An identity crisis: using genomics to determine species identities

This is the fourth (and final) part of the miniseries on the genetics and process of speciation. To start from Part One, click here.

In last week’s post, we looked at how we can use genetic tools to understand and study the process of speciation, and particularly the transition from populations to species along the speciation continuum. Following on from that, the question of “how many species do I have?” can be further examined using genetic data. Sometimes, it’s entirely necessary to look at this question using genetics (and genomics).

Cryptic species

A concept that I’ve mentioned briefly previously is that of ‘cryptic species’. These are species which are identifiable by their large genetic differences, but appear the same based on morphological, behavioural or ecological characteristics. Cryptic species often arise when a single species has become fragmented into several different populations which have been isolated for a long time from another. Although they may diverge genetically, this doesn’t necessarily always translate to changes in their morphology, ecology or behaviour, particularly if these are strongly selected for under similar environmental conditions. Thus, we need to use genetic methods to be able to detect and understand these species, as well as later classify and describe them.

Cryptic species fish
An example of cryptic species. All four fish in this figure are morphologically identical to one another, but they differ in their underlying genetic variation (indicated by the different colours of DNA). Thus, from looking at these fish alone we would not perceive any differences, but their genetic make-up might suggest that there are more than one species…
Cryptic species heatmap example
The level of genetic differentiation between the fish in the above example. The phylogenies on the left and top of the figure demonstrate the evolutionary relationships of these four fish. The matrix shows a heatmap of the level of differences between different pairwise comparisons of all four fish: red squares indicate zero genetic differences (such as when comparing a fish to itself; the middle diagonal) whilst yellow squares indicate increasingly higher levels of genetic differentiation (with bright yellow = all differences). By comparing the different fish together, we can see that Fish 1 and 2, and Fish 3 and 4, are relatively genetically similar to one another (red-deep orange). However, other comparisons show high level of genetic differences (e.g. 1 vs 3 and 1 vs 4). Based on this information, we might suggest that Fish 1 and 2 belong to one cryptic species (A) and Fish 3 and 4 belong to a second cryptic species (B).

Genetic tools to study species: the ‘Barcode of Life’

A classically employed method that uses DNA to detect and determine species is referred to as the ‘Barcode of Life’. This uses a very specific fragment of DNA from the mitochondria of the cell: the cytochrome c oxidase I gene, CO1. This gene is made of 648 base pairs and is found pretty well universally: this and the fact that CO1 evolves very slowly make it an ideal candidate for easily testing the identity of new species. Additionally, mitochondrial DNA tends to be a bit more resilient than its nuclear counterpart; thus, small or degraded tissue samples can still be sequenced for CO1, making it amenable to wildlife forensics cases. Generally, two sequences will be considered as belonging to different species if they are certain percentage different from one another.

Annotated mitogeome
The full (annotated) mitochondrial genome of humans, with the different genes within it labelled. The CO1 gene is labelled with the red arrow (sometimes also referred to as COX1) whilst blue arrows point to other genes often used in phylogenetic or taxonomic studies, depending on the group or species in question.

Despite the apparent benefits of CO1, there are of course a few drawbacks. Most of these revolve around the mitochondrial genome itself. Because mitochondria are passed on from mother to offspring (and not at all from the father), it reflects the genetic history of only one sex of the species. Secondly, the actual cut-off for species using CO1 barcoding is highly contentious and possibly not as universal as previously suggested. Levels of sequence divergence of CO1 between species that have been previously determined to be separate (through other means) have varied from anywhere between 2% to 12%. The actual translation of CO1 sequence divergence and species identity is not all that clear.

Gene tree – species tree incongruences

One particularly confounding aspect of defining species based on a single gene, and with using phylogenetic-based methods, is that the history of that gene might not actually be reflective of the history of the species. This can be a little confusing to think about but essentially leads to what we call “gene tree – species tree incongruence”. Different evolutionary events cause different effects on the underlying genetic diversity of a species (or group of species): while these may be predictable from the genetic sequence, different parts of the genome might not be as equally affected by the same exact process.

A classic example of this is hybridisation. If we have two initial species, which then hybridise with one another, we expect our resultant hybrids to be approximately made of 50% Species A DNA and 50% Species B DNA (if this is the first generation of hybrids formed; it gets a little more complicated further down the track). This means that, within the DNA sequence of the hybrid, 50% of it will reflect the history of Species A and the other 50% will reflect the history of Species B, which could differ dramatically. If we randomly sample a single gene in the hybrid, we will have no idea if that gene belongs to the genealogy of Species A or Species B, and thus we might make incorrect inferences about the history of the hybrid species.

Gene tree incongruence figure
A diagram of gene tree – species tree incongruence. Each individual coloured line represents a single gene as we trace it back through time; these are mostly bound within the limits of species divergences (the black borders). For many genes (such as the blue ones), the genes resemble the pattern of species divergences very well, albeit with some minor differences in how long ago the splits happened (at the top of the branches). However, the red genes contrast with this pattern, with clear movement across species (from and into B): this represents genes that have been transferred by hybridisation. The green line represents a gene affected by what we call incomplete lineage sorting; that is, we cannot trace it back far enough to determine exactly how/when it initially diverged and so there are still two separate green lines at the very top of the figure. You can think of each line as a separate phylogenetic tree, with the overarching species tree as the average pattern of all of the genes.

There are a number of other processes that could similarly alter our interpretations of evolutionary history based on analysing the genetic make-up of the species. The best way to handle this is simply to sample more genes: this way, the effect of variation of evolutionary history in individual genes is likely to be overpowered by the average over the entire gene pool. We interpret this as a set of individual gene trees contained within a species tree: although one gene might vary from another, the overall picture is clearer when considering all genes together.

Species delimitation

In earlier posts on The G-CAT, I’ve discussed the biogeographical patterns unveiled by my Honours research. Another key component of that paper involved using statistical modelling to determine whether cryptic species were present within the pygmy perches. I didn’t exactly elaborate on that in that section (mostly for simplicity), but this type of analysis is referred to as ‘species delimitation’. To try and simplify complicated analyses, species delimitation methods evaluate possible numbers and combinations of species within a particular dataset and provides a statistical value for which configuration of species is most supported. One program that employs species delimitation is Bayesian Phylogenetics and Phylogeography (BPP): to do this, it uses a plethora of information from the genetics of the individuals within the dataset. These include how long ago the different populations/species separated; which populations/species are most related to one another; and a pre-set minimum number of species (BPP will try to combine these in estimations, but not split them due to computational restraints). This all sounds very complex (and to a degree it is), but this allows the program to give you a statistical value for what is a species and what isn’t based on the genetics and statistical modelling.

Vittata cryptic species
The cryptic species of pygmy perches identified within my research paper. This represents part of the main phylogenetic tree result, with the estimates of divergence times from other analyses included. The pictures indicate the physiology of the different ‘species’: Nannoperca pygmaea is morphologically different to the other species of Nannoperca vittata. Species delimitation analysis suggested all four of these were genetically independent species; at the very least, it is clear that there must be at least 2 species of Nannoperca vittata since is more related to N. pygmaea than to other N. vittata species. Photo credits: N. vittata = Chris Lamin; N. pygmaea = David Morgan.

The end result of a BPP run is usually reported as a species tree (e.g. a phylogenetic tree describing species relationships) and statistical support for the delimitation of species (0-1 for each species). Because of the way the statistical component of BPP works, it has been found to give extremely high support for species identities. This has been criticised as BPP can, at time, provide high statistical support for genetically isolated lineages (i.e. divergent populations) which are not actually species.

Improving species identities with integrative taxonomy

Due to this particular drawback, and the often complex nature of species identity, using solely genetic information such as species delimitation to define species is extremely rare. Instead, we use a combination of different analytical techniques which can include genetic-based evaluations to more robustly assign and describe species. In my own paper example, we suggested that up to three ‘species’ of N. vittata that were determined as cryptic species by BPP could potentially exist pending on further analyses. We did not describe or name any of the species, as this would require a deeper delve into the exact nature and identity of these species.

As genetic data and analytical techniques improve into the future, it seems likely that our ability to detect and determine species boundaries will also improve. However, the additional supported provided by alternative aspects such as ecology, behaviour and morphology will undoubtedly be useful in the progress of taxonomy.

