The Basque Government approved our proposal for a Terrestrial Plant Diversity research group.  This was a required step to permitting me to sponsor and guide student doctoral research.  Our next step is to apply for space in the new María Goyri research building.

See the full proposal here.

Here is the text for the work I will lead with this group.  I welcome inquiries by potential doctoral students.

Research Direction: Evolutionary Ecology

Introduction

Due to rapid climate change caused by anthropogenic drivers, many terrestrial plants face new conditions that challenge their ability to maintain viable populations. Species with limited dispersal potential, or narrow habitat requirements and patchy distributions, will need to adapt in situ to new conditions, generally through evolutionary responses driven by natural selection. Understanding how species adapt when facing changing abiotic environments is crucial for developing a comprehensive understanding of the response of biodiversity to environmental change. The genetic basis of these responses, and the genetic record of previous responses, varies both among and within species. This variation can illuminate the adaptive evolution of populations, the historical demography of populations, and the lineage diversification that occur has occurred as barriers to gene flow arise and are later surmounted. In recognition of the importance of understanding the capacity of species to respond and persist in changing environments, this research aims to understand the adaptive and demographic responses that make possible the persistence and diversification of evolutionary lineages.

While understanding how adaptation and demography impact diversification could be approached from many angles, we propose an overarching RESEARCH QUESTION:

How can we use genetic patterns to help distinguish between alternative ecological mechanisms that impact species distribution, adaptation to environment, and evolutionary diversification?

We address this question through the elucidation of adaptive and demographic responses to environment, by seeking to generate alternative models of ecological and evolutionary mechanisms and compare them in their ability to account for empirical data. We develop this model-based approach strategically by establishing three research activities in which we:

 (i) develop genetic modelling of the coalescent process, distribution modelling based on environmental tolerances, and population dynamic simulations to test alternative demographic hypotheses using a quasi-Bayesian approach;

 (ii) employ Next Generation, High-Throughput sequencing and comparative genomic analyses, in combination with Bayesian statistical modelling of the relationships between genetic and environmental variation, to identify the demographic history of adaptive variation in diversifying intra-specific lineages; and

 (iii) conduct phylogenetic comparative analysis using Bayesian methods of tree construction and alternative models of rates of ecological diversification to identify tradeoffs involving adaptation along alternative ecological axes.

In these research activities, the questions define the primary objectives. These are operationalized in each of the three research areas as fine scale objectives. The research questions we address fall into three corresponding research areas: i. paleoecology and phylogeography, ii. population genomics, and iii. comparative phylogenetic analysis, each of which we address in turn below. In addition, our incorporation of Next Generation Sequencing and genomic analysis, which we have already begun, puts a premium on acquisition of basic DNA laboratory capacity for this research group.

Post-Glacial Recolonization of European Trees: Coalescent and Demographic Models

Background--Work by other researchers

Glacial periods of the late Cenozoic presented conditions that were cold and dry enough that most plant species in temperate northern latitudes were excluded from much of their modern ranges (Ehlers and Gibbard 2004, Ramstein et al. 2007). During these times, species were forced to disperse into refugia in southern areas of North America and Europe and were extirpated from areas with inimical climate (Jackson and Overpeck 2000, Ehlers and Gibbard 2004). For example, dominant tree species in Europe, in one line of thinking, were restricted to several refugia in southern Europe and recolonized Europe from these locations (Lang 1994, Willis and van Andel 2004). Further, pollen data suggests that Fagus sylvatica, now widely spread across Europe, was present in the Pyrenees, Cantabrian and southern Apennine mountains, and potentially in the Dinaric Alps. Similarly, pollen data suggest that Picea abies was distributed at the last glacial maximum (LGM, 24-20 ky BP) among the Russian plateau, the northern Carpathians, eastern Dinaric Alps, the Bohemian plateau and along the eastern Austrian Alps (Tollefsrud et al. 2008). Pollen data are not, however, the only evidence for the presence of refugia; additional evidence suggests P. abies may have had refugial areas in the far north of Europe (Kullman 2008, Parducci et al. 2012).

