Interview with the authors: Unparallel differentially expressed genes in parallel ecological divergence

In a recent paper in Molecular Ecology, Szukala et al. quantified the degree of gene expression and functional parallelism across polytopic divergence of montane and alpine ecotypes of in Heliosperma pusillum (Caryophyllaceae) and gained insights into the architecture of adaptive traits. They performed RNA-seq analyses of plants grown in a common garden and detected a large proportion of differentially expressed genes in each replicate ecotype pair. Functional enrichment of these genes, however, revealed that the traits affected by significant expression divergence are largely consistent across ecotype pairs, suggesting a polygenic architecture for the diverged adaptive traits and multiple routes for adaptation. A new genome assembly for H. pusillum was also presented in this study.

We sent a number of questions to lead authors of this work, Aglaia Szukala and Ovidiu Paun, to get more detail on this study.

Upper panels: Graphical representations of the alpine (A) and montane (M) ecotypes of Heliosperma pusillum and their ecological niches. Lower panel: Map showing the study sites of ecotype pairs that evolved in parallel.

What led to your interest in this topic / what was the motivation for this study? 

We are fascinated by the concept of parallel evolution and the molecular mechanisms behind this process. Given that drift is a major driver of evolution and due to the traditional focus on mono- or oligogenic traits, parallel evolution has been considered to be a rare process until recently. However, together with the increasing understanding of polygenic adaptation (Barghi et al., 2020), it has become clear that parallel evolution is relatively frequent, with implications across evolutionary biology and ecology. Specifically, for our study, previous works (Trucchi et al., 2017; Bertel et al., 2018) reported some evidence that altitudinal ecotypes in H. pusillum diverged in parallel. We wanted to rigorously test this hypothesis using demographic modeling and understand the level of molecular parallelism, with regard to divergent gene expression and outlier SNPs.

What difficulties did you run into along the way? 

For several reasons related to the planning of field work and the development of the wider project over years, we had to deal with uneven sampling sizes across populations, which needed to be taken into account, especially for the demographic modeling analyses. Fun fact: in reciprocal translocation experiments of a complementary study (Szukala et al., 2022) on the same species, whose data is also included in the present paper, we chose to use alpine microsites to plant our accessions that were fairly flat (in an otherwise steep area) and free of other plants. At the end of the vegetation season, those sites proved to be resting places for chamois which squeezed and munched most of our plants, while overfertilizing them. In previous years, when reciprocal transplantations were performed as preparation for this study, we faced droughts, poor germination and survival rates at some sites, leading to uneven sampling sizes across sites. Take-home message: experiments in the wild are always a challenge.

What is the biggest or most surprising innovation highlighted in this study? 

It is unclear how much overlap of divergence outliers is to be expected across natural evolutionary replicates. Our study showed a surprisingly low amount of shared molecular differentiation, which we did not expect given that the geographic range considered is relatively small and our study system is in a phase of incipient speciation with no reproductive isolation detectable (Bertel et al., 2016). The extremely low sharing of differentially expressed genes and outlier SNPs, but high similarity of GO terms involved across independent divergence events, indicates that the polygenic architecture of traits is relevant for adaptation of these populations to distinct altitudinal zones in the Alps.

Moving forward, what are the next steps in this area of research?

We further investigated the role of phenotypic plasticity for the development of parallel evolution in our system (Szukala et al., 2022) towards a better understanding of the relative role of genetic and environmentally-induced phenotypic variation in such replicated divergence events. We are also interested if other molecular mechanisms, which are sensitive to environmental input, such as epigenetic signals, could play an important role in parallel evolution. Further, we wish to understand how polygenic adaptation affects signatures of parallel evolution. Very interesting is to question if adaptation can use different genes to produce similar outcomes even in very closely related lineages, and how frequent this process takes place compared to the re-use of standing variation.

Alpine (blue) and montane (light orange) ecotypes of Heliosperma pusillum and their environments. In the alpine environment glabrous plants grow on more humid screes and meadows, also in proximity of streams. In the montane environment below the tree line, pubescent plants typically grow under rocks overhangs or on the rock as chasmophytes. Photo credit: Szukala A and Paun O.

What would your message be for students about to start developing or using novel techniques in Molecular Ecology?

Being curious and exploiting the most advanced and newest methods is good, but don´t forget to be robust, careful, and bias-aware when it comes to the interpretation of results.

What have you learned about methods and resource development over the course of this project?

It is often difficult to quantify and describe results relative to expectations in an objective way, because it is hard to formulate objective expectations in natural systems.

Describe the significance of this research for the general scientific community in one sentence.

Repeated evolution of similar phenotypes can involve different sets of genes.

Describe the significance of this research for your scientific community in one sentence.

Polygenic traits offer different genetic substrates for parallel evolution of similar phenotypes.


Barghi N, Hermisson J, Schlötterer C. 2020. Polygenic adaptation: a unifying framework to understand positive selection. Nature Reviews Genetics 21: 769–781.

Bertel C, Hülber K, Frajman B, Schönswetter P. 2016. No evidence of intrinsic reproductive isolation between two reciprocally non-monophyletic, ecologically differentiated mountain plants at an early stage of speciation. Evolutionary Ecology 30: 1031–1042.

Bertel C, Rešetnik I, Frajman B, Erschbamer B, Hülber K, Schönswetter P. 2018. Natural selection drives parallel divergence in the mountain plant Heliosperma pusillum s.l. Oikos 127: 1355–1367.

Trucchi E, Frajman B, Haverkamp THA, Schönswetter P, Paun O. 2017. Genomic analyses suggest parallel ecological divergence in Heliosperma pusillum (Caryophyllaceae). New Phytologist 216: 267–278.

Szukala A, Bertel C, Frajman B, Schönswetter P, Paun O. 2022. Parallel adaptation to lower altitudes is associated with enhanced plasticity in Heliosperma pusillum (Caryophyllaceae). bioRxiv 2022.05.28.493825; doi: 10.1101/2022.05.28.493825.

Featured study

Szukala A, Lovegrove-Walsh J, Luqman H, Fior S, Wolfe TM, Frajman B, Schönswetter P, Paun O. 2022. Polygenic routes lead to parallel altitudinal adaptation in Heliosperma pusillum (Caryophyllaceae). Molecular Ecology.

Interview with the authors: Genomic basis of Y‐linked dwarfism in cichlids pursuing alternative reproductive tactics

In a recent paper in Molecular Ecology, Singh et al. used genome sequencing, bioinformatics and population genetic analyses to gain insight into the genetics and evolution of a fascinating mating system. The species in question, Lamprologous callipterus, exhibits a mating system with two males morphs. Large “bourgeois” males carry empty snail shells that are inhabited and used as nests by the females. An alternate male morph, much smaller than the “bourgeois” males, also exists and inhabits shells along with the females. Previous genetic work had established that this mating system was Y-linked and that the male body size was a Mendelian trait, but the sex-determining locus had not been identified until this study.

