27 Environmental impacts

One of the clear lessons from time series fitting with Ecosim is that in order to fit models convincingly to time series data, we have to consider food web impacts, environmental change as well as human impacts, (see the On modelling chapter).

The basic structure of EwE is designed so that trophic impacts (be they direct or indirect) and direct human impacts (notably through exploitation) are dealt with through the food web and exploitation parts as detailed in the baseline Ecopath and Ecosim chapters of this text book. Indirect food web effects of a behavioural character, such as when one group impacts the feeding interactions between two other groups can be handled with mediation – as described in more detail in the Mediation chapter.  This leaves us with describing how environmental impacts are represented in EwE, and that’s the topic of this chapter.

As a starting point, environmental impacts are dynamic factors, i.e. they change over time, and they impact organisms in different ways. This in essence means that there are two steps to be considered for inclusion of environmental impacts: how they change and how they impact.

The first factor, how an environmental factor changes over time, is modelled with forcing functions. In Ecosim, these are temporal while in Ecospace they have to be both temporal and spatial, so reading in a spatial map for each time step. Other than that aspect, Ecosim and Ecospace work the same way with incorporating environmental impacts.

The second factor, how to model the direct impact of the forcing function on functional groups is done with environmental preference functions, The details of this are described in the Habitat Capacity chapter to which we refer.  That chapter describes the spatial implementation of habitat capacity, and this functionality was indeed developed for Ecospace initially, but it is now also implemented in Ecosim with the same approach and functionality as described in the chapter.

For details of how to use the forcing functions and define the habitat capacity preference functions, see the tutorial on Incorporating environmental forcing (available online and in pdf version only).

Inner workings in Ecosim

The description below of how forcing functions are considered in Ecosim is also applicable for mediation functions.

The basic Ecosim prediction for consumption or flow rate (unit: biomass/time, e.g., t km^-2 year^-1) of type i prey biomass to type j predator is of the functional form

[latex]\text{flow rate} = a_{ij} V_{ij} B_j \tag{1}[/latex]

where a_ij is a “rate of effective search” parameter, V_ij is vulnerable prey biomass, and B_j is effective predator abundance^[1]. If vulnerable prey were randomly distributed over the modelled area, and V, B_j were expressed per unit area, then a_ij can be interpreted as a volume (or area) swept per unit predator abundance per unit time, corrected for the proportion of time actually spent searching for food^[2].

To understand how effects of habitat changes may impact trophic flow rates, consider the a_ij parameter. For most trophic interactions, predators search for prey only over restricted spatial foraging arenas, and hence V_ij is distributed only over such areas rather than at random over the whole system.

Suppose the (practically unmeasurable) restricted area where foraging by j on prey i takes place is A_ij per unit total model area. Suppose that while in this area, each unit of predator abundance searches an effective area a_ij* for prey. On average, each such area searched should result in capture of V_ij/A_ij prey, since this ratio is prey density in the arena area. In other words, the flow rate could be modelled more precisely (if we could measure A_ij) as

[latex]\text{flow rate}=a_{ij}^{*} A_{ij} V_{ij} B_j \tag{2}[/latex]

i.e., the a_ij in Eq. 1 Equation  can be interpreted as a_ij =  a_ij*/A_ij. Expressed this way, we see that time forcing^[3] can influence the flow rate in at least three quite distinct ways:

• by altering the effective search rate a_ij* of the predator, for example by using a turbidity time forcing function^[4] of algal biomass that reduces a_ij* at high algal biomass.
• by altering the area A_ij over which vulnerable prey and/or predators are distributed^[5],
• by altering the vulnerability exchange rates v_ij that determine (along with a_ij*/ A_ij) V_ij from total prey biomass B_i^[6].^[7]

It is possible to apply multiple time forcing functions^[8] to each trophic flow (i,j) rate prediction, and to specify whether each function multiplies a_ij*, A_ij, and/or v_ij. Using these forms, one can choose the parameter that is multiplied by each forcing function^[9], i.e. one of the following choices:

• Multiply overall predator rate of effective search (a_i,j), for example to represent time-varying turbidity changes that affect predator search efficiency^[10].
• Multiply vulnerability exchange rate (v_ij), for example to represent increased movement rates of prey into vulnerable behavioural state at times when water mixing rates are higher.
• Multiply area of foraging arenas (divide A_ij by multiplier), for example to represent increase in habitat area available for juvenile fish refuges.
• Multiply area (divide A_ij) and also multiply v_ij, for example to represent increase in safe foraging habitat available to a predator that feeds on prey that become available in foraging arenas through passive drift/mixing processes such that increasing area used by predator results in higher proportion of total prey population being available in foraging areas at any moment.

How will you decide which of these options to use? Consider the description above or explore the impacts of the alternative settings.