Human interactions with tropical environments over the last 14,000 years at Iho Eleru, Nigeria

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Jacopo Niccolò Cerasoni ([email protected]).

Materials availability

The faunal and archaeological materials from Iho Eleru that were analyzed in this study are currently curated at the Max Planck Institute of Geoanthropology, Jena (DE). The paleobotanical materials from Iho Eleru that were analyzed in this study are currently curated for further analysis at Goethe-Universitat, Frankfurt (DE).

Radiocarbon dating (Table S4 and Figure S1)

Sampling and dating were carried out by the Curt-Engelhorn-Center Archaeometry gGmbH. For the bone samples collagen was extracted (modified Longin method), purified by ultrafiltration (fraction >30kD) and freeze-dried. For the plant macroremains pre-treatment using ABA-Method (Acid/Base/Acid, HCl/NaOH/HCl) was used. The insoluble fraction was used for further treatment. Following pre-treatment, the sample material was combusted to CO2 in an Elemental Analyzer (EA). CO2 was then converted catalytically to graphite. ¹⁴C were analyzed using a MICADAS-type AMS system in-house. The isotopic ratios ¹⁴C/¹²C and ¹³C/¹²C of samples, calibration standard (Oxalic Acid-II), blanks and control standards were measured simultaneously in the AMS. ¹⁴C-ages were ormalized to ¹³C = −25‰ and calibrated using the dataset IntCal20 and software SwissCal (L. Wacker, ETH-Zürich). Calibration graphs are generated using the software OxCal (small deviations between SwissCal and OxCal results are possible) (Table S4).

Archaeobotanical analysis (Table S6)

The archaeobotanical samples were excavated from Shaw and Daniels¹ from soil sieved with a mesh-size of 0.635 cm (0.25 inches) and mostly labeled (wood) charcoal sample. The samples contain fragments of wood charcoal and/or other arbonized botanical macro-remains. Many wood charcoal fragments show breaks that have occurred during sieving and storing and are too small for anthracological analyses. Wherever possible wood charcoal was analyzed using a reflected light microscope (Leica DM4000M) at different magnifications, 50× – 500x, in dark and bright field, after manually fracturing the charcoal fragments along the three planes: transverse, longitudinal tangential and longitudinal radial. The identification process follows the steps described in Höhn & Neumann⁵⁴ and relies on the Inside Wood database,⁵⁵ the wood reference collection of the Goethe University Frankfurt (JWGw), and wood anatomical atlases.^56,57,58 The names of the charcoal types are given in small capitals to clearly discriminate the anatomical charcoal types from the botanical taxa with which they are not necessarily identical.^54,59,60 Most type descriptions are published and cf. PIPTADENIASTRUM AFRICANUM resembles PARKIA spp. But has some septate fibers in addition,^54,60,61 ZANTHOXYLUM spp. Is documented in Figure S2. Carpological macro-remains were identified with a dissecting microscope (Leica S6D). Identification was achieved by comparison with specimens in the modern reference collection of the Laboratory of African Archaeobotany at Goethe University, Frankfurt am Main and with carbonized carpological material from Dibamba, Cameroon, identified by Stefanie Kahlheber (unpublished).

