Correspondence / Quaternary Science Reviews 110 (2015) 172e175 Fig. 1. Microsoft Excel graph of the number of the uncalibrated dates from Manning, Timpson (2014) for the time period 11,700e4300 BP. 173 Fig. 2. Microsoft Excel graph of the number of the uncalibrated dates from my more complete database for the time period 11,700e4300 BP. References Table 1 Thermoluminescence dating of Badarian pottery. Labcode Yrs BP SDV Sitename Caton-Thompson and Whittle 1975. Ox131b10 Ox131b11 Ox131b12 Ox131b13 Ox131b9 6400 6640 6460 7445 6280 365 365 475 405 355 Qau Qau Qau Qau Qau 4450 4690 4510 5495 4330 Badari Badari Badari Badari Badari ± ± ± ± ± 365 365 475 405 355 The authors in their references do not refer to some other pub€ n, 1996; Jennerstrasse 8 2002; Wendrich lications (Haas, 1989; Scho et al., 2009; Kindermann 2010; Barich et al., 2012; Phillipps et al., 2012) which have extensive lists of 14C dates. 9. According to the Refernces list, IntCal9 has been used for the calibration of the dates. However nowadays one should use the more recent IntCal13 (Bronk Ramsey, 2014). 10. Manning, Timpson (2014) presume that “by combining all 14 C dates in a region, the fluctuation in the density of dates through time can be used as a demographic proxy to investigate the timing of population change in northern Africa”. However, the density of dates is certainly also a result of the intensity of research in a specific region. For all those reasons the conclusion of Manning, Timpson (2014) should be considered as very preliminary. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.quascirev.2014.12.007. Barich, B.E., Lucarini, G., Gallinaro, M., Hamdan, M., 2012. Sheikh/Bir El Obeiyid: Evidence of Sedentism in the Northern Farafra Depression (Western Desert, Egypt). In: Prehistory of Northeastern Africa, New Ideas and Discoveries, Studies Archaeological Museum. in African Archaeology, vol. 11. Poznan Bronk Ramsey, C., 2014. Oxcal. https://c14.arch.ox.ac.uk/oxcal/OxCal.html. Caton-Thompson, G., Whittle, E., 1975. Thermoluminescence dating of the Badarian. Antiquity 49, 89e97. Haas, H., 1989. The radiocarbon dates from Wadi Kubbaniya. In: Close, A.E. (Ed.), The Prehistory of Wadi Kubbaniya, vol. 2. Southern Methodist University Press, Dallas, pp. 274e279. Jennerstrasse 8 (Ed.), 2002. Tides of the Desert e Gezeiten der Wüste. Heinrich€ ln. Barth-Institut, Ko €nen Besiedlungsgeschichte zwischen Kindermann, K., 2010. Djara. Zur mittelholoza € ln. Niltal und Oase. Heinrich Barth Institut, Ko € pelin, S., 2006. Climate-controlled Holocene occupation of the Sahara: Kuper, R., Kro motor of Africa's evolution. Science 313, 765e768. Phillipps, R., Holdaway, S., Wendrich, W., Cappers, R., 2012. Mid-Holocene occupation of Egypt and global climatic change. Quat. Int. 251, 64e76. €n, W., 1996. Ausgrabungen im Wadi el Akhdar. Gilf Kebir. Heinrich Barth InstiScho €ln. tut, Ko e du Nil et le Sahara oriental: une population Vermeersch, P.M., 2006. La valle historique fluctuante sous l’effet des variations climatiques. C.R. Pale vol 5, pre 255e262. Wendrich, W., Taylor, R.E., Southon, J., 2009. Dating stratified settlement sites at Kom K and Kom W: fifth millennium BCE radiocarbon ages for the Fayum Neolithic. Nucl. Instrum. Methods Phys. Res. B 268 (7e8), 999e1002. P.M. Vermeersch Department of Earth and Environmental Sciences, Katholieke Universiteit Leuven, Belgium E-mail address: pierre.vermeersch@ees.kuleuven.be. © 2014 Elsevier Ltd. All rights reserved http://dx.doi.org/10.1016/j.quascirev.2014.12.007 Response to “Comment on ‘The demographic response to Holocene climate change in the Sahara’, by Katie Manning and Adrian Timpson (2014)” 1. Introduction In volume 101 of Quaternary Science Reviews we published a summed probability distribution (SPD) analysis of radiocarbon DOI of original article: http://dx.doi.org/10.1016/j.quascirev.2014.12.007. data from 1011 Neolithic archaeological sites to examine the demographic response to Holocene climate change in the Sahara. Vermeersch (2015) is highly critical of our approach and makes 10 specific comments, which fall broadly into four categories: simple misunderstandings; inconsistencies in the data; our exclusion of Pharaonic samples; and criticisms of the method. We would 174 Correspondence / Quaternary Science Reviews 110 (2015) 172e175 like to thank Vermeersch for his comments and we appreciate the opportunity to provide clarity on these issues and respond in detail to all points below, with respect to these categories. 