SEGA Downloadable Datasets

As a condition for access to data provided by researchers of the Southwest Experimental Garden Array, the user agrees to abide by the following code of ethics.

A. The user agrees to notify the SEGA scientists who gathered data prior to use in any publication or presentation. The user will provide them with formal recognition that, at the researcher's discretion, may include co-authorship or acknowledgements on publications. The level of recognition should be negotiated between the data producer and end user before manuscripts are started.

B. The user realizes that the researchers who gathered these data may be using them for scientific analyses, papers or publications that are currently planned or in preparation, and that such activities have precedence over any that the user might wish to prepare. In this case, the user should be prepared to delay publication of their research until the data producer has published their own.

C. Because it may be possible to misinterpret a data set if it is taken out of context, the user will seek the assistance and opinion of those SEGA researchers involved in the design of a study and the collection of the data as the user analyzes the data. Moreover, the user realizes that these datasets may not be complete, and it may contain errors.


Dataset Title: Fremont Cottonwood Prelim RADSeq

Status:
Public

Abstract:
Three key hypotheses will be examined: First, landscape genetic connectivity in Fremont cottonwood determines community similarity and connectivity. We will characterize the arthropod and fungal communities of Fremont cottonwood and test whether genetic connectivity in Fremont influences the structure and composition of its dependent communities. Second, genotype x environmental interactions define arthropod and fungal communities associated with Fremont cottonwood. Using a combination of experimental gardens that all contain the same cottonwood source populations and individual genotypes from the species’ range, we will quantify tree genotype x environment interactions on dependent communities. Third, climate change and exotic species invasion will negatively impact Fremont cottonwood connectivity and its associated communities. Using the community data derived from hypotheses 1 & 2, we will develop genetics-based models to better inform conservation management of foundation species, their communities and associated ecosystem processes.

Methods: To investigate landscape genetic connectivity in Fremont we will use a population genomics approach employing genome-wide restriction site-associated DNA tags (RAD tags) in combination with single nucleotide polymorphism (SNP) discovery via next generation sequencing (Baird et al. 2008, Hohenlohe et al. 2010). Using existing population-level DNA collections of Fremont, along with additional planned collections (Fig. 2b) we will generate RAD tags for two individuals per population across 50 populations (100 individuals), essentially spanning the geographic range of Fremont. RAD tag libraries will be assembled using a double digest approach (Peterson et al. 2012) and GC-rich 4bp restriction site enzymes and barcodes that differ by at least two nucleotides with subsequent modification using P2 adapters (Davey et al. 2010). Each library will be sequenced using an Illumina MiSeq platform available in the Environmental Genetics & Genomics (EnGGen) facility at NAU to generate approximately 300 million reads comprising 150 bp per read (based on a 550 Mbp haploid genome). Using Stacks open source software for sequence alignment (Catchen et al. 2011) and appropriate filtering methods (e.g., Garvin et al. 2010) to examine both nuclear and chloroplast DNA in comparison with a reference cottonwood genome (Tuskan et al. 2006) (e.g., Bowtie; Langmead et al. 2009), we will identify SNP markers to assess both pollen (nuclear SNPs) and seed (cpSNPs) dispersal across the landscape. Once reliable SNPs are identified, we will then sequence a total of 15 individuals per population for a total of 750 individuals using the Illumina MiSeq platform. Population genomic data based on SNP markers from nuclear and chloroplast DNA (Schroeder et al. 2011) will be analyzed for genetic variation and structure based on standard population genetic metrics (e.g., nucleotide diversity [p], linkage disequilibrium [LD] population differentiation [FST])

On highest output, a miseq can generate 30M reads per run, and even then you have to get a little lucky.  If by each library you mean each pool of samples (number of samples unspecified here), revise that number downward to 25M (assuming the use of 150 cycle v3 kits in 1x150 mode - non-paired-end).  If by each library you mean each sample, then if we put 16 samples per run for 7 runs for the screening dataset described above (I think these were the numbers I was told), then we are sequencing each sample to 1.5M reads (1x150 or 2x75).


