Speciation Blog @speciationlog-blog - Tumblr Blog

Inferring from genomic ‘island’ of speciation

Reading Cruickshank &Hann 2014 Mol. Ecol. 23: 3133-3157

Genomic ‘island’ of speciation

After one lineage diverges into two, there is continuous sorting of ancestral genetic diversity and accumulation of species-specific diversity. A common genomic pattern among recently diverged lineages is the “island of speciation” pattern, in which there are only a small number of differentiated loci (‘island’) among the ‘ocean’ of undifferentiated loci in the genome.

Speciation-with-gene-flow model and others

The speciation-with-gene-flow model (Feder & Nosil 2010; Flaxman et al. 2012, 2013) claims that those genomic ‘islands’ represents regions of locally adaptive alleles or reproductive barriers thus are resistant to introgression/gene flow. This introgression effect could be captured by reduced relative genetic distance measures (Fst) on non-island regions.

In addition to the selection-migration mechanism, heterogeneous mutation rate, selection, and ancestral coalescence time can all lead similar ‘island’ pattern. Background selection and ancestral polymorphism in very recently diverged lineages can both lead to ‘island’ of differentiated loci without migration homogenizing the rest of the genomes. Mutation and recombination rate, selection pressure can be variable across the genome and through time, leading to heterogeneous patter of genetic diversity that might not be generated by processes under the model.

Dxy, Fst, Da, Df – what do they say?

This paper argues that we should look at absolute genetic distance (Dxy), and consider other possibilities in the organismic system under study. Especially for primary speciation-with-gene-flow scenario, migration among lineages should be tested before adapting the model for the data. Before thinking about the distance measures and their evolutionary implications, I should remember them properly.

Dxy: absolute measure of differentiation, independent of the diversity within the sub-lineages. Affected by: ancestral genetic diversity and substitution rate. Expected Dxy=variation within population + expected neutral site difference=(1-q)*a+ (r+4m)/8Nemr. It decreases as r (recombination rate) and m (migration rate) increases.

Da: net nucleotide difference between two populations, only includes the differences accumulated since the split of the lineages. Da=Dxy-(pi(x)+pi(y))/2

Df: the number of fixed differences between populations. It is a function of time since divergence and time to coalescence within populations.

Model selection

We compare values of Fst and Dxy in island and non-island genetic regions as a model-probing test. [To be added]

Reinforcement evolution: the balance of search cost and hybrid low fitness

Chap 14 of book by T. Price 2008

Reinforcement could be an important completing step of premating isolation, unless the mate preference has diverged well enough in allopatry, so that heterospecific mating is prohibited when individuals come together again. Preference for conspecifics should be selected if both the hybrid fitness and search cost for conspecifics are low. The graph on pp 305 shows when should heterospecific mating be expected considering relative hybrid fitness and search cost for conspecifics.

Often times when hybrid fitness is low, many factors could mitigate it. In the system I study, towsend’s warblers are better at keeping their territories than hermit and hybrids. Female hermit might benefit by mating with the good territory-holder, towsend’s, and potentially their offspring, if have gotten these good territorial genes form the townsend’s, could be more successful than hermit. Also, if hermits are hard to find, the females should be selected to mate with the more frequent type, if the hybrids fitness is no worse than hermits.

Sympatric speciation: still need solid evidence

Reading notes for Chap 4 Coyne and Orr 2004 book

This is the place where speciation occurs at free migration (m=0.5), the most miraculous speciation process (to me), if occurs for real. The initial reduction in gene flow was not due to geographical barriers or distance, but due to some features of the organism. Models for sympatric speciation have focussed on tackling two problems: competitive displacement and recombination.

Disruptive sexual selection

Models in this direction has been beset by competitive displacement, after divergence of lineages driven by sexual selection, one species, if became disadvantaged/rarer, can be lost by species drift. These models tend to have unrealstic assumptions. The Turner & Burrow (1995) model assumed dominance of genes for female mate preference and evaluate multiple males before mating, the “best of n” mating rule. Gavirlets & Waxman (2002) incorporated sexual conflict in this arena, argued that “chase-away” selection can subdivide a population and having the subdivisions evolving in different directions. However, these models can not explain how do the new lineages coexist in sympatry while maintaining reproductive isolation. Thus the more realistic models have to integrate ecological divergence. van Doorn (1998) showed that a combination of sexual selection and disruptive natural selection could work. Ecological divergence has to precede or follow the evolution of reproductive isolation through sexual selection to maintain species coexistence at sympatry.