We sent some questions to Pooja Singh, the author who led this work, to get more detail on this study.

Photo credit: Drawing by Pooja Singh, based on Barbara Taborsky’s original image.

What is the biggest or most surprising innovation highlighted in this study? 

The most novel aspect of this study is that we found an example of a young sex chromosome that may have evolved due to sexual antagonism over body size. While the sexual antagonism theory is considered the classical model of sex chromosome evolution, few empirical examples exist to support it. The other exciting finding was that the candidate body size/dwarfism gene that we propose for L. callipterus, GHRHR, is a well-known dwarfism gene in mammals. Fish and mammals shared a common ancestor over 440 million years ago, so the body size development pathway is genetically constrained through deep evolutionary time.

What difficulties did you run into along the way? 

The major challenge for me was that I knew little about sex chromosome evolution when I started this project, so I really had to do a lot of groundwork reading relevant literature and researching methods to be able to get things going. I had to start thinking beyond the classical XY and WZ old sex systems and familiarize myself with the workings of early stages of sex chromosome evolution.

When I read your paper, I had never heard of the fascinating mating system of these cichlids. They reminded me of the ruff, and the multiple inversions that seem to be involved in the different reproductive strategies in that system. You mention in the paper that you were not able to identify inversions based on the bioinformatic approaches you used. Is there a sense for how much chromosome evolution during the radiation? Could the use of the divergent reference genome have anything to do with the lack of a signal of inversions? 

To my knowledge the broad scale chromosomal structure of African cichlid species is similar. However, small scale structural variations (inversions, indels, translocations etc.) have not been investigated systematically. So yes, it is totally possible that our short read data and the divergent (and fragmented) reference genome assembly may have hindered our ability to detect inversions. The system really needs a long-read de novo genome assembly to resolve the inversion question.

In the Discussion, you talk about the possibility of different male Y-haplotypes. Is your data sufficiently high resolution that you could examine insertion/deletion polymorphisms in your dataset? 

Yes, we could technically identify small insertions/deletions in our data. Might certainly be something to investigate in the future, in combination with long-read Y assembly.

A recently proposed model of sex-chromosome evolution indicates that gene expression differences may predominate at the early stages of sex chromosome evolution (Lenormand and Roze 2022 Science – This is intriguing given that you didn’t find any smoking gun loci with signals of sexual antagonism. Do you have plans to look at patterns of gene expression across the different morphs?

While Lenormand and Roze’s theory is certainly exciting for the field of sex chromosome evolution, I think it is less plausible for the L. callipterus mating system because antagonistic body sizes in females and males are crucial shell-brooding success and fitness. And because we found the candidate sex-determining gene and body-size gene to be physically linked. I am certainly not a proponent of the ‘one classical theory explains all’ narrative and I really look forward to seeing what RNA-seq data reveal about sex chromosome evolution in this species. It would be especially interesting to see the landscape of cis/trans eQTLs of genes in our proposed L. callipterus sex chromosome and how much it reflects the expectations from Lenormand and Roze’s model. Beyond just this species, cichlid fishes are an excellent system to test the sexual antagonism vs gene regulation models of sex chromosome evolution.

Regarding the coverage analysis you used to identify the putative sex-linked locus. Given the hypothesis that the divergence of the sex chromosomes is recent, reads sampled from Y-linked regions may still map well to the orthologous region on the putative X-chromosome.  Did you tweak mapping quality filters at all?

I did run a less stringent mapping analysis, which overall had slightly higher mapping statistics, but the reduced coverage pattern on the L. callipterus sex chromosome was still significant.

Snail shell nest of L. callipterus with the nest owner in the left centre.
Photo credit: Koblmueller et al. 2007

Could you describe the significance of this research for the general scientific community in one sentence?

Sexual antagonism over body-size may have driven sex chromosome evolution in a shell-brooding cichlid fish where giant and dwarf male reproductive types have evolved.

Moving forward, what are the next steps in this area of research (unless otherwise covered)?

Our main priority right now is to keep the L. callipterus dwarf males alive and breeding. Our fish were recently moved from the University of Bern in Switzerland to the University of Graz in Austria, and it has proved difficult to get the dwarf males happy and breeding in the new facility. This is (probably) the only collection in the world of L. callipterus dwarf males outside Lake Tanganyika so they are very precious. Our next step is to write a convincing grant to get funding to build a long-read improved genome assembly, conduct RNA-seq, and sample and sequence more individuals from natural populations. I would like to use the RNA-seq data to map expression QTLs and investigate the regulatory interactions of candidate genes related to sex, morphology, physiology, and behaviour that we found in or around the L. callipterus sex region. It would also be interesting to study sex chromosomes in related lamprologine species, as our pre-liminary analysis in this manuscript suggests that the sex region may be shared across the lamprologine tribe

Interview with the authors: Whole-genome analysis of multiple wood ant population pairs supports similar speciation histories, but different degrees of gene flow, across their European ranges

In a recent paper in Molecular Ecology, Portinha et al. used population genomic data to analyse the speciation history of two closely related species of wood ants, Formica polyctena and F. aquilonia. Using a demographic modelling approach, the authors reconstruct the history of divergence for multiple heterospecific pairs of populations. In all cases, the authors found that there was evidence for divergence with gene flow. However, for a sympatric population pair sampled in Finland there was evidence for substantially elevated gene flow between the species. Their findings imply that population genomic analysis of speciation history may be geographically variable for particular species.

We sent some questions to Beatriz Portinha and Pierre Nouhaud, the corresponding authors of this work, to get more detail on this study.

Ant mound surface covered in ants. Photo credit: Jack Beresford

What led to your interest in this topic / what was the motivation for this study? 

Knowledge on the demographic and speciation histories is essential for understanding
contemporary genomic patterns in natural populations, which is why we wanted to
reconstruct it for the emerging Formica model system. Our study species, Formica polyctena
and F. aquilonia, are known to hybridize naturally in Southern Finland, where their hybrids
have been studied for over 10 years (Kulmuni et al., 2010; Martin-Roy et al. 2021). We
wanted to test whether a similar divergence history was consistently inferred across the
European ranges of both species, or whether the Finnish populations would stand apart,
possibly because of gene flow mediated by hybrid populations in the area.

What difficulties did you run into along the way? 

Formica polyctena and F. aquilonia had a limited genomic toolbox when we started the
project, and we initially relied on a distant and non-contiguous reference assembly.
Meanwhile, our group assembled a high quality reference genome (Nouhaud et al., 2022),
which improved the quality of our inferences.