Phytolith analysis method

The archaeobotanical samples containing sediment within their original storage bags were processed for sediment isolation using sieve meshes, with varying grid sizes between 1 and 0.25 cm. For the phytolith extraction, the samples were first deflocculated using a sodium hexametaphosphate and warm water solution. For every 10 mL of soil, 100 mL of solution was used (ratio 2:90, sodium hexametaphosphate:water). Any sample which did not correctly deflocculate after 72 h within the solution was manually crushed with a mortar and pestle. Clays were then removed from the samples by gravity sedimentation, with a 1:10 ratio of solution to water, letting it stand for 1 h and then removing supernatant liquids. The clay removal process was repeated 3 times. The resulting clay-free samples were then fractioned by wet sieving and dividing the samples into sands (mesh sieve with grid size of 250 μm) and silts (mesh sieve with grid size of 50 μm). A centrifuge (1500 rpm for 10 min) and distilled water were then used to repeatedly wash the silt samples. The sand samples were put aside and not used for the rest of the process. Given the high contraction of carbonates in the higher spits of the stratigraphy, HCl was used to remove any carbonate particles. This was done by carefully adding HCl to the samples until no reactions were visible. The samples were then rinsed three times with distilled water and a centrifuge at 1700 rpm for 10 min each. Organics were then removed using a 30% H2O2 solution and placing the samples within the solution in a hot water bath at 40°C. Given the low percentage of organic materials in the samples this step was concluded after 2 h in the hot water bath, with the second hour reaching a temperature of 80°C. The samples were then rinsed five times with distilled water and a centrifuge at 1700 rpm for 10 min each. The resulting samples were then processed for any phytolith extraction by flotation using Sodium Polytungstate (SPT-1). A heavy liquid was created using a hygrometer at a specific density of 2.3 g/mL. The SPT-1 solution was then mixed with the samples and centrifuged at 1700 rpm for 5 min. The materials which were floating at the top were then removed and then rinsed three times with distilled water and a centrifuge at 1700 rpm for 10 min each. Next, the samples were dried using acetone, which was mixed with the samples (approximately 10 mL for 1 mL of soil), stirred and centrifuged at 1500 rpm for 5 min. Supernatant acetone solution was removed, and the process repeated twice more. The final samples were then left to dry in a fume cupboard for 2 nights. Finally, the resulting materials were mounted on slides by mixing silicon oil with the samples, placing a few drops of the solution on each slide, and then covering with the slide cover and using clear nail polish to seal them. The slides were analyzed with a Olympus BX53M microscope at 40–100× magnification.

Taphonomic and taxonomic identification of vertebrate faunal assemblage (Tables S7 and S8)

Shaw and Daniels¹ report using ¼ inch sieves for excavated material, and sorting all vertebrate bone fragments according to context. To the best of our knowledge, all sieved material was preserved and no unidentifiable bone fragments were removed from the curated materials. The bone fragments were not cleaned in the present study, as no loose sediment adhered to bone surfaces. The initial phase of taphonomic recording was completed using an Olympus 10–40X zoom binocular microscope with high incident light. For final verification of cut marked and percussion marked bones, SEM (Scanning Electron Microscope) imaging was completed using a JEOL InTouch Scope JSM-IT100LA compact SEM Each bone fragment was analyzed for surface modification. This microscopic method of surface modification recording has shown 95% accuracy in blind tests,⁶² and allows zooarchaeologists to distinguish between cut marks, percussion marks, and tooth marks. Trampling and biochemical marks,^63,64 such as chemical, insect, and root etching,^65,66,67,68 were also recorded. Burning severity was recorded using bone surface colors that indicate burning in either a reducing or oxidizing atmosphere as: no burning, light (dark brown to black color), medium (black to dark gray color), heavy (gray color with shallow cracks), or calcined (white color with deep cracks). Burning extent was recorded as the proportion of a bone fragment displaying signs of burning. For each bone and tooth fragment that was not covered in an adhering carbonate matrix, the following variables were also recorded (see Table S7): measurements (length, width, thickness, or height for tooth crown height), skeletal element, element portion, side (left of right), taxonomic identification to the lowest level, circumference of fragment,⁶⁶ surface visibility, exfoliation of bone surface, dendritic etching,^65,68 pocking, sheen, smoothing, cut marks, percussion marks, percussion notches, tooth marks, tooth notches, rodent gnawing,⁶⁵ weathering,⁶⁹ fracture outline and angle,⁷⁰ presence of fresh break(s) from excavation or post-excavation,⁷⁰ and color. Taxonomic identifications of the vertebrate remains were completed using measurements and photographs collected in 2013 from modern reference collections stored at the American Museum of Natural History in New York. Published reference material^{71,72,73,74,75,76} was also used extensively to aid in vertebrate species identifications.

Scanning electron microscopy

SEM creates a high-resolution image of a surface by scanning it with a focused beam of electrons. In order to closely evaluate characteristics of the cut marks and other surface features, morphological analysis of the specimens was carried out in a scanning electron microscope (SEM) Jeol JSM-IT100 at the Microscopy and Paleobotany Laboratory of the Department of Archaeology at the Max Planck Institute of Geoanthropology, Jena.

