Pressure Impact Factors

Pressure-impact factors

Darwin produces two kinds of metrics: biodiversity pressure metrics (physical exchanges from product LCI) and biodiversity impact metrics (LCIA-characterized damages). Pressure metrics are methodology-agnostic; impact metrics depend on the LCIA model used.

Pressure-impact factors are drawn directly from third-party LCIA models (ReCiPe 2016, IMPACT World+ v2.1, GLOBIO 4 / IBIF v2). Darwin does not develop or adapt these factors itself: the platform's contribution is the structuring of input pressure data and output impact indicators, and the integration of LCIA-specific conventions into a unified pipeline.

Darwin supports six LCIA models — three standard frameworks and three Darwin-custom variants. See LCIA Models for the full catalog, output units and selection criteria.

Inputs

Pressure-impact factors take physical pressure data as input, in one of two forms:

• Physical flows — elementary exchanges from LCI databases used directly as pressure quantities (e.g., land use in m2.yr, water consumption in m3.yr).
• Aggregated midpoint indicators — pressures aggregated across multiple elementary flows into a single indicator, for example:
  • Climate change: total GHG emissions expressed in kg CO2-equivalent (kg CO2eq)
  • Ecotoxicity: toxic emissions aggregated as Toxic Unit equivalents for ecosystems (TUe)

Outputs

Pressure-impact factors translate these inputs into biodiversity impact indicators expressed in the aggregated metrics described above (PDF.m2.yr, species.yr, MSA.km2.yr). Each LCIA model has its native unit; Darwin offers custom variants to allow cross-model comparison where useful (e.g. ReCiPe PDF unit-converted from species.yr).

Focus: structuring input data for ecosystem use

The ecosystem use pressure is covered through physical flows. Structuring this input requires mapping LCI ecosystem use types to the relevant impact factor categories.

• Ecosystem use types are based on LCA databases LCIs.
• 54 ecosystem use types are managed covering the 3 biomes.
• 3 additional proprietary labels are added to allow extra accounting options such as contribution to artificialization or contribution to natural ecosystem conversion for instance.

Full list of ecosystem use types and labels

Ecosystem use type	Biome	Natural	Artificial	Ecosystem use category
lake, artificial	Freshwater	No	No	Lake
inland waterbody, unspecified	Freshwater	Yes	No	River
river, artificial	Freshwater	No	No	River
river, natural (non-use)	Freshwater	Yes	No	River
wetland, inland (non-use)	Freshwater	Yes	No	Wetland
seabed, drilling and mining	Marine	No	Yes	Industrial
seabed, infrastructure	Marine	No	Yes	Industrial
seabed, unspecified	Marine	Unknown	No	Sea
annual crop	Terrestrial	No	No	Crop
annual crop, greenhouse	Terrestrial	No	No	Crop
annual crop, irrigated	Terrestrial	No	No	Crop
annual crop, irrigated, extensive	Terrestrial	No	No	Crop
annual crop, irrigated, intensive	Terrestrial	No	No	Crop
annual crop, non-irrigated	Terrestrial	No	No	Crop
annual crop, non-irrigated, extensive	Terrestrial	No	No	Crop
annual crop, non-irrigated, intensive	Terrestrial	No	No	Crop
arable land, unspecified use	Terrestrial	No	No	Crop
cropland fallow (non-use)	Terrestrial	Yes	No	Crop
heterogeneous, agricultural	Terrestrial	No	No	Crop
permanent crop	Terrestrial	No	No	Crop
permanent crop, irrigated	Terrestrial	No	No	Crop
permanent crop, irrigated, intensive	Terrestrial	No	No	Crop
permanent crop, non-irrigated	Terrestrial	No	No	Crop
permanent crop, non-irrigated, extensive	Terrestrial	No	No	Crop
permanent crop, non-irrigated, intensive	Terrestrial	No	No	Crop
bare area (non-use)	Terrestrial	Yes	No	Desert
forest, extensive	Terrestrial	No	No	Forest
forest, intensive	Terrestrial	No	No	Forest
forest, primary (non-use)	Terrestrial	Yes	No	Forest
forest, secondary (non-use)	Terrestrial	Yes	No	Forest
forest, unspecified	Terrestrial	Unknown	No	Forest
grassland, natural (non-use)	Terrestrial	Yes	No	Grassland
grassland, natural, for livestock grazing	Terrestrial	Yes	No	Grassland
shrub land, sclerophyllous	Terrestrial	Yes	No	Grassland
construction site	Terrestrial	No	Yes	Industrial
dump site	Terrestrial	No	Yes	Industrial
dump site, inert material landfill	Terrestrial	No	Yes	Industrial
dump site, residual material landfill	Terrestrial	No	Yes	Industrial
dump site, sanitary landfill	Terrestrial	No	Yes	Industrial
dump site, slag compartment	Terrestrial	No	Yes	Industrial
industrial area	Terrestrial	No	Yes	Industrial
mineral extraction site	Terrestrial	No	Yes	Industrial
pasture, man made	Terrestrial	No	No	Pasture
pasture, man made, extensive	Terrestrial	No	No	Pasture
pasture, man made, intensive	Terrestrial	No	No	Pasture
traffic area, rail network	Terrestrial	No	Yes	Transportation
traffic area, rail/road embankment	Terrestrial	No	Yes	Transportation
traffic area, road network	Terrestrial	No	Yes	Transportation
unspecified	Terrestrial	Unknown	Unknown	Unknown
unspecified, natural (non-use)	Terrestrial	Yes	No	Unknown
urban, continuously built	Terrestrial	No	Yes	Urban
urban, discontinuously built	Terrestrial	No	Yes	Urban
urban, green area	Terrestrial	No	No	Urban
urban/industrial fallow (non-use)	Terrestrial	No	Yes	Urban