Related Publications:


Cushman, S., Max, T., Meneses, N., Evans, L. M., Honchak, B., Whitham, T.G. & G. J., Allan. 2014. Landscape genetic connectivity in a riparian foundation tree is jointly driven by climate gradients and river networks. Ecological Applications 24:1000-1014
.  
Dataset Contact: Dr. Gery Allan (Gery.Allan@nau.edu)

Download Dataset: Part 1 (6.4GB compressed) Part 2 (2.2GB compressed)





Dataset Title: Fremont Cottonwood Collection Trip Notes (Collection_Trip_Notes_All)

Status: Public

Abstract:
Using common garden studies and species distribution modeling, we will evaluate the impacts of climate change and exotic species invasion on Fremont cottonwood population connectivity, associated arthropod community connectivity and ecosystem processes. We expect results from this research to provide a comprehensive view of how genetic variation and structure in a foundation tree species impacts dependent communities, and how connectivity is affected by climate (altered temperature regimes) and exotic species invasion. The following hypotheses will be tested using natural populations (Hyp1), experimental common gardens (Hyp2),and modeling (Hyp3). Hyp1) Landscape genetic connectivity in Fremont cottonwood predicts community similarity and connectivity. Hyp2) Increased mean annual temperature and tamarisk presence will synergistically reduce the capacity for Fremont cottonwood to support biotic communities and alter ecosystem processes. 3) Climate change and tamarisk are amplified agents of selection that will negatively impact Fremont cottonwood connectivity and its associated communities. As a model tree system, our findings should be widely applicable to diverse systems and have major conservation implications (Whitham et al. 2010, 2012).

Methods:  The associated data set contains initial collection information for evaluating the above hypotheses regarding how genetic variation and structure in a foundation tree species impacts dependent communities, and how connectivity is affected by climate change and exotic species invasion. Previous work has shown that Populus fremontii is broadly structured into three ecoregions. Both population genetic structure and climate are significantly differentiated among the following three regions: 1) Arizona, 2) Utah, and 3) California and Nevada (Ikeda et al. in press). Initial field surveys and collections were begun during the summer 2014 field season (May-August). An average of 10 sites were surveyed from each of the three ecoregions to account for the influence of within and across region variation. In addition to ecoregion, sampling effort was further stratified to maximize latitudinal and climatic variation across the entire Fremont distribution.

At each site, 15 trees were selected for sampling. Several sites contained fewer than 15 trees due to heavy Tamarix invasion and/or very low population size, in which case all remaining trees were sampled. These latter sites typically occurred in far southern latitudes where climatic and biotic stressors were apparent. While high mortality led to unbalanced sampling design, the surviving trees may be highly informative for furthering our understanding of climate and exotic species impacts.

Each tree was given a Sample ID, tagged, and individually GPSed with a handheld GARMIN GPS unit. Latitude and Longitude were recorded for each sample in decimal degrees, using the North American Datum 1983 coordinate system. GPS accuracy varied +/- 2 to 18 feet. For each tree, 1) 5-7 leaves were collected, dried in Dri-rite®, and stored at room temperature until DNA was extracted for genetic analyses. Next, 2) the leaf modifying arthropod community was surveyed; species and abundance data were recorded for each tree. In addition to the arthropod community, 3) we also collected 10 twigs containing three years of growth in order to assess variation in the twig endophyte community. Twig samples were frozen until further analyses were conducted. Finally, sites were revisited during January 2015 to collect cuttings for propagation while P. fremontii was dormant. Ten cuttings per tree have been propagated and are currently growing at the Northern Arizona University greenhouse in preparation for planting common gardens.

Related Publications:
.  
Dataset Contact: Helen Bothwell (Helen.Bothwell@nau.edu)

Download Dataset: excel spreadsheet





Dataset Title: Populus angustifolia Microsatellite Master Spreadsheet (NL_MSAT)

Status: Pre-publication

Abstract:
Most current species distribution models (SDMs) suffer from the lack of a genetics perspective. They operate under the simplifying assumptions that 1) there are no barriers to gene flow (e.g., species are not genetically differentiated throughout their range), and 2) all populations share the same climate niche (e.g., no local adaptation). However, we know that many species are locally adapted in response to climate-related selection pressure, and barriers to migration and gene flow are common. This study addresses two main objectives. First, 1) we describe patterns of genetic diversity and population structure across the range of Populus angustifolia, an important foundation riparian tree species growing throughout the intermountain west from Arizona to Alberta. Measuring current levels of genetic diversity will establish a current baseline from which subsequent changes in response to climate can be assessed. How genetic variation is structured across the landscape is also valuable for delineating evolutionarily significant units for land management policy and future restoration efforts. Principal components analysis (PCA) revealed that six distinct P. angustifolia genetic lineages also occupy significantly different climate space (perMANOVA: R2 = 0.78, p = 0.001). After identifying populations with unique evolutionary trajectories that are also significantly differentiated by climate, 2) we produced six regional SDMs to predict how P. angustifolia distribution is expected to shift under future climate change scenarios. By integrating genetics and SDM, this research will provide more accurate predictions to help land managers prepare for the impacts of climate change on a key habitat type throughout the west, and importantly, this evolutionary-based modeling approach can be applied to any ecosystem worldwide. 