Samples were fixed on a 32 mm plain specimen holder with a conductive one-sided copper adhesive tape. For artifacts larger than 30 mm we used a specimen holder with a plain station; artifacts smaller than 30 mm were mounted on either aluminum cylinders or aluminum pin stubs. All samples were imaged uncoated. The SEM was operated in a high vacuum mode at an accelerating voltage of 10 kV. Specimens were visualised using an in-lens secondary electron detector (SED) at magnifications from 22× to 250x. Depending on the size of the artifact, we attuned working distance (WD) from 11 to 18 and probe current (P.C.) from 30 to 44. Brightness, contrast, stigma, and focus were adjusted manually for each sample in the InTouchScope program.

Stable isotope analysis

Enamel sampling for δ¹³C and δ¹⁸O analysis was conducted at the Max Planck Institute of Geoanthropology (MPI-GEA). Tooth samples were first cleaned by air abrasion to remove surface contaminants. Roughly 8 mg of enamel powder was then drilled from the buccal surface of the tooth to gain a sample representing the full growth period of the tooth. The drilled enamel powder was then pre-treated with a 1% NaOH solution for 1 h. Samples were then rinsed, vortexed, and centrifuged three times with MilliQ water. 0.1M acetic acid was then added to the samples for 10 min followed by the rinsing process with MilliQ water. After pre-treatment, samples were frozen overnight before being freeze dried for 4 h. Samples were analyzed on a Thermo Gas Bench 2 connected to a Thermo Delta V Advantage Mass Spectrometer. Roughly 3 mg of each sample was weighed into borosilicate glass vials. The vials were then flushed with helium at 100 mL/min for 10-min 20ul of 100% phosphoric acid was then added to each sample and left to react for 1 h. Samples were calibrated using a three-point calibration with international standards IAEA NBS 18: δ¹³C −5.04‰, δ¹⁸O −23.2‰, IAEA 603: δ¹³C + 2.46‰, δ¹⁸O −2.37‰, and IAEA CO8: δ¹³C −5.764‰, δ¹⁸O −22.7‰. Replicate precision of standards was used to measure machine error where δ¹³C ± 0.2‰ and δ¹⁸O ± 0.2‰. Overall measurement precision was studied through the measurement of repeat extracts from a bovid tooth enamel standard (n = 20, ±0.2‰ for δ¹³C and ± 0.4‰ for δ¹⁸O). Generally, tropical environments are expected to produce relatively low δ¹⁸O values due to the high intensity of rainfall⁷⁷ although values are heavily dependent on rainfall source.

Palaeoclimate reconstruction (Data S1)

We used the paleoclimate reconstructions based on the Hadley CM3 published in Beyer et al.24 to model changes in climate for the area surrounding Iho Eleru. We used the decimal coordinate location of Iho Eleru to extract climate variable values at 1.5° resolution from the Beyer et al.24 dataset in 1000-year intervals over the last 22,000 years (see Data S1). All climate variable values were then scaled together, and four climate variable values of interest were plotted in figure 6. These include BIO 1 (annual mean temperature), BIO4 (temperature seasonality), BIO 12 (annual precipitation), and BIO 15 (precipitation seasonality). Annual mean temperature and annual precipitation (BIO 1 and 12) were selected to have an average annual representation of the local climate and its effect on the local environment as precipitation and temperature are two main factors for vegetation growth and resilience. Temperature and precipitation seasonality were selected due to their effects on human resilience and agro-pastoral activities. Increased temperature and precipitation seasonality has a negative effect on agro-pastoral communities, as crop cultivation and livestock husbandry can be negatively impacted by drastic increases or decreases in temperature and precipitation.

