Scenario analysis

Scenario analysis in ekokrati.graph answers a counterfactual question: how would landscape connectivity change if one or more habitat patches were removed, added, or modified?

1. What a scenario is

A scenario in ekokrati.graph consists of:

1. An initial network — the baseline landscape as it exists today
2. A scenario network — the modified landscape (patches removed, added, or resized)

ekokrati.graph runs a full connectivity analysis on both networks under the same dispersal parameters and compares the results. The comparison is a static snapshot: it characterises the connectivity state of the modified landscape, not the dynamic process of how a population would respond to the change over time.

A scenario is not a simulation of population dynamics. It does not model extinction debt, time-to-colonisation, or demographic processes. For those questions, see the metapopulation documentation. What scenario analysis does model — precisely and reproducibly — is the structural connectivity consequence of a landscape change.

2. ΔEC(PC)%

The primary scenario output is the percentage change in Equivalent Connected Area:

$$ \Delta\text{EC(PC)}\% = \frac{\text{EC(PC)}_\text{scenario} - \text{EC(PC)}_\text{initial}}{\text{EC(PC)}_\text{initial}} \times 100 $$

Sign convention: negative values indicate a connectivity loss (habitat removal or degradation); positive values indicate a gain (restoration or creation).

ΔEC(PC)% is reported separately for each analysis distance, giving a dispersal-scenario profile of the impact.

Why percentage rather than absolute area? Absolute EC(PC) change depends on landscape size and cannot be compared across studies of different areas. A percentage change normalises for landscape size and is directly comparable: "a 5% connectivity loss" means the same thing in a 500 ha study area and a 50 000 ha region.

3. Patch removal scenarios

The most common scenario type in environmental assessment: a proposed development (road, housing, infrastructure) removes or fragments one or more habitat patches.

Workflow:

1. Upload the full existing habitat network as the initial network
2. Edit the scenario network to remove the affected patches
3. Run the scenario analysis
4. Review ΔEC(PC)% at ecologically relevant dispersal distances

Interpreting the result. "Removing patches X and Y causes a 3.2% loss of equivalent connected area at $d = 1\,000$ m" means the landscape effectively loses 3.2% of its functional habitat — not just the physical area of X and Y, but the connectivity those patches provided to the wider network.

The loss is often non-linear with area removed. A patch with high dPCconnector may contribute only 1% of total habitat area but 8% of total connectivity. Its removal severs paths between clusters that have no alternative route, causing a connectivity loss disproportionate to its physical size. dPC decomposition (see dPC decomposition) identifies such patches before a scenario is run.

Multiple distance profile. A development that removes a cluster of small patches close together will show a larger ΔEC(PC)% at short distances (where local connectivity is disrupted) than at long distances (where the removed patches were marginal stepping stones). Reporting ΔEC(PC)% at multiple distances gives a complete characterisation of which species groups are most affected.

4. Patch addition / restoration scenarios

The inverse case: planned habitat creation or restoration adds one or more patches to the landscape.

Typical use: ranking candidate restoration sites by their connectivity return. Given limited conservation funding, which location gives the largest ΔEC(PC)% gain per hectare of habitat created?

Workflow:

1. Upload the existing habitat network as the initial network
2. Add the candidate restoration patch(es) to the scenario network
3. Compare ΔEC(PC)% across candidate locations

A location that fills a structural gap — connecting two clusters that are currently isolated — will show a much higher ΔEC(PC)% than a location that adds habitat adjacent to already well-connected patches, even if the physical area added is identical. This is the patch-addition equivalent of dPCconnector: bridging patches are often the highest-value restoration targets.

5. Limitations

Static landscape. The model gives a connectivity snapshot, not a trajectory. A patch removal scenario shows how connectivity changes the moment the patch is lost; it does not predict how long it would take for the metapopulation to respond, or whether extinction debt would manifest within a policy-relevant timeframe.

Single-species parameterisation. Every result is conditional on the chosen dispersal parameters. A development that causes a 2% connectivity loss for a wide-ranging species may cause a 15% loss for a sedentary one under the same geometry. Always report and justify the chosen dispersal scenario, and consider running the analysis at multiple distances to characterise sensitivity.

Edge effects. Patches near the landscape boundary have fewer potential connections than centrally located patches of the same size — some of their neighbourhood falls outside the study area. The measured dPC of boundary patches may understate their true importance if the actual landscape extends beyond the study area. Where possible, include a buffer of surrounding habitat in the network to reduce boundary effects.

Patch area as a quality proxy. The model treats all hectares of habitat as equally valuable. Where habitat quality varies substantially between patches — for example, degraded intensively-managed grassland vs species-rich ancient meadow — area-based connectivity results should be supplemented with field assessment or a habitat quality layer.

6. Reporting standards

When citing scenario analysis results in publications, assessments, or reports:

Always include:

• Dispersal distance(s) used
• Dispersal probability at each distance
• Minimum edge probability (PC heuristic)
• Software version (ekokrati.graph vX.Y.Z)
• Area units (hectares or km²)

Recommended:

• ΔEC(PC)% at multiple distances, not just the single distance most favourable to the intended conclusion
• The initial EC(PC) values (absolute area, not just percentage change)
• dPC of the removed/added patches at each distance

For environmental impact assessments:

• Compare results to national or regional habitat connectivity baselines where available
• State whether the analysis covers only the directly affected patches or the broader landscape network
• Note any patches included in the network that lie outside the project boundary

Key references

• Saura, S. & Rubio, L. (2010). A common currency for the different ways in which patches and links can contribute to habitat availability and connectivity in the landscape. Ecography, 33(3), 523–537.
• Pascual-Hortal, L. & Saura, S. (2008). Integrating landscape connectivity in broad-scale forest planning through a new graph-based habitat availability methodology: application to capercaillie (Tetrao urogallus) in Catalonia (NE Spain). European Journal of Forest Research, 127(1), 23–31.
• Saura, S. et al. (2011). Network analysis to assess landscape connectivity trends: application to European forests (1990–2000). Ecological Indicators, 11(2), 407–416.