MiRA › Tutorials › Plant–pollinator
How do I visualize a plant–pollinator network in MiRA?
Plant–pollinator networks are bipartite: links run between two groups
(pollinators and plants), never within a group. When the same interaction is recorded at
several sites or times, it becomes a bipartite multilayer network. This
walkthrough uses a bundled empirical example so you can follow along in the browser with no
installation and no code.
The dataset
The Canary Islands pollination network
(Vitali et al. 2024) has 5 layers,
235 nodes, and 651 within-layer plus 154 between-layer links. Each layer is an island;
nodes are pollinators and plants; interlayer links connect the same species appearing on
more than one island. It is undirected, bipartite, and geographically explicit.
Walkthrough
- Open MiRA, click Load Example Data, and choose
Canary Islands pollination network. It loads directly into Network Mode.
- Because the dataset declares bipartite layers, MiRA draws each layer as two rows —
pollinators on one side, plants on the other. This is the layout that makes bipartite
structure legible; a single-row layout would hide it.
- In the Nodes panel, colour nodes by
node_type to separate
pollinators from plants at a glance, and size them by degree to spot generalist species.
- Use the Intralayer Links threshold slider to hide weak interactions and
reveal the backbone of strong, frequent visits.
- Switch to Grid View to see all five islands side by side as small
multiples — ideal for comparing which islands share structure and which are distinct.
- Type a pollinator's name in Search to highlight every copy of it across
all islands at once, showing how a single species links the layers.
- Open Dashboard Mode for the bipartite set-size ratios, degree
distributions, and layer-similarity (Jaccard) heatmaps.
Why bipartite matters. MiRA is one of the few multilayer visualisation
tools with a dedicated bipartite layout. If your data are bipartite, declare it (see the
data format page) so the two node sets are laid out and
coloured correctly.
Use your own data
Export your network to MiRA's JSON or CSV format — from the
emln R package,
multilayer_to_json(net, bipartite = TRUE) does this directly — then use
Load Data instead of the example loader. Everything above applies unchanged.