The number of functional genes coding for olfactory receptors differs markedly between species and has repeatedly been suggested to be predictive of a species’ olfactory capabilities. To test this assumption, we compiled a database of all published olfactory detection threshold values in mammals and used three sets of data on olfactory discrimination performance that employed the same structurally related monomolecular odor pairs with different mammal species. We extracted the number of functional olfactory receptor genes of the 20 mammal species for which we found data on olfactory sensitivity and/or olfactory discrimination performance from the Chordata Olfactory Receptor Database. We found that the overall olfactory detection thresholds significantly correlates with the number of functional olfactory receptor genes. Similarly, the overall proportion of successfully discriminated monomolecular odor pairs significantly correlates with the number of functional olfactory receptor genes. These results provide the first statistically robust evidence for the relation between olfactory capabilities and their genomics correlates. However, when analysed individually, of the 44 monomolecular odorants for which data on olfactory sensitivity from at least five mammal species are available, only five yielded a significant correlation between olfactory detection thresholds and the number of functional olfactory receptors genes. Also, for the olfactory discrimination performance, no significant correlation was found for any of the 74 relationships between the proportion of successfully discriminated monomolecular odor pairs and the number of functional olfactory receptor genes. While only a rather limited amount of data on olfactory detection thresholds and olfactory discrimination scores in a rather limited number of mammal species is available so far, we conclude that the number of functional olfactory receptor genes may be a predictor of olfactory sensitivity and discrimination performance in mammals.

Figure S1. Monotonic correlations between the number of functional olfactory receptor genes and olfactory detection threshold values for 1-octanol (A, rho = -0.975 p = 0.005), 2-octanone (B, rho = -1 p = 0.017), 2-nonanone (C, rho = -1 p = 0.017), n-butyl acetate (D, rho = 0.883 p = 0.009), and iso-butyl acetate (E, rho = -0.886  p = 0.033).

Figure S2. Plot of the 128 monotonic correlations performed against the number of functional olfactory receptor genes. A: results based on Spearman's rank-correlation test and B, results based on PGLS. 

Figure S3. Distributions of the likelihood ratio statistic for two evolutionary models (Brownian motion (BM) and white noise) fitted to the residuals of the regressions between the number of functional olfactory receptor genes and the threshold detection of the n-butanoic acid (A) as well as between the number of functional olfactory receptor genes and the success of odor pairs discrimination (B, C, D, E). For the success of odor pairs discrimination, we considered the enantiomers, level 1 (B), the discrimination of aliphatic with different carbon chain length, 1-Alcohols, level 2A (C), the Acetic esters C5 vs. C7, level 3A (D), and the Limonene, level 3C (E). The vertical line (sometimes too close form the Y-axis to be visible) indicates the observed value of likelihood ratio.

Figure S4. Significant pairwise differences in the detection threshold of monomolecular odorants across 12 and 10 mammal species. A negative delta value supports the hypothesis of a negative relation between the number of functional olfactory receptor genes and olfactory sensitivity, while a positive delta value indicates the opposite. Subplot A shows results without Bonferroni correction, and subplot B shows results with Bonferroni correction.

For this plot, we employed a Welch test to compare the series of threshold detection common to pairs of species, selecting species with at least four monomolecular threshold test comparisons in common. When the cell is colored, it indicates a significant difference between the paired species. Subsequently, the mean threshold detection of shared monomolecular odorants was calculated, and this value was compared between the paired species, specifically between the species with the highest number of olfactory receptor genes and the other species in the pair. This resulting value represents the delta. A negative delta p-value implies that, in the pairwise comparison, the species with a higher number of functional olfactory receptor genes exhibits a lower olfactory sensitivity threshold. Conversely, a positive delta p-value suggests that, in the pairwise comparison, the species with the lower number of functional olfactory receptor genes has the lower olfactory sensitivity threshold.

