Dating with different techniques: Consilience of divergence times between Bean Goose species

Several methods suggest that Taiga and Tundra Bean Goose diverged about 2.5 million years ago.

One of the strongest arguments for evolution is consilience, the principle that evidence from independent, unrelated sources converges upon the same conclusions. Numerous lines of evidence, from genetic analyses and comparative morphology to biogeography and embryology, all point to the same unescapable conclusion: life on Earth evolved over billions of years. Doubting evolution would just be silly.

The principle of consilience can also be applied to smaller questions. In my own work on the evolution of geese, I uncovered a nice example of consilience: the divergence between Taiga Bean Goose (Anser fabalis) and Tundra Bean Goose (A. serrirostris). Using different dating techniques, I always converged upon the same answer: these species diverged around 2.5 million years ago.

mtDNA vs. Genomics

Before we delve into my goose work, we start with a study in the Journal of Evolutionary Biology. In 2000, Minna Ruokonen and her colleagues compared about 1000 base pairs of the mitochondrial DNA for seven goose species. Although they only included a sample of the Tundra Bean Goose (and not the Taiga Bean Goose), we can still compare this species with other closely related goose species, such as the Pink-footed Goose (A. brachyrhynchus). Based on the level of genetic divergence in the mtDNA, the researchers provided “a rough estimate for the timing of speciation events of Anser species […] within approximately 2–2.5 million years.”

More than 15 years later, during my PhD at Wageningen University, I used genomic data to unravel the evolutionary history of these goose species. After constructing a phylogenetic tree (based on 6,630,626 base pairs), I ran a molecular clock analysis with the software MCMCtree. First, I estimated a mean substitution rate based on previous studies. Next, the split between the two goose genera – Anser and Branta – was constrained between 4 and 20 million years, the time period for which we have reliable goose fossils. And finally, the MCMCtree analyses were run multiple times to check for convergence of the results. Using this approach, the divergence between Taiga Bean Goose and Tundra Bean Goose was estimated ca. 2.5 million years ago. Very similar to the mitochondrial result by Minna Ruokonen and her colleagues.

Genomic data suggested that the Taiga Bean Goose (in green) and the Tundra Bean Goose (in orange) diverged around 2.5 million years ago. From: Ottenburghs et al. (2016).

Demographic Models

So, now we have two independent lines of evidence for the divergence time between Taiga Bean Goose and Tundra Bean Goose. But it gets even better. During my postdoc at Uppsala University, I focused on the evolution of these two species (and later adding the Pink-footed Goose to the mix). To understand their evolutionary dynamics, I opted for a demographic modelling approach with genomic data. Using the software package DADI, I compared different demographic scenarios, ranging from strict isolation to secondary contact with asymmetrical gene flow.

DADI simulates the change in allele frequencies using a diffusion equation, similar to gas molecules moving through a room. Depending on the interplay of genetic drift, selection and migration, genetic variants spread through a population at different speeds. The end result can be visualized in a square with different populations on the horizontal and vertical axes (in my case, the two Bean Goose species). Genetic variants that are unique for one of the species can be found in the lower left corner, whereas variants shared by both species are found in the top right corner. Gene flow between the populations mixes things up. Different demographic models lead to different squares which can be compared to the actual data.

My genomic analyses pointed to a scenario of allopatric divergence (about 2.5 million years ago) followed by recent secondary contact (about 60,000 years ago). Another independent line of evidence for the divergence time between Taiga Bean Goose and Tundra Bean Goose.

DADI uses a diffusion equation to simulate how genetic variants spread through a population. The result can be visualized in a square and compared with the actual data. For the Bean Geese, the close match between the data and the model suggested a scenario of allopatric divergence with secondary contact. From: Ottenburghs et al. (2020).


And there you have it. Three independent analyses that all converge upon the same conclusion. It does not matter if you use a simple calculation based on mitochondrial divergence, a molecular clock calibrated with fossils, or a demographic model using diffusion equations. The conclusion is always the same: Taiga Bean Goose and Tundra Bean Goose diverged ca. 2.5 million years ago. Now, that is consilience.


Ottenburghs, J., Megens, H. J., Kraus, R. H., Madsen, O., van Hooft, P., van Wieren, S. E., Crooijmans, R. P. M. A., Ydenberg, R. C., Groenen, M. A. M. & Prins, H. H. T. (2016). A tree of geese: A phylogenomic perspective on the evolutionary history of True Geese. Molecular Phylogenetics and Evolution101, 303-313.

