1........... Introduction to WG 2
2.2.......... Key physical and hydrological parameters
considered in habitat models
2.3........ Ecological Significance of key
attributes
3........... Interfaces with Ecology
3.1........ Habitat preference criteria
3.1.1..... Univariate Preference functions
3.1.2..... Multivariate statistical preference
functions
3.1.3..... Fuzzy-rule based preference
functions
3.2........ Biological process models
4.2........ State-of-the-art within COST-626
4.3........ Systematic overview
1
Introduction to
WG 2
2
Physics
2.1
Introduction
This overview focuses
on those stream models addressed to characterize the stream habitat. The
expected output of this type of models can vary from being purely descriptive
of the stream physical template to having some biological assessment
applications. Physical descriptive models are developed to evaluate the degree
of alteration of a given stream channel in relation to some reference
conditions. Biologically based models are developed to infer the standing stock
of a given species from the physical characteristics of a given stream.
Nevertheless, in between these two extremes there is a range of habitat models
addressed to obtain other outputs as shown in Figure 1.
Fig. 2‑1: Conceptual framework
of physical habitat models
Based on this
conceptual framework, habitat models (especially those having biological
implications) are mostly based on hydrological, morphological and hydraulic
parameters as the major factors influencing distribution and abundance of
organisms in the streams (see Figure 2).
Fig. 2‑1: Main factors affecting distribution and
abundance of organisms in running waters
2.2
Key physical and hydrological parameters
considered in habitat models
Habitat models can be
developed at different spatial scales for different predictive purposes.
Parameters used to describe hydrologic, morphologic and hydraulic aspects of
the streams vary according to the selected scale used for the model. This fact
is especially relevant to be considered in the development of the model to
simplify the array of parameters and select those which are more suitable for the scale at which the model will
operate. Table 1 summarizes some of the key parameters that should be/are
considered at different scales to model the physical and hydrological stream
template.
Table 2‑1: Key morphological
and hydrological parameters considered in habitat models at different scales
|
|
Parameters |
|||
|
Scope and scale |
Morphologic |
Hydraulic |
Hydrologic |
|
|
“Pico”-habitat ~ cm |
“Nose position” of fish |
§
Substrate size, type, shape §
Substrate “quality” for biological purposes §
Motion/no motion §
k/d (roughness / depth) |
§
Shear stresses §
Laminar/turbulent near-bed boundary layer §
Local flow velocity (nose) |
§
Baseflow Q §
Maximum peak flow and duration §
Drought events |
|
Micro- habitat ~ m |
Section |
§
Substrate size/type distribution §
Substrate stability §
Local elevation along cross-section
(geometry) §
Roughness §
Sediment porosity §
Bathymetry §
Roughness r (height of protruding rock) §
Embeddedness §
Porosity §
armour layer §
particle shape §
Wentworth scale (1..15), dominant/subdominant §
Macrophytes §
Overhanging braches §
Cover (Rocks) §
Percentage of fines |
§
Wetted perimeter (water width and depths) §
Local velocities §
Vertical hydraulic gradient §
Water transient storage zone §
Surface-subsurface lateral linkages §
Cover (pools) §
“Broken” water §
Turbulences §
Splashwater |
§
Temporal variation of discharge: daily,
seasonal, interannual §
Flood and drought regime: frequency,
magnitude, evenness |
|
Meso-habitat ~100 m |
Reach |
§
Topology §
Run/riffle/pool distribution §
Cross-section profiles §
Valley floor: constrained vs unconstrained §
Channel stability §
Bank stability §
Plan shape: meander vs braided §
Description of morph. Patterns by shape and
property §
Sinuosity §
Width/depth ratio §
Width/max. depth ratio §
Periphyton |
§
Mean cross-sectional velocity, water depths §
Spatial variance of velocity,
shear stress, depth |
§
Mean annual flow §
Average duration of the floods and droughts §
Spatial variation of discharge |
|
Macro- habitat ~1000 m |
Catchment |
§
Drainage area: stream length
ratio §
Frequency distribution of different stream
orders §
Branching degree and distribution §
Longitudinal gradient §
Presence of barriers §
Land-use activity §
Number of pools/100m |
§
Mean water residence time §
Channel vs uphill position of water table
(gaining or losing stream) |
§
Longitudinal variation of
cumulative water yield §
Seasonal variability in runoff §
Surface or subsurface runoff §
Flow continuity |
|
Ecoregion/Landscape |
§
|
§
|
§
|
|
|
Temporal Scale |
§
Disturbance frequency §
Disturbance duration (draughts,
suspended sediments e.g.) |
|||
|
Networking aspects |
§
Characteristic patch diversity §
Residual pool depths §
Availability and location of
refugia (from different threads) |
|||
Which of these can
be appropriate for upscaling?
