D.E. Hines - PowerPoint Presentation

Slide1

D.E. Hines

enaR

an package

for Ecological Network Analysis

Acknowledgements

Core Sound, Fall Closed

Oyster Reef

Lake Lanier

U.

Scharler

& B.D.

Fath

J.

Mejaski

Stuart R.

Borrett

University of North Carolina Wilmington

Matt Lau

Northern Arizona University

Shaoqing

ChenSlide2

General Approach

http://affordable-

tack.com

Nitrogen

Ecological Network Analysis

Wind in the Willows

Patterns reflect

development constraints

and

system function

Form



FunctionSlide3

Ecological Network Analysis (ENA)

Economic

Input–Output Analysis

(Leontief)

Networks of energy–matter

flows and storagestrace a single conservative currency

(e.g. nitrogen, carbon)usually a

ssume steady-state

“

Macroscope” to investigate organization &

connectivitydirect & indirect interactions

Resilience & sustainabilitySlide4

ENA Applications: Examples

Alligators enhance frog populations

- Indirect mutualisms

1999

Trophic Ecology

2012

Strength of coupling between BGC processes

Nitrification + Denitrification

Nitrification + Anammox

Biogeochemistry

2012

Urban metabolism of Vienna

Carbon perspective

Energy production, Construction key

Urban Metabolism/SustainabilitySlide5

Two Schools of ENA

Ulanowicz School

Patten School

Environment Focus

Photo courtesy of C Thomas

Trophic Focus

Information Theory

Photo from http://

www.cbl.umces.edu

/~

ulan/

University of Maryland

University of G

eorgia

Network Environ Analysis

Distinct but interwoven development of ENA

Ascendency Concept

Environ Concept

Phenomenological GraphDiff EqsSlide6

ENA Software

Tools for Science & Innovation

NETWRK 4.2

Fortran Code, DOS executable

Ulanowicz & Kay 1991Environmental Softwarehttp://www.cbl.umces.edu

/~ulan/ntwk/network.html

WAND

Allesina &

Bondavalli 2004Environ. Model. Softw.

Excel based, more user friendly

Ulanowicz School

Patten School

Ecopath

Christensen &

Pauly 1992

Focus on model construction

NEA.mFath & Borrett 2006Environ. Model. Softw

.

Matlab Function Collected NEA algorithmshttp//people.uncw.edu/borretts/research.html

EcoNetKazanci 2007Ecol. Mod.C++, Web Server

Particle Tracking Approachhttp://

eco.engr.uga.edu

No software covers both schools

Different

availability

,

use

,

extensibilitySlide7

Connect

Extensible

Software Design Objectives

Collect

Available

Tools for Science & Innovation

Algorithms from both

Ulanowicz

and

Patten

ENA Schools

Increase availability

Low cost software

Run on multiple OS (

PC, Mac, Unix)Users can access original code

add new functionalityAdditional network science tools

Social network analysisERGMSlide8

Connect

Extensible

enaR

package for

Collect

Available

Tools for Science & Innovation

Algorithms

Patten

structure, flow, storage,

environs,

control, utility

Ulanowicz

ascendency, mixed trophic impact, balance

R: a statistical programming environment

W

ide useFree, Open sourcePC, Mac, UnixCRAN

Existing Packages

network (Butts, 2008)sna (Butts, 2008)statnet (Handcock, 2008)iGraph

bioconductorIllustrate some ena package functionalitySlide9

Data Input: Model Data

Node Names

F: Flows from

i

to

j

Slide10

Data Input: Model Data

Node Names

F: Flows from

i

to

jSlide11

Data Input: Model Data

Node Names

F: Flows from

i

to

j

z: Inputs

e: Exports

r: Respirations

y: Outputs (e + r)

Living: logical (T or F) Data stored as a network

object

pack() and unpack()

(Butts 2008)Slide12

Data Input: read.xr

ead.scor()

read.wand()

Plain text file

NETWRK 4.2 - SCOR

WAND

CollectSlide13

Visualization> data(oyster)

> plot(oyster)

Simple Plot

Fancier Plot

Internal Fluxes

ConnectSlide14

Flow Analysis Example

Column-to-Row OrientationSlide15

Simultaneously Analyze Multiple Models

> ns =

lapply(model.list,

get.ns)> barplot(ns$id)> barplot

(ns$ASC.CAP)m

odel.list = list of Network Model data objects

get.ns() =

ena function to calculate network statistics

Low n

High n

50 trophic ecosystem models

(Salas & Borrett 2011)

Extend

Indirect / Direct

Ascendency / CapacitySlide16

sna: Centrality

Connect

Centrality – relative importance of node, many types

Core Sound, NC ecosystem Model (

Deheer

, 2012)

Centrality Calculations

degree()

betweenness()

gplot.target()Slide17

Future WorkAdd Additional AlgorithmsLindeman’s Spine

Cycle BasisSchramki’s control analysis revisionsConnection to Ecopath

Shiela Heymans, Villy Chistensen

What will you do with the package?How will you extend it?

Collect

Connect

ConnectSlide18

Connect

Extensible

Summary

Collect

Available

enaR

: package for Ecological Network Analysis

Ulanowicz

Patten

Free

PC, Mac, Unix

CRAN

Open Source

Users can grow

Existing Packages

n

etwork

snabioconductor

borretts@uncw.eduhttp://people.uncw.edu/borretts/