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

among species are neglected and all species are considered to have evolved according to the same dynamics. Specifically, in Volkov’s model the population dynamics of all species included in the ecosystem (i.e. the GM) is ruled by only three parameters: a birth rate (b), a death rate (d) and an immigration term (S) that represents a constant influx of individuals into the population, while inter-species interaction is neglected. The neutrality assumption is certainly an oversimplification of the GM ecosystem dynamics. Nevertheless, this model has been proven to well describe the RSA of various ecosystems [7, 31] and to well characterize their biodiversity [7, 31], while providing an evolutionary explanation of the current RSA configuration. According to Volkov’s model, indicated here as 1NB model, the RSA is described by a zero- truncated Negative Binomial distribution [31]. As detailed in the Materials and Methods sec- tion, such distribution characterizes the probability of having n individuals in a species (PRSA(n)), and it only depends on the three dynamic parameters b, d and S. Moreover, it holds true for all the species of the ecosystem, and it can be expressed as PRSAðnÞ ¼ y � NBðb; d; SÞ ð1Þ to highlight Hubbell’s biodiversity index [30] θ (see Eq (9) in the Materials and Methods section for the definition). The 1NB model fits well the RSA sampled from various ecosystems, including the coral reefs and the GM of several animals [7, 31]. However, incongruence between the neutral 1NB

model and empirical data has been previously reported for both ecosystems [7, 32], and sug- gests that the 1NB model is able to explain part of the GM community structure, but does not well describe the most abundant species that constitute the right tail of the distribution. These findings suggest that along with pure neutral models also hybrid niche-neutral models in which the neutrality assumption is partly relaxed should be taken into account [32]. For these reasons, in this work we try to describe the GM population using the neutral theory but also assuming that we might be observing multiple different ecological niches, each one driven by its own dynamic parameters. Analysing six publicly available data sets, we show that a model in which two non-interact- ing niches are considered, well fits the GM data. From this model we derive two biodiversity indices, one for each niche, that overall summarize the GM biodiversity. Finally, we show that such estimates achieve higher statistical power than classical biodiversity indices, in terms of the identification of the relationship between the GM biodiversity and aging and of the discri- mation of subjects with different diet and life style or with different health status. Results and discussion Data sets selection We selected and analysed publicly available 16S rRNA sequencing data sets related to aging and healthy aging in different populations that either: contained a wide age range of subjects; contained elderly subjects in various health conditions; contained Down Syndrome subjects, a

model of accelerated aging [33]. Moreover, a data set consisting of healthy Italians and Tanza- nian Hadza hunter-gathers was included to test the descrimintative ability of GM biodiversity (estimated through our modeling approach or using classical indices) under important diet and life-style differences. A brief description of the six selected data sets is provided in the fol- lowing, while further details about the data sets composition and the pre-processing are reported in the Materials and Methods section and in S1 and S2 Tables. The analysed data sets are: PLOS ONE Gut microbiota biodiversity and healthy aging PLOS ONE | https://doi.org/10.1371/journal.pone.0237207 October 30, 2020 3 / 23