Fig 1.
Sketch of the microbial community assembly model.
As shown in Panel A, the environmental pool contains two species, A (blue) and B (orange). Species A is present in the pool in abundance , while species B is present in abundance 1 −
. Microbial clusters of size n disperse from the pool into local microbial communities at rate c and replicate within these communities at rates
and
. Cluster composition and abundance are drawn from a binomial distribution
. Once each community reaches its carrying capacity, we analyze the abundance fluctuation distribution, which quantifies the number of communities with a given composition and abundance (B, C, and D). We characterize this distribution using its bimodality coefficient (BC) and mean relative abundance.
Fig 2.
Higher dispersal rates increase the contribution of dispersal to early community assembly.
Panels A and B show the mean number of clusters contributing to community assembly and the mean number of immigrants (i.e., dispersing microbes) n ×
as a function of the dispersal rate c for various cluster sizes n, respectively. In both panels, the simulated data are averaged over 104 microbial communities. The 95% confidence interval bars are smaller than the markers and therefore are not displayed in the figure. The solid lines represent our analytical predictions (see Eq 1). The two vertical dotted lines indicate key dispersal thresholds:
, where the mean time between the first and second dispersal events equals the time to reach carrying capacity via replication, and c = r, where dispersal and replication rates are equal. Parameter values: replication rates
, relative abundance of A in the pool
, carrying capacity K = 105.
Fig 3.
Cluster dispersal blurs the boundary between assembly regimes.
Panel A shows the bimodality coefficient BC as a function of the dispersal rate c for various cluster sizes n, whereas panel B shows it as a function of the cluster size n. In all panels, each data point corresponds to the bimodality coefficient of an abundance fluctuation distribution obtained from 104 simulated microbial communities. The solid lines represent our analytical predictions (see Eq 3). The two vertical dotted lines indicate key dispersal thresholds: , where the mean time between the first and second dispersal events equals the time to reach carrying capacity via replication, and c = r, where dispersal and replication rates are equal. Parameter values: replication rates
, dispersal rate c = 10−4 (in B), relative abundance of A in the pool
, carrying capacity K = 105.
Fig 4.
Cluster dispersal and within-community selection alter species abundance in a non-monotonic way.
Panel A shows the mean relative abundance of A as a function of the dispersal rate c for various cluster sizes n, whereas panel B shows it as a function of the cluster size n for different abundances in the pool
in the low-dispersal regime. In both panels, the simulated data are averaged over 104 microbial communities, whereas the solid lines represent our analytical predictions (Eqs 4 and 5). The 95% confidence interval bars are smaller than the markers and therefore are not displayed in the figure. In Panel A, the two vertical dotted lines indicate key dispersal thresholds:
, where the mean time between the first and second dispersal events equals the time to reach carrying capacity via replication, and c = r, where dispersal and replication rates are equal. In both panels, the horizontal lines show the relative abundance of species A in the microbial pool. Parameter values: replication rate of A
, replication rate of B
, dispersal rate c = 10−4 (in B), relative abundance of A in the pool
(in A), carrying capacity K = 105.
Fig 5.
Bimodality coefficient and mean relative abundance across multiple dispersal rates reveal patterns of selection and cluster dispersal.
In all panels, each data point shows the mean relative abundance of species A and the bimodality coefficient obtained from an abundance fluctuation distribution simulated over 104 microbial communities. Illustrative examples of these distributions for different parameter regimes are provided in S2 Fig. Each panel corresponds to a different cluster size n. Colors correspond to dispersal rate, with darker shades indicating higher dispersal. The bar denotes the gradient of dispersal rates from low (light) to high (dark). Parameter values: replication rate of A (red, green, yellow, purple)
(blue), replication rate of B
(red, blue, green, purple)
(yellow), relative abundance of A in the pool
(blue, green, yellow)
(red)
(purple), carrying capacity K = 105, dispersal rate
.
Fig 6.
Bimodality coefficient and mean relative abundance across multiple cluster sizes reveal patterns of selection.
In all panels, each data point shows the mean relative abundance of species A and the bimodality coefficient obtained from an abundance fluctuation distribution simulated over 104 microbial communities. Illustrative examples of these distributions for different parameter regimes are provided in S2 Fig. Each panel corresponds to a different dispersal rate c. Colors correspond to cluster size, with darker shades indicating higher cluster size. The bar denotes the gradient of cluster sizes from low (light) to high (dark). Parameter values: replication rate of A (red, green, yellow, purple)
(blue), replication rate of B
(red, blue, green, purple)
(yellow), relative abundance of A in the pool
(blue, green, yellow)
(red)
(purple), carrying capacity K = 105, cluster size n = 1−200.
Fig 7.
Cluster dispersal increases α-diversity and decreases β-diversity.
Panels A and C show the richness as a function of the dispersal rate and cluster size, respectively, where C focuses on the low-dispersal regime. The simulated data are averaged over 103 stochastic replicates (i.e., microbial communities). The 95% confidence interval bars are smaller than the markers and therefore are not displayed in the figure. Panels B and D represent the Jaccard distance as a function of the dispersal rate and cluster size, respectively, where D focuses on the low-dispersal regime. The data points are averaged over the comparison of each pair of 103 stochastic replicates (i.e., microbial communities). The solid lines represent our analytical predictions (Eqs 6 and 9). In panels A and B, the two vertical dotted lines correspond to key dispersal thresholds: , which approximates the mean time between the first dispersal event and reaching carrying capacity through replication alone [18], and c = r, where the dispersal rate equals the replication rate. In panels B and D, the dashed line represents the exact Jaccard distance for n = 1 (i.e.,
−
. Parameter values: replication rate r = 1, dispersal rate c = 10−4 (in C and D), number of species S = 7, relative abundance of each species in the pool p = 1/S, carrying capacity K = 105.