From mutation to speciation: the genetics of species formation

The genetics of speciation

Given the strong influence of genetic identity on the process and outcomes of the speciation process, it seems a natural connection to use genetic information to study speciation and species identities. There is a plethora of genetics-based tools we can use to investigate how speciation occurs (both the evolutionary processes and the external influences that drive it). One clear way to test whether two populations of a particular species are actually two different species is to investigate genes related to reproductive isolation: if the genetic differences demonstrate reproductive incompatibilities across the two populations, then there is strong evidence that they are separate species (at least under the Biological Species Concept; see Part One for why!). But this type of analysis requires several tools: 1) knowledge of the specific genes related to reproduction (e.g. formation of sperm and eggs, genital morphology, etc.), 2) the complete and annotated genome of the species (to be able to find and analyse the right genes properly) and 3) a good amount of data for the populations in question. As you can imagine, for people working on non-model species (i.e. ones that haven’t had the same history and detail of research as, say, humans and mice), this can be problematic. So, instead, we can use other genetic information to investigate and suggest patterns and processes related to the formation of new species.

Is reproductive isolation naturally selected for or just a consequence?

A fundamental aspect of studies of speciation is a “chicken or the egg”-type paradigm: does natural selection directly select for rapid reproductive isolation, preventing interbreeding; or as a secondary consequence of general adaptive differences, over a long history of evolution? This might be a confusing distinction, so we’ll dive into it a little more.

Of the two proposed models of speciation, the by-product of natural selection (the second model) has been the more favoured. Simply put, this expands on Darwin’s theory of evolution that describes two populations of a single species evolving independently of one another. As these become more and more different, both in physical (‘phenotype’) and genetic (‘genotype’) characteristics, there comes a turning point where they are so different that an individual from one population could not reasonably breed with an individual from the other to form a fertile offspring. This could be due to genetic incompatibilities (such as different chromosome numbers), physiological differences (such as changes in genital morphology), or behavioural conflicts (such as solitary vs. group living).

Certainly, this process makes sense, although it is debatable how fast reproductive isolation would occur in a given species (or whether it is predictable just based on the level of differentiation between two populations). Another model suggests that reproductive isolation actually might arise very quickly if natural selection favours maintaining particular combinations of traits together. This can happen if hybrids between two populations are not particularly well adapted (fit), causing natural selection to favour populations to breed within each group rather than across groups (leading to reproductive isolation). Typically, this is referred to as ‘reinforcement’ and predominantly involves isolating mechanisms that prevent individuals across populations from breeding in the first place (since this would be wasted energy and resources producing unfit offspring). The main difference between these two models is the sequence of events: do populations ecologically diverge, and because of that then become reproductively isolated, or do populations selectively breed (enforcing reproductive isolation) and thus then evolve independently?

Reinforcement figure.jpg
An example of reinforcement leading to speciation. A) We start with two populations of a single species (a red fish population and a green fish population), which can interbreed (the arrows). B) Because these two groups can breed, hybrids of the two populations can be formed. However, due to the poor combination of red and green fish genes within a hybrid, they are not overly fit (the red cross). C) Since natural selection doesn’t favour forming hybrids, populations then adapt to selectively breed only with similar fish, reducing the amount of interbreeding that occurs. D) With the two populations effectively isolated from one another, different adaptations specific to each population (spines in red fish, purple stripes in green fish) can evolve, causing them to further differentiate. E) At some point in the differentiation process, hybrids move from being just selectively unfit (as in B)) to entirely impossible, thus making the two populations formal species. In this example, evolution has directly selected against hybrids first, thus then allowing ecological differences to occur (as opposed to the other way around).

Reproductive isolation through DMIs

The reproductive incompatibility of two populations (thus making them species) is often intrinsically linked to the genetic make-up of those two species. Some conflicts in the genetics of Population 1 and Population 2 may mean that a hybrid having half Population 1 genes and half Population 2 genes will have serious fitness problems (such as sterility or developmental problems). Dramatic genetic differences, particularly a difference in the number of chromosomes between the two sources, is a significant component of reproductive isolation and is usually to blame for sterile hybrids such as ligers, zorse and mules.

However, subtler genetic differences can also have a strong effect: for example, the unique combination of Population 1 and Population 2 genes within a hybrid might interact with one another negatively and cause serious detrimental effects. These are referred to as “Dobzhansky-Müller Incompatibilities” (DMIs) and are expected to accumulate as the two populations become more genetically differentiated from one another. This can be a little complicated to imagine (and is based upon mathematical models), but the basis of the concept is that some combinations of gene variants have never, over evolutionary history, been tested together as the two populations diverge. Hybridisation of these two populations suddenly makes brand new combinations of genes, some of which may be have profound physiological impacts (including on reproduction).

DMI figure
An example of how Dobzhansky-Müller Incompatibilities arise, adapted from Coyne & Orr (2004). We start with an initial population (center top), which splits into two separate populations. In this example, we’ll look at how 5 genes (each letter = one gene) change over time in the separate populations, with the original allele of the gene (lowercase) occasionally mutating into a new allele (upper case). These mutations happen at random times and in random genes in each population (the red letters), such that the two become very different over time. After a while, these two populations might form hybrids; however, given the number of changes in each population, this hybrid might have some combinations of alleles that are ‘untested’ in their evolutionary history (see below). These untested combinations may cause the hybrid to be infertile or unviable, making the two populations isolated species.

DMI table
The list of ‘untested’ genetic combinations from the above example. This table shows the different combinations of each gene that could be made in a hybrid if these two populations interbred. The red cells indicate combinations that have never been ‘tested’ together; that is, at no point in the evolutionary history of these two populations were those two particular alleles together in the same individual. Green cells indicate ones that were together at some point, and thus are expected to be viable combinations (since the resultant populations are obviously alive and breeding).

How can we look at speciation in action?

We can study the process of speciation in the natural world without focussing on the ‘reproductive isolation’ element of species identity as well. For many species, we are unlikely to have the detail (such as an annotated genome and known functions of genes related to reproduction) required to study speciation at this level in any case. Instead, we might choose to focus on the different factors that are currently influencing the process of speciation, such as how the environmental, demographic or adaptive contexts of populations plays a role in the formation of new species. Many of these questions fall within the domain of phylogeography; particularly, how the historical environment has shaped the diversity of populations and species today.

Phylogeo of speciation
An example of the interplay between speciation and phylogeography, taken from Reyes-Velasco et al. (2018). They investigated the phylogeographic history of several different groups of species within the frog genus Ptychadena; in this figure, we can see how the different species (indicated by the colours and tree on the left) relate to the geography of their habitat (right).

A variety of different analytical techniques can be used to build a picture of the speciation process for closely related or incipient species. A good starting point for any speciation study is to look at how the different study populations are adapting; is there evidence that natural selection is pushing these populations towards different genotypes or ecological niches? If so, then this might be a precursor for speciation, and we can build on this inference with other complementary analyses.

For example, estimating divergence times between populations can help us suggest whether there has been sufficient time for speciation to occur (although this isn’t always clear cut). Additionally, we could estimate the levels of genetic hybridisation (‘introgression’) between two populations to suggest whether they are reasonably isolated and divergent enough to be considered functional species.

The future of speciation genomics

Although these can help answer some questions related to speciation, new tools are constantly needed to provide a clearer picture of the process. Understanding how and why new species are formed is a critical aspect of understanding the world’s biodiversity. How can we predict if a population will speciate at some point? What environmental factors are most important for driving the formation of new species? How stable are species identities, really? These questions (and many more) remain elusive for a wide variety of life on Earth.

 

The direction of evolution: divergence vs. convergence

Direction of evolution

We’ve talked previously on The G-CAT about how the genetic underpinning of certain evolutionary traits can change in different directions depending on the selective pressure it is under. Particularly, we can see how the frequency of different alleles might change in one direction or another, or stabilise somewhere in the middle, depending on its encoded trait. But thinking bigger picture than just the genetics of one trait, we can actually see that evolution as an entire process works rather similarly.