Methods:  Leaf samples were collected from 696 trees at 34 sites spanning the full latitudinal range of Populus angustifolia. Each tree was given a Sample ID, tagged, and individually GPSed with a handheld GARMIN GPS unit. Latitude and Longitude were recorded for each sample in decimal degrees, using the North American Datum 1983 coordinate system. GPS accuracy varied +/- 2 to 18 feet. For each tree, 5-7 leaves were collected, dried in Dri-rite® (Drierite Co. LTD, Xenia, OH), and stored at room temperature until DNA was extracted for genetic analyses. Dried leaves were powderized with a Geno/Grinder 2000 (BT&C/OPS Diagnostics), and genomic DNA was extracted using Qiagen 96-well mini plant kits (Qiagen Inc., Valencia, CA). Initial DNA concentrations were quantified using a NanoDrop® ND-1000 UV-Vis Spectrophotometer (Thermo Scientific, Wilmington, DE), then DNA was standardized to 15 ng/µL. I screened seventeen microsatellite (MSAT) markers previously developed under the Populus trichocarpa Genome Project (Tuskan et al. 2004 (http://www.ornl.gov/sci/ipgc/ssr_resource.htm), Tuskan et al. 2006). Of these, I selected twelve loci based on successful amplification across all five Populus species, substantial polymorphism, consistency of scoring, and broad coverage across the genome.

Microsatellite loci were amplified via touchdown PCR in 10µL reactions containing the following master mix: c 15ng DNA template, 200nM dNTPs, 1X KAPA Taq HotStart Buffer (KAPA Biosystems (KAPA), Wilmington, MA), 0.02 U/µl Taq DNA polymerase (KAPA), 3mM MgCl2,  6% glycerol, and 55-150nM primer. Forward primers were fluorescently labeled with FAM, NED, PET, or VIC (Applied Biosystems (ABI), Foster City, CA), and different fluorescently-labeled primers were amplified in multiplex. Thermal cycling conditions consisted of 1 cycle of 95°C for 5 min; 12 cycles of 95°C for 30 sec, 63°C for 1 min (1°C decrease per cycle), and 72°C for 30 sec; 20 cycles of 95°C for 30 sec, 54°C for 1 min, and 72°C for 30 sec; 1 cycle of 72°C for 10 min; final cycle to 4°C. PCR products were run on a 3730xl Genetic Analyzer (ABI) with Genescan LIZ500 internal size standard (ABI), and allele fragment sizes were scored in GeneMarker v2.2.0 (SoftGenetics LLC, State College, PA). All alleles were manually checked for accuracy.

The associated dataset consists of Sample IDs, Population (Pop) from which each sample was collected, Latitude and Longitude (X, Y) for each sample, and allele calls for 12 nuclear microsatellite loci. In row one, the information indicates that 12 MSAT markers are displayed below, consisting of 696 individual genotypes from 34 different populations (sampling sites). The remaining columns in row one indicate how many genotypes are in each population (i.e., ML contains 8 samples, AZR 12 samples, etc.).  In row three, columns 3:26 contain allele scores for each of the 12 markers (2 columns for diploid species).  The last two columns contain the geographic coordinates for each genotype.

Related Publications:
 
Dataset Contact: Helen Bothwell (Helen.Bothwell@nau.edu)

Download Dataset:
Contact Helen Bothwell




Dataset Title:  Populus angustifolia Chloroplast Microsatellite Master Spreadsheet (cpNL_GenAlEx_19 pops)

Status: Pre-publication

Abstract:
To identify potential refugia and understand how phylogeographic history has shaped present day patterns of population genetic diversity and structure in Populus angustifolia, we analyzed chloroplast DNA (cpDNA) across the species' range. The chloroplast genome is maternally inherited in Populus species. Its non-recombinant nature and slow mutation rate make cpDNA ideal for detecting ancient differentiation and dispersal events. We assessed length polymorphism at seven microsatellite (MSAT) loci in 344 individuals from 19 sampling locations, from which we discovered 127 unique haplotypes. Results from this dataset are being compared with nuclear MSAT data from the same individuals to understand the relative roles of ancient differentiation and dispersal events compared with more recent climatic selection pressure in structuring present day patterns of genetic diversity across the landscape. Additionally, results from this study will help land managers identify areas of high conservation priority. Refugia tend to harbor high levels of genetic diversity. Due to their long-term preservation of genetic resources, these hotspots of evolutionary potential are of high conservation value, particularly in light of global climate change.

