Development and Application of Analyses to Eastern Tigrai 

Eshetu and Terwilliger’s early calibration work (see Developing Analystical Methods) had included stops to sample soils in Tigrai.  There we gained an immediate appreciation of the potential in Tigrai to address key questions about effects of environmental variation on the formation and decline of complex human settlements (see Research Goals). We initiated the palaeoenvironmental research group and returned to eastern Tigrai to develop local-scale records of environmental change.

Reconstructing palaeoenvironmental variations first required identifying a source of proxies that is continuous over the time intervals of interest.  A high-resolution example is annual tree rings where the age of the outermost ring is known. The few trees from this predominantly deforested, tropical region could not provide a long enough environmental record to couple with Pre-Aksumite and Aksumite times and produced false as well as annual rings.  Lake cores containing datable materials are among the most ideal archives of proxies spanning several millennia during the Holocene.  Nonetheless, there are presently no lakes in Tigrai and buried palaeolakes had dried long before the beginning of Pre-Aksumite times.

Although only limited research had been done to reconstruct local-scale palaeoenvironmental history of Tigrai; happily, a potential archive for palaeoproxies spanning millennia had been identified: the deep gullies dissecting fluvial soils of the Tigrai plateau. The walls of gullies expose soils that demonstrated promise as well-ordered or at least identifiable progressions of deposition over time.  Aided by the authors themselves, we selected a series of soil sequences in gullies between Mek’ele and Adigrat that published studies suggested would span Pre-Aksumite through Aksumite times and, in some cases, other parts of the Holocene as well. No archaeological work was known, however, near these initial sequences.

Reconstructing past vegetation is a fundamental part of recording palaeoenvironments but we initially believed that we would not be able to get more information than the relative contributions of C₃ and C₄ plants that could be interpreted from δ¹³Csom analyses.The materials that have yielded the most detailed palaeobotanical records include pollen and charcoal.  Informative pollen records tend to come from lakes. Charcoal was not thought to be present in soils within Pre-Aksumite and Aksumite hegemonies. Furthermore, we had no expertise in charcoal identification and there was no charcoal reference collection in Ethiopia.  Finally, there had been no botanical inventories before the region had been deforested and assumptions about its natural vegetation were based on inventories at similar elevations elsewhere in Ethiopia.  What would become our study areas were thought to have been dominated by forests of the two fire intolerant gymnosperms: East African juniper (Juniperus procera) and Podocarp (Afrocarpus falcatus), although there is no sign of these species today.

Tsige Gebru, then a Master’s student in the group, saw abundant charcoalised wood fragments in the wall of the first and all subsequent gullies we observed (Figure 4).  She extracted charcoal from the sequences we decided to study, started a charcoal reference collection and used it to identify what she had extracted.  She found that even during the Holocene African Humid times several millennia before the Pre-Aksumite Period, forests were not exclusively juniper and podocarp dominated.  Angiosperm species indicative of fire-tolerant forests were also always present and sometimes dominated sample sets (Figure 5). This and the ubiquity of charcoal in all of the soil samples indicated a rich role of a perhaps variably human-influenced fire history on the region’s land cover.

In addition to easily analysed but coarse-scale changes in rainfall from soil δ¹⁵N values, we explored the possibility of obtaining high resolution information about rainfall from stable hydrogen isotopic (δ²H or δD) analyses of specific, land-plant derived n-alkanoic acids (n-Cn) in soil organic matter.  These compounds are the final products of decomposition, very resistant to further chemical or isotopic change and are thought to resist moving from the soil in which they were produced.  Although they were much less plentiful than in lakes, we were able to extract and analyse δ²Hn-Cn values of several n-alkanoic acid compounds, the most reliably abundant being Hexacosanoic acid (n-C26). Several findings suggested that analysing δ²Hn-C26 values gave high quality interpretations of variations in rainfall over time. One example is shifts in more than one sequence at times of known changes in rainfall such as the end of the Holocene African Humid period. A related example is high repeatability of trends between sequences over the same time interval. 

Additional proxy analyses showed some counter-intuitive trends from approximately 3600 calendar years before present (cal y BP) to what archaeologists studying around Aksum farther west had concluded was the early part of the Pre-Aksumite polity “D’MT”.  The following results (Figure 6) supported the possibility that much more fire activity had occurred then, than at other times represented by the soil sequence, and that increased fire activity rather than rainfall had influenced an increase in C₄ relative to C₃ vegetation cover.  Specifically, δ¹³Csom values suggested an increase in C₄ relative to C₃ vegetation that δ²Hn-C26 values indicated coincided---against expectations---with a slight increase in rainfall.  Micromorphological image analyses of soil thin sections suggested a higher percentage of burned material then, than at other times (%Char) and % total organic carbon (%TOC) was correspondingly high, likely due to the burned material input. A combination of indices from RockEval analyses (HI, OI) showed the organic matter to be severely altered as could also be a result of high fire activity.

δ²Hn-C26 values suggested that both Pre-Aksumite and Aksumite polities arose during wetter intervals of an arid period that succeeded the Holocene African Humid period. Indications from our early work in Tigray suggest that any roles humans played on land cover differed between emerging Pre-Aksumite and Aksumite Periods.  

The palaeoenvironmental group now needed to learn more about fire history in particular and palaeoenvironments closely associated with dated archaeological sites.