Written by Adam E. Naas, Eva Lieungh, Amanda M.A. Folland
Ecological Distance (ED), as estimated in this repository, is an approximation of the difference between two mapping units in terms of species composition, abiotic composition, and formative process. Its purpose is to give more information about differences where two assessments of the same area are not equal. Ecological Distance is given as numeric values between each mapping unit (ecosystem types) in 1:5000 and 1:20 000 scales in symmetrical matrices with mapping unit codes as row- and column names.
An ED of 0 means that two mapping unit assignments are identical. Where they are different, ED is given as an integer number. There is no predefined maximum ED, but a maximal ED is where the compared mapping units have (almost) no species in common, vastly different substrates, and/or different governing processes. Note that the estimated ED should be interpreted with caution because it is an approximation of difference and therefore often, if not always, inaccurate. Note also that the method for calculating ED for NiN version 2 and 3 is quite different, and the resulting ED is not directly comparable even between types that are unchanged between the two versions.
This repository hosts scripts to calculate ED from an input dataset with columns for short codes for major-type groups, major-types, and minor types, ecological space, 1:5000 units, 1:20 000 units, and 1:50 000 units. The output is a matrix of ED between mapping units, under ../output
. To use ED in analyses of e.g. rasterised maps, make pairwise comparisons using the mapping unit from each map as a 'coordinate' in the matrix to find pixel-wise ED between two maps. An example script to get the ED in a point dataset is provided under ../script
.
The NiN ecosystem typology (Halvorsen et al. 2020) is hierarchical, with three nested levels (major-type groups, major types, and minor types). In addition, mapping units are defined by aggregating minor types (within, but not across, major types) to be suitable for mapping at scales 1:500, 1:5000, 1:20 000 and 1:50 000. Ecological distance was first operationalised for NiN version 2 in Eriksen et al. (2018), and scripted and revised in Naas et al. (2023).
NiN version 3 includes eight major-type groups (MTGs): terrestrial non-wetland (T), wetland (V), snow and ice systems (I), limnic (L), marine (M), and river (O) bottom systems, and limnic (F), and marine (S) water body systems. This repository so far only covers ED for the terrestrial systems (T, V, and I) thoroughly. Marine, limnic and river bottom (L, O, and M) systems are also included, but criteria to define the ED between them properly are not considered. Updating the ED for these major-type groups will require some extra data curation and decisions about ecological distance.
Within major-type groups (e.g. terrestrial and wetland ecosystems), major types (MTs) are defined by a set of principles, e.g. separating ecosystems differing in type of disturbance or in dominance of ecosystem-engineering species groups (e.g. trees, helophytes).
Within major types, minor types exist in a theoretical ecological space along local complex environmental gradients (LECs). Within major types, each minor type spans an interval of standardised size along each main local complex environmental gradient (LEC; e.g. soil moisture gradient, lime richness gradient). This interval comprises a standardised amount of species compositional turnover, set at 25% change in species composition. This amount of turnover is defined as one ED unit.
LECs are divided into basic steps that are aggregated into major-type adapted steps within each major type for each spatial scale.The mapping units’ positions along major-type adapted steps can be translated into integer values if the mapping unit is fully contained within one major-type adapted step or decimal values if the mapping units span several major-type adapted steps. To find the position along major-type adapted steps for a mapping unit within another major type becomes more complicated. In those cases, the mapping units’ sometimes span a major-type adapted step only partly or not overlap with any major-type adapted step at the extreme parts of the LECs. To tackle the first situation, the major-type adapted step is estimated based on the basic steps. The second situation is tackled by adding two bordering major-type adapted steps by aggregating basic steps at both of the extreme ends of the LEC.
Many ecosystems are defined and differentiated based on principles other than species composition, or have no or unpredictable species occurrence. The lack of a usable species composition can stem from natural or anthropogenic disturbances, and result in substrates with different potential for succession and ecosystem composition and function. For these types, defining principles are used to estimate ED instead of, or in addition to, ED from species compositional turnover.
