---
title: "The AMBI index"
bibliography: references.bib
link-citations: true
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{The AMBI index}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  fig.path = "man/figures/"
)
```

The basis of the AMBI index is that soft-bottom macrofauna are divided into 
groups according to their sensitivity to increasing environmental stress. The 
distribution of counts of individuals or relative abundance between the different
groups is used to calculate a quantitative measure of the ecological quality of
the benthic environment. 

### Species Groups

Input to the `AMBI()` function is a dataframe of species counts with optional 
grouping variables, e.g. station or replicate IDs. The function matches species
names in the input data with names in the AMBI species list, in order to categorise
the observed species according to the AMBI method. The tool then calculates the
`AMBI` index resulting from the distribution of individuals between the groups.

The AMBI species list gives the groups (I, II, III, IV, V) in which each species 
is classified, as described by @BORJA20001100.

-   *Group I* \
    Species very sensitive to organic enrichment and present under
    unpolluted conditions (initial state). They include the specialist
    carnivores and some deposit- feeding *tubicolous polychaetes*.
-   *Group II* \
    Species indifferent to enrichment, always present in low
    densities with non-significant variations with time (from initial
    state, to slight unbalance). These include suspension feeders, less
    selective carnivores and scavengers.
-   *Group III* \
    Species tolerant to excess organic matter enrichment.
    These species may occur under normal conditions, but their
    populations are stimulated by organic enrichment (slight unbalance
    situations). They are surface deposit-feeding species, as
    *tubicolous spionids*.
-   *Group IV* \
    Second-order opportunistic species (slight to pronounced
    unbalanced situations). Mainly small sized *polychaetes*: subsurface
    deposit-feeders, such as *cirratulids*.
-   *Group V* \
    First-order opportunistic species (pronounced unbalanced
    situations). These are deposit- feeders, which proliferate in
    reduced sediments.

The list of species and their groups has been updated several times by the 
authors of the AMBI software. The version of the list used here is from 
8. October 2024.

After calculating the fractions $f_i$ of all individuals belonging to each group 
  $i \in{}  \{I, II, III, IV, V\}$, then the index is given by:

$$ AMBI = 0.0 * f_{I} + 1.5 * f_{II} + 3 * f_{III} + 4.5 * f_{IV} + 6 * f_{V} $$

So, the greater the proportion of sensitive species, the lower the resulting AMBI 
index. A sample consisting 100% of species from the most sensitive category 
(Group I) will have an AMBI index of 0.0. A population consisting entirely of 
species from Group V will have an index of 6.0.

## M-AMBI - the multivariate AMBI index

`MAMBI()` calculates M-AMBI the multivariate AMBI index, based on the 
three separate species diversity metrics:

 * AMBI index *AMBI*
 * Shannon Wiener diversity index *H'*
 * Species richness *S*

The principles of the M-AMBI index are described by @MUXIKA200716

 _"AMBI, richness and diversity, combined with the use, in a further
 development, of factor analysis together with discriminant analysis, is
presented as an objective tool (named here M-AMBI) in assessing
ecological quality status"_

It is, of course, possible to calculate M-AMBI using data generated in other 
analyses, outside the ambiR package but the `AMBI()` function can conveniently
provide all 3 of the metrics used as variables in the M-AMBI factorial analysis.

* from the input data with values of *AMBI*, *H'* and *S*, the variables are first
standardized, by subtracting by the mean and then dividing by the standard 
deviation.

* the analysis requires information on limits for each of the 3 variables: _(a)_ 
values corresponding to _reference_ or _undisturbed_ conditions. For the Shannon
diversity *H'* and species richness *S*, these are taken as the maximum values 
found in the data. This assumes that some of the observations are from 
_undisturbed_ sites so care should be given and suitable values provided by the
user if this assumption does not hold. For *AMBI*, the reference condition value
used is `0`, unless a different value is specified. _(b)_ default limit values 
corresponding to _bad_ conditions are `AMBI = 6`, `H = 0` and `S = 0`.

* factor analysis (FA) using the principal component analysis method
on the standardized variables generates 3 factors.

* the Varimax rotation method is applied to the results of FA.
The factor scores (`x`, `y` and `z`) are the new coordinates of 
each sampling station in the new factor space. 

* These coordinates are used to derive the EQR or M-AMBI values. The M-AMBI 
score is the mean of the distance along the zero to one scale in the three 
dimensions. Depending on specific regional conditions the M-AMBI value 
corresponding to the Good/Moderate and other class boundaries can be used to 
convert M-AMBI values to a normalised EQR value where the Good/Moderate boundary 
is at EQR = 0.6.

## AMBI software

The AMBI software was developed as a free standalone software to allow users 
to perform AMBI index calculations. Later versions were updated to include 
the multivariate index M-AMBI calculations and adjustments to the species list
used to assign species to ecological groups. The software is maintained and 
updated by AZTI [https://www.azti.es](https://www.azti.es/en/ambi-international-reference-for-marine-environment-assessment/),
where the latest version can be downloaded.

The ambiR package has been extensively tested and gives identical results to the
AMBI software, as long as the version of species list select corresponds to the
to the version used by the software.

## References