Genetic data indicate that the occupation of refugia by P. abies and the subsequent recolonization of Europe and Finoscandia by the species has been a complicated process. Spatial Analysis of Molecular Variance (SAMOVA) of mitochondrial sequence data shows that populations in the northern portion of the range of P. abies are genetically distinct from southern ones (Tollefsrud et al. 2008). These populations also have reduced genetic diversity and more-distinct geographic structuring of genetic variation. These differences suggest different demographic processes have differentially impacted populations during the LGM and subsequent post-glacial recolonization; compared to northern populations, southern population may have experienced more extreme, prolonged bottlenecks and loss of genetic variation. These genetic differences among northern and southern populations might have been sufficient to conclude that current northern populations are likely the result of recolonization from a single, continental northern refugia, while the southern populations recolonized from diverse sources. However, several observations, some using cutting-edge techniques for analysis of ancient DNA, call this scenario into question.

Three sources of data suggest the standard model of post-glacial colonization by P. abies from southern refugia is incomplete (Willis and van Andel 2004). First, recently acquired ancient DNA from a lake on the coast of Norway and phylogeographic analysis of P. abies populations in western Scandinavia suggest recolonization from a noatak on the northern Norwegian coast (Parducci et al. 2012). Second, dating of wood macro-remains of P. abies suggests the species was present in de-glaciated areas of north-central Scandinavia during the late glacial period (Kullman 2008, Paus et al. 2011). Third, sediment cores from unglaciated areas of southwest Britain provide Picea pollen during a restricted period that includes LGM, but does not continue beyond the late glacial (Kelly et al. 2010). These results suggest that the standard scenario of survival in southern refugia and subsequent re-colonization of P. abies exclusively from these areas is an incomplete picture. Unfortunately, widely distributed fossil pollen records only reach to the very late glacial period, and pollen records are up to 10 ky removed from the LGM (Lang 1994). They provide no contemporaneous record of refuge locations and tree macrofossil remains from the LGM are also very rare.

Own contributions--Phylogeography of European Trees

Refinements of species distribution models have been important in the assessment of climate impacts. In these models, the assumption of niche conservatism (Pearman et al. 2008a) is crucial when using species distribution models to infer the paleo-distributions of species. We have examined this assumption for major European tree species (Fagus sylvatica, Abies alba, and Picea abies) and several widely distributed understory species (Pearman et al. 2008b). Niche models were fitted with observed, interpolated data on current climate and recorded species distributions, then projected to gridded climate data from General Circulations Models of mid-Holocene. We tested the fit of the projected models by comparing the predicted mid-Holocene distributions to the species distribution represented by semi-fossil pollen from cores dated to the mid-Holocene. We found that the Holocene distribution of understory trees, represented by fossil pollen, was not well predicted (Pearman et al. 2008b). In contrast, the Holocene distribution of dominant tree was well predicted, as evaluated by AUC-ROC scores, suggesting niche conservatism in these species. The working assumption of niche conservatism seems reasonable for dominant European tree species.

The proposed use of distribution models to identify climate envelopes of phylogeographic lineages of dominant tree species is well supported by modelling results from our lab. Application of distribution models to intra-specific lineages is fairly new (Pearman et al. 2010). In fact, distribution models of intraspecific lineages of birds, reptiles, and large mammals may produce more-accurate predictions of potential impacts of climate change than modelling efforts that ignore intraspecific lineage structure (D'Amen et al. 2011). In addition, we have found that the application of models to identify LGM refugia can infer small but meaningful differences in the locations of glacial refugia of closely related species, and contribute to the identification of mechanisms of diversification (Schorr et al. 2012, Schorr et al. 2013). Our unpublished analyses indicate that, if one assumes that the climate niche envelopes of tree lineages are conserved, lineages of P. abies had distinct glacial refugia (Pearman, unpublished). Further, suitable climate for P. abies likely existed on the Atlantic coast of Europe in scattered locations between northern Norway and the southern coast of Britain.

A. Objectives These results suggest the following RESEARCH QUESTIONS, finding the answers to which serves as our objectives:

1. Does the current genetic structure of European tree populations provide differential support for models of refuge location and post-glacial colonization?
2. What geographic distribution of glacial refugia is most supported by current population genetic structure of Picea abies and other dominant trees?
3. Do demographic models of recolonization support the existence of northern refugia for European trees?

B. Research Plan

We will address the degree to which current population genetic structure supports different modeled refuge locations and recolonization dynamics by implementing a modeling effort composed of three key components:

• a. distribution models that fit climate envelopes of phylogeographic lineages;
• b. coalescent genetic models using the SimCoal software; and
• c. flexible population models developed in the SPLATCH software, which incorporates a wide variety of parameters that influence demography.