The demographic modelling software we used, fastsimcoal2, can simulate a large panel of
evolutionary scenarios. When planning this study, we wanted to design models that
considered alternative scenarios for the divergence of the species which would be as
biologically meaningful as possible, while keeping the number of models low enough that the
project 1) would not be a huge computational burden and 2) would be executable in the
available time frame (Beatriz’s MSc. project, funded by Erasmus+ and Societas pro Fauna et
Flora Fennica). This was an especially important aspect as we used four distinct population
pairs to reconstruct the history of the two species, so each model had to be run, at least, four
different times.

What is the biggest or most surprising innovation highlighted in this study? 

We found that there was already bidirectional gene flow occurring in Finland before the
hybridization events that led to the present-day hybrid populations. This was not suspected
before, as there is no evidence in the literature, and it suggests that F. polyctena in Finland
may be admixed, which is supported by the fact that we have not found non-admixed F.
individuals in Finland.

Moving forward, what are the next steps in this area of research?

The divergence history we inferred between F. polyctena and F. aquilonia can be used to
run simulations about the evolution of the hybrid populations, which is what we did in a
subsequent work (Nouhaud et al. 2022). In the longer run, it would also be important to
extend this work by reconstructing the divergence history of the whole F. rufa species group,
which encompasses 5 species (including F. aquilonia and F. polyctena) and where gene flow
is prevalent (Seifert, 2021).

Describe the significance of this research for the general scientific community in one sentence.

Genomes from individuals sampled thousands of kilometers apart tell the same ancient
history, while their most recent history may be different.

Describe the significance of this research for your scientific community in one sentence.

The divergence history between two species can be reliably and consistently inferred from a
small number of individuals sampled across the species’ ranges.

Portinha, B., Avril, A., Bernasconi, C., Helanterä, H., Monaghan, J., Seifert, B., Sousa, V. C., Kulmuni, J., & Nouhaud, P. (2022). Whole-genome analysis of multiple wood ant population pairs supports similar speciation histories, but different degrees of gene flow, across their European ranges. Molecular Ecology, 31, 3416– 3431.

Kulmuni, J., Seifert, B. & Pamilo, P. (2010). Segregation distortion causes large-scale
differences between male and female genomes in hybrid ants. Proceedings on the National
Academy of Sciences
, 107(16), 7371-7376.

Martin-Roy, R., Nygård, E., Nouhaud, P. & Kulmuni, J. (2021). Differences in thermal
tolerance between parental species could fuel thermal adaptation in hybrid wood ants.
American Naturalist, 198(2), 278-294.

Nouhaud, P., Beresford, J. & Kulmuni, J. (2022). Assembly of a hybrid Formica aquilonia× F.
ant genome from a haploid male. Journal of Heredity, esac019, 1-7.

Nouhaud, P., Martin, S. H., Portinha, B., Sousa, V. C. & Kulmuni, J. (2022). Rapid and
repeatable genome evolution across three hybrid ant populations. bioRxiv.

Seifert, B. (2021). A taxonomic revision of the Palaearctic members of the Formica rufa
group (Hymenoptera: Formicidae) – the famous mound-building red wood ants.
Myrmecological News, 31, 133-179.

Interview with the authors: Mega-fire in redwood tanoak forest reduces bacterial and fungal richness and selects for pyrophilous taxa that are phylogenetically conserved

In a recent paper in Molecular Ecology, Enright et al. examined how soil microbiomes are affected by extreme fires. The Soberanes mega-fire provided the authors with an opportunity to study how such extreme events, which are increasingly common with climate-change, can have lasting effects on ecology. By sampling the soil microbiome before and after the Soberanes mega-fire, Enright at al. demonstrated dramatically altered soil communities and a reduction in species richness associated with the mega-fire. There was a clear phylogenetic pattern to the particular microbes that increased or decreased abundance after the fire. Drawing from their results, Enright et al. propose a framework to predict the traits that post-fire microbial communities might exhibit.

We sent some questions to Sydney Glassman, one of the corresponding authors of this work, to get more detail on this new study.

Aerial view of the Soberanes mega-fire. Photo credit: Calfire

What led to your interest in this topic / what was the motivation for this study? 

I had originally been interested in sampling the redwood tanoak forests of Big Sur because I was interested in what the cascading effects of sudden oak death (SOD) induced mortality would be on soil fungal communities during my PhD at UC Berkeley. Prof Dave Rizzo at UC Davis had a large plot network investigating the effects of SOD on plant mortality. I teamed up with him in 2011 to select a subset of plots to collect soils to investigate the impacts on the soil microbial community via amplicon sequencing. Then, in 2016, I learned that half my plots burned in the catastrophic Soberanes Megafire. It’s extremely rare to have pre- and post-fire samples from the same sampling locations before and after a mega-fire. I was really curious about what the impact of a mega-fire would be on soil microbial communities especially since they had never been studied in redwood tanoak forests before. These forests are endemic and charismatic megflora of Califronia that are facing multiple global change factors and it is really unclear how the soil microbial communities will respond to wildfires and how that will influence the recovery of the vegetation. I had already moved to southern California at this time to start a post-doc at UC Irvine, so I asked Kerri Frangioso, who lived in Big Sur, if she would be able to re-sample any of the plots that burned. Using GPS, she was able to collect soils from the exact same sampling locations that I had sampled in 2011 from 3 of the plots (2 burned and 1 unburned) within 30 days of the fire being declared over. She mailed these soils to me, I extracted the DNA, and froze everything until I was able to start my own lab at UC Riverside in 2018.

What difficulties did you run into along the way? 

The terrain in Big Sur is notoriously challenging to traverse. It is extremely steep, lots of windy dirt roads, and there is a lot of poison oak. There is no cell reception in any of our plots and most are at least an hour from the nearest town.  Collecting the soil even before the fire was challenging enough. However, after fires, it is really challenging to access sites because roads are closed, landslides are common, and dead or dying trees are extremely hazardous especially in the case of wind. We were very lucky to be able to re-sample even 3 of our plots so fast after the fire.

What is the biggest or most surprising innovation highlighted in this study? 

I was really surprised that many of the same pyrophilous “fire loving” microbes that have been found to increase in frequency after pine forest fires also increased in frequency after redwood tanoak fires. That indicates that soil microbes are selected for by slightly different pressures than plants because the plants that regenerate post-fire in pine forests vs redwood tanoak forests are very different. It seems more likely that microbes instead survive via temperature thresholds and if fire is high severity enough, similar groups of microbes will respond. We collaborated with Kazuo Isobe to implement the CONSENTRAIT analysis and identified that microbial response to fire was indeed phylogenetically conserved, and it seemed that related groups of bacteria and fungi did indeed positively or negatively respond to fires. This will greatly enhance our ability to predict which microbes will respond to fire in any ecosystem since certain lineages seem evolutionarily adapted to survive fires. We also found that a basidiomycete yeast Basidioascus, dominated the fungal sequences at 30 days post-fire, and that had never been found before, probably because most post-fire sampling historically has been based on fruiting bodies.