Ottenburghs, J., Honka, J., Müskens, G. J., & Ellegren, H. (2020). Recent introgression between Taiga Bean Goose and Tundra Bean Goose results in a largely homogeneous landscape of genetic differentiation. Heredity125(1-2), 73-84.

Ruokonen, M., Kvist, L., & Lumme, J. (2000). Close relatedness between mitochondrial DNA from seven Anser goose species. Journal of Evolutionary Biology13(3), 532-540.

Featured image: Taiga Bean Goose (Anser fabalis) © Marton Berntsen | Wikimedia Commons

Decoupling of genetic and phenotypic evolution in the Tawny-Crowned Greenlet

Study finds deep genomic divergence despite little morphological changes.

South America houses high levels of cryptic bird diversity. Although some populations look morphologically similar, they often exhibit drastic genetic differences. Because the taxonomy of Neotropical birds is largely based on the study of museum specimens, we are probably missing a lot of species.

A notable example concerns the Tawny-Crowned Greenlet (Tunchiornis ochraceiceps). This small songbird can be found from the lowlands of Central America to the Amazon region. Based on subtle differences in plumage and morphometrics, ornithologists have recognized 10 subspecies. Genetic data, however, paint a drastically different picture. Analyses of mitochondrial DNA showed that some of these subspecies diverged ca. 10 million years ago. That would be a distinct species in anyone’s book.

A recent study in the journal Molecular Phylogenetics and Evolution undertook a more detailed analysis of the Tawny-Crowned Greenlet, combining plumage coloration, morphometrics, vocalizations and genomic data. How many cryptic species are hiding in the tropical rainforests?

Six Lineages

Let’s start with the genomic patterns. Based on more than 2000 ultraconserved elements (UCEs), the researchers could delineate six separate lineages (of which two showed signs of substructure). The main split in the phylogenetic tree – dated to about 9 million years ago – differentiates between the western and the eastern lineages. The western group falls apart into populations on both sides of the Andes, whereas the eastern group is subdivided by different Amazonian rivers. The figure below provides a nice overview of the complex geographical distribution of these genetic lineages.

Genomic analyses suggested six distinct lineages across Central and South America. From: Buainain et al. (2021).

Phenotypic Boundaries

But can we also discriminate between these lineages with phenotypic data or vocalizations? When it comes to morphometrics and song, the answer is no. The authors indicate that “no clear spatial separation between samples belonging to the two [genetic] clusters can be seen in the [morphological] classification graph” and “the quantitative analysis shows no diagnostic vocal characters for any of the populations of T. ochraceiceps.”

In terms of plumage coloration, the situation is slightly better. One plumage character – the coloration of the forehead and crown – can be used to distinguish between three groups, namely populations in the west, northeast of the Amazon and southeast of the Amazon. Other plumage traits were not diagnostic for different populations, but did show variation across the distribution of the Tawny-Crowned Greenlet. In most cases, the geographic extremes were clearly different with a large area of intermediate phenotypes in between. Drawing clear phenotypic boundaries between populations is difficult, at best.

Some examples of plumage traits that distinguished some populations of the Tawny-Crowned Greenlet. In general, however, it was difficult to determine clear phenotypic boundaries. From: Buainain et al. (2021).

A Conservative Species

These analyses reveal an interesting evolutionary pattern: the decoupling of genetic and phenotypic evolution. Although the Tawny-Crowned Greenlet is composed of genetically distinct populations, this differentiation is not reflected in the phenotypic variation. Why is the phenotype of this (group of) species so conserved? The researchers offer several explanations that require further investigation:

  • Stabilizing selection: there is no selective pressure – social or ecological – to develop morphological or vocal differences.
  • Habitat specialist: the Tawny-Crowned Greenlet follows a specific habitat type which does not require it to adapt morphologically.

Finally, the authors propose a taxonomic arrangement for the six genetic lineages. They divide them into four distinct species, of which two species are further separated into two subspecies: T. o. ochraceiceps, T. o. bulunensis, T. ferrugineifrons, T. luteifrons, T. r. rubrifrons and T. r. lutescens. This classification might change in the future. Stay tuned.

Geographical distribution of the six taxa (species and subspecies) in the Tawny-Crowned Greenlet. From: Buainain et al. (2021).


Buainain, N., Maximiano, M. F., Ferreira, M., Aleixo, A., Faircloth, B. C., Brumfield, R. T., … & Ribas, C. C. (2021). Multiple species and deep genomic divergences despite little phenotypic differentiation in an ancient Neotropical songbird, Tunchiornis ochraceiceps (Sclater, 1860)(Aves: Vireonidae). Molecular Phylogenetics and Evolution162, 107206.

