Protected-area connectivity (ProtConn)

International targets ask countries to protect a share of their land in systems of protected areas that are "well connected". The area protected is easy to measure; how well it is connected was not, until Saura and colleagues defined ProtConn, the percentage of a region that is protected and reachable from the other protected areas. In this last part of the series you use it on the landscape around Hannover: you make a network of hypothetical reserves from its largest forests, compute ProtConn and its fractions at five dispersal distances, find which reserves matter most, and test a plan that adds small stepping stones instead of large areas. At the end you can compute and report ProtConn for any region and any set of protected areas.

The idea#

Protected and connected#

In 2010 the Convention on Biological Diversity set Aichi Target 11: by 2020, at least 17% of land conserved through "ecologically representative and well-connected systems of protected areas". Its successor, the Kunming-Montreal framework of 2022, raised the area to 30% and kept the words "well-connected". Saura et al. (2017) proposed the indicator that is now used to follow that second half of the target, ProtConn, and computed it for every terrestrial ecoregion; Saura et al. (2018) did the same for every country and found large gaps, in many countries, between the area protected and the area protected and connected, and Saura et al. (2019) followed its change from 2010 to 2018.

ProtConn is the equivalent connected area of part 2, with the protected areas as the nodes, expressed as a share of the region:

ProtConn also splits into fractions that sum to 100%: ProtConn_Prot, the connectivity through protected land only (itself split into Within, inside each protected area, and Contig, between touching protected areas), ProtConn_Unprot, the part that depends on crossing unprotected land, and ProtConn_Trans, the part that depends on protected areas outside the region. A high Unprot fraction says that the network works only if the land between the reserves stays permeable: a target for land use policy, not for new reserves.

When to use it#

• Reporting on area-based targets: ProtConn is the indicator of the CBD's framework for "well-connected" systems of protected areas, computed for countries, ecoregions and provinces.
• Comparing regions with conn.protconn_regions, which gives a row per region of a layer.
• Evaluating a plan: ProtConn before and after adding (or degazetting) protected areas, and the contribution of each protected area (dProtConn), to choose where to protect next.
• Mapping it over a grid of hexagons with conn.connectivity_grid.

ProtConn is about protected areas, not habitat: a reserve counts with all its area, forest or not. For the connectivity of a habitat, use parts 1 to 4.

What you need#

• The region, as a polygon (or several, dissolved), in a projected coordinate system in metres. Here, region.geojson: the 50 × 50 km square of the series, in EPSG:3035.
• The protected areas, as polygons in the same coordinate system. Real studies use the World Database on Protected Areas or a national register. Here, to keep the example in one landscape, the reserves are hypothetical: the 18 forests of 500 ha or more, made polygons in reserves.gpkg (217 KB, 39,964 ha, 16.0% of the region). They are not the real protected areas of Lower Saxony.
• One or more median dispersal distances. Saura et al. used 10 km as the reference, with other distances to cover species that move less or more. Here 1, 2, 5, 10 and 20 km.

Enlarge

Build the model#

Making the reserves (optional)#

If your protected areas are already polygons, skip this. The reserves of the example were made from the forest map: conn.nodes gives every patch's area, conn.to_map paints the areas and the patch numbers on the patches, and raster.calc keeps the patches of 500 ha or more (None elsewhere: no value). The result is a map of reserves, which conn.protconn_raster reads directly.

format = "terraflux/1"

[[nodes]]
id = "land"
op = "raster.read"
params = { path = "clc_100m.tif" }
note = "CORINE Land Cover, cells of 100 m."

[[nodes]]
id = "graph"
op = "conn.graph_from_raster"
inputs = { map = "land.raster" }
params = { habitat = [23, 24, 25], area_unit = "ha" }

[[nodes]]
id = "patches"
op = "conn.nodes"
inputs = { map = "land.raster" }
params = { habitat = [23, 24, 25], area_unit = "ha" }
note = "Every forest patch and its area (ha)."

[[nodes]]
id = "area"
op = "conn.to_map"
inputs = { graph = "graph.graph", table = "patches.table" }
params = { column = "attribute" }

[[nodes]]
id = "patch"
op = "conn.to_map"
inputs = { graph = "graph.graph", table = "patches.table" }
params = { column = "id" }

[[nodes]]
id = "reserves"
op = "raster.calc"
inputs = { a = "area.map", p = "patch.map" }
params = { expression = "p if a >= 500 else None" }
note = "The forests of 500 ha or more, each with its number."

[[nodes]]
id = "region"
op = "raster.calc"
inputs = { l = "land.raster" }
params = { expression = "1" }
note = "The region: every cell of the map."

[[nodes]]
id = "protconn"
op = "conn.protconn_raster"
inputs = { protected = "reserves.map", region_map = "region.map" }
params = { distance = [1000, 2000, 5000, 10000, 20000], area_unit = "ha" }

The raster version is quick, but its nodes are the reserves' centroids, so the distances between reserves are measured centre to centre. ProtConn's definition uses the distance between the reserves' edges, which the vector version below computes. On this example the raster version gives ProtConn = 6.5% at 5 km, the vector version 9.3%: large reserves are far apart centre to centre and close edge to edge. Use the vector version for anything you report.

ProtConn of the protected areas#

format = "terraflux/1"

[[nodes]]
id = "reserves"
op = "vector.read"
params = { path = "reserves.gpkg" }
note = "The protected areas (polygons, EPSG:3035)."