Other factors that
could be considered in the model and that are not so scale-dependent are:
·
Temperature
·
Light
availability
·
Water quality
(oxygen, pH, conductivity, toxic substances, nutrient content, etc.)
·
Amount and type
of suspended particles
·
Food resources
availability
Fig. 2‑2: Relative importance of the habitat
template to limiting factors in relation to water quality
These factors may
override the importance of the stream physical template under certain
conditions; and thus, they can be considered as “limiting factors” for the
habitat models when used to predict habitat suitability for biological standing
stocks. For instance, the relative importance of the
habitat template for predicting fish abundance can vary as a function of the
water quality (see Figure 3).
2.3
Ecological
Significance of key attributes
To understand why
certain attributes or parameters are looked at within habitat models it is in
some cases important to understand their secondary effects, that is how they
act. For some attributes it is simply known from empirical experience that they
influence habitat quality but it is not well understood why and in which
manner.
An example for this is
the hydrologic regime, which consists of the mean annual flow with temporal
dynamics on top of that, the seasonality and the random characteristics of the
discharge. It is common belief that it is important but no quantitative data
can be found to specify this more clearly. One of the ways discharge dynamics
influence habitats is that they determine when sediment transport occurs and
which portion of the river bed is affected from sediment motion over a certain
period. The regime determines how often and how long particles of a certain
size move and how often and when incipient motion state is exceeded. For some
bottom dwelling species this could also be expressed as disturbance frequency.
Should there follow
a list of such examples?
3
Interfaces with
Ecology
Habitat models usually
consist of a physical part that analyses hydraulic and/or morphologic
attributes. Often these are considered as or linked with time series (e.g. of
the discharge). The result of this part is the description of the physical
environment that is available. Quality parameters or other information may be
added.
In a second step these
attributes are linked with or compared with what is called here “interfaces
with ecology” which describe how these physical attributes correspond with the
preference or the abundance (relative or absolute) of a certain specie. The
result of this part is “habitat quality” which can be expressed in different
terms. The traditional description (PHABSIM approach) for habitat quality is
Weighted Usable Area (WUA) or Suitability Index (SI). Often different life
stages (Spawning, larvae, juveniles and adult) and different seasons
(summer/winter) are treated separately. Additionally, certain “activities” can
be a criterion, such as feeding, resting, seeking shelter (“rufuge”), rearing
(salmonid fry still carrying their yolksacs that hide in the gravel), etc.
Table 3‑1: Source of
biological data used in habitat modelling
|
Correlation or process described |
Numerical interface |
Derivation from |
Outputs from model |
|
§
‘Association’ functions; resource functions
(This is data that records an organisms occurrence or ‘association’ with a
resource or physical variable) |
§
Habitat Suitability Indices (HSI’s) (Use or
Preferences) §
Regressionary models (univariate,
multivariate, direct gradient analysis, fuzzy logic and artificial
intelligence) |
§
Expert opinion §
Field measurement §
Biological knowledge |
§
Quantitative and qualitative measure of
habitat quality for organism (usually used as a surrogate for population
level) §
Index of habitat quality §
Probability of organism occurring |
|
§
Biological processes |
§
Physiology (growth, digestion, accumulation
of energy) §
Foraging behaviour §
Life-history strategies |
§
Experiment §
Observation |
§
Spatially explicit measure of habitat quality |
The results of this
type of approach usually is a pure prediction of habitat quality which is as a
consequence of the modeling approach not linked with the population dynamics.
However, often the results of these models are interpreted as a prediction of
future species abundance.