Divergent evolution

The classic view of the direction of evolution is based on divergent evolution. This is simply the idea that a particular species possess some ancestral trait. The species (or population) then splits into two (for one reason or another), and each one of these resultant species and populations evolves in a different way to the other. Over time, this means that their traits are changing in different directions, but ultimately originate from the same ancestral source.

Evidence for divergent evolution is rife throughout nature, and is a fundamental component of all of our understanding of evolution. Divergent evolution means that, by comparing similar traits in two species (called homologous traits), we can trace back species histories to common ancestors. Some impressive examples of this exist in nature, such as the number of bones in most mammalian species. Humans have the same number of neck bones as giraffes; thus, we can suggest that the ancestor of both species (and all mammals) probably had a similar number of neck bones. It’s just that the giraffe lineage evolved longer bones whereas other lineages did not.

Homology figure
A diagrammatic example of homologous structures in ‘hand’ bones. The coloured bones demonstrate how the same original bone structures have diverged into different forms. Source: BiologyWise.

Convergent evolution

But of course, evolution never works as simply as you want it to, and sometimes we can get the direct opposite pattern. This is called convergent evolution, and occurs when two completely different species independently evolve very similar (sometimes practically identical) traits. This is often caused by a limitation of the environment; some extreme demand of the environment requires a particular physiological solution, and thus all species must develop that trait in order to survive. An example of this would be the physiology of carnivorous marsupials like Tasmanian devils or thylacines: despite being in another Class, their body shapes closely resemble something more canid. Likely, the carnivorous diet places some constraints on physiology, particularly jaw structure and strength.

Convergent evol intelligence
A surprising example of convergent evolution is cognitive ability in apes and some bird groups (e.g. corvids). There’s plenty of other animal groups more related to each of these that don’t demonstrate the same level of cognitive reasoning (based on the traits listed in the centre): thus, we can conclude that cognition has evolved twice in very, very different lineages. Source: Emery & Clayton, 2004.

A more dramatic (and potentially obvious) example of convergent evolution would be wings and the power of flight. Despite the fact that butterflies, bees, birds and bats all have wings and can fly, most of them are pretty unrelated to one another. It seems much more likely that flight evolved independently multiple times, rather than the other 99% of species that shared the same ancestor lost the capacity of flight.

Parallel evolution

Sometimes convergent evolution can work between two species that are pretty closely related, but still evolved independently of one another. This is distinguished from other categories of evolution as parallel evolution: the main difference is that while both species may have shared the same start and end point, evolution has acted on each one independent of the other. This can make it very difficult to diagnose from convergent evolution, and is usually determined by the exact history of the trait in question.

Parallel evolution is an interesting field of research for a few reasons. Firstly, it provides a scenario in which we can more rigorously test expectations and outcomes of evolution in a particular environment. For example, if we find traits that are parallel in a whole bunch of fish species in a particular region, we can start to look at how that particular environment drives evolution across all fish species, as opposed to one species case studies.

Marsupial handedness.jpg
Here’s another weird example; different populations of marsupials (particularly kangaroos and wallabies) show preferential handedness depending on where the population is. That is, different populations of different species of marsupials shows parallel evolution of handedness, since they’re related to one another but have evolved it independently of the other species. Source: Giljov et al. (2015).

Following from that logic, it is then important to question the mechanisms of parallelism. From a genetic point of view, do these various species use the same genes (and genetic variants) to produce the same identical trait? Or are there many solutions to the selective question in nature? While these questions are rather complicated, and there has been plenty of evidence both for and against parallel genetic underpinning of parallel traits, it seems surprisingly often that many different genetic combinations can be used to get the same result. This gives interesting insight into how complex genetic coding of traits can be, and how creative and diverse evolution can be in the real world.

Where is evolution going?

Cat phylogeny
An example of all three types of evolutionary trajectory in a single phylogeny of cats (you know how we do it here at The G-CAT). This phylogeny consists of two distinct genera; one with one species (P. aliquam) and another of three species (the red box indicates their distance). Our species have three main physical traits: coat colour, ear tufts and tail shape. At the ancestral nodes of the tree, we can see what the ancestor of these species looked like for these three traits. Each of these traits has undergone a different type of evolution. The tufts on the ears are the result of divergent evolution, since F. tuftus evolved the trait differently to its nearest relative, F. griseo. Contrastingly, the orange coat colour of F. tuftus and P. aliquam are the result of convergent evolution: neither of these species are very closely related (remembering the red box) and evolved orange coats independently of one another (since their ancestors are grey). And finally, the fluffy tails of F. hispida and F. griseo can be considered parallel evolution, since they’re similar evolutionarily (same genus) but still each evolved tail fluff independently (not in the ancestor). This example is a little convoluted, but if you trace the history of each trait in the phylogeny you can more easily see these different patterns.

So, where is evolution going for nature? Well, the answer is probably all over the place, but steered by the current environmental circumstances. Predicting the evolutionary impacts of particular environmental change (e.g. climate change) is exceedingly difficult but a critical component of understanding the process of evolution and the future of species. Evolution continually surprises us with creative solution to complex problems and I have no doubt new mysteries will continue to be thrown at us as we delve deeper.

All the world in the palm of your hand: whole genome sequencing for evolution and conservation

Building an entire genome

If bigger is better, then biggest is best. Having the genome of a particular study species fully sequenced allows us to potentially look at all of the genetic variation in the entire gene pool: but how do we sequence the entirety of the genome? And what are the benefits of having a whole genome to refer to?

Whole genome assembly
A very, very simplified overview of whole genome sequencing. Similar to other genomic technologies, we start by fragmenting the genome into much smaller, easier to sequence parts (reads). We then use a computer algorithm which pieces these reads together into a consecutive sequence based on overlapping DNA sequence (like building a chain out of Lego blocks). From this assembled genome, we can then attach annotations using information from other species’ genomes or genetic studies, which can correlate a particular sequence to a gene, a function of that gene, and the resultant protein from these gene (although not always are all of these aspects included).

Well, assembling the whole genome of an organism for the first time is a very tricky process. It involves taking DNA sequence from only a few individuals, breaking them down into smaller fragments and multiplying these fragments into the billions (moreorless the same process used in other genomics technologies: the real difference is that we need the full breadth of the genome so that we don’t miss any spaces). From these fragments, we use a complex computer algorithm which builds up a consensus sequence like a Lego tower; by finding parts of sequences which overlap, the software figures out which pieces connect to one another. Hopefully, we eventually end up with one very long continuous sequence; the genome! Sometimes, we might end with a few very large blocks (called contigs), but this is also useful for analyses (correlated with how many/big blocks there are). With this full genome, we use information from other more completed genomes (such as those from model species like humans, mice or even worms) to figure out which sections of the genome relate to specific genes. We can then annotate these sections by labelling them as clear genes, complete with start and end point, and attach a particular physical function of that gene.

The benefits of whole genomes

Having an entire genome as a reference is an extremely helpful tool in conservation and evolutionary studies. The first, and perhaps most obvious benefit, is the sheer scale of the data we can use. By having the entirety of the genome available, we can use potentially billions of base pairs of sequence in our genetic analyses (for reference, the human genome is >3 billion base pairs long). Even if we don’t sequence the full genome for all of our samples, having a reference genome as basis for assembly our reduced datasets significantly improves the quantity and quality of sequences we can use.

Another very important benefit is the ability to prescribe function in our studies. Many of our processes for obtaining data, even for genomic technologies, use random and anonymous fragments of the genome. Although this is a cost-effective way to obtain a very large amount of data, it unfortunately means that we often have no idea which part of the genome our sequences came from. This means that we don’t know which sequences relate to specific genes, and even if we did we would have no idea what those genes are or do! But with an annotated genome, we can take even our fragmented sequence and check it against the genome and find out what genes are present.

Understanding adaptation

Based on that, it seems pretty obvious about exactly how having an annotated genome can help us in studies of adaptation. Knowing the functional aspect of our genetic data allows us to more directly determine how evolution is happening in nature; instead of only being able to say that two species are evolving differently from one another, for example, we can explicitly look at how they are evolving. Is one evolving tolerance to hotter temperatures? Are they evolving different genes to handle different diets? Are they evolving in response to an external influence, like a viral outbreak or changing climate? What are the physiological consequences of these changes? These questions are critical in understanding past and future evolution, and full genome analysis allows us to delve into them much deeper.