[[nodes]]
id = "region"
op = "vector.read"
params = { path = "region.geojson" }
note = "The region (a polygon)."

[[nodes]]
id = "protconn"
op = "conn.protconn"
inputs = { region = "region.vector", protected = "reserves.vector" }
params = { distance = [1000, 2000, 5000, 10000, 20000], area_unit = "ha", delta = true }
note = "A row per distance; delta: each reserve's importance."

[[nodes]]
id = "save_delta"
op = "vector.write"
inputs = { vector = "protconn.delta" }
params = { path = "delta.geojson" }

The other settings of conn.protconn:

• transboundary: protected areas outside the region within this distance of it (or of the nodes, transboundary_type) take part as stepping stones. Here the layer has only the region's reserves, so the Trans fraction is 0.
• bound = true: the region's own parts as nodes, for ProtConn_Bound (the connectivity a region of its shape could reach at most).
• distance_type = "least-cost" with a resistance map: least-cost distances between reserves, as in part 4.

Read the results#

ProtConn at five distances#

For a species of 10 km, 72% of the protected land is connected (RelConn); for a species of 1 km, 39%: the reserves are close enough to each other for wide-ranging mammals and birds, not for most forest invertebrates or small mammals. A target of 17% "well connected" would be missed at every distance even though Prot (16%) is close to it.

Through protected or unprotected land#

At short distances ProtConn is mostly the reserves' own area: there is little movement between them. At 10 km and more, most of it depends on the unprotected land between the reserves. That land is the network's weak point: if it became harder to cross (more towns, roads, intensive crops), ProtConn would fall without a hectare of reserve lost.

Which reserves matter most#

Enlarge

The delta output ranks the reserves. At 5 km the largest southern forest (9,151 ha) carries 22.9% of ProtConn, the south-western forest of 4,599 ha 13.3%, ahead of the larger north-eastern one (7,071 ha, 10.9%), probably because it is within reach of the southern forest while the north-eastern one is farther from the other large reserves. As in part 3, importance follows area first and position second.

A plan: stepping stones or more area?#

The reserves leave a gap across the north and the centre. The second scenario protects, in addition, the 17 forests of 40 to 500 ha that part 3 found to be stepping stones (dPCconnector of 0.5% or more at 2 km): 3,133 ha more, raising Prot from 16.0% to 17.2%.

At 5 km, ProtConn rises from 9.3% to 11.2% (+1.8 points) and RelConn from 58% to 65%; at 10 km, from 11.5% to 13.5%. The stepping stones add more connected area than their own area: they bring reserves that were isolated into reach of each other. At 1 km they add only 0.4 points: for a species that moves little, small reserves far from others count only for themselves. This is the reasoning behind connectivity-aware protected-area planning: the next hectare protected is worth most where it joins what is already there.

Settings that change the answer#

• The median dispersal distance changes ProtConn more than anything else (6.2% to 13.3% here). Give it at several distances, or at the reference distance and one shorter.
• Edge or centroid distances (vector or raster versions): 9.3% against 6.5% at 5 km.
• Transboundary areas. Ignoring the protected areas just outside the region underestimates ProtConn, especially for small regions; include them with transboundary when the data allow.
• Least-cost distances lower ProtConn where the land between reserves is hard to cross, and are closer to the target's spirit (connected for species), at the price of a resistance map to justify.
• The protected areas included: strict reserves only, or all designations. Say which.

Common pitfalls#

• Overlapping protected areas. A national park that overlaps a Natura 2000 site must not count twice: conn.protconn dissolves the protected areas clipped to the region, as Makurhini does.
• Geographic coordinates. Areas and distances need a projected, equal-area system: reproject the layers first.
• A region larger than the data. Protected areas missing from the layer count as unprotected land.
• Comparing ProtConn with habitat ECA. ProtConn is about protected land, whatever its habitat; a reserve of farmland counts as much as one of forest.

Going further#

• Part 2: ECA and the probability of connectivity.
• Part 3: the importance of each patch, used here to choose the stepping stones.
• Spatial prioritization: choosing new protected areas for many features at once.
• The operators: conn.protconn, conn.protconn_raster, conn.protconn_regions, conn.connectivity_grid, conn.make_grid.

References#

• Godínez-Gómez, O., Correa-Ayram, C., Goicolea, T., & Saura, S. (2026). Makurhini: an R package for comprehensive analysis of landscape fragmentation and connectivity. Environmental Modelling & Software, 201, 106981. https://doi.org/10.1016/j.envsoft.2026.106981
• Saura, S., Bastin, L., Battistella, L., Mandrici, A., & Dubois, G. (2017). Protected areas in the world's ecoregions: how well connected are they? Ecological Indicators, 76, 144–158. https://doi.org/10.1016/j.ecolind.2016.12.047
• Saura, S., Bertzky, B., Bastin, L., Battistella, L., Mandrici, A., & Dubois, G. (2018). Protected area connectivity: shortfalls in global targets and country-level priorities. Biological Conservation, 219, 53–67. https://doi.org/10.1016/j.biocon.2017.12.020
• Saura, S., Bertzky, B., Bastin, L., Battistella, L., Mandrici, A., & Dubois, G. (2019). Global trends in protected area connectivity from 2010 to 2018. Biological Conservation, 238, 108183. https://doi.org/10.1016/j.biocon.2019.07.028