Morphological diversity of soil microbes. Photo credit: Jenna Maddox

Moving forward, what are the next steps in this area of research?

I was able to leverage some of these results and results from my work sampling wildfires in Southern California chaparral to help me acquire a USDA grant from their Agricultural Microbiomes program (described here). The purpose of this grant is to characterize the traits of pyrophilous microbes and begin to get our knowledge of fire adaptation in microbes to that of plants. We understand a lot of the traits that enable plants to survive wildfires (like thick bark, vegetative resprouting, serotinous cones, etc) but we don’t have similar understanding of those traits in microbes. In order to understand these traits, Dylan Enright has begun performing biophysical trait assays on these microbes to determine their traits based on a large culture collection of pyrophilous microbes that I have been developing since I started my lab in July 2018. Over the last four years, 2 lab managers, one PhD student (Dylan Enright), 13 UCR undergraduates, and one part time laboratory technician have been involved in developing this culture collection of over 400 isolates of bacteria and fungi from burned soils from wildfires. Our goal is to characterize their traits with biophysical assays and eventually with genomics.

Have you gone back (or have you any plans to go back) to sample soils in the post-fire period? How long lasting do you think the effects of fire on microbial communities would be? 

Unfortunately, I have not been able to get this particular project funded (despite several attempts) and everything I did for this paper was completely unfunded. So I have not been able to return to these plots to sample again. I would be interested in returning to them eventually. I would predict the effects of the fire on the microbial communities could last decades if not longer, depending on if the plants themselves have been able to recover. Most of the literature on pyrophilous microbes suggests that high severity fire can have long term impacts on soil microbes that can last at least a decade or more. Given that the richness of both bacteria and fungi was reduced by up to 70% in one of our plots, I would predict it will take a long time to recover.

Describe the significance of this research for the general scientific community in one sentence.

Megafires have long lasting impacts on both plants and soil microbes alike, and it is important to understand the impacts on soil microbes since they drive plant and soil regeneration. 

Describe the significance of this research for your scientific community in one sentence.

The pyrophilous microbes that respond to a mega-fire in redwood tanoak forests are similar to those that respond to high severity wildfires in better studied pine forest systems, and the fact that they are phylogenetically conserved indicates that we will be able to predict what microbes will respond to wildfires in any system. Further, we are beginning to identify conserved trait responses that enable wildfire response that are analogous to plants and will help us bin and better understand fire adaptation traits in microbes.

Enright, D. J., Frangioso, K. M., Isobe, K., Rizzo, D. M., & Glassman, S. I. (2022). Mega-fire in redwood tanoak forest reduces bacterial and fungal richness and selects for pyrophilous taxa that are phylogenetically conserved. Molecular Ecology, 31, 2475– 2493.

Interview with the authors: Genetic data and niche modeling reveal complex interspecific interactions of invasive species with native congeners and help evaluate distribution pattern, range limits and invasion risk of the species

In a recent paper in Molecular Ecology, Espindola and Vázquez-Domínguez et al. combined comprehensive fieldwork, genetic analyses and a novel niche modeling approach to investigate population genetic patterns, distribution patterns of native and non-native red-eared slider turtle (Trachemys scripta elegans), one of the worst invasive species across the world, and its congeners. They found very little naturally occurring distribution overlap and genetic admixture between red-eared slider and other Trachemys species studied. In addition, they demonstrated that the native Trachemys species in Mexico have distinct climatic niche suitability, which probably prevents the invasion of red-eared slider in the area. However, major niche overlap was found between non-native red-eared slider and native species from different parts of the world, indicating that sites closer to ecological optima of invasive species have higher establishment risk than those closer to the niche-centre of the native species.

We sent a number of questions to lead authors of this work, Sayra Espindola and Ella Vázquez-Domínguez, to get more detail on this study.

What led to your interest in this topic / what was the motivation for this study? 

We have interest in population genetics of invasive species. In addition, Trachemys scripta elegans, one of the World’s 100 worst invasive species, is native to NE North America, and several native congeneric species are naturally distributed along the eastern coast of Mexico, which is an extraordinary scenario to test the effect of congeners on potential invasion patterns and evaluate their climatic and niche differences.

Trachemys spp (T. scripta, T. venusta, T. cataspila, T. taylori) and their distributions along the west coast of USA and Mexico. Trachemys scripta (red dots) in Mexico is non-native. A turtle trapping net is shown. Figure credit: Sayra Espindola/Ella Vázquez-Domínguez

What difficulties did you run into along the way? 

Maybe the most significant was that, at the time we did the molecular laboratory work, extracting DNA from samples that had been stored in formaldehyde (museum samples) was rather difficult, thus we could not obtain genomic data (SNPs) for those samples (extraction kits are much more efficient now). Nonetheless, we did sequence nuclear microsatellites loci, which provided adequate genetic information that enabled us to show the significant contemporary genetic differentiation present between native and non-native Trachemys scripta elegans individuals. 

What is the biggest or most surprising innovation highlighted in this study? 

There are two interesting findings. One is that non-native Trachemys scripta elegans individuals have very little naturally occurring distribution overlap and admixture with its congeners – they exhibited reduced gene flow and clear genetic separation despite having zones of contact. Also, we demonstrate that the native Trachemys species studied (T. cataspila, T. venusta) have distinct climatic niche suitability, which prevents the establishment of and displacement by the non-native Trachemys scripta elegans. Yet, as T. s. elegans has invaded and displaced native turtle species worldwide, we show that sites closer to T. s. elegans’ niche-center have higher establishment risk than those closer to the niche-center of the native species.

Moving forward, what are the next steps in this area of research?

We are working with our genomic data to identify loci under selection to evaluate the potential connection between specific genes and adaptive traits in these turtles. Considering the distinct climatic niches and distribution we found for the turtles, we are very kin to elucidate if there are adaptive differences among them. In addition, our results set the basis for future work – whole genome or gene-targeted sequencing, as well as a higher number of field-sampled individuals, would allow assessment of hybridization and specific gene introgression.

What would your message be for students about to start developing or using novel techniques in Molecular Ecology?

We would first tell them that molecular ecology research, combining ecological fieldwork and laboratory tasks, is absolutely amazing! We recommend choosing to work with the species/taxa that you more deeply like – this makes the journey very enjoyable; and also selecting a laboratory and research group with ample experience in molecular work and analyses, while at the same time not afraid of proposing novel questions and ways of analyzing them.

