sig_rep_zero {qfasar}

Replace invalid fatty acid signature proportions

Description

The utility function sig_rep_zero replaces fatty acid signature proportions that are less than or equal to zero or missing with a small constant and uses the multiplicative method (Martin-Fernandez et al. 2011) to scale the proportions to sum to 1.

Usage

sig_rep_zero(sig_data, zero_rep = 75)

Arguments

sig_data

A numeric matrix containing signature data as either proportions or percentages in column-major format.

zero_rep

A constant associated with the method and value to be used to replace invalid values. See Details. Default value 75.

Value

A list containing the following elements:

sig_adj

The signature data with non-positive or missing proportions replaced and scaled to sum to 1.0.

rep_val

The value used to replace invalid proporitons.

err_code

An integer error code (0 if no error is detected).

err_message

A string contains a brief summary of the execution.

Details

The function sig_rep_zero is an internal utility function.

The Kullback-Leibler (Iverson et al. 2004) and Aitchison (Stewart et al. 2014) distance measures are not defined for proportions of zero. Consequently, if either of these distance measures will be used in an analysis, the argument zero_rep should be strictly greater than 0. The chi-square distance measure (Stewart et al. 2014) is defined for proportions of zero, so if that distance measure will be used in the analysis, the argument zero_rep may equal zero. For simulation or other comparative work involving multiple distance measures, it may be advisable to use a common value to replace zeros.

The argument zero_rep must be either:

• Greater than or equal to 0 and no greater than 0.01, in which case the specified value is used to replace invalid proportions.

• Between 10 and 100, with an uninformed default of 75. In this case, zero_rep is interpreted as a percentage. The smallest non-zero proportion in sig_data is multiplied by the percentage and divided by 100. The result is used to replace invalid proportions.

Although Bromaghin et al. (2016) found that scaling signatures by varying constants introduces a bias in diet estimation, the slight distortion of the signatures caused by replacing invalid proportions with a small constant that varies between signatures is unlikely to introduce meaningful bias.

References

Bromaghin, J.F., S.M. Budge, and G.W. Thiemann. 2016. Should fatty acid signature proportions sum to 1 for diet estimation? Ecological Research 31:597-606.

Iverson, S.J., C. Field, W.D. Bowen, and W. Blanchard. 2004. Quantitative fatty acid signature analysis: A new method of estimating predator diets. Ecological Monographs 74:211-235.

Martin-Fernandez, J.A., J. Palarea-Albaladejo, and R.A. Olea. 2011. Dealing with zeros. P. 43-58 in V. Pawlowsky-Glahn and A. Buccianto, eds. Compositional data analysis: theory and application. John Wiley, Chichester.

Stewart, C., and C. Field. 2011. Managing the essential zeros in quantitative fatty acid signature analysis. Journal of Agricultural, Biological, and Environmental Statistics 16:45?69.