The other type of
interface between biological and physical attributes or processes are
describing the growth of an individual animal or a certain group or species
living under certain environmental conditions. This can include a large number
of individual processes that are each controlled by environmental conditions,
such as feeding, digestion, energy gain and consumption, reproduction. It also
can be a simple description of the dynamics of a certain species’ population
under certain environmental conditions that are based on empirical data and
integrate a large number of individual biological processes without
understanding the mechanics of these. This part of biological modeling can be
built upon the results of a pure physical habitat modeling approach or be
independent from that.
The following chapters
will describe both approaches in more detail.
3.1
Habitat
preference criteria
The list of physical,
chemical and biological variables that are related to an organisms presence or
probability of occurrence is enormous but can be summarized as in table …..
Table 3‑2: Attributes usually used for the description
of probability of use or occurrence in habitat models.
|
Aspect |
Attribute |
|
Micro-habitat |
Depth Velocity Substrate Threshold habitat size |
|
Meso-habitat |
Factors associated with channel
shape and slope (see Piotr) |
|
Macro-habitat / Catchment |
Riparian use Altitude Latitude Land –use Disturbance |
|
Ecoregion/Landscape |
Yann ? |
|
Chemical |
O2, toxicity |
|
Biological |
Traits, reproductive strategies |
Mechanisms for dealing
with the presentation of output vary according to the ecological reasoning behind
the modeling. Thus it can include the
· Aggregation of data into one index
· Amount of habitat above a certain
threshold
· Time series analysis
· Spatial distribution of the habitat
quality
Validation of habitat
models is problematic as there are many underlying assumptions which need to be
tested and the output is often in units which are difficult to measure
directly. Further, validation is often
seen as unnecessary and is rarely funded.
Most often attempts to validate habitat models relate the habitat
quality to population numbers.
(give a crisp definition, figure, how these
interfaces look and describe how you get them from field data)
3.1.1
Univariate
Preference functions
Univariate preference
functions are found by ……
3.1.2
Multivariate
statistical preference functions
(Piotr)
3.1.3
Fuzzy-rule based
preference functions
3.2
Biological
process models
Biological process
models are models that describe processes such as the dynamics of a population
of a given specie under certain environmental conditions. A very simple model
could be the growth of algae biomass in
a lake under certain trophic conditions and light and temperature, based on
empirical data. A more complex model would take individual processes within the
species metabolism into account, such as feeding, digestion, reproduction,
mortality etc. under given environmental conditions. These individual processes
can be either mechanistic and based on physical processes or they can be
empirical functions. A complex model can incorporate a number of individual
processes and combine these to a life cycle model or a multi species community
model.
These models can
either be linked upon the results of a plain physical habitat model or be
directly linked with certain data describing the physical and physiographic environment.
Biological process
models usually have predictive capacities regarding the development of
populations or species’ communities. What are the relationships between Habitat
and Production?
Who is
completing this
4
Models
4.1
Introduction
A model can be
anything from a very simple mathematical equation to very complex systems of
algorithms incorporating a large number of individual processes and
procedures. Models can be divided
roughly into several categories, however distiction may not always be very
precise because complex models consist of a large number of distinct submodels
which usually belong into different categories.
Table 4‑1: State-of-the-art
models
|
|
Statistical,
Stochastic |
Mechanistic |
|
Biological |
§
Suitability
index (BHABIM in CASIMIR) §
Multivariate
statistics §
Fuzzy-logic-suitability
(FHABIM in CASIMIR, HARPHA, HABSCORE) §
Time series
analysis §
Neural network §
Qn-type
models (flow duration curve based) |
§
PHABSIM type
models (EVHA, RHEHAMSIM, CASIMIR, FISO...) §
Bioenergetic §
Mult-Agent
(Moby Dick) §
Energy and
substance models §
Growth-temperature
based models §
5M7 §
Population
dynamics |
|
Hydraulic |
§
Frequency
distribution based models (FSTRESS Lamouroux, TAUSIM in CASIMIR §
Neural networks |
§
1,2,3-dim
steady state §
Time series of
steady state conditions §
1,2,3-dim
unsteady state (SSIM, MIKE, DELFT3D) §
Solute
transport models (konvection, advection, dispersion and diffusion) + reaction
kinetics |
|
Water quality |
§
Data processing
techniques |
§
> 600
processes & substances §
Dissolved
Oxygen §
Temperature §
Light §
Nutrients
(PO4, NO3, NH4) §
Conductivity §
Acidity §
Carbonates §
BOD, COD §
Algae §
Toxics §
Sediment
interface §
Passive/active
(salinity or temperature changes physical properties of water) |
|
Hydrology |
§
Statistical
hydrology §
Stochastic hydrology |
§
Precipitation-runoff
models §
Distributed
parameters §
Concentrated
parameters §
Point vs.