1.1. Simple misunderstandings (comments 1 and 6) A couple of Vermeesch's comments raise questions that we feel were fully answered in the publication. We regret if our language caused any ambiguity, and hope to resolve these here. Comment 1 questions the difference between Figure 3, which uses 3086 samples, compared to Appendix A, which has 3287 samples. Different scales of analysis, represented in the different figures, used different subsets of the data. Figure 2 illustrates the full SPD for the entire radiocarbon data set, inclusive of all 3287 samples, whilst Figure 3 illustrates the four sub-regional SPDs. Two hundred and one samples are from sites that were not included in the sub-regional analyses since they were not assigned to one of the four polygons (grey dots shown and explained in Figure 1). For example, the grey dots in West Africa, which are located predominantly to the south of the river Niger, display important social and economic differences to sites in both the Western littoral and Central Saharan polygons e.g. arable farming, technological specialisation and Sahelian pottery motifs (Manning, 2008). It was therefore considered preferable to exclude them from either polygon. Comment 6 includes a statement that we have not specified the time period of interest. We describe in Appendix A that the data ‘ … covers a calendar date range of 14,000e2000 yrs BP’. This is the same time range shown on the time-axis in Figures 2, 3 and 4. 1.2. Inconsistencies with the data (comments 2, 3 and 4) Efforts to clean large datasets (in this case thousands of radiocarbon dates) will never entirely eradicate errors, both human, and those inherited from other sources. The task therefore is always to minimise errors to acceptable levels. Even if we accept all Vermeersch's examples (n ¼ 22) this represents only 0.67% of the dataset, a rate so low that it can be considered a commendation of our data hygiene. An error rate this low will have a negligible affect on the analysis, and is dwarfed by the uncertainty associated with archaeological (or any empirical) sampling. Furthermore we do not agree with all his examples. Comment 2 notes the inclusion of 11 dates with an unknown lab code. These data were deliberately included in the analysis, having been obtained from a published source. The lack of a lab code is an obvious limitation of their provenance, which therefore increases the uncertainty associated with them, at worst adding noise to the signal. Their inclusion therefore promotes a more conservative approach. Comment 3 and 4 criticise the inclusion of five Thermoluminescence (TL) samples, and six duplicate samples. Vermeersch is indeed correct that both TL samples and duplicates should not have been included. Nevertheless the detrimental effects of this are to some extent mitigated by our methods, in two different ways. Firstly the TL samples have extremely large standard deviations, so their probability distributions are low and diffuse across a wide period. As with comment 2 this has a conservative effect of adding some noise to the signal. Secondly, the binning algorithm (further explained in Timpson et al. 2014) somewhat mitigates duplicate samples, since they will by definition be assigned to the same bin, after which their combined probability distribution is normalised. 1.3. Exclusion of Pharaonic samples (comments 6,7 and 8) This appears to be Vermeersch's primary criticism, regarding the inclusion of radiocarbon dates in the eastern Sahara polygon, and the specific culture history of that region. The key objective of our publication was to explore spatio-temporal fluctuations in “Neolithic” populations, and not the adaptive strategies of statelevel societies. This is an important distinction, since the former are characterised by a less intensive economic and political system and one that is essentially restricted by an environment's carrying capacity as a consequence of relatively low technological investment. In contrast, one of the key characteristics of state-level societies is their adoption of what Stuart and Gauthier (1981) termed “Power strategies” i.e. intensive resource exploitation including arable farming, complex technology, extensive trade, social stratification and extended political and economic alliances. These sociopolitical developments remove much of the environmental risk faced by Neolithic populations, and hence provoke a very different set of questions relating to resilience and climatic change (Weiss and Bradley, 2001; Turchin, 2003). We are explicit about this in the publication, stating that our data comes from 1011 “Neolithic sites” and specifically for the eastern Saharan polygon, that “This (abrupt population collapse) may be partly explained by the emergence of Pharaonic civilization along the Nile River leading to a spatial bias in our results as only Neolithic dates are included in this analysis”. Comment 6 notes a lack of predynastic and Pharaonic samples post c. 4000 BP. Vermeersch is correct that more Pharaonic dates are available in the Nile Valley. Similarly there are many more dates available from sites associated with the Garamantian civilization, the Kingdom of Kush and other state level societies, which flourished in northern Africa during the late Holocene, and within our time range of 14,000e2000 yrs BP. However, as we have explained, the objective of this paper was to explore the population history of “Neolithic” populations, and not the adaptive strategies of statelevel societies. Comments 7 and 8 again lament the exclusion of Pharaonic samples, which are present in Vermeersch's