What have you learned about methods and resource development over the course of this project?

In this project, we proposed and developed a novel modeling approach, in which by contrasting the niche suitability of the species, we were able to include, indirectly, the interactions that can occur when a species is introduced to habitats occupied by other species. The model is based on analyses of climatic niche suitability and the environmental centrality hypothesis, where fitness is expected to be highest in sites with environments closest to the center of the niche of the species. The development of this model and algorithms required an immense number of trials and errors, and once we had the final version, we had to again improve it after revision. The lesson then is that developing analytical models can take a lot, lot of time, but it is always worth the while!

Little climatic niche overlap between Trachemys scripta and two of its congeners, T. venusta and T. cataspila. Figure credit: Sayra Espindola/Ella Vázquez-Domínguez

Describe the significance of this research for the general scientific community in one sentence.

The distribution, range limits and potential risk of the invasion of invasive species can be evaluated with genetic information and ecological niche modeling.

Describe the significance of this research for your scientific community in one sentence.

Evaluating interspecific interactions between native and non-native closely related species with genetic information and niche modeling approach was key to determine the distribution patterns, range limits and invasion risks of Trachemys scripta elegans.

Wetlands system in the valley of Cuatrocienegas, Coahuila, Mexico, where the endemic Trachemys taylori lives. Photo credit: Ella Vázquez-Domínguez

Espindola S, Vázquez-Domínguez E, Nakamura M, Osorio-Olvera L, Martínez-Meyer E, Myers EA, Overcast I, Reid BN, Burbrink FT. 2022. Complex genetic patterns and distribution limits mediated by native congeners of the worldwide invasive red-eared slider turtle. Molecular Ecology.

Interview with the authors: Associations between MHC class II variation and phenotypic traits in a free-living sheep population

In a recent paper in Molecular Ecology, Huang, Dicks and colleagues analysed variation in the major histocompatibility complex (MHC) and phenotypic traits in an unmanaged population of sheep living on an island off the coast of Scotland. This population of sheep has been studied closely for more around 70 years, providing a very rare level of insight and statistical power to evolutionary genetic studies. The MHC is among the most variable parts of mammalian genomes and has long been known to be encode proteins central to the adaptive immune system. Through their analyses, Huang, Dicks and colleagues found associations with levels of circulating antibodies and variation at MHC loci.

We sent some questions to the corresponding author of this work, Wei Huang, to get more detail on this new study.

Rams in St Kilda. Photo credit, Martin Adam Stoffel.

Can you describe the significance of this research for the general scientific community in one sentence?

This study demonstrated the direct link between immune genes and antibody levels in wild populations.

What led to your interest in this topic / what was the motivation for this study? 

The major histocompatibility complex (MHC) contains a number of genes linked with immune defence in vertebrates. Associations between MHC variation and phenotypic traits or pathogens have been identified in many species. Also, selection on MHC genes has also been demonstrated in some studies. However, many previous studies only examined associations between MHC variation and a limited number of phenotypic traits or pathogens. Few of them have examined both MHC-fitness associations and MHC-trait associations. The longitudinal study of Soay sheep in St Kilda is a great system to study the associations between MHC variation and phenotypic traits and how the associations are linked with selection on MHC genes. Using three representative phenotypic traits monitored in thousands of sheep over decades, we are able to provide a full picture of MHC-trait associations in wild populations.

Can you describe the significance of this research for your scientific community in one sentence?

This study suggests associations between MHC and phenotypic traits are more likely to be found for traits more closely associated with pathogen defence than integrative traits and highlights the association between MHC variation and antibodies in wild populations.

What difficulties did you run into along the way? 

It is extremely hard to monitor populations and collect longitudinal data over decades. Thanks to our great field assistants and volunteers, the Soay sheep data has provided a good foundation. In terms of the specific study, the first difficulty is to genotype MHC in a large number of sheep. We used two steps to genotype the MHC genes. We first used genotype-by-sequencing to genotype hundreds of sheep. Then, benefiting from the high-density sheep SNP chip, we were able to use 13 SNPs to genotype MHC in the other thousands of sheep successfully.

Additionally, it is hard to choose the appropriate model. Some of our traits are not normally distributed and are also not closed to other common error structures. We instead used Bayesian statistical methods to run the analysis.

What is the biggest or most surprising innovation highlighted in this study? 

We used three representative traits to examine the associations between MHC variation and phenotypic traits. The traits included a fitness-related integrative trait, body weight, a measure of gastrointestinal parasites, faecal egg count, and level of three antibodies. All of the three traits are related to fitness. We only found associations between MHC variation and antibodies. Such results reflect the important role of MHC in immune defence in wild populations. Our study is one of the first studies to examine associations between MHC variation and multiple phenotypic traits. 

How do you think your results generalize to other systems?
Our study is based on the longitudinal study of Soay sheep. The large sample size provides great statistical power. Therefore, our results are reliable and solid. Also, we investigated phenotypic traits that have different links with immune defence. Therefore, our results can reflect the general pattern of MHC-trait associations.  

You conclude from your study that MHC variation is more likely to be associated with immune traits. How would you validate your findings for species with less rich data?

First, it is possible to use experiments to test the associations. In terms of wild populations, future studies can investigate multiple populations or multiple traits in a single population if they are restricted by the study length.

Moving forward, what are the next steps in this area of research?

Our study demonstrates that it is important to study MHC-antibody associations. Future studies should focus on immune traits rather than only examine MHC-pathogen associations. Also, previous studies are often restrained by small sample size. It would be nice if future studies could increase their sample size to strength the statistical power.

Huang, W.*, Dicks, K. L.*, Ballingall, K. T., Johnston, S. E., Sparks, A. M., Watt, K., Pilkington, J. G., & Pemberton, J. M. (2022). Associations between MHC class II variation and phenotypic traits in a free-living sheep population. Molecular Ecology, 31, 902– 915. 

*These authors contributed equally to this work

Interview with the authors

A holobiont view of island biogeography: Unravelling patterns driving the nascent diversification of a Hawaiian spider and its microbial associates

In their recent paper in Molecular Ecology, Armstrong and Perez-Lamarque et al investigated the evolution of the holobiont. The holobiont is the assemblage of species associated with a particular host organism. In the case of this study, the holobiont refers to the stick spider (Ariamnes), its microbiome and its endosymbionts. Taking advantage of the successive colonization of islands in a volcanic archipelego, Armstrong and Perez-Lamarque et al contrasted the evolutionary history of the host species to the different components of the holobiont on different islands in Hawai’i.

We sent some questions to the authors of this work and here’s what Benoît Perez-Lamarque, Rosemary Gillespie and Henrik Krehenwinkel had to say.

