Steve J. Sinclair, Peter Griffioen, David H. Duncan, Jessica E. Millett-Riley, and Matthew D. White. 2015. Quantifying ecosystem quality by modeling multi-attribute expert opinion. Ecological Applications 25:14631477. http://dx.doi.org/10.1890/14-1485.1


Supplement

Expert opinion data and regression trees.
Ecological Archives A025-089-S1.

Copyright


Authors
File list (downloads)
Description


Author(s)

Steve J. Sinclair
Arthur Rylah Institute for Environmental Research
Victorian Government Department of Environment, Land, Water and Planning
123 Brown St, Heidelberg, Victoria, 3084, Australia
E-mail: Steve.sinclair@delwp.vic.gov.au

Peter Griffioen
Ecoinformatics Pty Ltd
20 Alexander St, Montmorency, Victoria, 3094, Australia

David H. Duncan
NERP Environmental Decisions Hub
School of Botany
University of Melbourne
Parkville, Victoria, 3010, Australia

Jessica E. Millett-Riley
Arthur Rylah Institute for Environmental Research
Victorian Government Department of Environment, Land, Water and Planning
123 Brown St, Heidelberg, Victoria, 3084, Australia
E-mail: Jessica.Millett-Riley@delwp.vic.gov.au

Matthew D. White
Arthur Rylah Institute for Environmental Research
Victorian Government Department of Environment, Land, Water and Planning
123 Brown St, Heidelberg, Victoria, 3084, Australia
E-mail: Matt.White@delwp.vic.gov.au


File list

Expert_responses.csv (MD5: 4fc2478aea08e6be9451487596050078)

CLUS_code.csv (MD5: 3d7e79e881cacde17df47743d464f50f)

regression_tree_statements.txt (MD5: 1575afcb82a01c05cfa287faba5dabc7)

Description

responses.csv

This file contains the full raw data set used in our paper, including all expert evaluation of all synthetic sites, along with the regression tree model prediction for each site. Each row represents a single expert evaluation. The variables (column names) are as follows:

SiteNumber: Identifies each synthetic site. Not unique, given each site was evaluated by multiple experts.

Function: Identifies whether the evaluation was assigned to the test data set, the training data set, or whether it was ignored (being one of the upper and lower sites evaluated by every expert as calibration, see text).

ObsNumber: Unique identifier for the evaluation.

Themeda: The basal area cover of the grass species Themeda triandra in the synthetic grassland site. Expressed as a percentage.

OtherNativeGrassCover: The projective foliage cover of all native grasses other than Themeda triandra in the synthetic grassland site. Expressed as a percentage.

NativeHerbCover: The projective foliage cover of all native herb (i.e., non-woody dicot) species in the synthetic grassland site. Expressed as a percentage.

NativeHerbDivScale: The diversity of native herbs in the synthetic grassland site. 0= none, 1= low diversity, 2= high diversity.

ExoticPerennialCover: The projective foliage cover of all exotic perennials in the synthetic grassland site. Expressed as a percentage.

ExoticAnnualCover: The projective foliage cover of all exotic perennials in the synthetic grassland site. Expressed as a percentage.

Derocked: A descriptive variable, whether the synthetic grassland site has been de-rocked or not (0 = rocks removed, 1= rocks intact).

BareGround: The percentage cover of bare ground, expressed as a percentage.

Score: The score assigned by the expert. 0-100 in 5 point increments.

Re-scaled_Score: “Score” divided by 5, to fit a 0-20 scale.

Prediction_20_scale: Score prediction from CLUS (see main text), using 0-20 scale.

Prediction_100_scale: Score prediction from CLUS (see main text) using 0-100 scale.

Expert_name: The identity of the expert who made the evaluation. Names have been converted to “Expert1-n” to protect the anonymity of the experts.

Primary_Area: The primary area of expertise for the expert, described as either “Vegetation Ecology”, “Animal Ecology” or “Land Management”.

Primary_Affil: The primary affiliation / place of work of the expert, described as

The variables listed below are all annotated “1” if the expert self-described as being “somewhat knowledgeable” in the area, or “2” if the expert self-described as being “an expert”. Experts could annotate as many columns as they wished.

The variables listed below are all annotated “1” if the expert is professionally associated with the organization (or category of organization). DEPI is the former name of DELWP, the department of the senior author.


 

CLUS settings.csv

This file contains the raw code used to run CLUS for the final regression tree ensemble (20 trees) reported in the main text.


 

regression tree statements.txt

This file contains each of the 20 regression trees (and the average / ensemble), expressed as logical (IF/THEN) statement. These statements are arranged in tab-delimited txt format, arranged in a manner that they can be opened and used in a spreadsheet which allows formulae to operate with reference to other cells. They are in a syntax suitable for use in cells in Microsoft Excel.

The trees are numbered “Tree1” to “Tree20”.

The cell references in the trees should operate when opened in a spreadsheet. They are referenced as follows: