes some contribution to RI – by reducing gene flow from the Inlet to the Lake population. But why do Lake genotypes then not swamp the Inlet?  This is particularly worth asking given that the Lake population (in addition to outperforming Inlet fish in the inlet) is very large and the Inlet population is rather small… so by sheer numbers you’d expect gene flow to constrain adaptive divergence also in the Inlet (unless migrants are selected against, that is). So why does this not happen? There could, of course, be many different reasons. Firstly, given that our experiment only ran for three weeks (and not the whole life-cycle), it may be that we just didn’t pick up the whole suite of performance differences. Secondly, as piscivorous fish and diving birds were not allowed access to the enclosures, most of the performance differences we saw were likely mediated via diet and social interactions (e.g. competitive aggression) and it may be that we excluded some important selective factors.

Or maybe it could be that other reproductive barriers are important? One possibility is temporal isolation (different breeding times). Another is habitat choice… and here we are back where I started: maybe Lake fish just don’t like to move upstream, and thus stay put in the lake (when not venturing to the outlet). This could very well be the case given that several studies by the Hendry and Bolnick labs suggest that sticklebacks don’t readily venture against the flow.

The key, however, is – and here I come back to the word “mosaic” in the title – that maybe, instead of looking for one magic trait and one major reproductive isolating barrier (typically mate choice or genetic incompatibilities), we should be thinking of reproductive isolation as a composite of several different reproductive barriers that can vary in strength and direction. They may be asymmetric (as SAM seems to be in our case), but together with other barriers (possibly habitat choice in our case) they could result in strong TOTAL isolation.

Evidence for such mosaic isolation is certainly emerging from some systems (such as Timema walking sticks, Ischnura damselflies or Mimulus monkeyflowers)… so maybe the solution to the conundrum rests in understanding the relative strength of a multitude of reproductive barriers. There is only one way to know how common single major barriers versus a mosaic of weaker barriers are in nature. So, if you didn’t find evidence for RI in your pet system based on some single mechanism, keep looking!


IMAGE 3. Our mission (at least partially) accomplished! Now it’s your turn... © A.P. Hendry.

Sunday, March 16, 2014

Extinctive – a new addition to the lexicon of biodiversity science?

Much debate in science revolves around terminology – indeed, whole papers are written about specific words. A personal favorite – if only for the title – is Ontoecogenophyloconstraints by Antonovics and van Tienderen. For some reason, terminological issues seem particularly acute in the context of biodiversity science. What precisely are “ecosystem services”? What is sustainability? Tipping points? Earth system services?

As a result, we sometimes get into terminological debates at our bioGENESIS meetings – a few years ago in Cape Town, we even coined a new term “evosystem services.” This term arose from the realization that all ecosystem services are the product of organisms, and all organisms are the product of evolution. One plus one must mean that all past, present, and future ecosystem services are also EVOsystem services. This recognition is important because it makes clear the need to inject evolutionary thinking into biodiversity science, a goal that is – after all – the raison d'etre of bioGENESIS. We (mainly Dan Faith) invented this term over dinner on the edge of the Southern Ocean in Cape Town, and I can remember drawing a circle labelled “ecosystem services” surrounded and completely enveloped by another circle labelled “evosystem services”. The point was that not only are all recognized ecosystem services also evosystem services, but evolution provides many services (past, present, and future) that are not encapsulated by the usual view of ecosystem services. (We had imbibed enough to later draw another even more inclusive circle – geosystem services – and around that another circle – cosmosystem services – and around that yet another circle – theosystem services.)

Spring has sprung at Kew.
This week I attended another bioGENESIS meeting, this time at the Kew Royal Botanical Gardens in London, England – graciously hosted by Felix Forrest. And – yet again – we debated terminology, including the very same words we had debated at past meetings. This time, however, we ended in a different place. We were having dinner at a nice French restaurant (perhaps the best way to be sure of good food in England), and we started considering the meaning of the term “distinct,” which led us to contrast it with “distinctive.” In an effort to figure out how these terms differed, we started considering other words ending in “tinct” to which one could also add “ive” on the end. Instinct versus instinctive, indistinct versus indistinctive, and extinct versus extinctive. Huh? Extinctive? Is that even a word? We had never heard it before and thus reasonably assumed it did not exist in the English language.

