Measuring individual identity information in animal signals: Overview and performance of available identity metrics

Abstract

Identity signals have been studied for over 50 years but there is no consensus as to how to quantify individuality. While there are a variety of different metrics to quantify individual identity, or individuality, these methods remain un-validated and the relationships between them unclear. We contrasted three univariate and four multivariate metrics (and their different computational variants) and evaluated their performance on simulated and empirical datasets. Of the metrics examined, Beecher's information statistic (HS) was the best one and could easily and reliably be converted into the commonly used discrimination score (and vice versa) after accounting for the number of individuals and calls per individual in a given dataset. Although Beecher's information statistic is not entirely independent of sampling parameters, this problem can be removed by reducing the number of parameters or by increasing the number of individuals. Because it is easily calculated, has superior performance, can be used to describe single variables or signal as a whole, and because it tells us the maximum number of individuals that can be discriminated given a set of measurements, we recommend that individuality should be quantified using Beecher's information statistic.

Structure and functions of Yellow-breasted Boubou (Laniarius atroflavus) solos and duets

“…Tonal whistles are more efficiently propagated through dense forest habitat ( Boncoraglio & Saino, 2007 ) which, together with a higher amplitude, suggests that male phrases (A Wheeldon et al, 2020, unpublished data) are aimed at receivers at a further distance than the phrases produced by females. Although males share all phrase types they are clearly individually distinct ( Linhart et al, 2019 ). At the moment we do not know if this is also the case for females due to the complexity of the atonal harsh notes and limited amount of isolated female recordings in the field.…”

“…With male song phrases it was easy, despite a fully shared repertoire, to assign particular individuals as each male song from a particular category has its individual specificity reflected by small but consistent differences in frequency and duration. This time-frequency characteristic of male calls was already used in a methodological study on measuring individual identity in general ( Linhart et al, 2019 ). We compared the shape of phrases with the Peak Frequency Contour measurement of Raven Pro with measurements visible on screen and listening to the signal at a slow speed.…”

Peer Review #1 of "Structure and functions of Yellow-breasted Boubou (Laniarius atroflavus) solos and duets (v0.1)"

“…With male song phrases it was easy, despite a fully shared repertoire, to assign particular individuals as each male song from a particular category has its individual specificity reflected by small but consistent differences in frequency and duration. This time-frequency characteristic of male calls was already used in a methodological study on measuring individual identity in general (Linhart et al, 2019). We compared the shape of phrases with the Peak Frequency Contour measurement of Raven Pro with measurements visible on screen and listening to the signal with slow speed.…”

Identity signaling, identity reception and the evolution of social recognition in a Neotropical frog

“…For each species and acoustic property, we calculated a mean and standard deviation for each individual, the grand mean and standard deviation based on the average values for 20 individuals, and the "potential for individual coding" (PIC, [36]), which was the ratio of the among-individual coefficient of variation (CV a = grand SD/grand mean × 100%) to the within-individual coefficient of variation (CV w = individual SD/individual mean × 100%). The proportion of variance explained by individual differences was calculated as the effect size (partial η 2 ) of "individual" as a random effect in a separate ANOVA conducted for each property.…”

“…Note, conclusions did not change when running subsequent analyses using more restrictive sets of principal components (see Table S3). Principal components were used to compute Beecher's information statistic (H s ), which measures identity information as a signal's ability to reduce uncertainty about the identity of the signaler [7,36]. We also included these principal components as input variables in a linear discriminant analysis using the 'lda' function in the 'MASS' package in R and used a leave-one-out cross-validation procedure to evaluate the accuracy with which calls could be assigned to the correct individual.…”