Manhattan plot example
A (slightly edited) figure of full genome comparisons between domestic dogs and wild wolves by Axelsson et al. (2013), with the aim of understanding the evolutionary changes associated with domestication. For avid readers, this figure probably looks familiar. This figure compares the genetic differentiation across the entire genome between dogs and wolves, with some sections of the genome (circled) showing clear differences. As there is an annotated dog genome, the authors then delved into these genes to understand the functional differences between the two. By comparing their genetic differences to functional genes, the authors can more explicitly suggest mechanisms or changes associated with the domestication process (such as adaptation to a starch-heavy and human-influenced diet).

 

 

This includes allowing us to better understand how adaptation actually works in nature. As we’ve discussed before, more traditional studies often assumed that single, or very few, genes were responsible for allowing a species to adapt and change, and that these genes had very strong effects on their physiology. But what we see far more often is polygenic adaptation; small changes in a very large number of genes which, combined together, allow the species to adapt and evolve. By having the entirety of the genome available, we are much more likely to capture all of the genes that are under natural selection in a particular population or species, painting a clearer picture of their evolutionary trajectory.

Understanding demography

The much larger dataset of full genomes is also important for understanding the non-adaptive parts of evolution; the demographic history. Given that selectively neutral impacts (e.g. reductions in population size) are likely to impact all of the genes in the gene pool somewhat equally, having a full genome allows us to more accurately infer the demographic state and historical patterns of species.

For both adaptive and non-adaptive variation, it is also important to consider what we call linkage disequilibrium. Genetic sequences that are physically close to each other in the genome will often be inherited together due to the imprecision of recombination (a fairly technical process, so I won’t delve into this): what this can mean is that if a gene is under very strong selection, then sequences around this gene will also look like they’re under selection too. This can give falsely positive adaptive genes (i.e. sequences that look like genes under selection but are just linked to a gene that is) or can interfere with demographic analyses (since they often assume no selection, or linkage to selection, on the sequences used). With a whole genome, we can actually estimate how far away a base pair has to be before it’s not linked anymore; we call these linkage blocks, and they’re very useful additions to analyses.

Linkage_example
An example of linkage as a process. We start with a particular sequence (top); during recombination, this sequence may randomly break and rearrange into different parts. In this example, I’ve simulated four different ‘breaks’ (dashed coloured lines) due to recombination. Each of these breaks leads to two separate blocks of fragments; for example, the break at the blue line results in the second two sequence blocks (middle). If we focus on one target base pair in the sequence (golden A), then we can see in some fragments it remains with certain bases, but sometimes it gets separated by the break. If we compare how often the golden A is in the same block (i.e. is co-inherited) as each of the other bases, across all 4 breaks, then we see that the bases that are closest to it (the golden A is represented by the golden bar) are almost always in the same block. This makes sense: the further away a base is from our target, the more likely that there will be a break between it. This is shown in the frequency distributions at the bottom: the left figure shows the actual frequencies of co-inheritance (i.e. linkage) using the top example and those 4 breaks. The right figure shows a more realistic depiction of how linkage looks in the genome; it rapidly decays as we move away from the target (although the width and rate of this can vary).

Improving conservation management

In a similar fashion to demography, full genome datasets can improve our estimates of relatedness and pedigrees in captive breeding programs. The massive scale of whole genomes allows us to more easily trace the genealogical history of individuals, allowing us to assign parents more accurately. This also helps with our estimations of genetic relatedness, arguably the most critical aspect of genetic-based breeding programs. This is particularly helpful for species with tricky mating patterns, such as polyamory, brood spawning or difficult to track organisms.

Pedigrees
An example of how whole genomes can improve our estimation of pedigrees. Say we have a random individual (star), and we want to know how they fit into a particular family tree (pedigree). With only a few genes, we might struggle to pick where in the family it fits based on limited genetic information. With a larger genetic dataset (such as reduced-representation genomics), we might be able to cross off a few potential candidate spots but still have some trouble with a few places (due to unknown parents, polygamy or issues with genetic analysis). With whole genomes, we should be able to much better clarify the whole pedigree and find exactly where our star individual fits in the tree (red circle). It is thanks to whole genomes, we can do those ancestry analyses that have gone viral lately!

The way forwards

While many non-model species are still lacking in the available genomic information, whole genomes are progressively being sequenced for more and more species. As this astronomical dataset grows, our ability to investigate, discover and test theories about evolution, natural selection and conservation will also improve. Many projects already exist which aim specifically to increase the number of whole genomes available for certain taxonomic groups such as birds and bats: these will no doubt prove to be invaluable resources for future studies.

Why we should always pander to diversity

Diversity in the natural world

‘Diversity’ is a term that gets used a lot these days, albeit usually in reference to social changes and structures. However, diversity is not merely a human construct and reflects an extremely important aspect of the natural world at a variety of levels. From the smallest genes to the biggest ecosystems, diversity is a trait that confers a massive range of benefits to individuals, populations, species and even the entire globe. Let’s dissect this diversity down at different scales and see how beneficial it can be.

Hierarchy of diversity
The generalised hierarchy at life, with diversity being an important component of each tier. At the smallest tier, genes underpin all life. The collection of genetic diversity is often summarised into a population (as a single cohesive genetic unit). Several populations can be pooled together into a single (usually) cohesive speciesDifferent species are then components of a larger community (which in turn are components of a broader ecosystem).

Genetic diversity

At the smallest scale in the hierarchy of genetic differentiation, we have the genes themselves. It is a well-established concept that having a diversity of genetic variants (alleles) within a population or species is critical to their future adaptation, evolution and persistance. This is because different alleles will have different benefits (or costs) depending on the environmental pressure that influences them; natural selection might favour one allele over another at one time, but a different one as the pressure changes. Having a higher number of alleles within the population or species means that there is a greater chance at least a few individuals will possess an adaptive gene with the changing environment (which we know can be quite rapid and very, very strong). The diversity serves as a ‘buffer’ against extinction; evolution by natural selection functions best when there are many options to choose from.

Without this diversity, species run the risk of having no adaptive genes at the ready to deal with a selective pressure. Either a new adaptive gene must mutate (or come about in other ways, such as through gene flow from another population or species) or the population/species will suffer and potentially go extinct. As strong selection causes the species to dwindle, it enters what is referred to as the ‘extinction vortex’. Without genetic diversity, they can’t adapt: thus, more individuals die off, causing more genetic diversity to be lost from the population. This pattern is a vicious cycle which can inevitably destroy species (without serious intervention).

Extinction vortex
A very dramatic representation of the extinction vortex.

For this reason, captive breeding programs aim to maintain as much of the genetic diversity of the original population as possible. This reduces the probability of entering a downward extinction spiral from inbreeding depression and helps to maintain populations into the future (both the captive one and the wild population when we reintroduce individuals into the wild).

“Population”  diversity

Because genetic diversity is critically important for species survival, we must also try to preserve the diversity of the entire gene pool of a species. This means conserving highly genetically differentiated populations within a species as a priority, as they may be the only ones that possess the necessary adaptive genes to save the rest of the species. This adaptive genetic variation can then be introduced into other populations in genetic rescue programs and serve as a means to semi-naturally allow the species to evolve. Evolutionarily-significant units (ESUs) are one measure of the invaluable nature of genetically unique populations.

Although many more traditional conservationists strongly believe that ESUs should be managed entirely independently of one another (to preserve their evolutionary ‘pedigree’ and prevent the risk of outbreeding depression), it has been suggested that the benefit of genetic rescue in many cases significantly outweighs this risk of outbreeding depression. For some species, this really is an act of rescue: they are at the edge of extinction, and if we do nothing we condemn them to die out.

