Bestand wählen
Merken

Adding Phylogenies to QGIS and Lifemapper for Evolutionary Studies of Species Diversity

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
my name is Jeff Kavanagh and a software developer on life map a project of the Biodiversity Institute at the University of Kansas were housed in the natural history museum were bioinformatics of departments and we work a lot with species distribution modelling and all the species of modeling for a bunch of ecologist herpetologist and our soldiers to the museum In today I'd like to talk about a little bit about what life mapper is and how how we developed it to be a of distributed computational platform for doing species distribution modelling and also range and diversity modeling with multi-species models of what we've been striving to do with bringing phyla Jones's into our macroecological analysis is to bring 2 different communities together that being the phylogeographic community assembly of community with the bio geographers and what we started to do this we've extended the life map a platform to include what we call element rare range and diversity and those are analyses based on presence-absence matrices which are core data structure for macroecological research and so the software was designed to implement methodology that follows from several mathematical characteristics that are present In the presence absence matrix we also call presence-absence matrix opinion for sure so if you hear me say plan presence-absence matrix is what I mean and so so all these mathematical relationships were that developed at the University of Kansas the bison ecologists that I work with on an arbitrary which called arbor projects an avid tall NSF award for looking at at the tree of life and what this constitutes there are a group of people publishing open trees now in an open access platforms in graph databases for being able to pull down phylogenic trees 4 species of grafting trees resolving trees of In this kind of thing so we were doing the macroecological cyber decided to start looking at some this species based statistics that come out of the presence absence matrix can will start to look at them phyla genetically so we've been concerned up to this point a lot with of various all schoolmate measures of biodiversity like ad gamma alpha beta diversity species turnover and and these types of indices now we're starting to look at file genetic diversity and so what does that mean in a community you have a group of species cohabitating sharing which is we want the will start to look at how phyla genetically related or unrelated those species are within the community and so we took those up to presence-absence matrix what we calculate several different types of
statistics and so you can see really what a presence-absence matrix is is basically just a boolean matrix where you have 1 axis of the matrix constituting species for a large tax so very coarse-grained approach a resolution tax so let's say you do all mammals for the entire world or maybe 3 that the 3 thousand + species of birds in South America and so at macroecological skill we're talking about doing the geographic extent still like an entire continent and so in presence-absence matrix for South America for birds you would have 3 thousand columns 1 for each bird and then let's say you do it at 10 kilometer resolution grid so what the user does is starts out by defining a geographic area of interest and making a grid of regular polygons and that and then you intersect every 1 of those species range layers GIS layer can be vector arrested it can even be a model now show you how we use models here that in a 2nd with every 1 of those cells to provide these matrices which can become very large you know that you the 4 million elements is is not unheard of in some of these we have a 10 kilometer grid for example for North America that we've done 500 to about 800 species for a different tax end what we've done in this builder plugin for Q 2 so that exposes all this functionality and it acts as a thick line to a set of web services to the life mapper systems so what the life system is there really is a distributed code locatable and were looking at uh instances of life Mapper 2 different places like University of Florida and then it's just a distributed parallel system for district distributed computation both high-throughput and parallelisation for these types of matrix operations the it also has a computational pipeline that takes the Global Biodiversity Information Facility occurrence points for species if for every species in that it spits out models and we archive these distribution models and these models are based on climate prediction so we not only do just extinct species uh at the current time we also do for uh predictions for species will live given different climate scenarios for climate change those can be used as layers in these presence-absence matrix is so that you can get a future plans and knowledge is also currently in and so on 1 thing you end up having to do after you've intercepted the thousands of layers and that's a parallel process on the web server that you can get out of the web services the then we permute the matrix in order to be able build normal models from this boolean matrix and 1 standard ways of doing that in ecology is you have to keep the marginal totals intact while flipping bits inside of the matrix so that your row and column totals continue to add up to what they were originally and so you end up having to search the BN grids for checkerboard patterns in order to be able to swap bits bits out where be able permute the and you may end up having to permute and matrix several hundred times and so was the high throughput the parallel nature of life mapping to see where we offloaded this onto a server and then just use Q just as a client to do this we also have a python client library that basically abstracts all the details of the API away from the user having to know them and we use that in Q just state to interact with web services and you can use a Python client library individually to if you're a programmer you can just take the library program against it individual and so the other thing that life member does species distribution model which are 1 species at a time and so in the tool we have a way of being able to
pipe those archives species distribution models into the layers for opinion you can see going back to the
matrix basically how our the column totals or the range size of the species so it's it's presents across however many sites and that other marginal total then is the species richness of for each individual site in that grid and so here is an