Milá, B., Tavares, E. S., Munoz Saldana, A., Karubian, J., Smith, T. B., & Baker, A. J. (2012). A trans-Amazonian screening of mtDNA reveals deep intraspecific divergence in forest birds and suggests a vast underestimation of species diversity. PLoS One7(7), e40541.

Featured image: Tawny-Crowned Greenlet (Tunchiornis ochraceiceps) © Dominic Sherony | Wikimedia Commons

Do aggressive interactions lead to plumage divergence in chickadees?

Testing this hypothesis with some clever experiments.

Closely related species tend to look slightly different in areas where they co-occur (i.e. sympatry) compared to areas where they do not (i.e. allopatry). This pattern of greater divergence in sympatric than in allopatric species calls out for an explanation. One possibility relates to aggressive interactions. During the breeding season, for example, males of some bird species vigorously defend their territories. To avoid spending too much energy, they should direct their attacks to members of their own species. If they cannot recognize males from another species, they might engage in costly aggressive interactions. Hence, selection should favor males that are able to discriminate between species, leading to divergence in the traits that these males rely on (such as plumage patterns or songs).

In a recent study, Haley Kenyon and Paul Martin put this scenario to the test. They focused on three species of Poecile chickadee that mainly differ in their eyebrow stripe. Specifically, the researchers checked whether the Black-capped Chickadee (P. atricapillus) is able to recognize the Mountain Chickadee (P. gambeli) and the Mexican Chickadee (P. sclateri). If the aggression hypothesis is correct, you would expect that less attacks are aimed at birds with more divergent plumage patterns. That would be the Mountain Chickadee in this case.

The 3D-printed models of Black-capped Chickadee, (b) Mountain Chickadee, and (c) Mexican Chickadee. From: Kenyon & Martin (2021).

Allopatric Experiments

The researchers placed 3D-printed models of the three species in the field. They monitored the behavior of allopatric Black-capped Chickadee in response to three combinations of bird-models:

  • Black-capped Chickadee + Mountain Chickadee
  • Black-capped Chickadee + Mexican Chickadee
  • Black-capped Chickadee + control (i.e. Northern Cardinal)

Because Mountain Chickadees have a more pronounced eye-stripe, you would expect that Black-capped Chickadees are able to recognize them. Hence, they will mainly attack the Black-capped Chickadee model (which they see as a conspecific). Mexican Chickadees, however, look quite similar to Black-capped Chickadees. The birds might thus not discriminate between these two species and attack both models equally. And finally, the Northern Cardinal (Cardinalis cardinalis) looks so vastly different, that the chickadees will ignore it.

The results of this clever experiment did not support the aggression hypothesis. The researchers reported that “territorial Black-capped Chickadee males were equally likely to attack Mountain Chickadee and Mexican Chickadee models when they were paired with Black-capped Chickadee models.” Aggressive interactions do not seem to account for the divergent plumage patterns.

An overview of the expected patterns (in figures a and b) and the observed results (figure c) of the experiment. The Black-capped Chickadee males were equally likely to attack Mountain Chickadee and Mexican Chickadee models. From: Kenyon & Martin (2021).

Other Factors

The physicist Richard Feynman once said that “If it disagrees with experiment, it’s wrong. In that simple statement is the key to science.” That statement might be true, but it is too harsh for biological field experiments with many confounding factors. In this case, the unexpected outcome of the aggression experiments could provide insights into the behavior of the chickadees, and could inform future experiments.

The researchers offer two possible explanations why the Black-capped Chickadees were not able to discriminate between the 3D-printed species. First, the Black-capped Chickadees in the experiments were allopatric. Perhaps they need to co-occur with the other species in order to learn the difference between conspecifics and heterospecifics. Second, the experimental models only differed in plumage patterns. It is possible that the Black-capped Chickadees use additional signals (such as songs) to recognize other species.

Hybridization or Habitat?

However, it could still be that the aggression hypothesis is indeed wrong. It would be worthwhile to explore other explanations for the plumage divergence. One alternative mechanism is selection against hybridization (i.e. reinforcement). Females might be able to discriminate between males from different species, thereby avoiding maladaptive hybridization. And indeed, the levels of hybridization are very low in areas where Black-capped Chickadees and Mountain Chickadees co-occur.

Another alternative explanation relates to habitat use. Species that reside in different habitats could be subjected to contrasting selective pressures. Specifically, selection for optimal signal transmission might favor different plumage patterns in different habitats. Black-capped Chickadees and Mountain Chickadees do seem to inhabit different habitats (deciduous vs. conifer forests).