Conversion pressure rule (naturalness criterion)

The "natural" status of each ecosystem use type drives the conversion pressure indicator. A land-use change counts as a conversion, a restoration, or is ignored, depending on the natural status of the source and destination cover:

This rule is independent of the LCIA model: it alone determines each site's surface contribution to the pressure metric (loss, restoration or ignored). "Unknown" covers are resolved conservatively (worst case): an unknown source is counted as a loss unless it is offset by a confirmed natural gain, while an unknown destination is never credited as a gain. This avoids any spurious restoration credit.

Each LCIA model then applies its own characterization factors on top of these raw conversion and occupation pressures. Models differ in whether they stay coherent with the pressure rule:

The coherent models (IW+ LU, ReCiPe PDF) follow an SBTN-aligned logic: conversion only counts when a natural or semi-natural ecosystem is affected. The other models account for a continuous biodiversity gradient across all cover types, so a transition between two non-natural covers (e.g. extensive → intensive farming, cropland → urban) produces a non-zero impact even though the pressure rule ignores it — creating a visible divergence between a null pressure and a non-zero impact on IW+- and GLOBIO-based assessments.

Worked example — IMPACT World+ land-use bias (sunflower vs palm oil)

Converting natural forest can score lower than cultivating already-farmed land, because IW+ assigns a characterization factor of 0 to natural ecosystems as a conversion source. Comparing 1 tonne of sunflower oil (France) with 1 tonne of palm oil (Malaysia):

Despite palm oil driving primary-forest deforestation, its modelled impact is far lower — the forest source carries a zero CF and only the cropland destination is scored. This bias toward permanent crops is why Darwin reports the conversion pressure indicator (which flags the 24 m2 of natural conversion) alongside the LCIA impact.

Water use

Water pressure volumes are methodology-agnostic — identical across all LCIA models — while the biodiversity damage they translate into depends on the LCIA method.

Pressure (water volumes) follows the WFN / ISO 14046 categories:

• Blue water — freshwater withdrawn from a watershed and not returned (split into surface and ground).
• Green water — rainwater evapotranspired by plants.
• Grey water — volume needed to dilute pollutants to regulatory thresholds (not yet implemented — roadmap).

Damage varies by LCIA model:

• IMPACT World+ is pressure-damage coherent by construction (damage = characterization factor × net consumption).
• ReCiPe scores only the evapotranspiration fraction of water use, so its water-use damage is lower than IW+ for industrial uses (cooling, washing) and converges with IW+ for crops.
• GLOBIO (strict) does not cover water; GLOBIO Extended embeds the IW+ water characterization factors to recover coverage.

Volume alone does not explain impact: water results should be read together with the method choice and regional scarcity.

Alien Invasive Species (AIS) Impact Factors

Scientific basis

Alien invasive species constitute one of the five main IPBES pressures on biodiversity, yet their impacts had not previously been integrated into standard LCIA methodologies. Darwin's AIS model is based on the work of Borgelt et al. (2024), which developed country-to-country characterization factors (CFs) quantifying the time-integrated potentially disappeared fraction of native terrestrial species (PDF.yr) per kilogram of goods transported between two countries.

These CFs are derived from global datasets combining:

• First records of alien species introduction
• Native species distribution and conservation status
• Bilateral trade flows (1870-2019)

These characterization factors vary across several orders of magnitude, reflecting the strong spatial dependence of impact per trade route.

Key findings from the study confirm that importing goods from certain countries (e.g. China, South Africa, Madagascar into France) can generate higher terrestrial biodiversity impacts from alien species introductions than from the climate change associated with transport emissions — underscoring the importance of this pressure category.

Methodological choices

Based on expert consultation, the following choices were made:

• Average characterization factors are used (rather than marginal factors).
• Regional factors are preferred over global ones, given significant variability between the two approaches.

Metric: PDF.yr — an isolated indicator

The AIS model outputs results in PDF.yr (Potentially Disappeared Fraction of species x year). This metric is not compatible with PDF.m2.yr (used in Impact World+) or species.yr (used in ReCiPe), and no known conversion factors exist between them. CIRAIG experts confirmed this non-fungibility.