example of climate predictions used a model species distributions and then those distributions used as inputs to presence-absence matrix and this is just a species richness map but across 3 different climate scenarios for 50 years down the road of for about 600 species of mustards the but it turns out you know
there are a number of different spatial patterns and and biodiversity statistics there ecologists are concerned with and it turns out that most of them can be derived from the presence-absence matrix with a bunch of linear algebra applied but really starts out some very basic of calculations where everything is more or less derive from a couple different covariance matrices and for different vectors of data the species richness vector and the other species range vector for example in combination In the traces that several
of the different matrices in the parents characteristics that are interesting but you can derive the old-school beta diversity species turnover gamma diversity alpha diversity in something that we call the dispersion field and the diversity fuel and so the dispersion field is just the total sum of shared range sizes and where the diversity field is the total number of shared community compositions and so it turns out with little work you can get
just to all whole smear of of different statistics concerning the Power Rangers and the diversity of the cells within the species range and so there are some correlations here but there are 2 sides the statistics or statistics there are germane to the site that being each individual cell In presence-absence matrix the or this statistics that are germane to the individual species and those are mirror images of 1 another the so the
diversity of size and ranges of species are linked by to correlations the first one being between species diversity of sites and the mean range size of species occurring in those sites and then the 2nd correlation is between the range sizes species and species diversity within those ranges and we illustrate those in Q just using what we call range diversity plots and so uh dispersion points in those spots are defined as by the degree of co-occurrence of species when it comes to the diversity feel statistics statistic and then on the dispersion of field statistics or the site based statistics the plot is really kind of showing similarities sites welcome habitats those sites are a some interesting things started
server happening in the range diversity plus we use matplotlib inside cutest duties and we've made these plots interactive so that we can meet dataspaces and brush datapoints also you bit of that but just by plotting the mean proportion range size of a species and proportional species diversity you can start to see these these patterns like the little lines of ISO covariance down along the left hand side invariably end up being on mountain tops or along the coasts are on islands and so that's and this is a arranged diversity at but have degree resolution for but 400 mammals in Africa if you brush those data points along the left hand side of the plot there those sites are right along the east coast of Madagascar but so you can see the sky geometric straight stress start to show up what this plot ends up showing is that the very characteristic pattern here were species that have a large ranges tend to live in places where there's less diversity and so as you work to the right hand of the graph you start to get out into the Sahara and if you start on the left it starts out at the coast Mediterranean coast in words works its way down and to see here where there are fewer species the but the have larger ranges by supersede now upper left-hand so the graph you can start to see where species of a live that are in sites that have extremely high diversity but where the individual ranges of those species of small so you can see in equatorial Africa sites fall long-abandoned upper upper left the so we brush those sites and
you just in Ambrose inside here for Madagascar you can see the sights light up in this spaces are linked in the maps are based on presence-absence matrix so you can map all statistics in cutest that in the coming out of presence-absence matrix through statistics panel so these are the dispersion of dispersion feels statistics website space of statistics where we were their original on similarity sites but
then the next type plot we do this for each species is a data point on the plot and so these are these this is a species association spot where the average covariance of the species are or its average association of all species determines the tendency for the species to co-occur so what you can start to see here then our uh species communities and these illustrate that the diversity feels statistics so then how do we make these basis for the diversity field when these points of species well we will look at how far genetically related species in those communities are so we've used up my boss toxicity 3 then
to bring phylogenic trees rendered as SVG on into cutis so we have a achieved urine curious that you can use and where the user is able to ingest a tree the matches the species that went into the original presence-absence matrix and you can brush points that you you think we're species species there co occurring they've they fold out an open up inside the phylogenetic viewers then we allow the user once they've selected the community be able to say on test how far genetically related the species are to 1 another using brass links based on sequence data in the file genetic data structure the who in and so this kinds of
things that were interested in our community assembly processes you know whereby how related or unrelated species are to 1 another in a community can be affected a lot by of colonization dispersion competition excetera or did they have all their together and they have similar traits and therefore they're able to exploit the same type of niche so we test against a couple different concepts and and phylogenic I Clustering whereby species are more closely related in the community and you expect them to be phylogenetically or even where that they are less less related to 1 another than you would expect expect so then once you
select that subset of a species
you can calculate that mean nearest tax on distance which is a very basic a measure of how closely related as an average all species in your community or the and so in Q just you can kind of see where you can drive you range will the same thing I that of the in the it would that the side of that because in a hostile visual but the the the there go to about that