Methods: 
Leaf samples were collected from 344 trees at 19 sites spanning the full latitudinal range of Populus angustifolia. Each tree was given a Sample ID, tagged, and individually GPSed with a handheld GARMIN GPS unit. Latitude and Longitude were recorded for each sample in decimal degrees, using the North American Datum 1983 coordinate system. GPS accuracy varied +/- 2 to 18 feet. For each tree, 5-7 leaves were collected, dried in Dri-rite® (Drierite Co. LTD, Xenia, OH), and stored at room temperature until DNA was extracted for genetic analyses. Dried leaves were powderized with a Geno/Grinder 2000 (BT&C/OPS Diagnostics), and genomic DNA was extracted using Qiagen 96-well mini plant kits (Qiagen Inc., Valencia, CA). Initial DNA concentrations were quantified using a NanoDrop® ND-1000 UV-Vis Spectrophotometer (Thermo Scientific, Wilmington, DE), then DNA was standardized to 15 ng/µL. Length polymorphism was assessed at seven chloroplast microsatellite loci broadly distributed across the chloroplast genome, including three intergenic spacers - trnLUAA-trnTUGU, trnFGAA-trnLUAA, trnHGUG-psbA (Shaw et al. 2005) - and four microsatellite loci developed by Hersch-Green et al. (2014).

Microsatellite loci were amplified via touchdown PCR in 10µL reactions containing the following master mix: c 15ng DNA template, 200nM dNTPs, 1X KAPA Taq HotStart Buffer (KAPA Biosystems (KAPA), Wilmington, MA), 0.02 U/µl Taq DNA polymerase (KAPA), 3mM MgCl2,  6% glycerol, and 55-150nM primer. Forward primers were fluorescently labeled with FAM, NED, PET, or VIC (Applied Biosystems (ABI), Foster City, CA), and different fluorescently-labeled primers were amplified in multiplex. Thermal cycling conditions consisted of 1 cycle of 95°C for 5 min; 12 cycles of 95°C for 30 sec, 63°C for 1 min (1°C decrease per cycle), and 72°C for 30 sec; 20 cycles of 95°C for 30 sec, 54°C for 1 min, and 72°C for 30 sec; 1 cycle of 72°C for 10 min; final cycle to 4°C. PCR products were run on a 3730xl Genetic Analyzer (ABI) with Genescan LIZ500 internal size standard (ABI), and allele fragment sizes were scored in GeneMarker v2.2.0 (SoftGenetics LLC, State College, PA). All alleles were manually checked for accuracy. Based on analysis of nuclear microsatellite data, all clones were removed prior to analysis of the chloroplast dataset. Among 344 genotypes, we discovered 127 unique haplotypes. 

The associated dataset consists of Sample IDs, Population (Pop) from which each sample was collected, Latitude and Longitude (X, Y) for each sample, and allele calls for 7 chloroplast microsatellite loci. In row one, the information indicates that 7 MSAT markers are displayed below, consisting of 344 individual genotypes from 19 different populations (sampling sites). The remaining columns in row one indicate how many genotypes are in each population (i.e., ML contains 8 samples, ILC 14 samples, etc.).  In row three, columns 3:9 contain allele scores for each of the 7 markers.  The last two columns contain the geographic coordinates for each genotype.
 

Related Publications:

.  
Dataset Contact: Helen Bothwell (Helen.Bothwell@nau.edu)

Download Dataset:
Contact Helen Bothwell





Dataset Title:  Arthropod leaf modifying communities on Fremont cottonwood  

Status: Pre-publication

Abstract:
This dataset was collected on Fremont cottonwood throughout its range in the United States. A survey of leaf and twig modifying arthropods was conducted in the spring/summer of 2014. We found and recorded approximately 22 species of arthropods in this survey.

Methods:  All arthropods were visually censused and unknown arthropods were collected for later identification

Related Publications:
.  
Dataset Contact: Art Keith (Arthur.Keith@nau.edu)

Download Dataset: Contact Art Keith





Dataset Title:  Cibola Common Garden Plot Data

Status: Public

Abstract: In association with the Cibola National Wildlife Refuge (NWR), a cottonwood genetics research project designed by Northern Arizona University was implemented in 2006. The purpose of the study is to identify the range of genetic diversity in native Fremont Cottonwood (Populus fremontii) stands throughout the southwest and investigate the consequences of trait expression on the local community. By investigating how the gene expression of cottonwood influences community dynamics, it may be possible to select superior genotypes which best support genetic diversity for use in future restoration efforts.  Representative clones from various populations were harvested and planted in an experimental garden with replicates in other common gardens. Routine monitoring of growth, survival and gene expression between sites is ongoing.

