Resilient biotic response to long-term climate change in the Adriatic Sea

Abstract

Preserving adaptive capacities of coastal ecosystems, which are currently facing the ongoing climate warming and a multitude of other anthropogenic impacts, requires an understanding of long-term biotic dynamics in the context of major environmental shifts prior to human disturbances. We quantified responses of nearshore mollusk assemblages to long-term climate and sea-level changes using 223 samples (~71,300 specimens) retrieved from latest Quaternary sediment cores of the Adriatic coastal systems. These cores provide a rare chance to study coastal systems that existed during glacial lowstands. The fossil mollusk record indicates that nearshore assemblages of the penultimate interglacial (Late Pleistocene) shifted in their faunal composition during the subsequent ice age, and then reassembled again with the return of interglacial climate in the Holocene. These shifts point to a climate-driven habitat filtering modulated by dispersal processes. The resilient, rather than persistent or stochastic, response of the mollusk assemblages to long-term environmental changes over at least 125 thousand years highlights the historically unprecedented nature of the ongoing anthropogenic stressors (e.g., pollution, eutrophication, bottom trawling, and invasive species) that are currently shifting coastal regions into novel system states far outside the range of natural variability archived in the fossil record.

Keywords: Climate Change; Conservation Paleobiology; Glacial-Interglacial Cycle; Italy; Mediterranean Basin; Mollusk.

FIGURE 1

Conceptual framework. Idealized outcomes representing the patterns of community response to glacial–interglacial changes at the regional scale, evaluated by means of ordination analyses (NMDS) and correlation between abundances of species (black: pairwise comparison between the two interglacial units; green: comparisons between glacial and interglacial units). Each column shows one of the three idealized scenarios. Persistent pattern (a and d): Communities maintain species composition and diversity through environmental perturbations even though populations of constituent species shift spatially in concert with sea‐level changes. Resilient pattern (b and e): Communities shift to an altered state during the glacial period but return to previous composition with the re‐establishment of interglacial conditions. Stochastic pattern (c and f) unique species associations characterize communities from all three‐time periods

FIGURE 2

Gradient and rank correlation analyses: (a) NMDS ordination of nearshore samples containing at least 25 specimens (see also Figure S5 for an NMDS output based on sample size threshold of 60 specimens). Relative abundance of species was fourth root transformed. Samples are color‐coded according to the climatic interval: green—current interglacial (CIG), light blue—last interglacial (LIG), and dark red—last late glacial (LG). The size of each point is proportional to sample size. Convex hulls delimit the ordination space occupied by each group of samples. (b) Correlation between NMDS axis 1 sample scores (NMDS1) and species richness rarefied to 25 specimens. Standardized species richness for relatively small samples tends to be primarily driven by evenness, so the two measures are strongly correlated. (c) Correlation between NMDS1 and relative abundance of Mediterranean‐to‐Lusitanian and West African species recovered in each sample. (d) Correlation between NMDS1 and the sample water depth estimates based on species bathymetric preferences (see Materials and Methods for details). In b–d panels, rank correlation coefficient ρ is shown also for NMDS axis 2 sample scores

FIGURE 3

Distribution of pairwise Bray–Curtis (BC) distances between samples representing glacial and interglacial assemblages. (a) Current interglacial and last late glacial (CIG‐LG, based on 1170 pairs of compared samples). (b) Current interglacial and last interglacial (CIG‐LIG based upon 975 pairs of compared samples). (c) Last late glacial and last interglacial (LG‐LIG based on 270 pairs of compared samples). Red arrows mark the location of the observed mean values BC distances for each frequency distribution of the three pairwise comparisons. The x‐axis reports BC dissimilarity range, zero value indicates that two samples have the same faunal composition, one no species in common. In green sampling distributions of means based on randomization (based on 1000 iterations), under the null model that the samples came from the same system. Pairwise comparisons are based on the same species relative abundance matrix as the one used for the NMDS (n ≥ 25 specimens and rare species removed)

FIGURE 4

Pairwise comparisons of species total abundances (total counts in pooled data from each interval). (a) Current interglacial and last late glacial (CIG‐LG, upper left panel, x‐axis: CIG, y‐axis: LG). (b) Pleistocene interglacial and last late glacial (LIG‐LG, upper right panel, x‐axis: LIG, y‐axis: LG). (c) Holocene and Pleistocene interglacials (CIG‐LIG, lower panel, x‐axis: CIG, y‐axis: LIG). Species total abundances have been log‐transformed. The output of the randomization model based on 1000 iterations highlights the portion of two‐dimensional space in which the points should fall under the null model of a homogenous system. Spearman's rank correlation (ρ) for each pairwise comparison is reported on each panel; it is significant only for the interglacial pairwise comparison (i.e., CIG‐LIG, p < .001; see also Table S6)

FIGURE 5

Comparisons of species total relative abundances grouped according to their biogeographic distribution. (a) Current interglacial—CIG; (b) last late‐glacial—LG; (c) last interglacial—LIG. Information on the geographic range of mollusk species is after Poppe and Goto (1991, 1993). Abbreviations for biogeographic affinity of species distribution: BOR = species occurring in the Mediterranean, Lusitanian, and Boreal provinces; COS = species of cosmopolitan distribution (i.e., occurring from West African until Boreal provinces); MED/LUS = species occurring in the Mediterranean and/or Lusitanian provinces; WAF = species occurring in the Mediterranean, Lusitanian, and West African provinces