Synthesised regional palynological (NAP) record modeling (Data S2)

To provide additional context to the faunal, anthracological, and macrobotanical remains at Iho Eleru, multiple pollen records from Nigeria, Benin, and the Atlantic Ocean were summarized into indices tracking regional trends in vegetation cover. Pollen records were acquired from an archived version of the African Pollen Database64, harmonised, standardised, and plotted in the R statistical computing environment (R Core Team, 2021). The code, data, and a detailed R markdown document are provided in supplementary Data 4. Using a custom function, the datasets are harmonized with a master reference. This includes updating old taxonomic nomenclature as well as addressing spelling errors. Anne-Marie Lézine currently maintains the master list, which is a product of the African Pollen Database.⁴⁰ Data harmonisation includes associating pollen taxa with known PFTs (Plant Functional Types). Plant functional types can be used to group pollen taxa by growth form (arboreal trees/shrubs, herbaceous vegetation, lianas, arboreal palms, etc.) and make comparisons of records from different regions. Pollen records were harmonized and percentages of major plant functional types were calculated based on the total number of pollen identified (Figure S6). Our goal is to assess the relative proportion of forest and herbaceous vegetation cover (grasslands and savanna) and arboreal trees/shrubs and herbaceous PFTs are being used as proxies for these broad types of vegetation cover. Similar methods have been applied to regional assessments of vegetation change both for the African continent. After summarizing these records by PFT and calculating the percent of each major group, we compare the datasets by setting them on an equal scale using Z-scores, which represent the results as the difference from the mean in SD units. By doing so, we can compare fluctuations with small and large amplitudes on a uniform scale (Figures S3 and S4). The APD entry for Lac Sélé did not contain a radiocarbon chronology, so this code will either compute one using the Rbacon package or use an existing one. In order to create a single index to compare with the finds at Iho Eleru, we aggregated the Z-scores of arboreal and herbaceous PFTs from all of the records and binned them by chronological intervals. In Figures S7 and S8, the sample density for chronological bins representing 500 and 100 years are compared. This shows that, using 500-year bins, the minimum number of samples in a bin (at 20,000-19500 years BP) is two. The maximum is 14 and all of the Holocene bins (11,000 years BP to present) are represented by a minimum of six data points. amples were binned to 500-year intervals and the chronology was limited to the last 12,000 years. In order to evaluate whether our smoothing results were reasonable, multiple smoothing methods were applied to summarized results from both herbaceous and arboreal PFTs. All of the smoothing functions were calculated with 24 knots. Spline smoothing was calculated using R’s smooth.spline function. Kernel-density smoothing was calculated using R’s ksmooth function. Smoothed splines were also calculated with the ss function in the npreg package. These methods produced consistent results and produce a reasonable index of general patterns in vegetation cover as reflected in the representation of pollen in regional pollen records. n order to visualize our uncertainty and as a final check on the smoothing methods, the smoothed results are presented below with box-and-whisker plots of the binned sample results (Figure S8). This visualization reveals where outliers may have an over-sized influence on the curve (at 7500-7000 years BP and 5000-4000 years BP) and in which direction.

Historiographic methods

The site name (Ihò Eléérú) means “Cave of Ashes” in Yoruba language. Throughout this article the Yoruba version of the name has been anglicised into “Iho Eleru” for ease of spelling. The site was first published by the name “Iwo Eleru”, following an incorrect anglicised translation of the original Yoruba name. The Iho Eleru rock shelter was first reported as part of a large-scale survey of the hilly landscapes around Akure, conducted by Chief officer J. Akeredolu of the Department of Antiquities from Benin, Nigeria, in 1961. Following this report, in 1963 Thurstan Shaw led an exploration of the rock shelters and caves previously reported by J. Akeredolu. During the exploration of the research area, the officer of the Department of Antiquities, G. Connah, informed T. Shaw of the existence of Iho Eleru. Following the initial exploration, T. Shaw and S. Daniels organised and conducted excavations at the rock shelter between 1964 and 1965. Following two decades of research, where the material culture was studied, a final report of the excavation findings was published in 1984. Since then, several studies have been published concerning the human remains retrieved from the original Shaw and Daniels excavation, with only one article discussing the archaeological context since the original report.