and so I'm curious you can see a species based on diversity field the range diversity plot here and then what you can do use start to select species range diversity plot and through the magic of 43 the see them the appear in the tree of viceversa you can go down here and because the tree ends up being J. sign the and you can just do not so things like build their services for them we could find individual species or even at the genus level the so go they told me not to do like the the like that we and you have a little tree browser here the larger the names of these trees get huge army I've had the entire phylum of molluscan in here which had 85 thousand nodes you I don't know how many edges across India as a species of shrewd and you can see those see where those fall over here given you mean proportional species diversity in your proportional range size and arranged for support taken brush back and forth in both directions on the diversity field as C and then once you have a community selected the you go ahead and start calculating things like mean nearest tags on distance to see how closely related the species are based on sequence data the and that's about all i've got the questions the so I was wondering if you may have said this before but an so this has the of phylogenic trees for a and mammals and animals that is there any way to bring in plants it all user driven so you just use the plug life never put in which you can give you can use repository at and the directions are 1st complete as far as had started experiments and you got to have your species layers basically that was being said in order to make presence-absence matrix but yet I just that 1 example I have were my masters out some of the the hi of just a question about it seems like I'm using this presence absence matrices of the a sort of input for all of the functions that you due to the question I have is that of is there any conceivable way and you can definitely know when there's presence but you can notice absence absolute absences propagate uncertainty somehow in 3 D is linear algebra operations would it be some other kind of you know that you can actually do layout with continuous values the 1 we make in discrete lectures 2 0 1 0 1 what we do is we go through a thresholding process and there's a whole science behind where not what denotes a presence and an absence and so yeah but you can't do continues values if you have right now with the web services that are hooked up to this this is just basically for the Boolean version that yeah so that thresholding business is something we're working on right now in the same way that we make an archive now for all of the species and distribution models for those species for the cheapest data we intend to make a global picking you out of all models that we produce and so we will intercept always reasoning you be able to on-demand do that threshold so that the Imagen all mammals and you want just wrote denture out of out of those mammals and you only want them in this direction for South America so you'll be able to select that do that kind of thresholding on the and also the and thank you and
Matrizenrechnung
Distributionstheorie
Bit
Punkt
Gemeinsamer Speicher
Natürliche Zahl
Automatische Handlungsplanung
Gruppenkeim
Element <Mathematik>
Systemplattform
Computeranimation
Netzwerktopologie
Spannweite <Stochastik>
Informationsmodellierung
Software
Datentyp
Jensen-Maß
Indexberechnung
Datenstruktur
Softwareentwickler
ART-Netz
Einflussgröße
Analysis
Videospiel
Statistik
Matrizenring
Graph
Cybersex
Betafunktion
Datenhaltung
Softwareplattform
Elektronische Publikation
Natürliche Sprache
Arithmetisches Mittel
Mapping <Computergraphik>
Offene Menge
Speicherabzug
Projektive Ebene
Charakteristisches Polynom
Gammafunktion
Distributionstheorie
Matrizenrechnung
Programmiergerät
Bit
Punkt
Prozess <Physik>
Natürliche Zahl
Kartesische Koordinaten
Element <Mathematik>
Computerunterstütztes Verfahren
Computeranimation
Client
Prognoseverfahren
Web Services
Regulärer Graph
Mustersprache
Parallele Schnittstelle
Gerade
Bildauflösung
Lineares Funktional
Permutation
Statistik
Abstraktionsebene
Menge
Einheit <Mathematik>
Parallelrechner
Server
URL
Ordnung <Mathematik>
Instantiierung
Aggregatzustand
Subtraktion
Total <Mathematik>
Automatische Handlungsplanung
Zellularer Automat
Polygon
Gauß-Funktion
Informationsmodellierung
Datensatz
Benutzerbeteiligung
Datentyp
Programmbibliothek
Optimierung
Maßerweiterung
ART-Netz
Videospiel
Matrizenring
Physikalisches System
Vektorraum
Packprogramm
Mapping <Computergraphik>
Flächeninhalt
Boolesche Algebra
Mapping <Computergraphik>
Randverteilung
Distributionstheorie
Matrizenrechnung
Subtraktion
Web Site
Spannweite <Stochastik>
Prognoseverfahren
Ein-Ausgabe
ART-Netz
Computeranimation
Matrizenrechnung
Kovarianzfunktion
Statistik
Subtraktion
Gewichtete Summe
Total <Mathematik>
Matrizenring
Gemeinsamer Speicher
Betafunktion
Schaltnetz
Zahlenbereich
Vektorraum
Rechnen
Computeranimation
Spannweite <Stochastik>
Datenfeld
Mustersprache
Vererbungshierarchie
Jensen-Maß
Lineare Geometrie
Charakteristisches Polynom
Gammafunktion
Ablaufverfolgung
ART-Netz
Matrizenrechnung
Subtraktion
Statistik
Web Site
Punkt
Zellularer Automat
Plot <Graphische Darstellung>
Ähnlichkeitsgeometrie
Computeranimation
Arithmetisches Mittel
Spannweite <Stochastik>
Minimalgrad
Datenfeld
ART-Netz
Korrelationsfunktion
Bildgebendes Verfahren
Matrizenrechnung
Statistik
Web Site
Kovarianzfunktion
Bit
Punkt
Graph
Ähnlichkeitsgeometrie
Plot <Graphische Darstellung>
Raum-Zeit
Computeranimation
Mapping <Computergraphik>
Arithmetisches Mittel
Spannweite <Stochastik>
Minimalgrad
Verbandstheorie
Mustersprache
Server
Wort <Informatik>
Normalspannung
Geometrie
ART-Netz
Gerade
Bildauflösung
Softwaretest
Assoziativgesetz
Matrizenrechnung
Folge <Mathematik>
Statistik
Kovarianzfunktion
Punkt
Plot <Graphische Darstellung>
Elektronische Publikation
Binder <Informatik>
Viewer
Netzwerktopologie
Datenfeld
Benutzerschnittstellenverwaltungssystem
Basisvektor
Datentyp
Datenstruktur
ART-Netz
Teilmenge
Subtraktion
Datentyp
ART-Netz
Computeranimation
Distributionstheorie
Matrizenrechnung
Folge <Mathematik>
Prozess <Physik>
Browser
Versionsverwaltung
Computeranimation
Übergang
Richtung
Netzwerktopologie
Knotenmenge
Spannweite <Stochastik>
Informationsmodellierung
Web Services
Mittelwert
Lineare Geometrie
Abstand
Ganze Funktion
Analytische Fortsetzung
ART-Netz
Einflussgröße
Videospiel
Lineares Funktional
Nichtlinearer Operator
Schwellwertverfahren
Matrizenring
Plot <Graphische Darstellung>
Ein-Ausgabe
Packprogramm
Quick-Sort
Dienst <Informatik>
Datenfeld
Betrag <Mathematik>
Rechter Winkel
Ordnung <Mathematik>