In practice, ED is calculated based on two datasets. The main dataset contains rows for each major-type group, major type, and minor type and columns for short code, ecological space, 1:5000 unit, 1:20 000 unit, and 1:50 000 unit. The other dataset contains rows for each minor type and columns for principles other than ecological space that differentiate the ecosystem types (e.g., structuring species groups, vegetation presence, above/below forest line). These are the general rules for calculation:
When no local environmental complex gradients (LECs) are shared, add 1 ED (bLEC/“bLKM” excepted). Every major-type adapted step along a factor LEC gives an ED of 1. Between two mapping units, ED is calculated along each LEC by placing the mapping units within the first mapping unit’s major type to find their values with respect to major-type adapted steps for the first mapping unit’s major type. Then, the absolute difference between the two values is calculated. Next, the mapping units are placed within the second mapping unit’s major type to find their values with respect to major-type adapted steps for the second mapping unit’s major type. Then, the absolute difference between the values is calculated. The resulting ED is given as the mean of the two absolute differences. The LEC-based ED is scaled by a factor obtained by dividing the gradient length (i.e., the number of major-type adapted steps) for the major type at the minor-type level by the gradient length for the major type at the desired spatial scale (e.g., 1:5000). Differing presence/absence of a structuring species group adds an ED of 1. Difference in removal of structuring species groups gives an ED of 1. In that case the above rule (#4) is not taken into account. Position relative to the forest line differs by 1 ED. No difference if one mapping unit can be found both below and above. Difference in presence/absence of substrate is given an ED of 1. Difference in presence/absence of vegetation is given an ED of 1. The ED between major-type groups are defined by LECs.
Each Major-type group can be defined by the LECs and intervals in the table below (LEC_interval). For example, terrestrial non-wetland and wetland systems are separated by their position along the “VM” LEC Duration of period without inundation, where non-wetland major types span basic steps 0abc, and wetland major types span the z step.
T = SV_0abcdefg, TV_y, VM_0abc, SA_0ab |
V = SV_0abcdefg, TV_cdefghijk, VM_z, SA_0ab |
I = SV_y, TV_y, VM_0abc, SA_0ab |
L = TV_0ab, SA_0ab |
M = TV_0ab, SA_cdefghz |
O = TV_0ab, SA_0ab |
Each major type has its own set of LECs. The total compositional turnover along the LECs within the major type is used to differentiate major (hLEC; > 2 ED) and significant (bLEC; 1-2 ED) LECs. Within major types, ED is only taken into account for major LECs. Differences with respect to LECs and principles are only computed where they are relevant for both mapping units.
Several LECs (for example HH, HM, HA, and VF, TV, VM) are strongly correlated and could cause an inflation of ED with the default calculation method. In the group of correlated LECs, we only consider the ED for the LEC that gives the maximum value. In the current version of ED calculation, this correction has only been done for the HA, HM, and HH LECs, and for HR and HH. To change this, modify the cor_lec object in the create_ed_matrix.R script.
Halvorsen, R., Skarpaas, O., Bryn, A., Bratli, H., Erikstad, L., Simensen, T., & Lieungh, E. (2020). Towards a systematics of ecodiversity: The EcoSyst framework. Global Ecology and Biogeography, 29(11), 1887–1906. https://doi.org/10.1111/geb.13164
Naas, A.E., Halvorsen, R., Horvath, P., Wollan, A.K., Bratli, H., Brynildsrud, K., et al. (2023). What explains inconsistencies in field-based ecosystem mapping? Applied Vegetation Science, 26, e12715. https://doi.org/10.1111/avsc.12715
Eriksen, E.L., Ullerud, H.A., Halvorsen, R., Aune, S., Bratli, H., Horvath, P., Volden, I.K., Wollan, A.K., & Bryn, A. (2018). Point of view: Error estimation in field assignment of land-cover types. Phytocoenologia, 49(2), 135–148. https://doi.org/10.1127/phyto/2018/0293