Ecological survey methods

Today, the Iho Eleru rock shelter is situated approximately 50 km South of the modern boundary of the Guineo-Congolian rainforest phytogeographic region, with the local landscape being characterised by pioneer forest formations with a low to medium canopy cover (5-10m). Just North of this region is situated the Guinean Savanna zone, which is a large expanse of land extending West to East throughout West Africa and is characterised by forest islands and wooded savanna vegetation. The modern landscape is extensively modified by human activity, with agroforestry production focused on the cultivation of yams, bananas, cocoa, and cola nuts.

Geological and stratigraphy methods (Table S1)

The site is located within a lowland composed by eroded outcrops of ancient igneous deposits, with the presence of river terraces and well oxidised sandy deposits underlain by sedimentary formations of laterite. Within the region, various formations of complex metamorphic outcrops (crypto-crystalline silicates and siliceous stone formations) occur. Furthermore, Shaw and Daniels¹ (pg. 190, Figure 3) show the presence of quartzite veins which appear approximately 5-10 km West of Iho Eleru. Furthermore, various deposits of considerably eroded quartzite and vein quartz cobbles are present within local seasonal river valleys.

Iho Eleru is a rock shelter found toward the base of an igneous inselberg, found on the western margin of the Ikere Batholith. Topographically, the main human activity area would have been behind the drip line area of the shelter, within the main upper leveled platform area (trenches D & F, XVII-XXVII). On its southern and eastern edges, steep inclined taluses slope downwards. On its western edge, a very steep incline slopes upwards toward the upper levels of the inselberg, where other sheltered areas are found. On the northern edge of the platform area, the rock shelter closes following the two overhanging igneous outcrops, forming a small tunnel (Central Tunnel) with an SW-NE orientation. The stratigraphy at Iho Eleru shows very irregular igneous beds upon which the archaeological sediments sit. The depths of the stratigraphy vary from 0 cm (subsurface outcrop) to over 2m in depth (Trench S). The sediments have post-erosional origin, where materials from the upper levels of the inselberg fall during particularly wet periods, depositing within the platform area and surrounding slopes. In turn, similar activities cause an erosion of the platform area sediments, causing the sloped areas to have inverted stratigraphies. For this reason, during our re-analysis of the chronometric dates of the site, only charcoal samples that were retrieved in the upper level area were selected. Nevertheless, the resulting ¹⁴C dates still show discrepancies and inverted dates. This has been interpreted as being caused by heavy erosional processes, mostly during wet season periods, which cause Iho Eleru rock shelter to become a main channelling area for water moving downstream of the upper levels of the inselberg.

Shaw and Daniels state that the stratigraphical excavation was conducted with an initial layer (spit 1) that was excavated with a total depth of 25 cm, and subsequent spits of 15 cm. However, later descriptions of the excavation show that each and every spit per square was considered to be 15 cm in depth. This contradiction might be caused by a use of a 25 cm deep spit only in specific areas of the rock shelter, however such information could not be found by the authors of this paper. For these reasons, a consistent depth of 15 cm per spit was considered and applied for the upper plateau area which was evaluated in this study, and which is consistent with the original descriptions, interpretations and excavation diagrams by Shaw and Daniels (1984).

In the platform area of the rock shelter, Shaw and Daniels describe 4 main sediment types:

These units appear to be very distinct in terms of their sedimentary properties and likely reflect changes in the local environment surrounding the rock shelter. We propose some thoughts regarding possible site formation factors operating at Iho Eleru and describe how these may relate to changes observed in our paleoclimatic modeling.

Material culture imaging and figure design

Photographs of the faunal remains were taken following the protocol for photography of artifacts and small object (SOAP protocol).⁷⁸ Figures 1, 2, 3, 5 were designed and composed using Adobe © Illustrator © 2022. The maps on Figures 1A and 1B were originally designed on QGis ©. For the maps a digital elevation map of the region was merged and developed (utilising WGS 84 projection) from 1 km-resolution digital elevation models (DEMs),⁷⁹ and then categorised based on elevation and correlated with topographic bioregions. Map b on Figure 1 was created by mapping the distribution of modern ecoregions¹⁰ for the landscape surrounding Iho Eleru and geolocating the respective archaeological sites. Figure 4 was designed and created using Adobe © Photoshop © 2022.