Spring has sprung at Kew.
What might “extinctive” mean – if it were to mean anything at all? We decided it would likely mean “having the properties of being extinct without actually being extinct” – at least not yet. The term might thus apply to species that were sure to go extinct in the reasonably near future: that is, a species experiencing an “extinction debt” or “extinction in waiting”. After all, this definition seemed to fit fairly well with “instinct” versus “instinctive.” As an example, Pinta Island giant tortoises were extinctive for 40 years, right up until Lonesome George died just a few years ago. We were having a good time with this debate (at least I was) until Felix got out his smart phone and started looking up words. It turns out that the above four words are the only ones in English to end in “tinct” and, of these, extinct is the only one to which one cannot add an “ive” on the end. Cool! Oops – not true. It seems that extinctive is actually a real word (“tending or serving to extinguish or make extinct”), which presumably preempts our new definition. (Hell, while writing this, I can see that MSWord’s spell checker doesn’t even flag it. If Bill Gates says it is a word, it must be. Then again, while posting it, I see that Blogger does flag it - so Google says it isn't a word. Clash of the Titans!)

Felix showing us what might well be the world's most biodiverse square meter - the genomic DNA storage facility at Kew.
So the great debate ended in a whimper … until Dan pointed out that instinguish was definitely not a word – but should be. And so it went …

Dan demonstrating how to instinguish.

Some earlier posts about evosystem services:



The bioGENESIS crew at Kew: Luc, Felix, Geeta, Lucia, Anne-Helene, Keight, Dan, Melina, and Andrew

Wednesday, March 12, 2014

Evolution in small spaces: microgeographic adaptation and the spatial scale of evolution

Local adaptation has been the focus of intense study for many decades. Given how widely it has been observed across diverse taxa and ecological settings, one may not be too far out on a limb in saying that it is the default evolutionary scenario when suitable conditions exist. But we rarely reflect on what is “local” about this adaptation. In fact, the “local” moniker can be rather misleading. Local adaptation occurs when a population evolves traits that result in higher fitness of native individuals in the home environment relative to individuals from foreign populations, regardless of spatial scale. In a review article in this month’s issue of Trends in Ecology & Evolution, Mark Urban, Dan Bolnick, Dave Skelly, and I lay out the arguments for why we need to think about space more explicitly in studies of local adaptation, particularly at fine spatial scales.   

For example, the classic common garden experiment by Clausen, Keck and Hiesey in the 1930s was a beautiful demonstration of local adaptation in Potentilla glandulosa, a small flowering herb that shows clear divergence in several traits along a ~300 km transect (and 3.6 km elevation gradient). This is an example of local adaptation at a large spatial scale and, more importantly, with the low gene flow expected between the populations investigated. In this sense, adaptive divergence in Potentilla will not surprise most evolutionary ecologists – populations experience very different environments and share little gene flow, allowing them to adapt to their local natural selection regimes without intrusion of maladaptive gene flow.

Clausen, Keck and Hiesey’s 300 km transect and morphology of Potentilla glandulosa populations used in their common garden experiment. Individuals were transplanted to three locations (Stanford, Mather and Timberline) where the experiments were done, demonstrating a genetic basis for local adaptation at a large spatial scale.
Compare that example with local adaptation happening at much finer spatial scales. Evolutionary divergence at very small scales has largely been discounted as unlikely, due primarily to expectations of high gene flow that is expected to disrupt any incipient divergence. Yet more and more examples of fine-scale local adaptation are being documented and reported every year. The threespine stickleback (Gasterosteus aculeatus) provides important examples of this fine-scale adaptation. Data from Andrew Hendry’s lab suggest that stickleback body morphology diverges dramatically between fish separated by tens of meters between lake and adjoining stream habitats (e.g., Hendry & Taylor 2004; Moore & Hendry 2005). At even finer scales, Dan Bolnick’s group has found divergence in stickleback size, coloration, trophic position and diet across only a 1–2 meter cline of water depth in these highly mobile fish (e.g., Snowberg & Bolnick 2012).

Left: A typical inlet stream entering a lake habitat on Vancouver Island in British Columbia. Phenotypic divergence has been documented at this small spatial scale between stickleback inhabiting the inlet (and outlet) streams and the pond basin. Right: Within a lake, stickleback males can evolve divergent coloration, size and trophic position along a water depth gradient of only 1–2 meters. Photos by Dan Bolnick and Chad Brock.
One of the aims of our TREE review article is to highlight the empirical support for and theory behind evolutionary divergence at small spatial scales. However, “small spatial scales” is entirely relative to the study species being investigated and their dispersal attributes. For this reason, we advance three main concepts in our review:

1. We propose a metric called the “wright” that measures phenotypic divergence across space while standardizing this divergence based on the dispersal of the organism. The “wright” is scaled to the dispersal neighborhood of a species or population, representing all of the individuals located within a radius extending two standard deviations from the mean of the dispersal distribution (i.e., dispersal kernel). Scaling the metric based on dispersal allows comparisons of the degree of adaptive divergence among species with very different dispersal abilities. For example, significant phenotypic divergence between two populations of songbirds situated 100 meters apart will be far more unexpected than divergence between snail populations over the same Euclidean distance. However, by scaling the divergence observed by the number of dispersal neighborhoods separating those 100 meters, we can compare the differentiation between birds and snails in a meaningful way. We coined the term “wright” for this metric because of Sewall Wright’s development of the dispersal neighborhood (also called the “gene flow neighborhood”, “Wright’s neighborhood” or “panmictic unit”), and the fact that it is a direct analog to the previously defined “haldane”, a measure of divergence through time.