Two exciting explanations that require further investigation. Time to plan for the next experiments.


Grava, A., Grava, T., Didier, R., Lait, L. A., Dosso, J., Koran, E., Burg, T. M. & Otter, K. A. (2012). Interspecific dominance relationships and hybridization between black-capped and mountain chickadees. Behavioral Ecology23(3), 566-572.

Hill, B. G., & Lein, M. R. (1989). Territory overlap and habitat use of sympatric chickadees. The Auk106(2), 259-268.

Kenyon, H. L., & Martin, P. R. (2021). Experimental tests of selection against heterospecific aggression as a driver of avian colour pattern divergence. Journal of Evolutionary Biology34(7), 1110-1124.

Featured image: Black-capped Chickadee (Poecile atricapillus) © USFWS | Wikimedia Commons

What causes taxonomic disagreement between bird checklists?

Study identifies a combination of research bias and differences in divergence rates.

Why can’t we all just get along? This question often comes to my mind when I follow the heated debates between taxonomists. Personally, I find the quest to pigeonhole the immense biodiversity into species and subspecies not that interesting. I prefer to focus on the ecological and evolutionary processes that gave rise to all this diversity. For conservationists and policy makers, however, a reliable overview of the species in an area can be crucial. It is thus certainly important to continue with the correct classification of life.

But what is the correct way to classify organisms? The species problem has been haunting biologists for centuries. Charles Darwin already indicated that “No one definition has satisfied all naturalists; yet every naturalist knows vaguely what he means when he speaks of a species.” And the struggle continues to this day (see this blog post). Taxonomists often disagree about the species status of a certain population. A first step in solving these taxonomic disputes is to understand what factors cause the disagreements. And that is exactly what Montague Neate-Clegg and his colleagues did. They scoured four bird checklists for taxonomic disagreements and tried to identify the underlying causes. Their findings recently appeared in the journal Global Ecology and Biogeography.

Research Bias

The researchers examined four commonly used checklists for the world’s avifauna:

  • Howard and Moore Checklist of the Birds of the World
  • eBird/Clements Checklist of Birds of the World
  • Birdlife International Digital Checklist of Birds of the World
  • International Ornithological Community (IOC) World Bird List

This extensive comparison culminated in 11,389 extant species names of which 9,894 were recognized by the four checklists. Some simple math reveals 1,495 problematic cases. A minority of these (18 cases) were newly discovered species, whereas the rest were genuine taxonomic disagreements, such as the redpolls (genus Acanthis) and the crossbills (genus Loxia).

A detailed look at the disputed taxa showed a clear geographical research bias. The researchers noted that “taxonomic agreement was lowest for species in Southeast Asia/Australasia and the Southern Ocean, understudied regions where islands have driven high levels of cryptic diversification. In contrast, agreement was highest in the temperate Northern Hemisphere where diversity is lower and research is more extensive.” Luckily, ornithologists are working hard to stabilize the taxonomy in Southeast Asia and Australasia (see for example here and here).

Taxonomic agreement in birds across regions of the world for (a) all checklists and (b) excluding Howard & Moore (which has not been updated since 2014). From: Neate‐Clegg et al. (2021).

Ecological Reasons

Apart from the research bias, the analyses also pointed to several ecological traits. First, bigger species – especially with a body mass over 500 grams – were less likely to be in dispute. Obviously, big species are easier to observe and study. But there are also ecological reasons: larger body size is often associated with longer lifespans, smaller clutch sizes and larger home ranges. Together, these factors might lead to lower diversification rates, and consequently less diversity which is easier to classify.

Second, taxonomic agreement was higher in open-habitat species and migratory species. These traits tend to result in high levels of dispersal, resulting in gene flow between populations. The homogenizing exchange of genetic material prevents the formation of new species, resulting in clearly delineated species. This situation contrasts with altitudinal migrants that are partly depended on forest habitats. Here, the researchers found more taxonomic disagreements which they explained as follows:

Our results may, therefore, support the theory that intermediate dispersal ability leads to higher rates of diversification whereby sufficient dispersal capability is required to colonise new areas but not so much dispersal ability that gene flow prevents speciation. The greatest potential for cryptic lineage differentiation may, therefore, occur in lineages with intermediate forest dependence and intermediate mobility.

Ecological predictors of taxonomic agreement for the world’s birds for all checklists (in black) and excluding Howard & Moore (in grey, which has not been updated since 2014). From: Neate‐Clegg et al. (2021).