Introducing genetic material across populations (or even species!) can generate new functional genes that allow the recipient species to adapt to selective pressures. This might sound very strange, and could be extremely rare, but examples of adaptive genetic material in one species originating from another species through hybridisation do exist in nature. For example, the black coat of wolves is a highly adaptive trait in some populations and is encoded for by the Melanocortin 1 receptor (Mc1r) gene. However, the specific mutation in Mc1r gene that generates the black coat colour actually first originated in domestic dogs; when wild wolves and domestic dogs interbred, this mutation was transferred into the wolf gene pool. Natural selection strongly favoured this new variant, and it very rapidly underwent strong positive selection. Thus, the adaptiveness of black wolves is thanks to a domestic dog mutation!

Species diversity

At a higher level of the hierarchy, the diversity of species within a particular community or ecosystem has been shown to be important for the health and stability of said community. Every species, however small or seemingly unimpressive, plays a role in the greater ecosystem balance, through interactions with other species (e.g. as predator, as prey, as competitor) and the abiotic environment. While some species are known to have very strong impacts on the immediate ecosystem (often dubbed ‘keystone species’, such as apex predators), all species have some influence on the world around them (we’re especially good at it).

Species interactions flowchart

The overall health and stability of an ecosystem, as well as the benefits it can provide to all living things (including humans) is largely determined by the diversity of species. For example, ‘habitat engineers’ are types of species that, by altering the physical environment around them (such as to build a home), directly provide new habitat for other species. They are a fundamental underpinning of many incredibly vibrant ecosystems; think of what a reef system would look like if there were no corals in it. There’d be no anemones growing colourfully; no fish to live in them; no sharks to feed on these non-existent fish. This is just one example of a complex ecosystem that truly relies on its inhabiting species to function.

Ecosystem jenga
Much like Jenga, taking out one block (a species) could cause the entire stack (the ecosystem) to collapse in on itself. Even if it stands up, however, the system will still be weaker without the full diversity to support it.

Protecting our diversity

Diversity is not just a social construct and is an important phenomenon in nature, at a variety of different levels. Preserving the full diversity of life, from genetic diversity within populations and species to full species diversity within ecosystems, is critical to maintaining healthy and robust natural systems. The more diversity we have at each level of this hierarchy, the greater robustness and security we will have in the future.

Surviving the Real-World Apocalypse

The changing world

Climate change seems to be the centrefold of a large amount of scientific research and media attention, and rightly so: it has the capacity to affect every living organism on the planet. It’s our duty as curators and residents of Earth to be responsible for our influences on the global environmental stage. While a significant part of this involves determining causes and solutions to our contributions to climate change, we also need to know how extensive the effects will be: for example, how can we predict how well species will do in the future?

Predicting the effect of climate change on all of the world’s biodiversity is an immense task. Climate change itself is a complicated system, and causes diverse, interconnected and complex alterations to both global and local climate. Adding on top of this, though, is that climate affects different species in different ways; where some species might be sensitive to some climatic variables (such as rainfall, available sunlight, seasonality), others may be more tolerant to the same factors. But all living things share some requirements, so surely there must be some consistency in their responses to climate change, right?

Apocalypse 2
Lucky for Mr Fish here, he’s responding to a (very dramatic) climate change much, much better than his bird counterpart.

How predictable are species responses to climate change?

Well, evidence would surprisingly suggest not. Many species, even closely related ones, can show very different responses to the exact same climatic pressures or biogeographical events. There are a number of different traits that might affect a species’ ability to adapt, particularly their adaptive genetic diversity (which underpins ‘adaptive potential’). Thus, we need good information of a variety of genetic, physiological and life history traits to be able to make predictions about how likely a species is to adapt and respond to future (and current) climate changes.

Although this can be hard to study in species of high extinction risk (getting a good number of samples is always an issue…), traditional phylogeographic methods might help us to make some comparisons. See, although the modern Earth is rapidly changing (undoubtedly influenced by human society), the climate of the globe has always varied to some degree. There has always been some tumultuousness in the climate and specific Earth history events like volcano eruptions, sea-level changes, or glaciation periods (‘ice ages’) have had diverse effects on organisms globally.

Using comparative phylogeography to predict species responses

One tool for looking at how different species have, in the past, responded to the same biogeographical force is the domain of ‘comparative phylogeography’. Phylogeography itself is something we have discussed before: the ‘comparative’ aspect simply means comparing (with complex statistical methods) these patterns across different and often unrelated species to see how universal (‘congruent’) or unique (‘incongruent’) these patterns are among species. The more broadly we look at the species community in the region, the more we can observe widespread effects of any given environmental or geographical event: if we only look at fish, for example, we might not to be able to infer what response mammals, birds or invertebrates have had to our given event. Sometimes this still meets the scale we wish to focus; other times, we want to see how all the species of an area have been affected.

Actual island diagram
An (very busy) example of different species responses to a single environmental event. In this example, we have three species (a fish, a lizard, and a bird) all living on the same island. In the middle of the island, there is a small mountain range (A). At this point in time, all three species are connected across the whole island; fish can travel via lakes and wetlands (green arrows), lizards can travel across the land (blue arrow) and birds can fly anywhere. However, as the mountain range grows with tectonic movements, the waterways are altered and the north and south are disconnected (B). The fish species is now split into two evolutionarily separate groups (green and gold), while lizards and birds are not. As the range expands further, however, the dispersal route for lizards is cut off, causing them to eventually also become separated into blue and black groups (C). Birds, however, have no problems flying over the mountain range and remain one unified and connected orange group over time (D). Thus, each species has a different response to the formation of the mountain range.
Evol history of island diagram
The phylogenetic history of the three different species in the above example. As you can see, each lineage has a slightly different pattern; birds show no divergences at all, whereas the timing of the lizard and fish N/S splits are different (i.e. temporally incongruent).

Typically, comparative phylogeographic studies have looked at the neutral components of species’ evolution (as is the realm of traditional phylogeography). This includes studying the size of populations over time, how well connected they are and were, what their spatial patterns are and how these relate to the environment. Comparing all of these patterns across species can allow us to start painting a fuller picture of the history of biota in a region. In this way, we can start to see exactly which species have shown what responses and start to relate these to the characteristics that allowed them to respond in that certain way (and including adaptation in our studies). So, what kinds of traits are important?

What traits matter? Who wins?

Often, we find that life history traits of an organism better dictates how they will respond to a certain pressure than other factors such as phylogeny (e.g. one group does not always do better than another). Instead, individual species with certain physical characteristics might handle the pressure better than others. For example, a fish, bird and snake that are all able to tolerate higher temperatures than other fish, birds or snakes in that region are more likely to survive a drought. In this case, none of the groups (fish, birds or snakes) inherently do better than the other two groups. Thus, it can be hard to predict how a large swathe of species will respond to any given environmental change, unless we understand the physical characteristics of every species.

Climate change risk flowchart
A generalised framework of various factors, and their interactions, on the vulnerability of species under current and future climate changes by Williams et al. 2018. The schematic includes genetic, ecological, physical and environmental factors and how these can interact with one another to alleviate or exacerbate the risk of extinction.

We can also see that other physiological or ecological traits, such as climatic preferences and tolerance thresholds, can be critical for adapting to climatic pressures. Naturally, the genetic diversity of species is also an important component underlying their ability to adapt to these new selective pressures and to survive into the future. Trying to incorporate all of these factors into a projected model can be difficult, but with more data of higher quality we can start to make more refined predictions. But by understanding how particular traits influence how well a species may adapt to a changing climate, as well as knowing the what traits different species have, might just be the key to predicting who wins and who dies in the real-world Game of Thrones.

Age and dating with phylogenetics

Timing the phylogeny

Understanding the evolutionary history of species can be a complicated matter, both from theoretical and analytical perspectives. Although phylogenetics addresses many questions about evolutionary history, there are a number of limitations we need to consider in our interpretations.

One of these limitations we often want to explore in better detail is the estimation of the divergence times within the phylogeny; we want to know exactly when two evolutionary lineages (be they genera, species or populations) separated from one another. This is particularly important if we want to relate these divergences to Earth history and environmental factors to better understand the driving forces behind evolution and speciation. A traditional phylogenetic tree, however, won’t show this: the tree is scaled in terms of the genetic differences between the different samples in the tree. The rate of genetic differentiation is not always a linear relationship with time and definitely doesn’t appear to be universal.