Ariames waikula (on the island of Hawaii). Photo credit: George Roderick

What led to your interest in this topic / what was the motivation for this study? 

Gut microbiota play multiple roles in the functioning of animal organisms. In addition, host-associated microbiota composition can be relatively conserved over time and the concept of the “holobiont” has been proposed to describe the ecological unit formed by the host and its associated microbial communities. Yet, it remains unclear how the different components of the holobiont (the hosts and the microbial communities) evolve. This is what spurred our interest. Taking advantages of the chronologically arranged series of volcanic mountains of the Hawaiian archipelago, we were able to tackle this question and could investigate how the different components of the holobiont have changed as the host spiders colonized new locations.   

Can you describe the significance of this research for the general scientific community in one sentence?

The evolution of Hawaiian spider hosts and their associated microbes are differently impacted by the dynamic environment of the volcanic archipelago.
Can you describe the significance of this research for your scientific community in one sentence.

The host and its associated microbiota may not act as a single and homogeneous unit of selection over evolutionary timescales.

Ariames waikula (on the island of Hawaii). Photo credit: George Roderick

What difficulties did you run into along the way? 

All the different components of the holobiont are not as easy to study. For instance, for the host spiders, we used double digest RAD sequencing (ddRAD) to obtain genome-wide single nucleotide polymorphism data. With such data, we could precisely reconstruct the evolutionary histories of the different spider populations in the last couple of million years and tracked the finest changes in their genetic diversity. In contrast, characterizing the composition of the microbial components is much more challenging. We used metabarcoding of a short region of the 16S rRNA gene to identify the bacteria present. However, over such short evolutionary timescales, this DNA region is too conserved to accumulate many differences between isolated populations. Therefore, we had high-resolution data for the spider hosts but comparably low-resolution data for the bacterial communities. To ensure that the observed patterns were not artefactually driven by such differences of resolutions, we complemented our analyses with a range of simulations to assess the robustness of our findings.

What is the biggest or most surprising innovation highlighted in this study? 

We find that the different components of the holobiont (the host spiders, the intracellular endosymbionts, and gut microbial communities) respond in distinct ways to the dynamic environment of the Hawaiian archipelago. While the host spiders have experienced sequential colonizations from older to younger volcanoes, resulting in a strong (phylo)genetic structuring across the archipelago’s chronosequence, the gut microbiota was largely conserved in all populations irrespectively of the archipelago’s chronosequence. More intermediately, we found different endosymbiont genera colonizing the spiders on each island. This suggests that this holobiont does not necessarily evolve as a single unit over long timescales.

In the conclusion to your study, you point out how different components of the holobiont likely contribute differently to selection/colonization history in this system. If you had unlimited resources, what would you do to strengthen this conclusion? 

We indeed suspect that the different components of the holobiont probably did not act as a single and homogeneous unit of selection during the colonization of the Hawaiian archipelago. First, it would be ideal to perform an even broader sampling, targeting more Ariamnes populations and species from older islands, to better characterize the long-term changes of the different holobiont components. Using sequencing technics with better resolution (as detailed below) would also improve our characterization of the microbial component(s) of the holobiont. Second, to properly test for selection, we should perform transplant experiments of the bacterial communities between spider populations/species and measure whether or not it impacts holobiont fitness. We would expect to find a significant impact of the transplant for the endosymbionts, but no or low impact for the gut bacterial communities of these spiders.

The geological history of Hawai’i provides a powerful system to build understanding of the evolution of holobiont. Are you aware of other systems where similar studies could be performed? (I appreciate that this is related to the previous question!).

Many other archipelagos, with similar island chronosequences, like the Canary Islands or the Society Islands, are also ideal for testing hypotheses on the evolution of holobionts. Within the Hawaiian archipelago again, we could replicate our work on other holobiont systems. For instance, among arthropods, plant feeders might rely more importantly on their microbiota for their nutrition, and this might likely translate into different patterns of holobiont evolution.

Moving forward, what are the next steps in this area of research?

As previously said, one main limitation is the low resolution of the 16S rRNA metabarcoding. This prevented us to look at the evolutionary history of the individual bacterial lineages. Using a new model, we have recently tackled this issue of low resolution ( and we reported little evidence of microbial vertical transmission in these holobionts. Yet, the next step would be to move from classical metabarcoding to metabarcoding with longer sequencing reads (e.g. the whole 16S rRNA gene) or even metagenomics. It would provide more resolution for looking at bacterial evolution and would also bring more information on the functioning of these bacterial communities (e.g. are gut microbiota contributing to the digestion of these Hawaiian spiders in natural environments?).

Armstrong, E. E.*, Perez-Lamarque, B.*, Bi, K., Chen, C., Becking, L. E., Lim, J. Y., Linderoth, T., Krehenwinkel, H., & Gillespie, R. G. (2022). A holobiont view of island biogeography: Unravelling patterns driving the nascent diversification of a Hawaiian spider and its microbial associates. Molecular Ecology, 31, 1299– 1316. 

*Authors contributed equally

Interview with the authors: Transposable elements mark a repeat-rich region associated with migratory phenotypes of willow warblers

In a recent paper in Molecular Ecology, Caballero-López and colleagues investigated the genetics of migratory behaviour in a two subspecies of willow warbler (Phylloscopus trochilus trochilus and Phylloscopus trochilus acredula). Previous work had identified several genetic markers associated with migratory behaviour in this species, but a particularly important candidate marker was unable to be mapped to previous genome assemblies. This suggested to Caballero-López et al, that the important marker may lie in a highly repetitive, and thus difficult to assemble, genomic region. Leveraging a recent genome assembly based on long-read technology and a quantitative PCR approach, Caballero-López et al found that the elusive migration marker is located in a genomic region rich in remnants of transposable elements.

We sent some questions to the primary author of this work, Violeta Caballero-López, to get some more insight and details about this exciting study.

Willow warbler male, 2017. Photo credit: Harald Ris.

What led to your interest in this topic / what was the motivation for this study? 

My research aims to shed some light on our understanding of the genetics underpinning bird migration, which is currently very poor. Passerine birds migrate alone, and they follow the same routes to wintering grounds as their parents, fully relying on genetic mechanisms.

The motivation for this specific study was to try to characterize a region in the genome which varies between two subspecies of willow warbler that present differential migration to Africa. Until now, this region was only identified as an AFLP-derived marker which failed to be mapped to the genome. However, with the use of molecular techniques such as qPCR in combination with a good quality genome assembly, we could understand the nature of this element better.

AFLP: Amplified Fragment Length Polymorphism

qPCR: Quantitative Polymerase Chain Reaction

Can you describe the significance of this research for the general scientific community in one sentence?

