Classify indicators as sensitive or resistant to degradation

Compares how far each indicator has moved between a degraded soil and a reference soil, relative to how far soil organic carbon moved. Indicators that decline faster than carbon are sensitive – they register degradation early – and those that decline more slowly are resistant.

Usage

sensitivity_resistance(degraded, reference, carbon = "SOC", tolerance = 0.1)

Arguments

degraded

Named numeric vector of indicator values in the degraded soil, or a one-row data frame.

reference

Named numeric vector of the same indicators in the reference soil, or a one-row data frame. Must cover every name in degraded.

carbon

Name of the carbon indicator against which the others are compared. Defaults to "SOC".

tolerance

Half-width of the band around 1 within which an indicator is called "proportional" rather than sensitive or resistant. Defaults to 0.1.

Value

A data frame with one row per indicator, ordered from most sensitive to most resistant:

indicator

Indicator name

degraded, reference

The two input values

change

degraded / reference

ratio

change divided by the carbon indicator's change

class

"sensitive", "proportional" or "resistant"

Details

This is Kuzyakov et al.'s (2020) second and much less used approach. Each indicator's relative change is divided by the relative change in the carbon indicator. On the 1:1 identity line the indicator degrades at exactly carbon's rate; below it, faster; above it, slower.

The expected pattern is that (micro)biological properties are sensitive and physical properties resistant, which is why an early-warning monitoring programme should watch the biological ones.

It does not always separate. Kuzyakov reports the classification resolving cleanly on a Luvic Phaeozem and failing to separate on a Calcic Chernozem. Treat a tidy result as informative and an untidy one as a fact about the soil rather than a failure of the analysis – a ratio near 1 for everything means the indicators are degrading together, which is itself worth reporting.

Both inputs are single soils summarised to one value per indicator – a mean over the plots of each condition, typically. The function does not propagate the uncertainty in those means, so a ratio close to 1 should not be over-read.

References

Kuzyakov, Y. et al. (2020). Frontiers of Agricultural Science and Engineering 7(3):282-288. doi:10.15302/J-FASE-2020338

See also

standardize_to_reference

Examples

degraded  <- c(SOC = 1.1, OM = 1.9, N = 0.09, CEC = 9.0, BD = 1.55)
reference <- c(SOC = 2.0, OM = 3.4, N = 0.18, CEC = 13.0, BD = 1.25)

sensitivity_resistance(degraded, reference)
#>   indicator degraded reference    change     ratio        class
#> 1         N     0.09      0.18 0.5000000 0.9090909 proportional
#> 2       SOC     1.10      2.00 0.5500000 1.0000000 proportional
#> 3        OM     1.90      3.40 0.5588235 1.0160428 proportional
#> 4       CEC     9.00     13.00 0.6923077 1.2587413    resistant
#> 5        BD     1.55      1.25 1.2400000 2.2545455    resistant