 

Anatomy of phylogenies.jpg
The general anatomy of a phylogenetic tree. A phylogeny describes the relationships of tips (i.e. which are more closely related than others; referred to as the topology), how different these tips are (the length of the branches) and the order they separated in time (separations shown by the nodes). Different trees can share some traits but not others: the red box shows two phylogenetic trees with similar branch lengths (all of the branches are roughly the same) but different topology (the tips connect differently: A and B are together on the left but not on the right, for example). Conversely, two trees can have the same topology, but show differing lengths in the branches of the same tree (blue box). Note that the tips are all in the same positions in these two trees. Typically, it’s easier to read a tree from right to left: the two tips who have branches that meet first are most similar genetically; the longer it takes for two tips to meet along the branches, the less similar they are genetically.

How do we do it?

The parameters

There are a number of parameters that are required for estimating divergence times from a phylogenetic tree. These can be summarised into two distinct categories: the tree model and the substitution model.

The first one of these is relatively easy to explain; it describes the exact relationship of the different samples in our dataset (i.e. the phylogenetic tree). Naturally, this includes the topology of the tree (which determines which divergences times can be estimated for in the first place). However, there is another very important factor in the process: the lengths of the branches within the phylogenetic tree. Branch lengths are related to the amount of genetic differentiation between the different tips of the tree. The longer the branch, the more genetic differentiation that must have accumulated (and usually also meaning that longer time has occurred from one end of the branch to the other). Even two phylogenetic trees with identical topology can give very different results if they vary in their branch lengths (see the above Figure).

The second category determines how likely mutations are between one particular type of nucleotide and another. While the details of this can get very convoluted, it essentially determines how quickly we expect certain mutations to accumulate over time, which will inevitably alter our predictions of how much time has passed along any given branch of the tree.

Calibrating the tree

However, at least one another important component is necessary to turn divergence time estimates into absolute, objective times. An external factor with an attached date is needed to calibrate the relative branch divergences; this can be in the form of the determined mutation rate for all of the branches of the tree or by dating at least one node in the tree using additional information. These help to anchor either the mutation rate along the branches or the absolute date of at least one node in the tree (with the rest estimated relative to this point). The second method often involves placing a time constraint on a particular node of the tree based on prior information about the biogeography of the species (for example, we might know one species likely diverged from another after a mountain range formed: the age of the mountain range would be our constraints). Alternatively, we might include a fossil in the phylogeny which has been radiocarbon dated and place an absolute age on that instead.

Ammonite comic.jpg
Don’t you know it’s rude to ask an ammomite her age?

In regards to the former method, mutation rates describe how fast genetic differentiation accumulates as evolution occurs along the branch. Although mutations gradually accumulate over time, the rate at which they occur can depend on a variety of factors (even including the environment of the organism). Even within the genome of a single organism, there can be variation in the mutation rate: genes, for example, often gain mutations slower than non-coding region.

Although mutation rates (generally in the form of a ‘molecular clock’) have been traditionally used in smaller datasets (e.g. for mitochondrial DNA), there are inherent issues with its assumptions. One is that this rate will apply to all branches in a tree equally, when different branches may have different rates between them. Second, different parts of the genome (even within the same individual) will have different evolutionary rates (like genes vs. non-coding regions). Thus, we tend to prefer using calibrations from fossil data or based on biogeographic patterns (such as the time a barrier likely split two branches based on geological or climatic data).

The analytical framework

All of these components are combined into various analytical frameworks or programs, each of which handle the data in different ways. Many of these are Bayesian model-based analysis, which in short generates hypothetical models of evolutionary history and divergence times for the phylogeny and tests how well it fits the data provided (i.e. the phylogenetic tree). The algorithm then alters some aspect(s) of the model and tests whether this fits the data better than the previous model and repeats this for potentially millions of simulations to get the best model. Although models are typically a simplification of reality, they are a much more tractable approach to estimating divergence times (as well as a number of other types of evolutionary genetics analyses which incorporating modelling).

Molecular dating pipeline
A (believe it or not, simplified) pipeline for estimating divergence times from a phylogeny. 1) We obtain our DNA sequences for our samples: in this example, we’ll see each Sample (A-E) is a representative of a single species. We align these together to make sure we’re comparing the same part of the genome across all of them. 2) We estimate the phylogenetic tree for our samples/species. In a Bayesian framework, this means creating simulation models containing a certain substitution model and a given tree model (containing certain topology and branch lengths). Together, these two models form the likelihood model: we then test how well this model explains our data (i.e. the likelihood of getting the patterns in our data if this model was true). We repeat these simulations potentially hundreds of thousands of times until we pinpoint the most likely model we can get. 3) Using our resulting phylogeny, we then calibrate some parts of it based on external information. This could either be by including a carbon-dated fossil (F) within the phylogeny, or constraining the age of one node based on biogeographic information (the red circle and cross). 4) Using these calibrations as a reference, we then estimated the most likely ages of all the splits in the tree, getting our final dated phylogeny.

Despite the developments in the analytical basis of estimating divergence times in the last few decades, there are still a number of limitations inherent in the process. Many of these relate to the assumptions of the underlying model (such as the correct and accurate phylogenetic tree and the correct estimations of evolutionary rate) used to build the analysis and generate simulations. In the case of calibrations, it is also critical that they are correctly dated based on independent methods: inaccurate radiocarbon dating of a fossil, for example, could throw out all of the estimations in the entire tree. That said, these factors are intrinsic to any phylogenetic analysis and regularly considered by evolutionary biologists in the interpretations and discussions of results (such as by including confidence intervals of estimations to demonstrate accuracy).

Understanding the temporal aspects of evolution and being able to relate them to a real estimate of age is a difficult affair, but an important component of many evolutionary studies. Obtaining good estimates of the timing of divergence of populations and species through molecular dating is but one aspect in building the picture of the history of all organisms, including (and especially) humans.

The many genetic faces of adaptation

The transition from genotype to phenotype

While evolutionary genetics studies often focus on the underlying genetic architecture of species and populations to understand their evolution, we know that natural selection acts directly on physical characteristics. We call these the phenotype; by studying changes in the genes that determine these traits (the genotype), we can take a nuanced approach at studying adaptation. However, our ability to look at genetic changes and relate these to a clear phenotypic trait, and how and why that trait is under natural selection, can be a difficult task.

One gene for one trait

The simplest (and most widely used) models of understanding the genetic basis of adaptation assume that a single genotype codes for a single phenotypic trait. This means that changes in a single gene (such as outliers that we have identified in our analyses) create changes in a particular physical trait that is under a selective pressure in the environment. This is a useful model because it is statistically tractable to be able to identify few specific genes of very large effect within our genomic datasets and directly relate these to a trait: adding more complexity exponentially increases the difficulty in detecting patterns (at both the genotypic and phenotypic level).

Single locus figure
An example of a single gene coding for a single phenotypic trait. In this example, the different combination of alleles of the one gene determines the colour of the cat.

Many genes for one trait: polygenic adaptation

Unfortunately, nature is not always convenient and recent findings suggest that the overwhelming majority of the genetics of adaptation operate under what is called ‘polygenic adaptation’. As the name suggestions, under this scenario changes (even very small ones) in many different genes combine together to have a large effect on a particular phenotypic trait. Given the often very small magnitude of the genetic changes, it can be extremely difficult to separate adaptive changes in genes from neutral changes due to genetic drift. Likewise, trying to understand how these different genes all combine into a single functional trait is almost impossible, especially for non-model species.

Polygenic adaptation is often seen for traits which are clearly heritable, but don’t show a single underlying gene responsible. Previously, we’ve covered this with the heritability of height: this is one of many examples of ‘quantitative trait loci’ (QTLs). Changes in one QTL (a single gene) causes a small quantitative change in a particular trait; the combined effect of different QTLs together can ‘add up’ (or counteract one another) to result in the final phenotype value.

Height QTL
An example of polygenic quantitative trait loci. In this example, height is partially coded for by a total of ten different genes: the dominant form of each gene (Capitals, green) provides more height whereas the recessive form (lowercase, red) doesn’t. The cumulative total of these components determines how tall the person is: the person on the far right was very unlucky and got 0/10 height bonuses and so is the shortest. Progressively from left to right, some genes are contributing to the taller height of the people, with the far right person standing tall with the ultimate 10/10 pro-height genes. For reference, height is actually likely to be coded for by thousands of genes, not 10.

The mechanisms which underlie polygenic adaptation can be more complex than simple addition, too. Individual genes might cause phenotypic changes which interact with other phenotypes (and their underlying genotypes) to create a network of changes. We call these interactions ‘epistasis’, where changes in one gene can cause a flow-on effect of changes in other genes based on how their resultant phenotypes interact. We can see this in metabolic pathways: given that a series of proteins are often used in succession within pathways, a change in any single protein in the process could affect every other protein in the pathway. Of course, knowing the exact proteins coded for every gene, including their physical structure, and how each of those proteins could interact with other proteins is an immense task. Similar to QTLs, this is usually limited to model species which have a large history of research on these specific areas to back up the study. However, some molecular ecology studies are starting to dive into this area by identifying pathways that are under selection instead of individual genes, to give a broader picture of the overall traits that are underlying adaptation.

Labrador epistasis figure
My favourite example of epistasis on coat colour in labradors. Two genes together determine the colour of the coat, with strong interactions between them. The first gene (E/e) determines whether or not the underlying coat gene (B/b) is masked or not: two recessive alleles of the first gene (ee) completely blocks Gene 2 and causes the coat to become golden regardless of the second gene genotype (much like my beloved late childhood pet pictured, Sunny). If the first gene has at least one dominant allele, then the second gene is allowed to express itself. Possessing a dominant allele (BB or Bb) leads to a black lab; possessing two recessive alleles (bb) makes a choc lab!
Labrador epistasis table
The possible combinations of genotypes for the two above genes and the resultant coat colour (indicated by the box colour).

One gene for many traits: pleiotropy and differential gene expression

In contrast to polygenic traits, changes in a single gene can also potentially alter multiple phenotypic traits simultaneously. This is referred to as ‘pleiotropy’ and can happen if a gene has multiple different functions within an organism; one particular protein might be a component of several different systems depending on where it is found or how it is arranged. A clear example of pleiotropy is in albino animals: the most common form of albinism is the result of possessing two recessive alleles of a single gene (TYR). The result of this is the absence of the enzyme tyrosinase in the organism, a critical component in the production of melanin. The flow-on phenotypic effects from the recessive gene most obviously cause a lack of pigmentation of the skin (whitening) and eyes (which appear pink), but also other physiological changes such as light sensitivity or total blindness (due to changes in the iris). Albinism has even been attributed to behavioural changes in wild field mice.

Albinism pleiotropy
A very simplified diagram of how one genotype (the albino version of the TYR gene) can lead to a large number of phenotypic changes via pleiotropy (although many are naturally physiologically connected).

Because pleiotropic genes code for several different phenotypic traits, natural selection can be a little more complicated. If some resultant traits are selected against, but others are selected for, it can be difficult for evolution to ‘resolve’ the balance between the two. The overall fitness of the gene is thus dependent on the balance of positive and negative fitness of the different traits, which will determine whether the gene is positively or negatively selected (much like a cost-benefit scenario). Alternatively, some traits which are selectively neutral (i.e. don’t directly provide fitness benefits) may be indirectly selected for if another phenotype of the same underlying gene is selected for.

Multiple phenotypes from a single ‘gene’ can also arise by alternate splicing: when a gene is transcribed from the DNA sequence into the protein, the non-coding intron sections within the gene are removed. However, exactly which introns are removed and how the different coding exons are arranged in the final protein sequence can give rise to multiple different protein structures, each with potentially different functions. Thus, a single overarching gene can lead to many different functional proteins. The role of alternate splicing in adaptation and evolution is a rarely explored area of research and its importance is relatively unknown.

Non-genes for traits: epigenetics

This gets more complicated if we consider ‘non-genetic’ aspects underlying the phenotype in what we call ‘epigenetics’. The phrase literally translates as ‘on top of genes’ and refers to chemical attachments to the DNA which control the expression of genes by allowing or resisting the transcription process. Epigenetics is a relatively new area of research, although studies have started to delve into the role of epigenetic changes in facilitating adaptation and evolution. Although epigenetics is still a relatively new research topic, future research into the relationship between epigenetic changes and adaptive potential might provide more detailed insight into how adaptation occurs in the wild (and might provide a mechanism for adaptation for species with low genetic diversity)!

 

The different interactions between genotypes, phenotypes and fitness, as well as their complex potential outcomes, inevitably complicates any study of evolution. However, these are important aspects of the adaptation process and to discard them as irrelevant will not doubt reduce our ability to examine and determine evolutionary processes in the wild.

Fantastic Genes and Where to Find Them

The genetics of adaptation

Adaptation and evolution by natural selection remains one of the most significant research questions in many disciplines of biology, and this is undoubtedly true for molecular ecology. While traditional evolutionary studies have been based on the physiological aspects of organisms and how this relates to their evolution, such as how these traits improve their fitness, the genetic component of adaptation is still somewhat elusive for many species and traits.

Hunting for adaptive genes in the genome

We’ve previously looked at the two main categories of genetic variation: neutral and adaptive. Although we’ve focused predominantly on the neutral components of the genome, and the types of questions about demographic history, geographic influences and the effect of genetic drift, they cannot tell us (directly) about the process of adaptation and natural selective changes in species. To look at this area, we’d have to focus on adaptive variation instead; that is, genes (or other related genetic markers) which directly influence the ability of a species to adapt and evolve. These are directly under natural selection, either positively (‘selected for’) or negatively (‘selected against’).

Given how complex organisms, the environment and genomes can be, it can be difficult to determine exactly what is a real (i.e. strong) selective pressure, how this is influenced by the physical characteristics of the organism (the ‘phenotype’) and which genes are fundamental to the process (the ‘genotype’). Even determining the relevant genes can be difficult; how do we find the needle-like adaptive genes in a genomic haystack?

Magnifying glass figure
If only it were this easy.

There’s a variety of different methods we can use to find adaptive genetic variation, each with particular drawbacks and strengths. Many of these are based on tests of the frequency of alleles, rather than on the exact genetic changes themselves; adaptation works more often by favouring one variant over another rather than completely removing the less-adaptive variant (this would be called ‘fixation’). So measuring the frequency of different alleles is a central component of many analyses.

FST outlier tests

One of the most classical examples is called an ‘FST outlier test’. This can be a bit complicated without understanding what FST is actually measures: in short terms, it’s a statistical measure of ‘population differentiation due to genetic structure’. The FST value of one particular population can determine how genetically similar it is to another. An FST value of 1 implies that the two populations are as genetically different as they could possibly be, whilst an FST value of 0 implies that they are genetically identical populations.

Generally, FST reflects neutral genetic structure: it gives a background of how, on average, different are two populations. However, if we know what the average amount of genetic differentiation should be for a neutral DNA marker, then we would predict that adaptive markers are significantly different. This is because a gene under selection should be more directly pushed towards or away from one variant (allele) than another, and much more strongly than the neutral variation would predict. Thus, the alleles that are way more or less frequent than the average pattern we might assume are under selection. This is the basis of the FST outlier test; by comparing two or more populations (using FST), and looking at the distribution of allele frequencies, we can pick out a few alleles that vary from the average pattern and suggest that they are under selection (i.e. are adaptive).

There are a few significant drawbacks for FST outlier tests. One of the most major ones is that genetic drift can also produce a large number of outliers; in a small population, for example, one allele might be fixed (has a frequency of 1, with no alternative allele in the population) simply because there is not enough diversity or population size to sustain more alleles. Even if this particular allele was extremely detrimental, it’d still appear to be favoured by natural selection just because of drift.

Drift leading to outliers diagram
An example of genetic drift leading to outliers, featuring our friends the cat population. Top row: Two cat populations, one small (left; n = 5) and one large (middle, n = 12) show little genetic differentiation between them (right; each triangle represents a single gene or locus; the ‘colour’ gene is marked in green). The average (‘neutral’) pattern of differentiation is shown by the dashed line. Much like in our original example, one cat in the small population is horrifically struck by lightning and dies (RIP again). Now when we compare the frequency of the alleles of the two populations (bottom), we see that (because a green cat died), the ‘colour’ locus has shifted away from the general trend (right) and is now an outlier. Thus, genetic drift in the ‘colour’ gene gives the illusion of a selective loci (even though natural selection didn’t cause the change, since colour does not relate to how likely a cat is to be struck by lightning).