Methods:  Cottonwood seedlings were mapped and attributed as the garden was planted.

Related Publications:

Dataset Contact: Kevin Grady (Kevin.Grady@nau.edu)

Download Dataset: Zipped ESRI Shapefile






Dataset Title:  Tree genotype mediates covariance among communities from microbes to arthropods and lichens

Status: Public

Abstract: 

  1. Community genetics studies frequently focus on individual communities associated with individual plant genotypes, but little is known about the genetically based relationships among taxonomically and spatially disparate communities. We integrate studies of diverse communities living on the same plant genotypes to understand how the ecological and evolutionary dynamics of one community may be constrained or modulated by its underlying genetic connections to another community.
  2. We use preexisting data sets collected from Populus angustifolia (narrowleaf cottonwood) growing in a common garden to test the hypothesis that the composition of pairs of distinct communities (e.g. endophytes, pathogens, lichens, arthropods, soil microbes) covary across tree genotypes, such that individual plant genotypes that support a unique composition of one community are more likely to support a unique composition of another community. We then evaluate the hypotheses that physical proximity, taxonomic similarity, time between sampling (time attenuation), and interacting foundation species within communities explain the strength of correlations.
  3. Three main results emerged. First, Mantel tests between communities revealed moderate to strong community-genetic correlations in almost half of the comparisons; correlations among phyllosphere endophyte, pathogen and arthropod communities were the most robust. Second, physical proximity determined the strength of community-genetic correlations, supporting a physical proximity hypothesis. Third, consistent with the interacting foundation species hypothesis, the most abundant species drove many of the stronger correlations. Other hypotheses were not supported.
  4. Synthesis. The field of community genetics demonstrates that the structure of communities varies among plant genotypes; our results add to this field by showing that disparate communities covary among plant genotypes. Eco-evolutionary dynamics between plants and their associated organisms may therefore be mediated by the shared connections of different communities to plant genotype, indicating that the organization of biodiversity in this system is genetically based and non-neutral.


Methods:  See notes pages in dataset and related papers.

Related Publications:

Dataset Contact:   Jamie Lamit (ljlamit@mtu.edu)

Download Dataset: Excel Spreadsheet





Dataset Title:   Overhead photographs of
experimental mesocosms from four ecosystems subjected to long-term climate
change treatments
(Images)


Status: Public

Abstract: 


Methods:  The archived photographs depict overhead views of experimental mesocosms used in the Merriam-Powell Climate Change Experiment. This experiment occurs across five life zones along an elevation gradient in northern Arizona: mixed confer forest, 2620 m above sea level (masl), mean annual temperature (MAT) of 6.6 °C, and mean annual precipitation (MAP) of 543.3 mm, Ponderosa Pine Forest (2344 masl, MAT 8.9, MAP 392.8), Piñon-Juniper Woodland (2020 masl, MAT 10.8 , MAP 272.0), and Great Basin Desert (1556m, MAT 12.8, MAP 127.3). Each site is equipped with a weather station. From the mixed conifer, ponderosa pine, piñon-juniper, and high desert grassland life zones, 40 mesocosms (30 cm diameter x 30 cm deep) were removed from grass-dominated areas in 2002. Each mesocosm was placed intact into PVC cylinders. 20 mesocosms from each life zone were transplanted down-slope to the adjacent, lower elevation, life zone to effect a warming treatment. The 20 remaining mesocosms were transplanted within the site of origin as controls. The great basin desert site served as the warming treatment from the grassland life zone. Precipitation treatments at each site were effected using passive interceptors to divert precipitation off of each individual mesocosm (30% reduction), or to divert additional precipitation onto each mesocosm from the area around it (50% increase), with ambient precipitation as a control (n=6 or 7). These treatments were selected to span the range of projected precipitation changes according to Christensen et al. (2007). The photographs archived here are overhead photographs of each mesocosm taken using a digital camera (Canon PowerShot A620, Canon U.S.A., Inc., Lake Success, NY, USA). These photographs are used to estimate early and late peak aboveground biomass. Data from outside the plots were collected to develop a relationship between per cent cover and aboveground biomass.  These datasets are available as both downloadable archives and as file directories within the iPlant Colaborative Researchers may join iPlant and then request access to the images through iPlant sharing.  iPlant includes the BisQue set of image analysis tools.

Related Publications:

Dataset Contact:    Paul Dijkstra (paul.dijkstra@nau.edu)

Download Images by Year (Gzip Tar Files): 2014, 20122011, 2010, 2009, 2008, 2007, 2006, 2005, 2004, 2003, 2002

Download Spreadsheet with Above Ground Primary Productivity  (AGPP) Excel Spreadsheet (includes metadata) This dataset is provisional, it has not undergone QA/QC or statistical analysis.  Please contact Rachel Rubin for more information on using this data.




