• 1 Introduction
• 2 Detailed Description of the Functions’ In- and Outputs
  • 2.1 Initialization and Input
  • 2.2 Solver and List with results
  • 2.3 Post-Processing and additional Functions
• 3 Example Application
• 4 Batch Application and Sensitivity Analysis
  • 4.1 Solving for Multiple Uncertainty Values
  • 4.2 Selecting Specific Indicator Bundles - Investigating the Indicator’s Sensitivities
    • 4.2.1 Socio-economic
    • 4.2.2 Ecological
    • 4.2.3 Immediate Economic Success
  • 4.3 Pay-Off Matrix
  • 4.4 The Use of fixDistance
• 5 Suggested citation
• 6 Literature

1 Introduction

This README is closely related to a publication in Methods in Ecology and Evolution (Husmann et al. 2022). When managing ecosystems, the question arises as to how to simultaneously combat biodiversity loss and maintain ecosystem functioning while improving human welfare. Designing multifunctional landscapes means deciding upon the composition and configuration of land-cover types, given that landscape patterns drive the landscape`s ecological value and its ability to satisfy societal demands. Land-cover allocation models, particularly optimization approaches, have proven helpful in revealing trade-offs between multiple objectives and services provided by different land-cover compositions.

The R package optimLanduse provides functions for easy and systematic applications of the robust multiobjective land-cover composition optimization approach of Knoke et al. (2016). It includes tools to determine the land-cover composition that best balances the multiple functions and services a landscape can provide, as well as tools for a deeper understanding and visualization of the contributions of the distinct indicators. The method has been developed and applied previously in a couple of studies, with some examples to be found in the Literature section. The methodological background of the approach can, e.g., be found in Knoke et al. (2016) and Husmann et al. (2022). We here refer to the nomenclature of Husmann et al. (2022). The package opens the approach of Knoke et al. (2016) to the community of landscape planners and provides opportunities for systematic or batch applications. To further enhance this, we have designed a shiny dashboard for the package to get a brief idea of its functionalities and to facilitate the usage of the package, see http://rshiny.gwdg.de/apps/optimLanduse/. The current package version optimizes land-cover compositions, but aspects of configuration may be added in the future.

2 Detailed Description of the Functions’ In- and Outputs

This chapter provides a brief overview of the package functions (Fig. 1). For detailed information on methodological background, functions, and workflow please also refer to Husmann et al. (2022). We further refer the reader to the respective help pages of the package for more information.

The stable version of the package can be installed using the CRAN server. The development version can be found on the GitHub project page.

# If not already installed
#install.packages("optimLanduse")

Fig. 1: Overview of the functions of the optimLanduse package. Green diamonds: input and output data; blue rectangles: functions; gray parallelograms: optional function settings.

2.1 Initialization and Input

The initScenario() function combines the user settings with the data into an optimLanduse-object ready for solving. The following input data are required:

• coefTable: The package is only capable of processing a long-oriented type of data structure (Table 1). All combinations of land-cover (landUse) alternatives and indicators must be listed vertically. Each row must contain the average expectation, the uncertainty, and the direction (more is better/ less is better) of the respective land-cover and indicator combination. The column names of the table must exactly follow the nomenclature displayed below. You can also find this format in the built-in example tables exampleGosling.xlsx or exampleEmpty.xlsx. The help files of the exampleData() and initScenario() functions contain more details regarding the required data format. An empty template incl. the predefined headings can be accessed via exampleData(“exampleEmpty.xlsx”). All further columns in the coefTable will be dropped.

Table 1: Example of the data set from Gosling et al. (2020) to illustrate the required data structure.

• uValue: The argument for the uncertainty level (