Secondly, the cut-off for a ‘significant’ vs. ‘relatively different but possibly not under selection’ can be a bit arbitrary; some genes that are under weak selection can go undetected. Furthermore, recent studies have shown a growing appreciation for polygenic adaptation, where tiny changes in allele frequencies of many different genes combine together to cause strong evolutionary changes. For example, despite the clear heritable nature of height (tall people often have tall children), there is no clear ‘height’ gene: instead, it appears that hundreds of genes are potentially very minor height contributors.

Polygenic height figure final
In this example, we have one tall parent (top) who produces two offspring; one who is tall (left) and one who isn’t (right). In order to understand what genetic factors are contributing to their height differences, we compare their genetics (right; each dot represents a single locus). Although there aren’t any particular loci that look massively different between the two, the cumulative effect of tiny differences (the green triangles) together make one person taller than the other. There are no clear outliers, but many (poly) different genes (genic) acting together.

Genotype-environment associations

To overcome these biases, sometimes we might take a more methodological approach called ‘genotype-environment association’. This analysis differs in that we select what we think our selective pressures are: often environmental characteristics such as rainfall, temperature, habitat type or altitude. We then take two types of measures per individual organism: the genotype, through DNA sequencing, and the relevant environmental values for that organisms’ location. We repeat this over the full distribution of the species, taking a good number of samples per population and making sure we capture the full variation in the environment. Then we perform a correlation-type analysis, which seeks to see if there’s a connection or trend between any particular alleles and any environmental variables. The most relevant variables are often pulled out of the environmental dataset and focused on to reduce noise in the data.

The main benefit of GEA over FST outlier tests is that it’s unlikely to be as strongly influenced by genetic drift. Unless (coincidentally) populations are drifting at the same genes in the same pattern as the environment, the analysis is unlikely to falsely pick it up. However, it can still be confounded by neutral population structure; if one population randomly has a lot of unique alleles or variation, and also occurs in a somewhat unique environment, it can bias the correlation. Furthermore, GEA is limited by the accuracy and relevance of the environmental variables chosen; if we pick only a few, or miss the most important ones for the species, we won’t be able to detect a large number of very relevant (and likely very selective) genes. This is a universal problem in model-based approaches and not just limited to GEA analysis.

New spells to find adaptive genes?

It seems likely that with increasing datasets and better analytical platforms, many more types of analysis will be developed to delve deeper into the adaptive aspects of the genome. With whole-genome sequencing starting to become a reality for non-model species, better annotation of current genomes and a steadily increasing database of functional genes, the ability of researchers to investigate evolution and adaptation at the genomic level is also increasing.

Pseudo or science? Interpreting scientific reports

Telling the real from the fake

The phrase ‘fake news’ seems to get thrown around ad nauseum these days, but there’s a reason for it (besides the original somewhat famous coining of the phrase). Inadvertently bad, or sometimes downright malicious, reporting of various apparent ‘trends’ or ‘patterns’ are rife throughout nearly all forms of media. Particularly, many entirely subjective or blatantly falsified presentations or reports of ‘fact’ cloud real scientific inquiry and its distillation into the broader community. In fact, a recent study has shown that falsified science spreads through social media at orders of magnitude faster than real science: so why is this? And how do we spot the real from the fake?

It’s imperative that we understand what real science entails to be able to separate it from the pseudoscience. Of course, scientific rigour and method are always of utmost importance, but these can be hard to detect (or can be effectively lied through colourful language choices). When reading a scientific article, whether it’s direct from the source (a journal, such as Nature or Science) or secondarily through a media outlet such as the news or online sources, there’s a few things that you should always look for that will help discern between the two categories.

Peer-review and adequate referencing

Firstly, is the science presented in an objective, logical manner? Does it systematically demonstrate the study system and question, with the relevant reference to peer-reviewed literature? Good science builds upon the wealth of previously done good science to contribute to a broader field of knowledge; in this way, critical observations and alternative ideas can be compared and contrasted to steer the broader field. Even entirely novel science, which go against the common consensus, will reference and build upon prior literature and justify the necessity and design of the study. Having written more than one literature review in my life, I can safely assure you that there is no shortage of relevant scientific studies that need to be read, understood and built upon in any future scientific study.

 

Methods, statistics and sampling

Secondly, is there a solid methodological basis for the science? In almost all cases this will include some kind of statistical measure for the validity (and accuracy) of the results. How does the sample size of the study measure up to what the target group? Remember, a study size of 500 people is definitely too small to infer the medical conditions of all humans, but rarely do we get sample sizes that big in evolutionary genetics studies (especially in non-model species). The sampling regime is extremely important for interpreting the results: particularly, keep in mind if there is an inherent bias in the way the sampling has been done. Are some groups more represented than others? Where do the samples come from? What other factors might be influencing the results, based on the origin of the samples?

Cat survey comic 2
Despite having a large sample size, and a significant result (p<0.05), this study cannot conclude that all dogs are awful. It can conclude, however, that cats are statistically significant assholes.

Presentation and language of findings

Thirdly, how does the source present the results? Does it make claims that seem beyond a feasible conclusion based on the study itself? Even if the underlying study is scientific, many secondary sources have a tendency to ‘sensationalise’ the results in order to make them both more appealing and more digestible to the general public. This is only exacerbated by the lack of information of the scientific method of the original paper, actual statistics, or the accurate summation of those statistics. Furthermore, a real scientific study will try to (in most cases) avoid evocative words such as ‘prove’, as a fundamental aspect of science is that no study is 100% ‘proven’ (see falsifiability below). Proofs are a relevant mathematical concept though, but these fall under a different category altogether.

Here’s an example: recently, an Australian mainstream media outlet (among many) shared a story about a ‘recent’ (six months old) study that found that second-born children are more likely to be criminals and first-born children have higher IQ. As you might expect, the original study does not imply that being born second will make you a sudden murderer nor will being the first born make you a prodigy. Instead, the authors suggest that there may be a link between differential parental investment/attention (between different age order children) as a potential mechanism. They ruled out, based on a wealth of statistics, the influence of alternative factors such as health or education (both in quality and quantity). Thus, there is a correlative (read: not causative) effect of age on these characteristics. If you directly interpreted the newscast (or read some of the misguided comments), you might think otherwise.

Falsifiability 

Fourthly, are the hypotheses in the study falsifiable? One of the foundations of the modern scientific method includes the requirement of any real scientific hypothesis to be falsifiable; that is, there must be a way to show evidence against that hypothesis. This can be difficult to evaluate, but is why some broad philosophical questions are considered ‘unscientific’. A classic example is the phrase “all swans are white”, which was apparently historically believed in Europe (where there are no black swans). This statement is technically falsifiable, since if one found a non-white swan it would ‘disprove’ the hypothesis. Lo and behold, Europeans arrive in Australia and find that, actually, some swans are black. The original statement was thus falsified.

Swan comic 2
Well, I’ll be damned falsified. Just pretend the swan is actually black: I don’t have enough ink to make it realistic…

The role of the peer: including you!

Peer-review is a critical aspect of scientific process, and despite some conspiracy-theory-esque remarks about the secret Big Science Society, it generally works. While independent people inevitably have their own personal biases and are naturally subjective to some degree (no matter how hard we may try to be objective), a larger number of well-informed, critical thinkers help to broaden the focus and perspective surrounding any scientific subject. Remember, nothing is more critical of science than science itself.

Peer review comic
One of the most apt representations of peer-review I’ve ever seen, from Dr. Nick D. Kim (PhD). Source: here.

While peer-review is technically aimed at other scientists as a way to steer and inform research, the input of outsider, non-specialist readers can still be informative. By closely looking at science, and better understanding both how it is done and what it is showing, can help us evaluate how valuable science is to broader society and shift scientific information into useful, everyday applications. Furthermore, by educating ourselves on what is real science, and what is disruptive drivel, we can aid the development of science and reduce the slowing impact of misinformation and deceit.