Dataset Title:  Narrowleaf cottonwood (Populus angustifolia) genetic distance matrices

Status:  Pre-publication

Abstract:  Gene flow is a fundamental evolutionary process that underlies species' ability to adapt to changing environments. Yet, for most species, little is known about the specific factors that facilitate or inhibit dispersal through complex landscapes, and the effects these factors have on genetic diversity and differentiation. We applied a landscape genetic approach to understand how environment and climate influence the movement of genes in a foundation riparian tree (Populus angustifolia), and their relationships with species-wide patterns of genetic diversity and differentiation. Functional connectivity, or the ability of organisms or propagules to travel through a landscape, is dependent upon the quality of the space between individuals, with the intervening matrix either hindering or facilitating connectivity depending on the cost of dispersal (Spear et al. 2010). In order to test how landscape resistance influences genetic connectivity, we first parameterized models of landscape resistance in geographic information system (GIS) raster layers by assigning values to each grid cell based on hypothesized relationships between landscape variables and cost of movement (Cushman et al. 2006, Spear et al. 2010). We then used multivariate restricted optimization (Shirk et al. 2010) in a reciprocal causal modeling framework (Cushman et al. 2013) to compare matrices of genetic distance with matrices of cost distance.

Methods:  We calculated pairwise genetic distance among 696 individuals (see Populus agustifolia Microsatellite Master Spreadsheet (NL_MSAT)) using two different methods for comparison. First, we utilized the principal components analysis (PCA)-based method of Shirk et al. (2010; R Code S1). We converted microsatellite data into a matrix of individual allele frequencies (0, 0.5, or 1) using the import2genind function in the ADEGENET package (Jombart 2008, Jombart & Ahmed 2011) in R 3.1.2 (R Development Core Team 2014). We then derived eigenvectors from the allele frequency data and generated a pairwise genetic distance matrix based on distance among individuals along the first eigenvector using the Euclidean distance function in ECODIST (Goslee & Urban 2007). For comparison, we also calculated inter-individual AMOVA genetic distance (Excoffier et al. 1992, Dyer et al. 2004) using the GSTUDIO package in R (Dyer 2014). Individualistic approaches based on allele frequency distributions have been shown to provide greater power of detection and better reflect contemporary influences on genetic connectivity, compared to population- and heterozygosity-based metrics such as FST (Murphy et al. 2008, Murphy et al. 2010).

Related Publications: 

Bothwell HM, Cushman SA, Woolbright SA, Hersch-Green EI, Evans LM, Allan GJ, Whitham TG. Connectivity models predict genetic diversity of a foundation riparian tree, driven by seasonal precipitation gradients and riparian network connectivity. Submitted to Molecular Ecology.

Cushman SA, McKelvey KS, Hayden J, Schwartz MK (2006) Gene flow in complex landscapes: testing multiple hypotheses with causal modeling. American Naturalist, 168, 486–499.

Cushman SA, Wasserman TN, Landguth EL, Shirk AJ (2013) Re-evaluating causal modeling with Mantel tests in landscape genetics. Diversity, 5, 51-72.

Dyer RJ, Westfall RD, Sork VL, Smouse PE (2004) Two-generation analysis of pollen flow across a landscape V: a stepwise approach for extracting factors contributing to pollen structure. Heredity, 92, 204-211.

Dyer RJ (2014) gstudio: Analyses and functions related to the spatial analysis of genetic marker data. R package version 1.3. Available at: http://CRAN.R-project.org/package=gstudio.

Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics, 131, 479-491.

Goslee SC, Urban DL (2007) The ecodist package for dissimilarity-based analysis of ecological data. Journal of Statistical Software, 22, 1-19.

Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics, 24, 1403-1405.

Jombart T, Ahmed I (2011) adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics, 27, 3070-3071.

Murphy MA, Evans JS, Cushman S, Storfer A (2008) Evaluation of a novel approach for representing "populations" as continuous surfaces in landscape genetics. Ecography, 31, 685-697.

Murphy MA, Evans JS, Storfer A (2010) Quantifying Bufo boreas connectivity in Yellowstone National Park with landscape genetics. Ecology, 91, 252-261.

Shirk AJ, Wallin DO, Cushman SA, Rice CG, Warheit KI (2010) Inferring landscape effects on gene flow: a new model selection framework. Molecular Ecology, 19, 3603-3619.