own Egyptian dataset, and many of which are published in Kuper and €pelin (2006). The additional Pharaonic samples are irrelevant Kro for the objectives of our analysis, as they would conflate the available Neolithic data with samples derived from Pharaonic contexts, and tell us little about the response of Neolithic populations to the €pelin (2006) is AHP. Similarly the data provided in Kuper and Kro included in our database, but we have simply removed the Pharaonic samples. Whilst many of the dates from the additional suggested references were included, it is possible that some were unintentionally missed in our data collection. The size of our dataset is simply the consequence of practical limits in our research budget. To what extent any empirical sample is fairly representative of the true population is a fundamental problem to be explored through appropriate statistics and inference. Hence, the MCSPD method presented in this paper, specifically tests the degree to which a population curve deviates from what would be expected if the distribution was merely a result of random sampling. In this sense, Vermeersch's description of our data set as “… very incomplete …” and his rhetoric “ How can the database [ … ] be trusted as being representative for the region?” are unhelpful, since by definition all samples are incomplete approximations of the true population. 1.4. Criticisms of the method (comments 5, 9 and 10) In comment 5, Vermeersch suggests the Nile valley should have been excluded from the analysis, and asserts “… the Nile valley population was not subjected to the same environmental conditions as the Saharan population”, and the “River Nile cannot be considered as a Saharan river”. Clearly this is a simple matter of scale. No two points are ever subjected to exactly the same environmental conditions, although we can expect closer points to have more similar conditions. As our analysis was at the sub-continental scale, we defined Correspondence / Quaternary Science Reviews 110 (2015) 172e175 our region of interest, and the River Nile finds itself flowing through it. Ultimately our results suggest that at this scale, both the eastern Saharan populations (inclusive of the Nile Valley), the Central Sahara and the Atlas/Hoggar were subjected to similar climatic drivers. The inclusion of the Nile Valley data is also well justified on archaeological grounds. There are clear affinities in the material culture and economic practices of the earliest Holocene populations both in this region and subsequent settlements to the west, providing some evidence for the Nile Valley being the source of Neolithic expansion into the Sahara (Sutton 1974; Gabriel, 1978; Haaland, 1992). To exclude these samples would essentially remove a key component of Saharan population history, giving the false impression that Neolithic populations did not inhabit the Nile Valley. Furthermore, Vermeersch plays fast and loose with the problem of variation in environmental conditions, by implying that the rest of the Sahara was ecologically homogeneous. This was clearly not the case, and there has been extensive research documenting the ecological and palaeohydrological variation of the Saharan Hololy et al., 2014), as cene (Drake et al., 2011; Francus et al., 2013; He well as the adaptive strategies adopted by local populations in response to that variation (Cancellieri and di Lernia, 2013). At the root of this lies a common issue-the incorrect view that research at different scales are in some way incongruent. Our publication is not a local scale study of regionally specific demographic histories, as Vermeersch appears to have interpreted it. Instead we have provided a broad-scale view of demographic changes at the sub-continental level. By assessing the similarity in regional population fluctuations across the entire width of Africa we were able to show strong support for a climatic cause, without needing to rely on palaeoclimate proxies, and the inherent difficulties they bring at this scale of analysis. Comment 9 correctly notes that a more recent calibration curve has now been published, and it is of course always better to use the most up-to-date data for any analysis. Nevertheless, subsequent reanalysis using the 2013 intcal. curve, has made no difference to the inferences and conclusions and has a negligible effect on the SPD plots. Comment 10 raises the point that the shape of the SPD will be influenced by differences in research intensity in different areas. This is certainly true, not just for this analysis but for any archaeological analysis, since we are always limited to data from where someone decided to excavate. This is one of several biases initially discussed by Rick (1987) when he first proposed constructing summed date distributions, and has since been dealt with in detail by many authors including Timpson et al. 2014. Generally, we are limited in the inferences that can be drawn from regions where there are simply no data. However, a key aspect of the method used in this analysis is to account