Species Sorting of Benthic Invertebrates in a Salinity Gradient - Importance of Dispersal Limitation

Abstract

The relative importance of environment and dispersal related processes for community assembly has attracted great interest over recent decades, but few empirical studies from the marine/estuarine realm have examined the possible effects of these two types of factors in the same system. Importance of these processes was investigated in a hypothetical metacommunity of benthic invertebrates in 16 micro-tidal estuaries connected to the same open sea area. The estuaries differed in size and connectivity to the open sea and represented a salinity gradient across the estuaries. The Elements of Metacommunity Structure (EMS) approach on estuary scale was complemented with a mechanistic variance partitioning approach on sample scale to disentangle effects of factors affecting assembly of three trait groups of species with different dispersivity. A quasi-Clementsian pattern was observed for all three traits, a likely response to some latent gradient. The primary axis in the pattern was most strongly related to gradients in estuary salinity and estuary entrance width and correlation with richness indicated nestedness only in the matrix of the most dispersive trait group. In the variance partitioning approach measures of turnover and nestedness between paired samples each from different estuaries were related to environmental distance in different gradients. Distance between estuaries was unimportant suggesting importance of factors characterizing the estuaries. While the high dispersive species mainly were sorted in the salinity gradient, apparently according to their tolerance ranges towards salinity, the two less dispersive traits were additionally affected by estuary entrance width and possibly also area. The results exemplify a mechanism of community assembly in the marine realm where the niche factor salinity in conjunction with differential dispersal structure invertebrates in a metacommunity of connected estuaries, and support the idea that dispersive species are more controlled by the environment than less dispersive species.

Fig 1. Map over the investigation area showing locations of 16 estuaries with areas indicated by different colors and id numbers as in Table 1.

Open sea is indicated by blue color and the 242 sampling sites by black dots. Copyright on the base map by the Danish Geodata Agency.

Fig 2. Sorted site-by-species incidence matrices for invertebrate species in 16 estuaries, with species on the x-axis and sites (estuaries) on the y-axis.

Matrices for HD trait (A), ID trait (C) and LD trait (D) result from reciprocal averaging and show species ranges (black columns) hiding embedded absences. Matrix B shows the incidence matrix for the HD trait with the species ordered as in matrix A and with sites (estuaries) ordered after falling salinity from the top and with gradient end values indicated. Horizontal solid line indicates the 18 psu limit. Species recorded from the low saline SW Baltic Sea, south of the Danish Straits, are indicated by grey boxes. Black boxes indicate species never recorded south of the Danish straits i.e. at salinities < ca. 18 psu. Matrices with estuary id and species names are given in S2 File.

Fig 3. Plots of β_sim and β_nes for each trait group (panel rows) vs 4 predictors (panel columns).

Solid regression lines indicate significant effect (P<0.05) and broken line no effect (P>0.05) of the predictor using marginal tests. HD = High dispersive, ID = Intermediate dispersive and LD = Low dispersive trait groups. Salinity difference = Absolute difference between estuaries of average estuary salinity, Entrance difference = Log10 of Absolute difference in estuary entrance width (km), Area difference = Log10 of Absolute difference between estuary areas (km²), Distance = the shortest water way distance (km) between the centers of the estuary entrances.

Fig 4. Results of marginal and sequential tests in DistLM modeling of β_sim with three predictors (A = Salinity difference, B = Log10 Entrance width difference and C = Log10 Estuary area difference) of paired samples from different estuaries.

HD = High dispersive, ID = Intermediate dispersive and LD = Low dispersive species trait group. Variance proportion = the proportion of total variance explained by the predictor (S²). Top sequence (A,B,C) shows results from marginal tests and the following three sequences (A1,B1,C1 and B2,C2,A2 and C3,A3,B3) show results from sequential tests, where A1, B2 and C3 are first fitted to data and B1,C1,C2,A2, A3, B3 gives the remaining independent variance explained after the previous variable has been fitted. Dashed vertical line indicates the total variance proportion explained by the variables. *** = P = 0.001, ** = P <0.01, * = P < 0.05 and ns = P > 0.05.

Fig 5. Results of marginal and sequential tests in DistLM modeling of β_nes with three predictors (A = Salinity difference, B = Log10 Entrance width difference and C = Log10 Estuary area difference) on paired samples from different estuaries.

For more information see legend of Fig 4.

Fig 6. Plots of β_sor vs salinity difference of the three dispersal groups.

a): Beta vs differences between estuaries and b): Beta vs differences between estuaries and adjacent open sea.

Fig 7. Alpha of dispersal trait groups at two spatial scales plotted against average estuary salinity.

a): Sample scale and b): Estuary scale. Regression lines are from ordinary linear regression (OLR, n = 16). ** = P<0.01, * = P<0.05, ns = P>0.05).