Repeat-rich regions which are often considered “junk DNA” might have a larger role on phenotypes and function than previously thought.

Can you describe the significance of this research for your scientific community in one sentence?

It is important to revise the role of repeat DNA on the determination of a complex trait such as the determination of bird migratory routes.

Willow warbler singing in Siberia, 2017. Photo credit: Harald Ris.

What difficulties did you run into along the way? 

For more than 20 years “WW2” has been an elusive AFLP marker, observed to be fixed in the “northern” subspecies P. t. acredula. It could only be amplified in PCR as a 154 bp fragment and then sequenced, but its nature was totally unknown. The identification and curation of this sequence as a transposon (TE) was challenging because it is an old, degraded “LTR portion” of the full element. This required a willow warbler genome built with long read sequencing techniques that provided regions of the genome rich in repeat DNA. Locating the ends of this transposon was also complicated. Alignment “breaks” serve as a detection method for the target site duplications that mark the edge of these elements. However, they could not be used in our system because these TEs appear consistently embedded within a larger block of repeats. This interfered with our estimation of age and theories about the origin of the repeat.

LTR: Long terminal repeat

What is the biggest or most surprising innovation highlighted in this study? 

The most surprising finding here is the presence of a large repeat-rich region (>12 mb) that segregates in both willow warbler subspecies. This region is characterized by several copies of the WW2 derived variant, which turned out to be part of a transposable element belonging to the endogenous retrovirus family. Furthermore, we provide solid evidence of its independence from the other polymorphic regions in chromosomes 1 and 5. As this TE seems to be inactive, and no clear functional genes have been detected on its surroundings, it remains puzzling why this region correlates with migration in the willow warbler so strongly.

You end your paper describing how it’s premature to think that the association of the WW2 derived variant has a causal role on the trait. Based on your knowledge of the warbler genome, would you care to speculate as to the actual causal basis of the phenotype?

The most supported hypothesis is that migration is a complex trait influenced by gene packages. In the case of the willow warblers, I would speculate that the repeat rich region, and not necessarily the WW2 derived variant itself, could affect migration indirectly through 1) the formation of a structural variant in a chromosome that affects gene expression 2) the trans-regulation from this region of some gene(s) elsewhere in the genome 3) the presence of an adjacent gene outside this region that we have not been able to detect in the current genome assemblies so far 4) a missed single copy gene within the repeat rich region. However, the last one is the least likely given that areas with such a repeat density rarely contain functional genes.

Have you got any ideas of how you might test the hypothesis that chromosomal rearrangements were facilitated by the presence of TEs

The most exciting possibility is to visually confirm if these rearrangements have taken place. A way to test this empirically would be to obtain a karyotype of each subspecies and combine it with fluorescence in situ hybridization (FISH). First, a probe labelling the WW2 derived variants would signal the location of the repeat-rich region. Once the location of this region is resolved, it is possible to design several fluorescent probes outside of it to determine if the chromosomal arrangement around it is maintained both in the genome of P. t. acredula and its orthologue region in P. t. trochilus.

Moving forward, what are the next steps in this area of research?

The biggest mystery within this study is the location in the genome of this repeat-rich region that contains several copies of the WW2 derived variant. One of the biggest challenges of genome assemblies is the mapping and correct location of repeat-dense sequences, and therefore future effort should be focused on targeting empirical evidence of the location of this region. Then we could get a better hint on if and/or how this region affects migration. Is it downstream or upstream of any gene complex? Is it silenced? how does its orthologue look in P. t. trochilus?

Typical working setup for the willow warbler team, 2021. Photo credit: Harald Ris.

Caballero-López, V., Lundberg, M., Sokolovskis, K., & Bensch, S. (2022). Transposable elements mark a repeat-rich region associated with migratory phenotypes of willow warblers (Phylloscopus trochilus). Molecular Ecology, 31, 1128– 1141.

Interview with the authors: How genomic data reveal cryptic species and how migration patterns maintain genetic divergence in birds?

In a recent paper in Molecular Ecology, Tang et al. investigated genetic divergence of different subspecies of pale sand martin (Riparia diluta) using genome-wide data. They found that the subspecies in Central and East Asia, which vary only gradually in morphology, broadly represent three genetically differentiated lineages. No signs of gene flow were detected between two lineages that met at the eastern edge of the Qinghai-Tibetan Plateau, which is likely due to largely different breeding and migration timing. Limited mixed ancestries were found in Mongolian populations between two lineages that might take divided migration routes around the Qinghai-Tibetan Plateau, and the authors hypothesize that selection against hybrids with nonoptimal migration routes might restrict gene flow. See the full article for more details of the study and the interview with lead author Manuel Schweizer below for more stories behind this exciting work.

Pale sand martin Riparia diluta tibetana, Mongolia, June 2018. Photo Credit: Manuel Schweizer

What led to your interest in this topic / what was the motivation for this study? I studied pale sand martin in Central Asia as part of the work on a field guide to the birds of Central Asia, which was published in 2012. I was then fascinated by the fact that the different subspecies described for this species breed in completely different environments: Central Asian steppes and semi deserts, high altitude grasslands on the Qinghai-Tibetan Plateau, or lowland subtropical China. Although it was evident that morphological identification of single individuals of the different subspecies without context is not possible, I suspected that cryptic diversity might be involved in this complex. This was corroborated by mtDNA data that we published in 2018. Together with Gerald Heckel and our PhD student Qindong Tang, I wanted to investigate this further using genome-wide data and test if gene flow is reduced in areas of potential contact between evolutionary lineages.

Breeding site of pale sand martin on the east edge of Qinghai-Tibetan Plateau in Zoige (Sichuan Province, China). Photo Credit: Qindong Tang

What difficulties did you run into along the way? The biggest challenge was to get a comprehensive geographic sampling together. As pale sand martins breed in low densities only, this meant a lot of travelling. Fortunately, we could count on the great support of our collaboration partners and their network – Yang Liu from Sun Yat-sen University in Guangzhou, China, and Gombobaatar Sundev from the National University of Mongolia. Moreover, Qindong Tang made an incredible effort and did an excellent job during the fieldwork. 

What is the biggest or most surprising innovation highlighted in this study? Given the absence of obvious sexually selected traits and only gradual morphological differentiation between the different evolutionary lineages of the pale sand martin, the level of genetic differences and the fact that they behave like different species at least at the eastern edge of the Qinghai-Tibetan Plateau is indeed surprising. So, we were left with the following question: what processes and mechanism prevent a complete mixing at secondary contact zones? We think that seasonal migration behavior might be an essential factor in maintaining genetic integrity of these morphologically cryptic evolutionary lineages.