Metadaten

Formale Metadaten

Titel Adding Phylogenies to QGIS and Lifemapper for Evolutionary Studies of Species Diversity
Serientitel FOSS4G 2014 Portland
Autor Cavner, Jeffery
Lizenz CC-Namensnennung 3.0 Deutschland:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/31684
Herausgeber FOSS4G, Open Source Geospatial Foundation (OSGeo)
Erscheinungsjahr 2014
Sprache Englisch
Produzent Foss4G
Open Source Geospatial Foundation (OSGeo)
Produktionsjahr 2014
Produktionsort Portland, Oregon, United States of America

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Phylogenetic data from the "Tree of Life" have explicit spatial and temporal components when paired with species distribution and ecological data for testing contributions to biological community assembly at different geographic scales of species interaction. Important questions in biology about the degree of niche suitability and whether the history of a community's assembly for an area can affect whether the species in a community are more or less phylogenetically related can be answered using several different spatially-filtered measures of phylogenetic diversity. Phylogenetic analyses which support the description of ecological processes are usually achieved in a handful of software libraries that are narrowly focused on a single set of tasks. Very few applications scale to large datasets and most do not have an explicit spatial component without relying on external visualization packages. This prompted us to explore bringing phylogenetic data into an open-source GIS environment. The Lifemapper Macroecology/Range & Diversity QGIS plug-in is a custom plug-in which we use to calculate and map biodiversity indices that describe range-diversity relationships derived from large multi-species datasets. We describe extensions to that plug-in which expand the Lifemapper set of ecological tools to link phylogenies to spatially-derived 'diversity field' statistics that describe the phylogenetic composition of natural communities.
Schlagwörter QGIS
Biodiversity
Phylogeography
Biogeography
Parallel Computing
WPS

Ähnliche Filme

Loading...
Feedback