for ‘ascertainment’ bias in regions where were do have data i.e. to mitigate the effect of single sites with disproportionate numbers of radiocarbon dates. This is achieved using a systematic binning process, which ensures that dates are only assigned to a new bin if there is at least a 200-year gap since the previous date. This addresses the excavator’s ‘ascertainment’ or ‘wealth’ bias, and ensures that each site-phase is equally weighted when generating the SPD. Whilst the issue of sampling bias certainly adds some complexity therefore, it falls far short of undermining the utility of the methods used. 2. Conclusions To conclude, a few of Vermeersch's comments are readily answered in the text itself, nevertheless we hope our reply offers some additional transparency, particularly in regards to the objectives of our analysis. Other comments are correct in revealing an extremely low-level error rate in the data, and we are most grateful for being given the opportunity to amend these. However, it appears 175 that Vermeersch fully grasped neither the objectives of the paper, nor the methods we have used. The use of radiocarbon dates both as a valid population proxy and a potential palaeoclimate proxy, rely on utilising an appropriate scale of observation in both time and space. This research examined the broad-scale fluctuations in Neolithic Saharan populations in relation to changes in climate associated with the onset and termination of the AHP. To achieve this we have deliberately disregarded much of the specific culture histories of each region, as well as the evident variation in Saharan paleoecology. Whilst this was necessary to reveal the broad-scale fluctuations in Neolithic population levels, it is not our intention to negate specific regional culture histories. On the contrary, a key objective of this research was to establish a background trend in the demographic response to Holocene climate change in the Sahara in order to provide a framework for more localised studies of demographic change and related ecosystem dynamics. References Cancellieri, E., di Lernia, S., 2013. Re-entering the central Sahara at the onset of the Holocene: a territorial approach to Early Acacus hunter-gatherers (SW Libya). Quat. Int. http://dx.doi.org/10.1016/j.quaint.2013.08.030 (available online 12.09.13.). Drake, N.A., Blench, R.M., Armitage, S.J., Bristow, C.S., White, K.H., 2011. Ancient watercourses and biogeography of the Sahara explain the peopling of the desert. PNAS 108, 458e462. http://dx.doi.org/10.1073/pnas.1012231108. Francus, P., von Suchodoletz, H., Dietze, M., Donner, R.V., Bouchard, F., Roy, A.-J., €pelin, S., 2013. Varved sediments of Lake Yoa Fagot, M., Verschuren, D., Kro (Ounianga Kebir, Chad) reveal progressive drying of the Sahara during the last 6100 years. Sedimentology 60, 911e934. http://dx.doi.org/10.1111/j.13653091.2012.01370.x. Gabriel, B., 1978. Gabrong-achttausendjahrige keramik im Tibesti-Gebirge. In: Kuper, R. (Ed.), Sahara: 10000 jahren zwischen Weide und Wusten. Museen der Stadt, Cologne, pp. 189e196. Haaland, R., 1992. Fish, pots and grain in Early and Mid-Holocene adaptions in the Central Sudan. Afr. Arch. Rev. 10, 43e64. ly, C., Le zine, A.-M., APD contributors, 2014. Holocene changes in African vegetaHe tion: tradeoff between climate and water availability. Clim. Past 10, 681e686. € pelin, S., 2006. Climate-controlled Holocene occupation of the Sahara: Kuper, R., Kro motor of Africa's evolution. Science 313, 803e807. Manning, K., 2008. Mobility, Climate Change and Cultural development. A Revised View from the Lower Tilemsi Valley, Northeastern Mali (unpublished Phd thesis). University of Oxford. Rick, J.W., 1987. Dates as data: an examination of the Peruvian Preceramic radiocarbon record. Am. Antiq. 52, 55e73. Stuart, D.E., Gauthier, R.P., 1981. Prehistoric New Mexico. Background for Survey. University of New Mexico Press, Albuquerque. Sutton, J.E.G., 1974. The African aqualithic. Antiquity 51, 25e34. Timpson, A., Colledge, S., Crema, E., Edinborough, K., Kerig, T., Manning, K., Thomas, M.G., Shennan, S., 2014. Reconstructing regional population fluctuations in the European Neolithic using radiocarbon dates: a new case-study using an improved method. J. Arch. Sci. 52, 549e557. Turchin, P., 2003. Historical Dynamics: Why States Rise and Fall. Princeton University Press, Princeton. Vermeersch, P.M., 2015. Comment on “The demographic response to Holocene climate change in the Sahara”, by Katie Manning and Adrian Timpson (2014). Quat. Sci. Rev. 110, 172e175. Weiss, H., Bradley, R.S., 2001. What drives societal collapse? Science 291, 609e610. Katie Manning* Institute of Archaeology, University College London, 31-34 Gordon Square, London, WC1H 0PY, UK Adrian Timpson Research Department of Genetics, Evolution and Environment, University College London, Darwin Building, Gower Street, London, WC1E 6BT, UK * Corresponding author. E-mail address: k.manning@ucl.ac.uk (K. Manning). © 2014 Elsevier Ltd. All rights reserved http://dx.doi.org/10.1016/j.quascirev.2014.12.008