With that said, sexual selection can still and very likely play an important role in sympatric (?) speciation. In many radiation systems (e.g. cichlids in lake lake Victoria and Mlawi), species differ strongly in sexually selected traits than ecological traits. It’s thought in this cichlid system, ecological adaptation for feeding efficient and the succeeding sexual selection in breeding color contributed to radiation.

species drift: loss of the rarer species by random

Disruptive natural selection

There are two types of models: discrete and continuous resource models. The discrete model requires the evolution of niche preference, niche adaptation, and assortative mating. To allow for connection of niche adaptation and assortative mating, habitat selection has to be strong, and the linkage of genes underlying adaptive traits and mating preference. There also has to be high genetic variation of all the three traits to start with.

However this model doesn’t help the organismic systems that do not have drastic ecological differentiation. For example, the cichlids diverged in almost identical habitat. So a continuous model of natural selection was proposed. Dieckmann and Doebeli (1999) proposed unimodel distribution of both resource and ecological trait for this resource, frequency-dependent selection favors individuals of extreme trait values, as they can exploit open resources. Selection favors coupling of mating and trait genes which splinter population into discrete groups. The critic for this model is the puzzling destine of opened central niche after divergence, as the subsequent filling-in of new population might smooth out the trait distribution again. It is also criticized for assuming strong trait-based preference that might be rare in nature. A more realistic model was the Kondrashov & Kondrashov (1999) model that tried “neutral-trait” for female preference. Genetic drift allows week linkage of genes underlying the neutral-trait (e.g. color) and ecological-trait. Then females will be selected for mating with males of similar color.

Evidence-seeking

Comparative analysis: most recently formed species have sympatric ranges and the more distantly related species have less overlap in ranges, due to post-speciation range expansion. However, this is restricted to groups that have fossil evidence that range shifts were small. Also we should focus on whether the recently diverged lineages overlap in ranges or not, and not use regression intercept and slope for this interest, because the older groups might not be informative.

#sympatric speciation

Birdsong and divergence

What are the causes of song evolution? How are song signals used in social interaction among birds? Would song evolution lead to the formation of premating barriers? The chapter 9 and 10 of T. Price, Speciation in Birds book attempted to answer these questions.

First there are a couple key terms to clarify:

Social selection: “selection on traits that result in advantages in social interactions, usually between member of the same species.”

genetic assimilation: “the process by which a phenotypic trait that is selected by some environmental stimuli, is taken over by genotype, so that is found even in absence of the environmental influence that had at firs been necessary to produce it.”

Song learning and divergence

The three clades of birds that can do vocal learning, learning/improvisation of songs, encompass more than 50% of the total number of bird species, which casts on a hint that song learning might accelerate speciation. Vocal learning is such a cool process in which birds invent orders and compositions of syllables, pitch, among other aspects of the song. Some of the innovations can spread in a population, making the species-recognition signals different than before.

The book considered three ways that could predispose populations to discrete song types : 1) some song variant might produce more stimulus to the receivers than others; 2) new variant can arise through copy errors, and these variants can be established through drift; 3) big scale disruption in song transmission can happen through “withdraw of learning”, when insufficiently tutored birds colonized a new population.

Once the frequency of new song variant is high enough, through drift or selection, there is usually frequency dependent selection, favoring the common song type, speeding up its spread. Moreover genetic assimilation can kick in sometimes following cultural evolution. Notice that genetic assimilation only occur in species that undergo vocal learning as the species that don’t learn their songs starts song evolution with genetic evolution. Repertoire size is likely to evolve genetically in response to selection in the environment (large repertoires are costly, but attractive to the females). Physiological and morphological constraints in the ability to learn certain can also lead to song differences that are underlay by genetic modulation of these constraints. These constrains could differentiate through various level ecological selection pressures, in parapatry, for example.

