Gut microbiota ecology: Biodiversity estimated from hybrid neutral-niche model increases with health status and aging

holds true for both θrare and θabundant and indicates that biodiversity increases with age in both young and old subjects while staying constant in the middle ages (see Figs 2 and 3, Table 1). For some of the data sets we observe a deviation from the general trend (Table 1). For example, in the Kong data set a statistically significant decrease of biodiversity during young age is observed, albeit this could be due to the small number of young subjects. We remark that the data set specific deviations from the global trend are related to the effect of all the variables that differ between one data set and the others, including sequencing tech- nology, 16S targeted region, pre-processing of data, population ethnicity, and all the other data set specific unknown sources of variation. Indeed, the global model is not affected by such sources of variation, nor by the different sample size of the data sets. In the model, in fact, the categorical variable relative to the data set of origin was treated as a Contrast coding. This means that the results for the general model are defined as the grand mean (average of the means) of the effects, and they are therefore independent from the numerosity of each individual data set. This allowed us to include data sets with different sam- ple size without distorting the results. To corroborate our results, we fitted the splines regression model considering five among the most common classical biodiversity indices: Hill1 [28], Hill2 [28], Pielou [27], Shannon [25]

and Simpson [26] index. The mathematical definitions of these indices are reported in the Materials and Methods section. Here, we only recall that for all indices but Simpson, higher values correspond to higher biodiversity. Table 1 shows that the increase of GM biodiversity with age is detected not only by θrare and θabundant, but also by all other indices, with the only exception of Pielou, that does not identify a statistically significant increase of biodiversity for old subjects. Notice that θrare does not detect the increase of GM biodiversity with age in young subjects. However, according to the 2NB model θrare and θabundant describe the biodiver- sity of two different niche of the GM ecosystem and should therefore be taken into account simultaniously. 2. GM biodiversity and healthy aging. We tested the discriminative ability of the biodi- versity indices θrare and θabundant considering the Biagi & Schnorr data set, that includes indi- viduals from 3 groups: i) Italian healthy subjects [35] (Healthy Italian), ii) age-matched Italian DS subjects (DS Italian), that represent a model of accelerated aging [33]; and iii) Hadza Fig 1. Natural cubic spline model with 3 degrees of freedom. (left) Basis functions for a natural cubic spline with 3 degrees of freedom. (right) Natural cubic spline with 3 degrees of freedom and coefficients equal to: coef(intercept) = 0.0; coef(spline1) = 0.6; coef(spline2) = 0.5; coef(spline3) = 1.0. The black line represents the overall model, while the colored lines are the basis functions rescaled by the model coefficients.

https://doi.org/10.1371/journal.pone.0237207.g001 PLOS ONE Gut microbiota biodiversity and healthy aging PLOS ONE | https://doi.org/10.1371/journal.pone.0237207 October 30, 2020 6 / 23