Spear SF, Balkenhol N, Fortin MJ, McRae BH, Scribner KIM (2010) Use of resistance surfaces for landscape genetic studies: considerations for parameterization and analysis. Molecular Ecology, 19, 3576-3591.

Dataset Contact:  Helen Bothwell (Helen.Bothwell@nau.edu)

Download Datasets:  AMOVA_GD (CSV Text File), PCA_GD (CSV Text File), R Code S1_Calculating PCA-based genetic distance (Word Document)






Dataset Title:  Populus angustifolia Landscape Resistance Model (GIS raster layer)

Status:  Pre-publication

Abstract:  The ability of organisms or propagules to travel through a landscape depends on the quality of intervening space between individuals. Landscape features can either hinder or facilitate gene flow among individuals depending on the cost or relative resistance they incur on movement. We developed a landscape resistance model for narrowleaf cottonwood (Populus angustifolia) based on hypothesized cost-dispersal relationships between gene flow and key environmental predictor variables. Specifically, we hypothesized that river network connectivity, terrestrial upland resistance, and climate gradients jointly control genetic connectivity in P. angustifolia. Using multivariate restricted optimization in a reciprocal causal modeling framework, we quantified the relative contributions of each environmental predictor variable to cottonwood genetic connectivity. We found that (1) all riparian corridors, regardless of river order, equally facilitate connectivity, while terrestrial uplands provide 2.5× more resistance to gene flow than riparian corridors. (2) Cumulative differences in precipitation seasonality and precipitation of the warmest quarter are the primary climatic factors driving genetic differentiation; together these are 45× more influential than riparian corridors. This landscape resistance model was then used as input to map landscape genetic connectivity corridors across P. angustifolia's range (see Populus angustifolia Landscape Genetic Connectivity Model available for download), that if disrupted could have long-term ecological and evolutionary consequences. These data were developed with the goal of informing conservation and restoration management of threatened riparian ecosystems throughout the western U.S and the valuable biodiversity they support.

Methods:  We collected leaf material from 696 trees at 40 different sampling locations spanning the full range of P. angustifolia within the U.S. and recorded geographic coordinates for each sample. Genetic analysis of leaf material was performed with a panel of 12 microsatellite loci (see Populus angustifolia Microsatellite Master Spreadsheet available for download). To test how landscape resistance (e.g. relative permeability of landscape variables to dispersal and gene flow) influences genetic connectivity, we first calculated pairwise genetic distances using a principal components analysis (PCA)-based method (Shirk et al. 2010); for comparison, we also calculated inter-individual AMOVA genetic distance (Excoffier et al. 1992; Dyer et al. 2004). Landscape resistance models were then parameterized in geographic information system (GIS) raster layers by assigning values to each grid cell based on hypothesized relationships between environmental variables and cost of movement. Next, we calculated matrices of pairwise cumulative cost distances to travel through each of our hypothesized resistance landscapes using the distmat function in the sGD R package (Shirk & Cushman 2011; Shirk 2014). Finally, we used reciprocal causal modeling (Cushman et al. 2013) to identify which landscape resistance model best explained spatial genetic patterns in cottonwood. For each set of resistance hypotheses, we calculated a matrix of partial Mantel correlation coefficients (r; Mantel 1967; Legendre & Legendre 1998) between genetic distance (GD) and each landscape resistance hypothesis (Hypi = 1to n) while partialling out the effect of all other hypotheses (Hypj = 1to n) following the general format:  GD ~ Hypi = 1to n | Hypj = 1to n. The best-supported model was then selected by comparing relative support, or the difference between reciprocal partial Mantel tests (e.g., (GD ~ H1 | H2) – (GD ~ H2 | H1)). The best-supported, multivariate resistance hypothesis, in which river network connectivity, terrestrial upland resistance, and climate gradients have been optimized with respect to each other and cottonwood gene flow, is available for download below.

**The GIS raster layer is spatially referenced to the North American Datum 1983 and Equidistant Conic Projection, with Central Meridian set to -107.7°, Standard Parallel 1 at 33°, and Standard Parallel 2 at 45°.

