Ecological dynamics in palaeo records provide insight into climate-change processes occurring over longer timescales than those recorded in modern or recent-historical observations. This long view is becoming increasingly important as current climate change accelerates and moves the Earth system into states that are unprecedented for the Quaternary Period (past ca. 2 M years), as responses of the biota to dramatic changes of the distant past expand the envelope of observations used to anticipate future change. 

Palaeo data have been used to assess the performance of a range of models that aim to project future conditions, from large-scale climate models to detailed process models. The palaeo data typically take the form of long time-series of biodiversity in a specific location, or spatially extensive summary maps of change, based, for example, on fossil-pollen records. Species-envelope models project future spatial distributions in changed environments. These generally rely on patterns of association with contemporary physical conditions, characterised by climate and soil variables; it is less easy to determine how distributions are modified by biological processes, such as dispersal characteristics or biotic interactions.  Such interactions are difficult to identify from palaeo data based just on ecological principles without good knowledge of taxon-specific ecologies. 

What if the qualities inherent to a long time series of floristic change can reveal different modes of interaction amongst taxa, to reconstruct a temporal understanding of processes such as competition, niche construction and extinction cascades?  A high-quality, taxon-rich dataset accompanied by knowledge of changing environmental conditions would be a necessary first requirement.

By retrieving ancient DNA from a variety of Quaternary deposits (sedimentary DNA: ‘sedaDNA’), we can learn about past floristics in greater detail than is normally possible using pollen or macrofossils. Pollen-assemblage diversity is often restricted by poor—or even no—representation of insect-pollinated taxa. Macrofossil assemblages, though often species-rich, can be restricted spatially, for example, if they are derived from animal middens, and they are sometimes virtually absent from lake sediments. Lake-sediment DNA records could unlock a “Goldilocks” approach to floristic reconstruction, having the taxonomic resolution of high-quality macrofossil records and representing a clearly defined, observable spatial area: the lake’s hydrologic catchment.

Of course, such records have their own constraints and biases: for example, the effect of sediments on DNA preservation, biased representation by DNA fragments of a taxon’s biomass, the way the material is prepared in the laboratory, and the completeness—or otherwise—of look-up databases. It turns out that arctic and sub-arctic localities have produced some of the most compelling sedaDNA records to date. Cooler temperatures undoubtedly play a role in slowing degradation of DNA prior to its incorporation in a stable state of preservation, and highly minerogenic sediments, typical of northern ecosystems, appear to provide a particularly good matrix for molecular attachment. Furthermore, floristic databases are typically more complete where regional floras are small, and this is particularly the case with the high northern latitudes. Relatively complete DNA catalogues have been developed for arctic and sub-arctic regional floras.   

One of the longest and most detailed records of changing floristics (and by inference, vegetation communities) comes from the largest and deepest lake in the Polar Urals of Russia. The sediments of Lake Bolshoye Shchuchye  were cored by a team from the University of Bergen led by John-Inger Svenson and Haflidi Haflidason. Their core has yielded a suite of palaeoenvironmental records covering the past 25,000 years from the peak of the last glaciation until recent time. The sedaDNA record, comprising 153 samples, was developed by Charlotte Clarke, then a postgraduate student at the University of Southampton, in collaboration with a team at the Arctic University of Norway led by Inger Greve Alsos. It contains 162 taxa of vascular plants, ferns and fern allies and bryophytes. It closely matches an accompanying pollen record developed by Anne Bjune at the University of Bergen, and it shows a gradual increase in richness and diversity as changing climate following after the last glaciation led to migration and establishment of new suites of taxa.  The accumulation of diversity waxed and waned over the period of record.  Such a long, highly detailed and taxon-rich time series invites investigation of more subtle patterns—such as the nature of species interactions and sequences of accumulation and loss.

Using an innovative statistical approach based in information theory, developed by Patrick Doncaster at the University of Southampton, we assessed how sedaDNA assemblages vary through time, particularly whether the observed sample composition progressed from one time-step to the next in an orderly or disorderly fashion. To put it somewhat simplistically, was there an unpredictable coming and going of taxa, or did taxa that joined a growing community establish within it, or did those that were lost from a shrinking community subsequently not return to it? Computing a statistical time series with the somewhat challenging, but accurate, name of Relative Entropy of Community Assembly (‘RECA’) allowed us to visualize the orderliness, or otherwise, of temporal trends in floristic composition. The analysis tests for signals of orderliness amongst the potentially random appearances and disappearances of taxa typical of any palaeorecord. The resultant pattern can then be compared against time series of independently derived palaeoenvironmental data (Figure 1).