Because these modelling paradigms are well developed, the novelty of this work lies in the synthetic combination of paradigms, their application to unresolved questions of post-LGM recolonization from refugia, and implementation of an approach that includes comparison of alternative models to address persistent questions in paleobiogeography. The analysis framework follows a pseudo-Bayesian framework in which a large parameter space is surveyed during the modelling of coalescent processes and population expansion. The output of the combination of models is a geographic distribution of populations of varying population genetic structure. These distributions are compared to the observed geographic and genetic distributions of modern populations. We intend to develop and implement approximate Bayesian computation (ABC) to identify the distribution of parameters on which to base inferences.

The research will proceed along the following work plan, which consists of a series of concrete fine-scale objectives:

B. 1. We will acquire genetic and pollen data from collaborators (in existing collaborations). An agreement for data acquisition has already been reached in regard to Picea abies. Sources for genetic data on Abies alba and Fagus sylvatica have been identified and will be approached. We have every reason to believe that these people will also be willing to provide data as they are already published.

B. 2. We will organize and map in a geographic information system (GIS) species occurrence and pollen data as they are acquired.

B. 3. We will then conduct paleoclimate envelope modelling in R, using existing modeled LGM and Holocene data from general circulation models (GCMs). These data are already in our possession. Based on the projected climate envelope models, and previously published data on the distribution of fossil pollen, we will define a suite of alternative hypotheses for refugia location and recolonization pathways.

B. 4. We will then implement SimCoal/SPLATCH modeling over defined parameter space in the ABC framework. This work will make use of a parallelized computational approach, taking advantage of the computers in the Department of Plant Biology and Ecology’s Laboratory for Computational Ecology and Evolution.

B. 5. We will calculate descriptive population genetic statistics using the Arlequin software and range statistics for the modelled genetic lineages in spatial analysis functionality in R. The statistics will be used to develop an approximate posterior distribution for these statistics in relationship the same statistics generated from the observed genetic data.

B6. Based on these posteriors we will synthesize and evaluate the relative support for alternative scenarios (refugia location and migration rates). Once this has been begun, we will also begin manuscript production.

Risk

The various data sources are either in public databases or are have been published in the scientific literature. The only risk is that the number of eventual co-authors will increase. We do not have a thorough knowledge of the computation time that will be involved. This is counterbalanced by the availability of 4 workstations with a total of 0.8TB of RAM. The University also has a computation cluster on which we could purchase computation time. Designing the workflow for ABC will require knowledge transfer from the lab of L. Excoffier at the University of Bern, Switzerland. We know of no reason why this would not be possible. Other labs with ABC experience, ones with which we have had previous contact, could also be approached.

C. Internationalization

This is a highly collaborative, international project. Providers and participants so far reside in five nations (UK, Norway, Sweden, Switzerland and France), and the Basque Country. Additional data acquisition will result in several additional participants. We will further expand the collaboration network by contacting groups that are responsible for the development and maintenance of the simulation and population genetic software. We will arrange that the doctoral student will visit several of the collaborating laboratories, and will make an extended stay in a lab known for ABC implementation. All manuscripts will be submitted to journals that are indexed in the ISI Web of Science. We foresee participation in the following international meetings: 2017 invited talk Norwegian Inst. Bioeconomic Research, As, Norway; 2018 invited talk University S. Hampton, UK; 2018 invited talk U. Montpellier, France; 2020 invited talk Univ. Bern, Switzerland; 2017, 2019 meetings of European Society for Evolutionary Biology; 2017-2021 meetings of the Society for the Study of Evolution (USA); 2018, 2020 meetings International Biogeography Society.

D. Activities for group integration

We will begin to examine how a model-based approach might be applied to additional research in the other three research lines. In particular, we will examine how demographic models, such as those used here, can be applied to understand the dynamics of species invasions (L2).  We will examine the possibility of coordinating this activity with Activity 4, in order to develop a demographic modelling component that will explain the origin of phylogeographic diversity (L1).

9. Adaptive diversification in a widely distributed, arid-adapted perennial: A NextGeneration Sequencing approach. (Current financing of 70.000 € through ERC Project ADAPT)

Background--Work by other researchers.

Widely distributed species may face a range of environmental conditions, especially variation in soil characteristics and climate. Regarding climate, Darwin expressed a certain ambivalence regarding the importance of climate in limiting species distributions, writing on the same page that each species is adapted to climate, but that range limits were determined by competition with other species. In the intervening period of more than 150 years, the importance of adaptation of species to the range of environments they encounter has developed into a major research direction (Balding 2006, Parisod and Holderegger 2012, Schoville et al. 2012). The ability of plants, for instance, to adapt to local conditions is important given rapid global climate change; plants with limited dispersal ability or restricted habitat requirements will not be able to disperse to follow the changing geographic distribution of suitable conditions, and will have to adapt in situ to new conditions (Visser 2008, Hoffmann and Sgro 2011). Thus, while local adaptation to environment may be frequent in plants (Savolainen et al. 2007), it is unclear how the existence of local adaptation in a plant species is related to its evolutionary potential when faced with environmental change.