2. We establish a threshold for distinguishing evolution at fine spatial scales, and formalize the definition of microgeographic adaptation. The microgeographic term has long been used, albeit inconsistently, to describe local adaptation at small spatial scales. This includes divergence across 25 meters in snails to 40 kilometers in brown trout. In order to be applied consistently across studies and species, we define microgeographic adaptation as adaptive divergence occurring within one dispersal neighborhood, an area where dispersal is expected to be frequent enough to prevent genetic drift. In this way, microgeographic adaptation is defined as a special case of local adaptation occurring at spatial scales where populations should experience high gene flow based on the expected levels of dispersal.

Top: A hypothetical landscape with three forest patches of the ‘light’ and ‘dark’ variety supporting populations of a moth species with two distinct color morphs. Each morph has higher fitness on the trees more closely matching their color and providing better camouflage. Bottom: The dispersal distribution (i.e., kernel) for this moth overlaid on the focal ‘light’ forest patch. The red circle delineates the dispersal neighborhood proposed by Wright, with a radius of two standard deviations from the mean of the kernel. Microgeographic adaptation occurs when two populations separated by less than one neighborhood radius adaptively diverge (e.g., the moth morphs diverge between the two forest patches under the kernel). Divergence between sites outside of this neighborhood would be considered local adaptation, but not microgeographic (e.g., between the light forest and the dark forest patch to the left). Adapted from Richardson et al. 2014.
3. We evaluate seven mechanisms that can either initiate or amplify adaptive divergence at fine spatial scales. Microgeographic adaptation is of particular interest because it occurs despite the high potential for mixing between nearby populations, making it unlikely that neutral processes can generate appreciable variation at this scale. This divergence requires some process that increases the strength of natural selection or reduces maladaptive gene flow relative to dispersal ability. The full list can be found in the paper; however, non-random gene flow (e.g. habitat choice and phenotypic sorting), spatially autocorrelated selection regimes, and selective barriers against migrants are three mechanisms that may commonly contribute to microgeographic adaptation.

We also highlight notable examples of microgeographic adaptation, consider broader implications of fine-scale adaptation for ecology and evolutionary biology, and conclude with a discussion of the immense opportunities that exist to more explicitly integrate spatial scale into evolutionary ecology. This includes specific recommendations for evaluating evolutionary processes at fine spatial scales.

The most salient messages we hope result from this article are that (1) we need to start considering space explicitly by incorporating spatial considerations into any study design, (2) understanding the role of dispersal and gene flow is critical to understanding the scale of evolution, and researchers should make a more concerted effort to characterize and quantify the dispersal distributions of our study species, (3) researchers should integrate observations of natural selection, standard experimental methods (e.g, common garden and transplant experiments) and innovative approaches (e.g., introduction and tracking of maladapted genotypes) with an eye towards understanding the minimum scale of evolutionary divergence and the mechanisms driving this divergence, and (4) a standard measure of evolution across space is needed to compare divergence across multiple species. Our hope is that the “wright” will catalyze the collection of data needed to evaluate the generality of microgeographic adapation in nature. Abundant opportunities also exist for creative manipulations of natural selection, dispersal and the genetic makeup of populations in the wild in order to understand how evolution operates at small scales.

After our review was in press, we had a chance to present these ideas as part of a symposium at the January meeting of the American Society of Naturalists at Asilomar in California. The response then and since the article was published has been exceedingly positive. The one objection that has come up several times is from researchers asserting that we have known about and appreciated fine-scale adaptation for a long time. With some probing, however, what they are generally referring to is divergence occurring at spatial scales that are “surprising” to the investigator in that system. Perhaps that’s a necessary starting point, but with this article and the standardized “wright” metric we are trying to move away from subjective assessments of what scales are surprising, to quantitative evaluations of the spatial scale of evolution.

The paper:

Richardson JL, Urban MC, Bolnick DI, Skelly DK. 2014. Microgeographic adaptation and the spatial scale of evolution. Trends in Ecology & Evolution 29 (3): 165-176.

http://www.cell.com/trends/ecology-evolution/abstract/S0169-5347(14)00015-9