Patagonia case study: STM Model

By: Graciela Rusch (NINA), Veronica Rusch (INTA), Santiago Varela (INTA), and Anders L. Madsen (HUGIN)
14 September 2016

Introduction

A state-and-transition model for the Patagonia case study.

State-and-Transition models (STMs) when applied to managed systems are graphical, heuristic models that make use of ecological knowledge from diverse sources to describe quantitatively ecosystem responses to management. STMs are models of ecosystem condition that integrate use and management as ecological factors driving ecological change. We built an STM of native Nothofagus antarctica (Ñire) forest under silvopastoral use (cattle production and fuel wood extraction) in northern Patagonia, Argentina. There are indications that some levels of use are likely to be unsustainable, i.e., they can drive the silvopastoral system to a condition with low tree cover where critical ecosystem services are lost with time. Logging can exceed the capacity of the system to produce wood and cattle grazing damages tree seedlings and saplings, hindering the recruitment of new trees. Ñire trees are thought to live ca 150 years.

According to the National Law for Conservation of Native Forest, the tree cover and the ecosystem services generated by forest should be maintained in time. We have modelled the capacity to generate five ecosystem services by this system, i.e., fuel-wood production, habitat quality for native fauna, grass production, protection of soil and water and recreation quality. We model how this capacity may change with different levels of use.

By implementing the STM as a Bayesian Belief Network (BBN), we have incorporated levels of uncertainty linked our knowledge about the system.

The STM implemented as a BBN.

Interactive Front-end

Below are some HUGIN widgets for interacting with the model.

State Variables (t)

Regeneration density

Mature stem density

Mature BA

Mature stem volume

Mature stem diameter

Cane cover

Herbaceous cover

Shrub cover

Dead wood

Acaena cover

Management Actions and Time

Logging

Grazing pressure

Ñire planting

Time

State Variables (t + time)

Regeneration density_1

20.56%0
27.50%0-400
51.94%400-600

Mature stem density (1)

10.10%0
8.99%0-50
15.14%50-200
25.57%200-400
23.90%400-600
16.30%>600

Mature BA (1)

16.67%Impossible
16.67%0
16.67%1-15
16.67%15-25
16.67%25-40
16.67%40-70

Mature stem volume (1)

37.81%Impossible
15.39%0-50
12.18%50-100
16.82%100-200
15.26%200-400
2.55%>400

Mature stem diameter (1)

10.00%0-10
14.00%10-20
20.00%20-30
20.00%30-40
36.00%>40

Cane cover (1)

17.95%<1
82.05%>1

Herbaceous cover (1)

20.83%Impossible
21.25%<15
23.17%15-50
20.58%50-75
14.17%>75

Shrub cover (1)

16.67%Impossible
16.25%0-10
17.02%10-30
14.89%30-60
35.18%>60

Dead wood (1)

41.67%<1
58.33%>1

Acaena cover (1)

25.00%Impossible
37.95%Low
37.05%High

Ecosystem Services

Production fuel wood

13.30%0
20.73%0.1-1.25
18.48%1.25-2.5
27.54%2.5-5
19.95%5-7.5

Habitat quality for native fauna

20.00%No habitat
20.00%<3
20.00%3-5
20.00%5-7
20.00%>7

Production grass

20.83%Impossible
35.50%<1000
26.95%1000-2500
16.71%>2500

Soil protection & Water quality

46.87%Impossible
9.38%Very low
9.07%Low
10.80%Intermediate
23.88%High

Recreation quality

47.19%Impossible
24.07%Low
17.39%Intermediate
11.35%High

References

Rusch, V. E., A. Goijman, P. Peri, D. Lopez, L. Claps, M. Sarasola, A. Cardozo, and G. M. Rusch. 2015. State-and-Transition models: A tool to analyse decision-making and ES delivery. Fourth International Congress of Ecosystem Services in the Neotropics, Mar del Plata, Argentina.

Rusch, V. E., D. Lopez, L. Cavallero, G. M. Rusch, P. Peri, A. Cardozo, N. Hansen, A. von Muller, L. A. Garibaldi, and M. Sarasola. 2015. An ecological framework to establish management boundaries. The case of ñire forests of northern Patagonia, Argentina. VIII Congreso Internacional de Sistemas Agroforestales, Puerto Iguazu, Misiones, Argentina.

Useful references for those interested in BBN include:

Kjærulff, U. B. and Madsen, A. L. (2013) Bayesian Networks and Influence Diagrams: A Guide to Construction and Analysis. Springer, Second Edition.

Contact information

For further details on the paper: Graciela Rusch

For further details on the use of Bayesian networks and web deployment of models contact: Anders L Madsen (alm(at)hugin(dot)com)