Rapid environmental change can impose strong selection on plant populations. When particular genetic variants express phenotypes that are highly advantageous in new environments, selection can remove much or all of the genetic variability at the loci with major effects on the phenotype, a process known as selective sweep (Muir and Filatov 2007). The effect of selective sweeps on genomic variation depends on recombination rates and the distribution of affected loci on chromosomes, with the result that diversity may be reduced only in a restricted region surrounding the locus (Bierne 2010); the rest of the genome and much of the affected chromosome may not experience a reduction in diversity, with the result that at the population level, some regions of the genome may be fixed for a sequence while others remain unfixed and segregating.

While selective sweeps and the resultant impoverishment of genomic regions is a result of previous bouts of selection, on-going selection may alter gene frequencies at segregating loci and contribute to local adaptation (Günther and Coop 2013). Adaptive variation of this kind is potentially more important for adaptation to changing environment than that based on selective sweeps because populations still segregate loci that can be targeted by selection. These loci may be especially important when gene flow is low, as in the case of species found in isolated habitats. In these cases, adaptive responses to altered environment depend on local reservoirs of genetic variation at segregating loci that, nonetheless, are involved in adaptation to environment. Adaptive potential can be limited with populations lack genetic variation at loci involved in adaptation (Kelly et al. 2012). The distributions of genetic variation at segregating loci and in genomic regions that have experienced selective sweeps are both potentially important to determining geographic variation in potential evolutionary response in the face of rapid climate change. In order to model the potential for evolutionary response to changing climate, one needs to know how adaptive variation is distributed across relevant climate gradients and whether loci that are important to adaptation have heritable variation within populations, as indicated by a lack of fixation.

Own contributions -- Next generation sequencing and phylogeographic analysis

--Adaptive divergence in a South African endemic plant

We have spent parts of the last six years investigating diversification through adaptation to arid environments in multiple plant systems, at both the intra- and interspecific levels. Intraspecific studies across broad geographic areas are necessary for untangling mechanisms, such as selection by environment and interruption of gene flow, that are responsible for the diversification of lineages (Lexer et al. 2013). As part of an interdisciplinary study of factors that impact diversity in the Cape Floristic Region (CFR), we employed Next Generation Sequencing to examine intraspecific structure within a broadly distributed species in the grass-like family Restionaceae (Poales). We collected samples from 10 populations of Restio capensis that were broadly scattered across the CFR. We conducted Restriction site Associated DNA sequencing (RAD-seq) of genomic DNA of 10 individuals per population and Sanger sequencing of plastid DNA from 5-8 individuals. We used pair-wise Fst values of individual SNP loci and environmental variables to construct linear mixed models of the association between SNP polymorphism and environmental variation. We identified 156 outlier SNP loci, based on Bayesian analysis of pairwise Fst values. We also mapped plastid haplotype frequencies and conducted analysis of molecular variance (AMOVA) to identify barriers to gene flow (Lexer et al. 2014).

Population genomic analysis identified annual precipitation and temperature seasonality as major contributors to population differentiation at outlier SNP loci (Lexer et al. 2014). Variation among populations at individual loci could be attributed variously to climate differences among populations, geography and distance, and membership in different phytogeographic provinces. We detected a pattern of population differentiation along the North-South axis in both total SNP and plastid haplotypes. These geographic patterns in the distribution of genetic variation likely indicate that at least two phylogeographic lineages arose at some point in the history of this species. Differences in climate were the strongest drivers of variation at adaptive SNP loci, as detected in mixed-models of divergence at SNP outlier loci (61 of 156 loci). Climate differences also impacted among-population divergence of plastid DNA, suggesting that plastid genes also respond to environmental selection. Of the outlier SNP loci identified from RAD-seq, nearly 10% obtained annotations in BLAST2GO search, providing target genes for future research. Overall, these results indicate that differences in environment were strongly associated with population differentiation (Lexer et al. 2014).

