Computer modelling ecosystem processes and change

Lesson 8Presentation 1

Suggested reading Harris, M. 1998. Lament for an

ocean. The collapse of the Atlantic cod fishery. McClelland and Stewart Inc., Toronto.

Chapter 5 documents the errors made in the inventory and calculation of cod populations

Why computers Ecosystems are complex Can vary the attributes and

processes in complex ways with computers

To manage our activities Assess management actions across many

spatial scales Predict long term effects Understand management on biodiversity Predict influences on specific components

of ecosystem (e.g. climate, biological legacies)

Predict population dynamics of wide range of species

Compare to natural change

Complex task

What types of models Many different types of models to project

population trends, demographics, service & product supply, nutrient and energy fluxes in space and time

Models used to understand interactions (e.g. researchers to explain results)

Models used to predict future condition (e.g. used by managers to predict future condition of resources)

Three types of models: Deterministic (single outcome) Probabilistic (chance) Process (based on ecosystem

process)

Deterministic Most common Based on rules that show how some

change occurs E.g. tree growth based on data from

permanent plots. We know how the dia. and ht changed over time. We use this info to grow a tree of the same species by changing dia and ht. with time.

Probabilistic Second most common Uses probabilities of chance events

to show changes E.g. probability of fire in a forest

landscape, what proportion of the forest will burn in a given time period

Process Least common Requires a lot of data and

information to develop Uses ecological processes to show

how attributes in an ecosystem will change.

E.g. uses sunlight period, nutrient & moisture flux to grow a plant.

Models today are a hybrid of these 3 types

Basic components of models

Attributes of interest

Logic how attributes change

Change in attributes of interest

Inventory or start population

Projected inventory or population

Limits: Data used

Is the inventory/population correct Garbage in = garbage out

Basic rule but often overlooked E.g

cod stocks: population estimates were wrong

Forest inventories in Ontario: no requirement to quantifying accuracy

How to overcome limit Monitor and update inventory

Recognize limit of monitoring or inventory method,

develop verification methods E.g. creel and index netting for fish

population Aerial photos and ground surveys for

forest inventory

Limits: Logic used Incorrect rules result in incorrect

outputs E.g.

incorrect growth rate of trees will result in incorrect volume growth of timber

Incorrect growth rate of prey population will result in incorrect growth rate for population of predators

How to overcome limit Know what the precision levels are

for the rules used Use sensitivity analyses to

determine how they affect outputs

Sensitivity analyses Finding out how sensitive the outputs

are to variation in the rule set Small changes in some rules may

result in huge changes in outputs. E.g.

succession rules in how forest stand composition changes make a large difference in timber volume for each tree species

What about processes we do not know? We will never know everything about

ecosystems Need validation studies

Does the real changes on time match the computer output

Need studies in ecosystem processes to discover key factors currently not known

Use Adaptive Management

Cod stocks

Adaptive management Based on social science and

ecosystem knowledge Implementing policy or practice as

experiments Involves systematic monitoring to

detect surprise (what we do not know)

Integrated assessment to develop knowledge system

Contrast Adaptive Management

Most policy uses casual observation no system to capture new information from current practice and use it.

Conventional experimentation results in new theory but applicability may be narrow

Trial & error: may result in new knowledge but hit and miss

Adaptive management cycle

New knowledge

Apply operationally as experiment

Hypotheses

Monitor results

Questions