Related Publications: 
Bothwell HM, Cushman SA, Woolbright SA, Hersch-Green EI, Evans LM, Allan GJ, Whitham TG. Landscape genetic model identifies connectivity corridors for conserving threatened riparian ecosystems in the American West. Molecular Ecology. (In review)

Dataset Contact:  Helen Bothwell (Helen.Bothwell@nau.edu)

Download Dataset:  Zipped ESRI Shapefile




Dataset Title:  Populus angustifolia Landscape Genetic Connectivity Model (GIS raster layer)

Status:  Pre-publication

Abstract:  Gene flow is a fundamental evolutionary process that contributes to species' capacity to adapt to changing environments. Yet, for most species, little is known about the specific environmental factors that facilitate or inhibit genetic connectivity through complex landscapes, and the effects these have on genetic diversity and differentiation. We applied a landscape genetic approach to understand how geography and climate influence gene flow in a foundation riparian tree (Populus angustifolia), and their relationships with species-wide patterns of genetic diversity and differentiation. Using multivariate restricted optimization in a reciprocal causal modeling framework, we quantified the relative contributions of riparian network connectivity, terrestrial upland resistance, and climate gradients on genetic connectivity. We found that (1) all riparian corridors, regardless of river order, equally facilitate connectivity, while terrestrial uplands provide 2.5× more resistance to gene flow than riparian corridors. (2) Cumulative differences in precipitation seasonality and precipitation of the warmest quarter are the primary climatic factors driving genetic differentiation; together these are 45× more influential than riparian corridors. (3) Genetic diversity (He) was positively correlated with connectivity (R2 = 0.3744, p = 0.0019), illustrating the utility of resistance models for identifying landscape conditions that can support a species’ ability to adapt to environmental change. (4) We present an interactive map (link to download below) highlighting key genetic connectivity corridors across P. angustifolia's range that if disrupted could have long-term ecological and evolutionary consequences. In combination, our findings provide recommendations for conservation and restoration management of threatened riparian ecosystems throughout the western U.S and the valuable biodiversity they support.

Methods:  We collected leaf material from 696 trees at 40 different sampling locations spanning the full range of P. angustifolia within the U.S. and recorded geographic coordinates for each sample. Genetic analysis of leaf material was performed with a panel of 12 microsatellite loci (see Populus angustifolia Microsatellite Master Spreadsheet available for download). To test how landscape resistance (e.g. relative permeability of landscape variables to dispersal and gene flow) influences genetic connectivity, we first calculated pairwise genetic distances using a principal components analysis (PCA)-based method (Shirk et al. 2010); for comparison, we also calculated inter-individual AMOVA genetic distance (Excoffier et al. 1992; Dyer et al. 2004). Landscape resistance models were then parameterized in geographic information system (GIS) raster layers by assigning values to each grid cell based on hypothesized relationships between environmental variables and cost of movement. Next, we calculated matrices of pairwise cumulative cost distances to travel through each of our hypothesized resistance landscapes using the distmat function in the sGD R package (Shirk & Cushman 2011; Shirk 2014). Finally, we used reciprocal causal modeling (Cushman et al. 2013) to identify which landscape resistance model best explained spatial genetic patterns in cottonwood. For each set of resistance hypotheses, we calculated a matrix of partial Mantel correlation coefficients (r; Mantel 1967; Legendre & Legendre 1998) between genetic distance (GD) and each landscape resistance hypothesis (Hypi = 1to n) while partialling out the effect of all other hypotheses (Hypj = 1to n) following the general format:  GD ~ Hypi = 1to n | Hypj = 1to n. The best-supported model was then selected by comparing relative support, or the difference between reciprocal partial Mantel tests (e.g., (GD ~ H1 | H2) – (GD ~ H2 | H1)). The best-supported, multivariate resistance hypothesis (Populus angustifolia Landscape Resistance Model available for download), in which river network connectivity, terrestrial upland resistance, and climate gradients have been optimized with respect to each other and cottonwood gene flow, was then used to construct a map of key genetic connectivity corridors across P. angustifolia's range. This map was generated with the program UNICOR (Landguth et al. 2012). Using the optimized landscape resistance model and a habitat-suitability weighted random sample of data points across our study region as input, UNICOR then calculates cumulative resistance kernel densities. Cumulative resistance kernel densities represent the expected density of dispersing individuals or propagules at any given location, weighted by the cumulative cost to traverse the modeled resistance landscape between all individuals (MacDonald et al. in review). The resulting landscape genetic connectivity map is available for download below.

**The GIS raster layer is spatially referenced to the North American Datum 1983 and Equidistant Conic Projection, with Central Meridian set to -107.7°, Standard Parallel 1 at 33°, and Standard Parallel 2 at 45°.


Related Publications:  Bothwell HM, Cushman SA, Woolbright SA, Hersch-Green EI, Evans LM, Allan GJ, Whitham TG. Landscape genetic model identifies connectivity corridors for conserving threatened riparian ecosystems in the American West. Molecular Ecology. (In review)

Dataset Contact:  Helen Bothwell (Helen.Bothwell@nau.edu)

Download Dataset:  Zipped ascii + projection file