Without Bonferroni correction, out of the 56 tested pairwise comparisons, 32 are significant, comprising 20 negative delta values and 12 positive delta values. With Bonferroni correction, out of the 56 tested pairwise comparisons, 14 are significant, comprising 10 negative delta values and 4 positive delta values.

We express skepticism about this approach as it minimally considers the actual difference in the number of functional olfactory receptor genes. It solely identifies whether one species has a higher number than another, without accounting for the magnitude of the difference. Consequently, drawing meaningful conclusions from this approach is challenging. We find greater confidence in the classical approach, where we plot and test the overall raw threshold of detection against continuous values of the number of functional olfactory receptor genes. This is exemplified by methods such as linear regression and logistic regression (Figure 1 and Table S8).

Table S1: Tables including the lists of the references used to extract olfactory performance data.

Table S2: Raw data and results of tests of correlation between the number of functional olfactory receptor genes and a species’ olfactory sensitivity. In this table and all the others, 'NA' indicates the value is not available or not applicable. Yellow highlights indicate the significance of the test.

Table S3: Raw data and results of tests of correlation between the number of functional olfactory receptor genes and the proportion of successfully discriminated odor pairs according to the level 1 with data set A, B, and C (see Materials and Methods).

Table S4: Raw data and results of tests of correlation between the number of functional olfactory receptor genes and the proportion of successfully discriminated odor pairs according to the level 2 with data set A, and B (see Materials and Methods).

Table S5: Raw data and results of tests of correlation between the number of functional olfactory receptor genes and the proportion of successfully discriminated odor pairs according to the level 3 with data set A (see Materials and Methods).

Table S6: Raw data and results of tests of correlation between the number of functional olfactory receptor genes and the proportion of successfully discriminated odor pairs according to the level 3 with data set B (see Materials and Methods).

Table S7: Raw data and results of tests of correlation between the number of functional olfactory receptor genes and the proportion of successfully discriminated odor pairs according to the level 3 with data set C (see Materials and Methods).

Table S8: Results of tests of correlation and model fitting for the unbalanced overall data sets and balanced subsets for olfactory detection thresholds as well as proportion of successfully discriminated monomolecular odor pairs.

Table S9: Results of the phylogenetic inertia tests.

Table S10: Results of the tests presented in the Table S8 but in this case, the number of functional olfactory receptor genes was transformed into numerical rankings. These rankings range from 'one,' corresponding to the species with the lowest number of functional olfactory receptor genes, to the maximum number.

Table S11: Results of the tests presented in the Table S10 but in this case, the data exclude trials wherein all species successfully discriminated odor pairs. 

Table S12: Acknowledgments and credits for the Creative Commons silhouettes used in the graphical abstract from the main research article: Does the number of functional olfactory receptor genes predict olfactory sensitivity and discrimination performance in mammals? The silhouettes herein referred to are only presented in the graphical abstract of the main article and not in the Dryad. All the files uploaded in the Dryad are under CC0 license terms (without any restriction).

Data S1: CSV files and R scripts. The root folder contains five folders for olfactory discrimination tests with three levels (1, 2, 3 corresponding respectively to large scale, small scale, and fine scale) tested and the three datasets (A, B, and C). In these respective folders, various subfolders contain R scripts and CSV files used to perform the analyses specified in the folder names. These analyses include evolutionary model comparison (BM and white noise), PGLS analyses, tests of statistical power, Spearman's rank-correlation test, and tests of phylogenetic inertia. The root folder also contains similar analyses for olfactory sensitivity, the comparison of phylogenetic inertia between a dataset of 20 species and a dataset of 396 species, Welch two-tailed tests, and the previous analyses transforming the data into numerical ranking from 1 to 5 with or without all the success. All these scripts were run with R studio version 2023.06.2+561 and the following packages: caper; lme4; ggplot2; tidyverse; lmtest; nlme; pmc; ggplot2; tidyr; dplyr; pwr; phytools; reshape2.