Parameters

This page describes every parameter that controls a ekokrati.graph network analysis. All parameters are set in the submission form or, for scripted access, in a parameters.json file submitted alongside the input files.


Dispersal parameters

Distances

Type: list of positive integers (metres) Example: 300, 1000, 3000

The dispersal distances to analyse. ekokrati.graph runs one full connectivity analysis per distance and returns all results together.

Each distance represents a separate dispersal scenario — a species for which the probability of dispersing that far equals the Dispersal probability value below. Comparing EC(PC) across distances gives a connectivity profile across a spectrum of dispersal abilities, from sedentary to wide-ranging species.

Choosing distances: start with the range that brackets the focal species or species guild. Distances below the median nearest-patch distance will produce sparse, fragmented networks; distances above the 90th percentile will produce nearly complete graphs. The pre-submission edge histogram shows the distribution of pairwise distances in your patch layer and helps calibrate this choice.

Dispersal probability

Type: float in (0, 1) Default: 0.5 Field label: Probability of dispersal

The probability that an individual disperses the stated Distance. This value, together with the distance, determines the dispersal kernel: a species that has a 50% probability of crossing 1 000 m has θ = −ln(0.5)/1000 ≈ 0.000693 m⁻¹.

The same probability value applies to all distances in the list. Because θ is derived per distance from (prob, d), each distance represents a different species scenario — see Dispersal kernel.

Default guidance: 0.5 (the threshold is the median dispersal distance) is the conventional choice following the Conefor standard and makes dPC values comparable across studies that use the same convention.


Graph construction parameters

Minimum edge probability

Type: float in (0, 1) Default: 0.001 Field label: Min. edge probability

Edges (potential dispersal connections) with a dispersal probability below this threshold are excluded from the graph before the analysis runs. This is a computational cutoff — it reduces graph size and speeds up analysis without affecting ecologically meaningful results, since connections with p < 0.001 have negligible influence on dPC.

This is not the same as Dispersal probability. Dispersal probability parameterises the kernel (what species are we modelling); minimum edge probability prunes the graph (which connections are negligible). See the Progressive refinement workflow for how to use both.

Adjusting: increase to 0.01–0.05 for a faster first run on large networks (100 000+ edges). Decrease below 0.001 only if your landscape is very sparsely connected and you are concerned about missing long-distance stepping-stone routes.


Input format

Distance method

Type: enum — edge or centroid Default: edge

How pairwise distances between patches are measured.

Method Description When to use
edge Nearest point on each patch boundary Preferred — ecologically correct for most species
centroid Patch centroid to centroid Faster; appropriate for small patches relative to dispersal distance; matches Conefor default

For most landscapes, edge and centroid produce similar rankings but different absolute dPC values. If replicating Conefor results use centroid.


Analysis scope

Include all metrics

Type: boolean Default: true

When true, ekokrati.graph computes the full decomposition: dPC, dPCflux, dPCconnector, dPCintra, dIIC, EC(PC), and betweenness centrality (dBC_PC).

When false, only EC(PC) and dPC (without decomposition) are computed — roughly 50% faster. Use for exploratory runs on large networks.


Scenario parameters

These parameters apply only to scenario analyses (Edit tab → Run as scenario, or /jobs/from-layer). They are not present in the standard submission form.

Edge overrides

Type: object with remove list and optional force_include list

Manual edge overrides applied on top of the computed graph:

  • remove — centroid-index pairs to exclude regardless of dispersal probability
  • force_include — centroid-index pairs to include regardless of min_edge_prob

These are set interactively in the Edit tab and do not need to be specified manually in normal use.