--Adaptive divergence in Eriogonum (Polygonaceae)

Simultaneous with work on the Restionaceae, we investigated adaptation to environment in a completely different system of arid-adapted plants in the genus Eriogonum, which are broadly distributed across western North America. These plants inhabit a broad range of environments, from Pacific coastal dunes to alpine areas, and to the Mojave, Sonoran and Great Basin Deserts. We used sequences from GenBank and new Sanger sequencing of phylogenetically informative loci to obtain phylogenetic trees from a posterior Bayesian distribution. These were used in modelling the evolution of ecological traits, including the climate optima and tolerances (niche breadth) of a representative sample of 68 species, covering each of three sub-genera. We modelled trait evolution with a selection of evolutionary models, including neutral models (Brownian motion evolution) and models in which selection acts to direct and constrain trait variability (Ornstein-Uhlenbeck processes). Evolutionary rates of these traits were estimated for annual and perennial eriogonoids (Kostikova et al. 2013, Kostikova et al. 2014a).

The relevant result from this work is that while annual eriogonoids have experienced relatively rapid evolution of environmental optima, environmental tolerances have evolved more rapidly in perennial species (Kostikova et al. 2013). Perennial species, many of which are polyploidy, also have on average broader environmental tolerances than do annuals. Environmental preferences are associated with leaf and full-plant morphology (Kostikova et al. 2014a). These results suggest that environmental differences that have evolved in annual species are consolidated taxonomically at the species level. In contrast, differences in environment among populations of perennial species tend to remain within species-level designations. This suggests that perennial species may have greater intra-specific differentiation than do annuals; annuals may become reproductively isolated more quickly than lineages within perennial species (Kostikova et al. 2013). This is consistent with the large number of varieties that have been described for some perennial species, based on differences in morphology.

A. Objectives.

These results lead to the following RESEARCH QUESTIONS, the answering of which serves as our objectives:

• A1. How are adaptive and demographic processes reflected in the among-population genetic structure of species?
• A2. How are the genes that convey adaptation to environmental variation in arid environments distributed across the genome and what genes are most important?
• A3. Do selective sweeps that have arisen from selection by the environment vary in age?
• A4. How is the distribution of adaptive variation likely influenced by current and future climate change?

While patterns of genomic adaptation have been modelled in intensively studied tree species and Arabidopsis, the novelty of this work is that adaptation in sparsely distributed meta-populations of non-model organisms have never been addressed. This line of research has the possibility of integrating demographic and adaptive responses in a widely-distributed non-model organism.

B. Research Plan

 We will address the distribution of adaptive genetic variation to populations and phylogeographic lineages in a widely distributed, arid-adapted species, Eriogonum umbellatum. This species has 40 described varieties and occurs over an area three times the size of Europe in western North America. Only some of these varieties may have experience strong selection, as among sunflower ecotypes (Andrew and Rieseberg 2013). Eriogonum umbellatum presents an excellent opportunity to conduct research across the entire range of a widely distributed species because of the ease of transport by car in North America and the simple logistics of permit acquisition for collection on public lands. Our work plan involves accomplishing these fine-scale objectives in this research are:

B1. Collect leaf samples from 30 individuals in 60 populations. These populations span a maximum of 6.5 degrees of longitude (720 km) and 8.6 degrees of latitude (930 km), a distance equivalent to that between Gibraltar in southern Spain and Mont-de-Marsan in southern France. We also have collected leaf tissue in liquid nitrogen, as a basis for sequencing and assembling de novo an E. umbellatum transcriptome. Two advantages of working with this species are that we are able to (a) sample from the entire species range under the same requirements for permitting, collection and export of specimens, and (b) study a species that spans an enormous range of environmental conditions, from the rainy California coast to the arid Mojave Desert. (Fully accomplished)

B. 2. We will extract and sequence whole gDNA, a process that has begun and will continue through much of the research. An initial priority is to conduct sequencing on Illumina MiSeq in preparation for genotyping by sequencing (GBS), a Next Generation process similar to the RAD-seq used with Restionaceae. To accomplish this we have already formed a collaborative relationship with the National Center for Genomic Analysis (CNAG) in Barcelona.

B. 3. We will conduct RNAseq for subsequent exome assembly and annotation (already started).

B. 4. We will then conduct GBS on 60 populations and 10 samples per population. This will allow us to determine genotypes at thousands of SNP loci.

B. 5. We will (i.) conduct bioinformatics analysis of GBS products and (ii.) assemble the exome by aligning gDNAseq and RNAseq products.