Species Concepts

The ecological traits discussed above could push some taxa into the “grey zone” of the speciation process where different species concepts support different taxonomic decisions. This conflict between species concepts is apparent when comparing certain check lists. Birdlife, Clements and Howard and Moore follow the Biological Species Concept (focusing in reproductive isolation) whereas the IOC adheres to the Evolutionary Species Concept (focusing on lineage differentiation, an approach that I also prefer). The reliance on and application of these different species concepts partly explains some taxonomic disagreements.

Hence, aligning the bird checklists will thus involve finding a consensus on the “best” species concept. It seems that we are right back from where we started. Those dreaded species concepts…


Neate‐Clegg, M. H., Blount, J. D., & Şekercioğlu, Ç. H. (2021). Ecological and biogeographical predictors of taxonomic discord across the world’s birds. Global Ecology and Biogeography30(6), 1258-1270.

Featured image: Common Redpoll (Acanthis flammea) © Jyrki Salmi | Wikimedia Commons

Hubbs Principle: Loneliness can lead to hybridization

The genetic consequences of unbalanced species numbers.

Imagine that you are a Hawaiian Goose (Branta sandviciensis) in a Dutch waterfowl collection. One day, your owner is not paying attention and forgets to lock the door to your cage. You sneak out and disappear into the spacious fields of the Netherlands. When the breeding season starts, you look for a potential mate. Unfortunately, there are no other Hawaiian Geese around. In the end, you become so desperate that you settle for another species, pairing up with a – perhaps slightly confused – Barnacle Goose (B. leucopsis). A few months later, Dutch birdwatchers report a peculiar hybrid goose (see picture above).

This short story illustrates the so-called “Desperation Hypothesis”. An individual cannot find a conspecific partner and eventually mates with another species. This scenario was first described in fish by Carl L. Hubbs. He noted that “Great scarcity of one species coupled with the abundance of another often leads to hybridization: the individuals of the sparse species seem to have difficulty in finding their proper mates.” This phenomenon – now known as Hubbs Principle – does not only apply to fish. It has also been observed in birds.

Disproportionate Ducks

The scenario of the lonely Hawaiian Goose is an extreme case. Hubbs Principle also applies when one species is much more abundant than another. On the Falkland Islands, for example, Speckled Teal (Anas flavirostris) outnumber Yellow-billed Pintails (A. georgica) by about ten to one. This numerical imbalance resulted in hybridization between these two duck species. Given the preponderance of Speckled Teals in the area, hybrids are more likely to backcross with this species. Genes are thus expected to flow from the Yellow-billed Pintail into the Speckled Teal. And indeed, genetic analyses by Kevin McCracken and Robert Wilson confirmed this pattern of gene flow. The genetic signature of Hubbs Principle in action.

A first-generation hybrids between Speckled Teal and Yellow-billed Pintail. From: McCracken & Wilson (2011).

Invasion Genetics

It becomes even more interesting when we consider the expansion of one species into the distribution of another one. During such an invasion scenario, the numerical imbalance between the species shifts, resulting in some peculiar genetic patterns. Let’s walk through the scenario step by step. In the beginning, the expanding species is outnumbered and is thus more likely to hybridize with members from the local population. As the expansion proceeds, the resident species and previously produced hybrids are engulfed by the expanding species, thereby overturning the numerical imbalance. Consequently, hybrids have a higher chance of backcrossing with members of the expanding species, resulting in gene flow from the resident into the expanding species.

A literature survey by Mathias Currat and his colleagues showed that these patterns regularly occur in nature. They found 44 studies that quantified gene flow during a species invasion. In 36 cases (82%), gene flow was highly asymmetrical from the local into the invading species (as expected). Only seven cases (16%) reported gene flow in the reverse direction. Exceptions to this general pattern provide exciting avenues for further research. Genes that are going against the flow might confer an advantage to their carriers. The invasion scenario assumes neutral genetic variation, but sometimes you need to invoke selection. But I will leave that topic – adaptive introgression – for another blog post.


Currat, M., Ruedi, M., Petit, R. J., & Excoffier, L. (2008). The hidden side of invasions: massive introgression by local genes. Evolution62(8), 1908-1920.

Hubbs, C. L. (1955). Hybridization between fish species in nature. Systematic Zoology4(1), 1-20.

McCracken, K. G., & Wilson, R. E. (2011). Gene flow and hybridization between numerically imbalanced populations of two duck species in the Falkland Islands. PLoS One6(8), e23173.

Featured image: Hawaiian Goose (Branta sandviciensis) x Barnacle Goose (B. leucopsis) © Luc Bekaert |