Spatio-temporal dynamic modeling of plant communities responses to hydrological pressures in a semiarid Mediterranean wetland

J. Martínez-López    J. Martínez-Fernández    B. Naimi    M.F. Carreño    M.A. Esteve
1Ecology and Hydrology Department - University of Murcia (Murcia, Spain)
2Applied Biology Dept. University Miguel Hernandez (Elche, Spain)
3ITC - University of Twente (Enschede, The Netherlands)
[Uncaptioned image]

1 Introduction

Study area

[Uncaptioned image]

Marina del Carmoli wetland (300 ha)

[Uncaptioned image]

Wetland plant communities

Semiarid Mediterranean saline wetlands are semi-terrestrial ecosystems

[Uncaptioned image]
  • Salt steppe (left) - priority habitat by the Habitats Directive

    Salt marsh (center) - habitat of interest by the HD

    Reed beds (right) (Phragmites australis) - invasive

External water inputs

Percentage of irrigated areas has increased in the last decades due to the opening of a water transfer (Martínez-López et al., 2013)

[Uncaptioned image]

Plant communities change

Important plant communities are being lost!

[Uncaptioned image]

Carreño et al., 2008; Martínez-López et al., 2012

Objective

  • Spatially explicit wetland model of how irrigated agriculture is affecting plant community composition in this semiarid Mediterranean wetland

2 Methods

Modelling environment

  • R as a modelling environment:

    • GIS capabilities

      source code is flexible

      free availabity and growing user community

[Uncaptioned image]

State variables

  • Wetland is divided into pixels (25 m)

    Plant communities are modelled separately pixel by pixel (4 maps)

    The total abundance of plant communities within a pixel is limited so:

    • competition among plant communities mediated by

      • total drainge water input to the wetland

        spatial environmental variables influencing water availability and growth

      the dispersion of other PC from the surrounding pixels

[Uncaptioned image]

Initial and validation maps of plant communities

Model was tested by means of remote sensing data for the period 1992-2008

[Uncaptioned image]

Carreño et al., 2008; Martínez-López et al., 2012

Model assumptions I

  • Increasing water input

    Only conversion to more humid / less saline plant communities

[Uncaptioned image][Uncaptioned image]

Model assumptions II

native vs. invasive taxa
  • invasive reed beds are potentially present in all pixels

    salt marsh is able to disperse into neighbour pixels

[Uncaptioned image][Uncaptioned image]

Non spatial forcing input

Drainage water input

WARP index (Martínez-López et al., 2014a,b)

[Uncaptioned image]

Wetland environmental spatial parameters

  • (A) distance map to ephemeral river 1 (reed beds)

    (B) distance map to ephemeral river 2 (reed beds)

    (C) Flow accumulation map (salt marsh)

[Uncaptioned image]
  • All parameters are on a relative 0–1 scale.

Model diagram

[Uncaptioned image]

Model development/execution

  1. 1.

    Initial dynamic model was developed using Stella (1 pixel)

  2. 2.

    Conversion to R using ’StellaR’ script (Naimi and Voinov, 2012)

  3. 3.

    State variables and spatial environmental variables as matrices

  4. 4.

    Model wrapped as a R function

  5. 5.

    ode.2D("euler" method, time = 24 year, TS = 0.25) (library "deSolve")

3 Conclusions

Conclusions

  1. 1.

    The model serves as a tool for

    • 2.

      wetland conservation and management studies (habitat loss)

    • 3.

      testing plant community interactions

    • 4.

      testing relationships between plant communities and environmental variables in space and time

  2. 5.

    The library undergoes further developments in order to become a flexible tool for the development of new spatio-dynamic models