B 6. A post-doc will develop a Bayesian algorithm to streamline the assembly of sequences from homeologous genes and gene families in this polyploid species.

B. 7. The subsequent objective will be to identify (i) selectively neutral SNPs with the program BayesEnv2 (Günther and Coop 2013) and (ii) identify evolutionary clusters of related populations with Bayesian algorithms in the software STRUCTURE.

B. 8. We will conduct phylogeographic and demographic analyses to describe the demographic history of genetic effective population size for each sampled population, allowing us to reconstruct the size, spatial structure, and diversification of the entire species through time. Neutral loci from GBS provide an estimate of population relatedness that is essential to reaching a later objective. We will then begin to develop a second Next Generation sequencing exercise.

The GBS process samples only SNP loci across about 1% of the genome. While these SNPs provide a great deal of information on demography and population history, they are of somewhat limited value in accomplishing our next objective: identify loci that are impacted through selection by the environment.

B. 9. To accomplish this, we will re-extract gDNA from the collected samples because of the greater amount required (about 3 ug from each of 60 pops x 10 samples, using a manual CTAB extraction and chloroform/phenol cleanup).

B. 10. Then in collaboration with CNAG, we will conduct Target Capture NextGen sequencing with probes (‘baits’) that are designed using the species exome (developed earlier in the research, B. 3.).

B. 11. Subsequently, we will address bioinformatics of these products, including alignment of NextGen sequence products and SNP calling.

B. 12. These SNPs together with the SNPs from GBS will us to conduct a full analysis of the relationship between genetic variation and environment through implementation of the linear models available in the software BayesEnv2. Additional analyses will identify additional selected loci, selective sweeps, and determine the extent and distribution of local adaptation in the previously identified phylogeographic lineages.

B. 13. Finally, we will conduct mapping and distribution modeling of climate change impacts on adaptive variation, using GCM climate model data and gene community algorithms (Fitzpatrick and Keller 2015). These algorithms are appropriate for modeling and projecting the potential climate envelops of suites of genomic loci, and subsequently predicting the future distribution that would confer no loss of adaptation (i.e., under unconstrained population dispersal and migration).

Risks

This cutting edge research faces several risks. De novo assembly of the transcriptome and exome of E. umbellatum may require more sequencing that we have conducted so far. If gene families turn out to be very large, hiring of the post-doctoral scholar and development of the Bayesian approach to aligning gene families may need higher prioritization. Extraction of 3ug of DNA from leaf samples of E. umbellatum is challenging and may require troubleshooting of existing extraction protocols. We need to acquire a basic level of capacity in genetic lab work because current facilities, including those of SGIker, are inadequate for working with the hazardous materials necessary for extracting DNA from wild plants (e.g., SGIker cannot perform manual CTAB extractions or chloroform-phenol sample cleanup. These standard protocols for plant population genomics work, but beyond the capacity of SGIker). This necessity is also crucial for phylogeographic aspects of L1 and research activity 4. Sufficient support to equip the Botany section of the Department of Plant Biology and Ecology with basic laboratory capacity will greatly enable us to sidestep existing limitations on laboratory capacity within the Faculty of Sciences and Technology.

C. Internationalization

This research avenue is highly collaborative, with several Spanish and international participating organizations. Two collaborating institutions are particularly important. The Rancho Santa Ana Botanic Garden in Claremont, California, USA will continue to provide substantial support for the collecting effort, including assistance in shipping cryogenic samples (for RNA), expertise in variety and species identification, and as a repository for herbarium vouchers from the collected populations. The Mojave Desert Research Center of the University of California, USA has assisted with the acquisition of samples from populations within a 100 km radius of the center. The National Center for Genomic Analysis (CNAG, Barcelona) is providing access and reduced rates of high-throughput sequencing on both Mi-seq and Hi-seq Illumina platforms. They are also providing bioinformatics and population genomics expertise. Further, we will give the following invited talk during the course of the research: Norwegian Inst. Bioeconomic Research, As, Norway (2017); University S. Hampton, UK (2018); U. Montpellier France (2018); and Univ. Bern, Switzerland (2020). We will contribute oral presentations at the biannual meetings of European Society for Evolutionary Biology in 2017 and 2019. Once preliminary results are available, we will present at the meetings of the Society for the Study of Evolution (USA; 2017-2021). We will also present at the biannual meetings of the International Biogeography Society (2018, 2020). All papers will be published in ISI-indexed journals.