Moving forward, what are the next steps in this area of research? The next step is evident: we need to study in detail migration behavior of the different lineages. The ranges of two of them meet in the area of a well-known avian migratory divide, where western lineages take a western migration route around the Qinghai-Tibetan Plateau to winter quarters in South Asia, and eastern lineages take an eastern route to Southeast Asia. This might also be the case in the pale sand martins and we hypothesize that hybrids might have nonoptimal intermediate migration routes and selection against them might restrict gene flow. This will need quite some field work and application of up-date technologies such as modern data loggers. Let’s hope that the development of the pandemic will allow field work again soon.

What would your message be for students about to start developing or using novel techniques in Molecular Ecology? It is best to get started and not be intimidated or even afraid. The easiest way to learn new methods is to start using them. It is also important to build a network of people who can be asked for support when problems arise.

What have you learned about methods and resource development over the course of this project? As always in studies with a phylogeographic background, sampling matters most. Try to organize a complete geographic sampling in the beginning of a project. Sampling in parts of the distribution area of our study system was planned in the third year of Qindong’s PhD, however, this could not be achieved due to the pandemic. As a consequence, we still lack samples from western Mongolia which would have been important and made the overall picture more comprehensive. This work did not include any development of new methods, however, a knowledge of state-of-the art methodological approaches is obviously always crucial.

Describe the significance of this research for the general scientific community in one sentence. Our study points towards contrasting migration behavior as an important factor in maintaining evolutionary diversity under morphological stasis.

Describe the significance of this research for your scientific community in one sentence. Our discovery of cryptic diversity in the pale sand martin indicates that evolutionary diversity might be underestimated even in such well-studied groups such as birds, and it suggests that it is worth having a closer look at widespread species occurring in different environments.

Photo of the first author Qindong Tang during the field work. Photo Credit: Qin Huang
Sampling team in Qinghai, PR China, June 2016. From left to right: Manuel Schweizer, Paul Walser Schwyzer, Yang Liu, Qin Huang, Yun Li and our driver. Photo Credit: Manuel Schweizer
Sampling team in Mongolia, June 2018. From left to right: Tuvshin Unenbat, Turmunbaatar Damba,  Gombobaatar Sundev, Paul Walser Schwyzer, Manuel Schweizer, Silvia Zumbach, Sarangua Bayrgerel. Photo Credit: Manuel Schweizer

Tang Q, Burri R, Liu Y, Suh A, Sundev G, Heckel G, Schweizer M. 2022. Seasonal migration patterns and the maintenance of evolutionary diversity in a cryptic bird radiation. Molecular Ecology.

Interview with the authors: How can we use machine learning and genomics to make predictions about the effects of climate change?

In a recent paper in Molecular Ecology Resources, Fitzpatrick et al. used a combination of common garden experiments, genome sequencing, and machine learning analyses to understand how genomic offsets (a measure of maladaptation) can be used to predict how organisms might respond to future environmental change. They found that genetic offset was negatively associated with growth and was a better predictor of performance than the difference in sampling site and common garden environmental variables were alone. See the full article for more details on how these trends aligned with panels of putatively informative and randomly selected SNPS, and the interview with lead author Matthew Fitzpatrick below for even more insight into this exciting work.

What led to your interest in this topic / what was the motivation for this study? My research focuses on spatial modeling of biodiversity and involves forecasting how climate change may impact natural systems. Demand for such forecasts continues to grow given the threats facing biodiversity. However, a major – and often overlooked – challenge is assessing forecasting models, which is really important given their potential to (mis)inform conservation. 

The motivation for this study was to test a type of genomics-based forecast founded on an idea that my coauthor Steve Keller and I developed a few years ago that we termed “genetic offsets”. Genetic offsets are in essence a forecast of climate maladaptation based on existing relationships between (adaptive) genomic variation and climate gradients. We tested how well genetic offsets correspond to biological responses to rapid climate change – in this case by transplanting trees from their home climate to a common garden experiment and measuring their response.

What difficulties did you run into along the way? There were all sorts of challenges one might expect with setting up and running common gardens experiments in two countries, which as the modeler on the project I, thankfully, was largely isolated from. We were lucky to have Raju Soolanayakanahally on our team to help with common garden logistics in Canada, along with Steve’s lab running the Vermont common garden. Additionally, there was the challenge of how best to evaluate the population genomic data for signatures of local adaptation prior to the genetic offset modeling. This can always be a challenge to ensure you’re minimizing the effects of population structure and false positives. Steve and his former postdoc Vikram Chhatre approached this from several angles to make sure we had a robust set of selection outliers. From the modeling perspective, we had to be creative about fitting and summarizing a very large number of machine learning models.  

What is the biggest or most surprising innovation highlighted in this study? We found pretty solid evidence that genetic offsets can serve as a meaningful estimate of the degree of expected maladaptation of populations exposed to climate change. It was nice to get some confirmation of our idea, but what was really surprising was that sets of randomly selected SNPs predicted performance of trees as well as or slightly better than did our set of carefully selected candidate SNPs, which was the opposite of what we expected. We’ve seen some other evidence in our simulation studies that also suggest SNPs from the genomic background can be predictive of maladaptation, although the reasons for this are still being investigated.

Moving forward, what are the next steps in this area of research? Ours is a single study on a single species of tree. Many more tests are needed in other study systems before we can fully understand the situations in which genetic offsets can serve a useful purpose. Also, our study tested genetic offsets derived from the machine learning method Gradient Forest, but Gradient Forest is just one of several statistical methods that can be used to estimate offsets. An important next step in my lab is to perform similar testing using another promising method known as generalized dissimilarity modeling.

What would your message be for students about to start developing or using novel techniques in Molecular Ecology? Take good notes and document the process! You will thank yourself later. If you are developing a new method, it is important to thoroughly test it to be sure you understand how it behaves in different circumstances and to make clear its intended uses before publishing on it. And last, teach others to use your method! 

What have you learned about methods and resource development over the course of this project? I thought I knew a lot about Gradient Forest and its behavior, but this study – and another we have in review on testing genetic offsets using simulated data – taught me that methods do not always behave the way we might expect or hope. And even when we have simulated “truth known” data, it can be difficult to understand why methods are behaving a certain way.

Describe the significance of this research for the general scientific community in one sentence. This study shows that for some organisms it may be possible to use genetic data to inform climate change impact assessments.

Describe the significance of this research for your scientific community in one sentence. This study provides evidence that spatial patterns of adaptive genomic variation along climatic gradients can be used to estimate the magnitude of expected maladaptation of populations exposed to rapid climate change through time.  

Fitzpatrick MC, Chhatre VE, Soolanayakanahally RY, Keller SR. 2021. Experimental support for genomic prediction of climate maladaptation using the machine learning approach Gradient Forests. Molecular Ecology Resources.