Negative values of RECA indicate periods when community assembly was more ordered than disassembly (species entered and stayed); this is characteristic of emergent processes of a community—competition, facilitation, and niche construction, i.e., endogenous processes that can exert control over community function. Such ordered gains may be counterbalanced by relatively more disordered losses that characterize occasions when a tightly structured community collapses. We found that the negative RECA state is characterized by assemblages of high distinctiveness in the temporal record. In contrast, when RECA is positive, the losses are more ordered (predictable) than the gains, as happens in extinction cascades. These occur when environmental conditions move beyond the envelope of variability in which the community developed, for example, during the onset of a marked shift in climate. Self-organization gives way to unpredictable arrivals and departures of taxa, and those that drop out tend to stay out, reflecting the dominance of exogenous factors.

The climate in Northwest Europe was characterized by strong fluctuations at the end of the glacial period, beginning about 14,000 years ago. Values of the Greenland ice-core isotope record the Bølling-Allerød warming, followed by reversion to cold, dry conditions during the Younger Dryas period, and then strong warming into the Holocene interglacial about 11,500 years ago (Figure 1). Comparison of the RECA values from Lake Bolshoye Shchuchye with the general progression of climate change shows downward shifts that signal niche construction and endogenous organization following periods of climate change. The most distinctive shift, from strongly positive to strongly negative, occurs a few millennia into the Holocene, and the negative status prevails until recent time, reflecting both a stabilization of the climate and establishment of a dominance hierarchy related to the complex structure and species interactions within woody-dominated Holocene plant communities. 

Prior to this long dominance of self-organizing community assemblies, positive values of RECA indicate a period of disruption with an abrupt onset, which coincides with the extremely rapid temperature rise at the end of the Younger Dryas. This warming is often cited as the past temperature change that best approximates the rate and magnitude of anthropogenic warming expected over the coming decades in the absence of concerted mitigation efforts. The historical disruption continued for several millennia, reflecting an extended period of adjustment subsequent to the initial temperature rise.

Before the Holocene, the RECA curve shows two sharp increases indicating disruption, probably due to environmental change, about 16,000 and 14,000 years ago, respectively. The former, while echoing changes in Greenland (and hence probably more broadly in northern Europe), may also reflect a major local change, as glaciers disappeared from the higher elevations of the catchment at this time, which would have affected catchment climate and hydrology. About 14,000 years ago, a rapid temperature increase, almost as rapid as that at the end of the Younger Dryas, marked the onset of the Bølling-Allerød interstadial and warming-related compositional disruption. Subsequently, temperatures gradually declined into the Younger Dryas cold stage, and communities eventually stabilized, as indicated by negative RECA values, this time as cold-adapted assemblages. 

What can these findings tell us about contemporary climate change and northern plant communities? Current and projected rates and amplitudes of climate change are higher than those experienced at the site during any period of the last 20,000 years, with the possible exception of rate of the Younger Dryas warming. The RECA time series points to the possibility that warming might force a state change towards exogenous regulation of community structure for the first time in over 8000 years. Previously, cold-adapted arctic-alpine taxa were out-competed by woody taxa at lower elevations in the catchment, but the future will see both sets of taxa confronting temperatures and weather extremes possibly without precedent in the past two million years, implying a long period of plant community instability that will have an impact on broader ecosystem structure and function. 

The RECA methodology should be applicable beyond the example provided here, to other high-diversity time-series such as marine and freshwater microfossil records and DNA-based time series of microbial dynamics. For microbes, it may be possible to infer the ecological functions of taxa from the (dis)orderliness of waxing and waning community assemblages, even with little direct knowledge of taxon-specific ecologies.

Read more about the Bolshoye Shchuchye DNA record in Scientific Reports https://doi.org/10.1038/s41598-019-55989-9  

Photo credit: John-Inger Svenson and Haflidi Haflidason