D. Activities for group integration

Much of the work in this Research Avenue attempts to identify aspects of population structure. We will coordinate to develop research to identify population structure in species relevant to the conservation of biodiversity (L3) and to understand the expansion dynamic in species invasions (L2). This activity will be coordinated with Activity 4 to develop the phylogeographic component.

10. Morphological and physiological constraints on adaptive diversification.

Background--Work by other researchers.

The increasing availability of sequence data, stored in publically accessible databases, has enabled the application of comparative phylogenetic analysis to ever-larger clades of species. The availability of these data, when combined with the increasing number of algorithms for comparative analysis of ecological traits, provides great opportunity to examine the factors that impact lineage radiation and ecological diversification, the ultimate sources of biological diversity. Understanding the ways in which rates of diversification have responded to changing environmental conditions can provide an additional, longer-term perspective on the processes that create, maintain, and limit biological diversity. For example, episodes of geological diversification can provide new ecological opportunities for species ecological diversification, as has happened in the occupation of different elevation ranges by closely related species of Himalayan passerine birds (Price et al. 2014). Further, changes in atmospheric conditions have promoted the repeated evolution of adaptive metabolic pathways, such as C4 metabolism in grasses (Edwards et al. 2010) and fluctuating temperatures can lead to the evolution of generalism and preadaptation to novel climates (Ketola et al. 2013). In some cases, species may be able to radiate into new environments with little cost in response to ecological opportunity, as in Liolaemus lizards of South America (Pincheira-Donoso et al. 2015) and in grasses (Wueest et al. 2015). On the other hand variation in life history or functional traits may constrain evolution of environmental tolerances in plants (Smith and Beaulieu 2009). On the other hand, adaptation to along one environmental axis may impact evolutionary potential along another, as in Alyssum (Mengoni et al. 2003, Warwick et al. 2008, Cappa and Pilon-Smits 2014, Li et al. 2014). These examples suggest that changing environment is an important impetus for the diversification of lineages; adaptation may, however, be influence by a range of aspects of species biology. Fortunately, recent development of algorithms that can identify changing evolutionary rates along trees, either as changes in strength of selection, or as changes in rates of lineage formation (birth-death processes), can help to identify key innovations and the origin of constraints.

Own contributions--Evolutionary modelling, algorithm development and application

We have been active in the development and application of modern algorithms for examining how traits interact to affect tempo and mode of evolution during diversification of clades. Recent development of algorithms for phylogenetic comparative analysis have provided the tools to examine how selection influences evolutionary rates associate with trait states (Salamin et al. 2010). In this light, one key issue in the application of a modelling approach to determining the ecological factors that impact mechanisms evolutionary diversification has been the need to account for both the inter- and intraspecific variation simultaneously. This is because error in sampling species characteristics creates a bias that favors selection of more-complicated models, ones in which selection plays an important role (Silvestro et al. 2015). We have contributed to recent advances in the study of ecological diversification and have directly addressed this issue with the development of new functionality in an R package (Kostikova et al. 2016).

In empirical studies we find that evolution of trophic characters and diversification of environmental preferences are tightly related in fish (Litsios et al. 2012), birds (Pearman et al. 2014), and in some plants (Kostikova et al. 2014a). For example, in temperate Polygonaceae species, increased dispersal ability and broader climatic tolerance over those in tropical clades has increased net rates of diversification, primarily through dampening rates of extinction of temperate species (Kostikova et al. 2014b). Within the Polygonaceae, the genus Eriogonum has radiated extensively in western North America. Environmental tolerances of Eriogonum species are dependent on an important life history character, the perennial/annual dichotomy. Perennial species have been able to evolve broader climate tolerances, potentially because rhizomes and woody tissue provide energy storage and are advantageous in colder, high-elevation environments (Kostikova et al. 2013). This is similar to the advantages held by resprouting species over species that must reseed in some fire-prone ecosystems, such as the Restionaceae of the Cape Floristic Region of South Africa (Litsios et al. 2014b). In that system, the life history dichotomy influences the distribution of species over a broad range of environmental aridity and soil types (Wueest et al. 2016). This may also hold for Eriogonum species in North America, some of which are endemic or nearly endemic to ultramafic and serpentine soils.

A. Objectives

Our empirical results indicate that rates of diversification are impacted by adaptations that allow colonization of new environments. In particular, life history variation and habitat selection are associated with subsequent rates of lineage formation and ecological diversification.