I can imagine that such song evolution started with vocal learning could lead to allopatric (especially through founder effect), and parametric speciation. After a founder or bottleneck event, the founders might sing insufficiently tutored song, which will be transmitted in the following generations in this population, eventually lead to premating isolation between the founder population and the initial source population. In allopatric populations isolated by certain geographical barriers, the origin and spread of cultural mutations might happen in parallel, leading to divergence in species-recognition signals among isolated populations. At parapatry, ecological selection gradient might result in a gradation of songs (e.g. certain songs might stand out more over certain ecological background) that lead to “regiolects”, termed by Martens (1996), which is a step further than dialect in the process of speciation. These regiolects can be more distinct than dialects among groups, and partially reproductively isolate the groups.

However, song learning might not facilitate sympatry speciation unless there is some other force that reproductively isolate these song variant groups. For example, if some features of female mating bias is linked to the song variants somehow, male song variants can assortatively mate with female bias variants, and cause premating isolation among song clusters. The song variants arise in the population that undergo song learning is insufficient to cause differentiation, because the new variant might either be lost or become common enough to be stabilized with frequency dependent selection, which does not subdivide the population.

Song and social interaction

There are a few interesting facts about how birds use their songs that I should take a note of.

In terms of development of the song, after fledging from the nest, birds started singing plastic songs, in which the order and structure of the syllables are highly variable, and different song types sang by different tutors can be heard in plastic songs. Then at adult song is the crystallized song that deleted many syllable types. Young birds tend to learn the song types that they heard repetitively.

As for response of the songs, birds tend to respond more strongly to the songs that belong to their own dialects, and the intention to retreat at an aggressive encounter is reflected by sing the song type that is not in the opponent’s repertoire; males singing rare songs are more likely to lose their territories. Birds also change their songs among breeding seasons to match their neighbors. There is a relationship between song and population. In larger populations, birds tend to sing songs with greater diversity of syllables.

#song learning #speciation #genetic assimilation #cultural evolution #premating isolation

Behavior ecology, island biogeography and endemism Chap 7 & 8, Speciation in Birds by T. Price

Behavior innovation and population differentiation

This chapter introduced a non-ecological model of population differentiation: behavioral innovation leading to population differentiation. In particular, feeding innovation/flexibility tend to be associated with diet generalism, which lead to range expansion and speciation. Food generalists tend to colonize a new habitat more at ease than food specialists. The ability to colonize a new location tend to be measured by range size (Sol et al. 2005). Foraging innovation (cultural mutation) can spread through social learning and undergo cultural evolution.

As far as I understand, this new culture will put population to a novel adaptive peak, away from the original culture norm (at the original adaptive peak; see Figure 7.5). If such new culture restrains the individuals to certain habitat/niche away from the original population, this new cultural subpopulation will reproduce and pass on their cultural practice within this new niche, thus accumulate genetic difference that eventually contribute to reproductive isolation.

However, I’m skeptical about whether such transient behavioral innovations could be sufficient to differentiate the populations. If this behavior can be socially learned, migrants among the niches could easily learn the trick and fit in to this new niche. Thus there won’t be a reduction of gene flow among the cultural populations.

Island biogeography and endemism

This chapter discussed around the island biogeography theory proposed by MacArthur & Wilson (1967). The number of species on an island is a product of colonization-extinction balance. Two properties of the island are mostly concerned here: its size and its distance to source. The closer islands to the sources have greater colonization rates. The larger island tend to hold larger population size, thus lower extinction rates. The large island that is closet to the source have the most species number, while small distant island has the fewest number of species.

A complication to this theory was proposed by Diamond (1980, 1984) that the number of species should influence extinction rate, which was only influenced by island size in MacArthur & Wilson’s model. This model was behind the rationale that with every addition of species on the island, the existing species has to be confined to smaller population size, thus will have increased extinction rate. In this model colonization increases extinction, instead of being independent to extinction. Colonization leads to extinction due to that it increases turnover rate on the island.

Since colonization causes extinction, the closer islands should have fewer endemism even though they have more species in total, because the extinction rate of them is too high. So, island endemism is boosted by island area and isolation from the source. This island biogeography model predicts that this endemic species on distant island should be fairly old, which has not been tested. When estimating time, notice that migration will underestimate divergence time, and overestimation extinction rate.

#island biogeography #endemism #behavioral innovation #cultural evolution

Quantitative Genetics; Chap 6 Population Genetics by J.H. Gillespie

Finally my reading welcomed my beloved statistical applications: law of large number and regression, in addition to the fact that nonrandom mating chapter had probability theory application that I enjoyed. Reading on this book has becoming easier and more fun. This chapter first formulated trait variation in the population then showed what happen when there is selection and how does it evolve.

