Artificial selection of communities drives the emergence of structured interactions

Abstract

Species-rich communities, such as the microbiota or environmental microbial assemblages, provide key functions for human health and ecological resilience. Increasing effort is being dedicated to design experimental protocols for selecting community-level functions of interest. These experiments typically involve selection acting on populations of communities, each of which is composed of multiple species. Numerical explorations allowed to link the evolutionary dynamics to the multiple parameters involved in this complex, multi-scale evolutionary process. However, a comprehensive theoretical understanding of artificial selection of communities is still lacking. Here, we propose a general model for the evolutionary dynamics of species-rich communities, each described by disordered generalized Lotka-Volterra equations, that we study analytically and by numerical simulations. Our results reveal that a generic response to selection for larger total community abundance is the emergence of an isolated eigenvalue of the interaction matrix that can be understood as an effective cross-feeding term. In this way, selection imprints a structure on the community, which results in a global increase of both the level of mutualism and the diversity of interactions. Our approach moreover allows to disentangle the role of intraspecific competition, interspecific interactions symmetry and number of selected communities in the evolutionary process, and can thus be used as a guidance in optimizing artificial selection protocols.

Overlaps between eigenvectors of spiked, correlated random matrices: From matrix principal component analysis to random Gaussian landscapes

“…[6][7][8][9][10][11][12][13][14]. Deformed random matrices and their outliers have been shown to play a relevant role in a variety of contexts: examples can be found in finance [15], inference and detection problems [16,17], constraint satisfaction problems [18], quantum chaos [19], localization of polymers by defects [20], and theoretical ecology [21,22]. The eigenvectors associated to the outliers play a relevant role in these applications: their projection on the subspace spanned by the low-rank perturbations remains large in the limit of large matrix size [7,23], a phenomenon akin to condensation [24].…”

Non-Gaussian random matrices determine the stability of Lotka-Volterra communities

“…We note that the statistics of the reduced interaction matrix elements are determined by the extinction dynamics in the GLVE system, and are consequently vastly different to those of the original interaction matrix [28,29]. For instance, they are non-Gaussian (even when the a ij are Gaussian), and there are correlations between elements sharing only one index (see SM Section S6).…”

Breakdown of Random-Matrix Universality in Persistent Lotka-Volterra Communities

“…S2 in the Supplemental Material). We note that the statistics of the reduced interaction matrix elements are determined by the extinction dynamics in the GLVE system, and are consequently vastly different from those of the original interaction matrix [33,43]. For instance, they are non-Gaussian (even when the a ij are Gaussian), and there are correlations between elements sharing only one index (see the Supplemental Material [39], Sec.…”