Species parameter estimation using traits4models

Miquel De Cáceres, Rodrigo Balaguer

Ecosystem Modelling Facility, CREAF

Outline

  1. Why package traits4models?
  2. Harmonization of plant trait databases
  3. Harmonization of allometry databases
  4. Creating species parameter tables for medfate

M.C. Escher - Babel tower, 1928

1. Why package traits4models?

Purpose

  • Process-based models like medfate are challenging to parameterize for new areas!
  • Package traits4models is designed to facilitate creating plant species parameter tables for process-based models of forest function and/or dynamics.
  • Utility functions are provided to:
    1. Harmonize plant trait databases.
    2. Harmonize plant allometry databases.
    3. Populate model species parameter tables from harmonized data.

  1. Why package traits4models?

2. Harmonization of plant trait databases

Why harmonization?

  1. Species parameter values cannot be drawn from a single data source.
  2. The availability of plant trait data progressively increases, as additional efforts are made in observational or experimental studies.
  3. Multiple data sources need to be harmonized before species parameter tables are build.
  1. Harmonization of plant trait databases

Harmonization procedure

General procedure

For each trait data source, harmonization needs to be conducted in terms of:

  • Task a - Definition and nomenclature of measured traits
  • Task b - Trait measurement units
  • Task c - Taxonomy of the plant entities

How it is done in traits4models?

  • The package provides a data table called HarmonizedTraitDefinition that presents plant trait definitions and their required notation and units.
  • Tasks a and b should be performed manually using tidyverse functions.
  • Task c is conducted with the help of the taxonomic backbone provided by World Flora Online 1 and using function harmonize_taxonomy_WFO().
  • Function check_harmonized_trait() allows checking that the harmonized source is acceptable.
  • Results are stored as csv or rds (compressed) format.
  • The procedure is illustrated in vignette Trait database harmonization.
  1. Harmonization of plant trait databases

3. Harmonization of allometry databases

Harmonization procedure

General procedure

For each allometry data source, harmonization needs to be conducted in terms of:

  • Task a - Definition and nomenclature of response and predictor variables
  • Task b - Formulation of allometric equation
  • Task c - Taxonomy of the plant entities

How it is done in traits4models?

  • Tasks a and b should be performed manually using tidyverse functions but following a harmonized vocabulary to describe the equation, the response variable, the predictor variables and the parameter names.
  • Task c is conducted with the help of the taxonomic backbone provided by World Flora Online 1 and using function harmonize_taxonomy_WFO().
  • Function check_harmonized_allometry() allows checking that the harmonized source is acceptable.
  • Results are stored as csv or rds (compressed) format.
  • The procedure is illustrated in vignette Allometry database harmonization.
  1. Harmonization of allometry databases

4. Creating species parameter tables for medfate

Parameter estimation procedure

The following sections describe the main steps used to create a species parameter table for medfate and the functions in traits4models that assist the process:

# Step Function(s)
1 Initialize the parameter table with target taxonomic entities init_medfate_params()
2 Populate species parameters from forest inventory data fill_medfate_inventory_traits()
3 Populate plant allometric coefficients from suitable databases fill_medfate_allometries()
4 Populate plant functional traits from harmonized data bases fill_medfate_traits()
5 Checking the final parameter table check_medfate_params()
6 If necessary, completing strict parameters complete_medfate_strict()

The procedure is illustrated in vignette Building species parameter tables for medfate.

Warning

  • Unfortunately, this procedure is not sufficient, as it does not include metamodelling and calibration exercises that are needed for some specific parameters!
  • In the future, function check_medfate_params() should include verification of the relationships between physiological traits.
  1. Populating species parameter tables for medfate

SpParamsMED vs. traits4models

Species trait parameter table SpParamsMED was created before traits4models and is provided in medfate as default parameter table:

dim(medfate::SpParamsMED)
[1] 217 157

However, the taxonomy of SpParamsMED is a simplification of that of the Spanish National Forest Inventory (IFN).

Package traits4models provides SpParamsES which has more taxonomic entities, matching IFN taxonomy and ready to be used when parsing forest plot data with package forestables:

dim(traits4models::SpParamsES)
[1] 530 156

Data sets SpParamsES, SpParamsFR and SpParamsUS provided in traits4models follow the populating procedure explained here.

SpParamsMED is still supported, but SpParamsES, SpParamsFR and SpParamsUS are more likely to be maintained and curated in the long term.

  1. Populating species parameter tables for medfate

M.C. Escher - Babel tower, 1928