Trait variation

Phenotypic variation was broken down into additive contributions of maternal, paternal and environmental components (P= Xm + Xp + e). As for the genetic contribution, Xm or Xp, it introduced an interesting representation of population trait value X, by the allele frequencies (pi) that corresponds to different phenotypes (xi), so that the probability of X being xi is pi. So the mean allelic contribution of this locus is the sum of xi * pi, which should be zero.

Heritability

The proportion of additive effect over the total phenotypic variation is heritability, h. Another way of getting at this heritability is to look at covariance of phenotypes between offspring and parent. The phenotype of the parent and offspring is formulated respectively: Pp= Xm +Xm’ + ep and Po= Xm +Xp +eo. The covariance, after removing the zero covariance terms (assuming alleles are independently inherited), it comes to Var {Xm} only. Var {Xm} is half of VA, additive genetic variance, assuming Var {Xm} and Var {Xp} are equal. So Cov{Pp, Po}= VA/2. Then the correlation of the Pp and Po is VA/(2*Vp), thus half heritability. To generalize this relation, the correlation between relatives are just r * h. This is also happen to be the slope of regression line between phenotypes of relative Y against relative X.

Does natural history traits tend to have lower heritability? It’s mentioned that such traits might have higher environmental variance, or they tend to be the under strong natural selection, thus have lower additive variance.

Selection

Clayton (1957) did a selection experiment in which he bred the extreme flies in terms of bristle number. Over 24 generations, there was indeed separation in trait distribution of the low bristle line versus high bristle line. However such divergence did’t continue but leveled off. It was suspected that there might be a optimal number of bristle number and natural selection counteract artificial selection bringing the population back around the optima.

There are a few key terms for selection: S, selection differential; R, selection response, i(p), intensity of selection, Kh, the rate of phenotypic evolution (Haldane).

R=h^2 * S = h^2 *i(p) * sqrt(Vp).

Kh= ln(x2/x1)/T, where x1 and x2 are measures of some traits for two species that are separated for T millions of years.

#quantitative trait #selection #evolution

Reading log on mating barriers

Chap6 &10, Speciation by Coyne and Orr; Chap11-13; Speciation in Birds by T, Price

Today’s reading was about behavioral mechanisms of reproductive isolation.

Key concepts are reinforcement, character displacement, species recognition, sexual imprinting, filial imprinting, cultural drift.

Mating behaviors that mess up hardy-weinberg expectation

The deviance of heterozygosity from hardy-weinberg expectation can be caused by many behavioral mechanisms, other than inbreeding and population subdivision (explained yesterday). For instance, assortative mating can increase F in the genotype frequency. Also, rigid monogamy of breeding pairs over multiple breeding seasons should contribute to greater homogeneity than random mating across breeding seasons? In pp 264 of the speciation in Birds book, it described this pattern of greater pair bonding stability on island, and birds became less sexually dimorphic than the mainland. So there should be more reduction in heterozygosity than Fisher’s island equilibrium 1/(1+4Nm), considering birds are getting more monogamous after they migrated to the island. However, such behavioral pattern is likely limited by N (less monogamous as N increases), With more migrants (2Nm), the heterozygous reduction due to pair bonding behavior should be less prominent. This occurs when migration is predominant, heterozygosity is low, and isolation disappears. With low number of migrants (2Nm) however, drift predominates, heterozygosity should be high, but less than predicted due to the alteration of pair bonding behavior, which producing more similar offspring than random mating among breeding individuals over breeding seasons.

##Added notes, on Jan 21, 2016.

After a conversation with Darren on Jan 21, 2016, I changed my mind about the impact of monogamy on heterozygosity. We worked out the thought experiment on arbitrary 8 individual population with genotype AA, Aa, aa, and Aa, a female and a male each. Monogamy doesn’t decrease heterozygosity, as long as the initial pairing decision was random. If the pairing as assortative (individuals of same genotype mate), it decreases heterozygosity; whereas if the initial pairing was dis-assortative (AA mates with aa), it increases heterozygosity. Thus monogamy depends on the initial pairing decision, it might enlarge the variance of heterozygosity, without altering the mean of it. ##

Behavioral contributors to reproductive isolation

The table 6.1 on page 217 in the Speciation book summarized all the

Both selfing and assortative mating can lead to reproductive isolation. However selfing does not give rise to new biological species, as it is merely isolated lineages of self-replicating individuals, or a collection of “microspecies”, instead of interbreeding individuals that comprise a population.

As for assortative mating, there are many causes: sexual imprinting, runaway selection, cultural species, etc. Cultural species is specifically dealing with parasitic species (e.g. cowbirds), when young birds imprint on certain host species, and form the host “cultural species”. This could happen when the common host birds became hard to find for certain patches of the original population; and when laying eggs in a new host outweigh the searching cost of the “correct” host. Then these hatchlings would be imprinted in a way such that they tend to lay eggs in the new host, which spreads this new host culture, and isolate from the original population that lay eggs in the old host. This could be like the emergence of a new mutation--very hard to spread and likely get lost initially. I’m also concerned with the inbreeding depression if the initial spread of this new host culture depends on the individuals from clutches lay by a few females.

Chap5 nonrandom mating; Population Genetics, Gillespie 2004

This chapter modeled scenarios in which there is deficiency of heterozygosity under hardy-weinberg expectations, due to inbreeding and subdivisions. Before diving into the forces, there are three important parameters the reflect excess homozygosity: F, fxy, and r:

F parameter is used to estimate excess of homozygosity. So the frequency of A1A1 is p^2 * (1-F) + p* F, which is a sum of two mutually exclusive events: 1) get two A1 not due to special circumstances (F, probability of homozygosity due to special circumstances); and 2) given the first A1, get the second A1 not due to the special circumstance (1-F).

Another important parameter is fxy, the kinship coefficient, a measure of relatedness, which is the probability of two alleles (one from individual x and one from individual y) are identical by descent.

The third parameter, coefficient of relatedness, r, is related to fxy. It is 2 times fxy. This coefficient of relatedness is simply half of the mean number of shared alleles: 1/2*(0*r1+1*r1+2*r2)=1/2 * (r1+2r2).

Inbreeding: bad homozygosity but more efficient selection

During inbreeding, inbreeding coefficient FI is the probability that two alleles in an individual are identical by descent. FI=fxy. This FI decreases population mean fitness (=1-a-bFI). We can calculate overall survival chance of a child by considering non genetic effect and all the loci in the genomes. This survival chance is approximately a linear function that decreases with increasing FI. Interestingly, Morton (1956) found the relationship between this inbreeding coefficient and children viability was indeed linear. However, in other system, there might be greater synergistic effect of mutations, causing sharper decline of survival rate or fitness as inbreeding coefficient increases.

Another interesting thought in this chapter was about the evolution of selfing. When should selfing mutant be maintained? The answer is wired in the interplay between selection against inbreeding and the benefit of greater number of progenies generated by selfers. This threshold is determined by B <2log2, where B is fitness decrease due to inbreeding. B is approximated as 1.734 (>2log2) for human population, thus we don’t self. Plants should be either complete selfers or complete outcrossers.

However inbreeding is not always bad, it can enhance the effect of selection against partially recessive deleterious alleles, those that have h<0.5. The equilibrium frequency of deleterious mutation q is approximately u/hs, where mutation increases it, while selection acting against heterozygotes (hs) decreases it. With FI, the bottom became (1-FI)*hs+FI *s, which is greater than the original hs, thus putting more pressure on deleterious heterozygotes.

Subdivision: homogenize things --shaped by drift-migration balance

Subdivision increases homozygosity as long as there is genetic variance in allele frequency across subdivisions.

Fst=Gs-Gt/(1-Gt)

Gs: the probability that two alleles that are randomly drawn with replacement from a randomly selected subdivision are identical by state. Gt:...........................................................................................................................from the entire species.........................................................

Migration can homogenize things up, as equilibrium variation in the island is 1/(1+4Nm), which is zero when N reaches infinity, under Wright’s island model. 2Nm is the number of diploid individual migrate at each generation. If 2Nm is greater than 1/2, isolation disappears.

#nonrandom mating #subdivision #inbreeding #migration

Reading plan 3.1 mating and speciation

Today I read about nonrandom mating in Chapter 5 of Population Genetics by J.H. Gillespie, then I species recognition and mate choice in Chapter 13 and 14 of Speciation in Birds, by T. Price.

#speciation #mating

Trending Blogs

Recently Viewed Blogs

Speciation Blog