Landscape Model of a WeatherDriven Process the case

Landscape Model of a Weather-Driven Process: the case of wind dispersed Conyza canadensis seed Ed Luschei University of Wisconsin – Madison Department of Agronomy

Outline n Preaching to converted: Importance – Example: GR Conyza canadensis § § § Population movement via aerial seed dispersal Biology and weather linked/correlated in several ways Implicit treatment of weather – Extrapolating phenology n Future priorities - Cooperative development – SBML standards and Weed. ML § § Why don’t standards already exist? Why doesn’t the weather “back end” exist (in the public sphere)?

Horseweed (C. canadensis) Project Glyphosate resistant Conyza canadensis, problem for no-till soybean producers n Spatial modeling at several scales n – Field scale & LDD via boundary layer escapees – Regional -- with fields as “units” – Landscape w/ counties as units n More details elsewhere… – Dauer, J. , E. Luschei, and D. Mortensen. 2009. Effects of landscape composition on spread of an herbicideresistant weed. Landscape Ecology 24: 735 -747. – Dauer, J. T. , D. A. Mortensen, E. C. Luschei, S. Isaard, E. Shields, M. Van. Gessel. 2009. Release, Escape and

Elson Shields - Cornell Aerial Sampling ~6 m Tower of Dauer Horseweed

Conyza canadensis Vertical Seed Distribution

5 Year Simulation 76 98 34 0 22

Spatial Movement (First Detection) n Data: Presence/absence by county by year n Distribution of “high quality” habitat – Pct of county area in no-till GR soybean – Map to follow [Data from CTIC] of Absent Present is logistic function of distance and habitat quality n Probability – Use scenario-based Monte-carlo

HR No-Till Soybean Acreage

Where is the Weather Part? n Implicit – Observed patterns and then modeled as diffusive process with major assumptions – Perhaps best described as a “thought experiment” on how heterogeneous land-use affects rate of spread n Why can’t we do better than that?

Phenology & Forecasting n Weedometer (weedometer. net) – including a rough spatial extrapolation via Hopkin’s “Law”

u

Multispecies “Gantt Charts”

Claim n Data-driven Bio-Clim modeling needs two things in order to work – 1. Modeling standards, transparent, modifiable, that can interface with… – 2. Publically supported data “archiving” and a delivery system

Example of Model Standards n Symbolic-Biology Markup Language (SBML) – http: //sbml. org n Weed. ML (paper forthcoming from Niels Holst, Denmark, in “Weed Science” journal) – http: //weedml. org