Our previous empirical work raises central RESEARCH QUESTIONS, the answering of which serves to define our objectives:

• A. 1. Does adaptation to physiologically stressful environments, characterized by extremes of temperature, seasonality, or exposure to toxic soils, provide ecological opportunity for diversification, as in our studies of adaptive radiations in the Clownfish and Polygonaceae (Litsios et al. 2012, Kostikova et al. 2014b, Litsios et al. 2014a) or, as we find with lack of further diversification upon adoption of the reseeding strategy in species of Restionaceae (Litsios et al. 2014b), resign lineages to evolutionary dead ends?
• A. 2. Do adaptations to climate and stressful environments, such as presented by ultramafic soils, occur independently or do evolutionary tradeoffs exist?
• A. 3. How does adaptation along one resource or environmental axis affects mode and tempo of evolution along other axes?

The novelty of this work lies in identifying trade-offs among traits that effect ecological diversification.

B. Research Plan

In response to these questions, we propose to continue the research program we have developed in comparative phylogenetics. Our work plan will proceed by accomplishing the following series of fine-scale objectives:

B. 1. We will identify study clades with multiple species that occupy extreme environments, including ultramafic and serpentine soils. Our initial system will be to examine occupation of ultramafic soils in the Brassicaceae and the genus Alyssum. We will continue to search for potential study clades that inhabit a diverse range of environments, or which display substantial life history or morphological variation.

B. 2. Once identified, we will expand taxon coverage through field collecting and sequencing of useful loci. These loci will include those already represented in GenBank and new loci as necessary.

B. 3. We will develop trait matrices for clades using herbarium collections and our own fieldwork.

B. 4. We will determine vectors of plant functional traits by assembling trait and distribution data using public databases and our own herbarium visits.

B. 5. We will then determine phylogenetic relationships through Sanger sequencing of phylogenetically informative loci and, as possible, we will use Next Generation sequencing approaches to provide SNPs for construction of trees based on coalescent models.

B. 6. Subsequently, we will construct and expand phylogenetic trees using Bayesian algorithms, such as BEAST2, and produce dated chronograms when dated fossils are available for tree calibration.

B. 7. Once we have produced expanded trees, we will implement phylogenetic modelling to simulate trait evolution using neutral process models, models of diversification with selection, and models incorporating life-history variation (Ornstein-Uhlenbeck processes).

B. 8. We will analyse the stability of evolutionary diversification rates using all-birth Yule process models, Bayesian models that incorporate both speciation and extinction (BISSE models and similar) and 2-rate Brownian motion (BM) models, disparity plots of trait diversification, and randomized gamma-statistic distributions.

This research activity is, as described above, more general than the previous two activities. We will begin work on a new plant clade every two years, incorporating an additional Ph.D. student each time. This will provide a parallel research trajectory in the lab, one complementary to the other nine research avenues. Publications will be initiated simultaneously with the continuation of analyses, so as to establish and enhance the volume of our publication stream

Risks

The increasing availability of sequence data, distribution data, and trait data make this an exciting time to conduct research in comparative phylogenetics. However, this is especially the case because new analytical approaches provide an increased range of hypotheses for investigation. The field is growing rapidly, which implies increasing competition with other groups to identify useful study groups. This suggests that understudied groups, and those poorly represented in databases are the systems to be mined for opportunity. Increasingly, difficult groups, those with high levels of polyploidy or rates of hybridization, will be open for work. Phylogenetic and bioinformatics methods to deal with these complications will be well received, but empiricists may have to wait for their development. Species with high levels of secondary compounds, which complicate DNA extraction, will put a premium on our ability to modify extraction protocols ourselves, in a laboratory designed to meet the challenges presented by work on wild plants.

C. Internationalization

This research will gain international exposure by initiating the following activities: In the period 2017-2019 we anticipate conducting trips for international collection and herbarium visits, extended stays in foreign lab for Ph.D. students, Eurasian countries; 2017 invited talk Norwegian Inst. Bioeconomic Research, As, Norway; 2018 invited talk University S. Hampton, UK; 2018 invited talk U. Montpellier France; 2020 invited talk Univ. Lausanne, Switzerland; 2017, 2019 meetings of European Society for Evolutionary Biology; 2017-2021 meetings of the Society for the Study of Evolution (USA).

D. Activities for group integration

We will integrate a phylogenetic approach into the research line that addresses vegetation composition from the perspective of phylogenetic structure of plant assemblages. We can also contribute to understanding the local and regional processes that influence vegetation composition by developing phylogenies of assemblages and dominant vegetation groups. These will facilitate study of the historical, biogeographic processes that have impacted vegetation.