non-point sources (sediment and nutrients) |
|
Morphodynamic |
§
Classification
& description based on data o
Shear power o
Bed forms o
Plan forms o
Shields number |
§
Stable
channel conditions / Incipient motion §
Duration of
motion related to partikel size §
Loose boundary
hydraulics, bedforms §
Suspended/bedload
transport models §
Single vs.
Multi-fraction models §
Armouring layer
and sublayers §
Sediment
tansport in multiple layers §
Morphodynamic
models |
|
Spatial analysis |
§
GIS based
technologies §
Landscape
Ecology ·
Contagion ·
Juxtaposition ·
Interspersion ·
Patch size ·
Minimum
distance, area ·
Spatial
analysis technique (integration of output from other models on a higher
level) ·
·
…. |
|
4.2
State-of-the-art within
COST-626
The models used within
the COST-626 network are not a complete list of available models but represent
only the ones in use by members of this group. Some of the models described are
elementary modules that are components of more complex models or
toolboxes.
See appendix 1
4.3
Systematic
overview
Brainstorming:
Different categories
in terms of
Approach
Scale
Target species
Userfriendliness
Data requirements
Scale-aspect
Ecoregion
Published, validated
Make this a
systematic overview
In a given
situation, a certain question asked: with a given amount of time/money and a
certain background experience, which is the choice between models you have.
5
Expected Results
To be discussed with
end-user group.
Traditional fish
habitat models take only the physical habitat into account. It is well known
that population dynamics do in many cases not follow habitat availability
because other factors might be limiting. There is a number of possible reasons
for that:
Species interaction is
neglected
Food availability is
neglected
…
Mortality rates at
different life stages are not well known and mechanisms are not well
understood. What are physical habitat criteria to increase survival rates e.g.
during the winter or in the transition phase from yolk-sac to fry (starvation
problem)
….
Therefore it is clear
that habitat models can only provide a limited perspective of the reasons for
success or failure of a species community.
Aspects
(Brainstorming):
- Predicted physical habitat for certain
taxa/life stages
- Fish per usable area expected
- Predicted and measured populations
dynamics
- Communities or taxa
- Results are scale dependent
- The scope is a function of the scale
6
Upscaling
Some major aspects:
Fig. 6‑1: Relevant aspects for floodplain ecological processes
(Wentworth 2001)
Using a representative reach of a river
(quantitative proof for this assumption?) for a modeling program and then
transferring the results to other reaches of the river is not upscaling but
simply working with random samples representing the entity.
Different models are
applicable at certain scales only. A model must be able to represent its
appropriate scale. The results of such models can be incorporated in larger
scale management tools. This means that complex models can be used to directly
generate generic information to be used within management + decision making
tools to be given to basin managers or decision makers. This is important but
not really upscaling.
Physical upscaling
means to use results gained from a model with a certain (spatial) resolution on
a certain scale, e.g. a river reach with a given length, and generate results
that are applicable and valid to an area with a wider scale, e.g. a longer
river reach or a higher organizational level (floodplain, other river
reaches). Does
anyone do that?
Multi-scale models
include some principles from large scale models (e.g. temperature variability)
into habitat models. Consider connectivity over space and time between small
scale habitat units.
Is a direct upscaling
needed or not
- Which are relevant parameters for an
upscaling process (quantitative links across scales)
- Modeling approaches used for upscaling
- Is downscaling ever required?
|
Physical habitat modeling (small
scale) aspect |
Large scale aspect |
|
|
7
Validation
Quantitative
validation needs much biological data