While our method is not affected by biases introduced by regional differences in research intensity, the SPD could showcase spurious patterns if there is evidence of a temporally heterogeneous dating process
When examined on its own, we can consider a number of alternative causes that might have led to the observed fluctuations in the SPDs. A general time-dependent taphonomic loss would not affect our analyses (the exponential model mimics the process and the permutation-based comparison of the SPDs uses the observed 14 C dates directly, hence already integrating the effects of time-dependent loss), but a spatially divergent, inhomogeneous thinning process might produce some fluctuations that are not related to the underlying population dynamics. This problem is clearly not limited to SPDs and applies to all count-based time-series, but we are not aware of any study suggesting and quantifying this type of bias.
For instance, scholars might focus on dating sites of specific chronological interval rather than others, effectively leading to a higher density of 14 C dates
We are confident that this research bias is not affecting the samples from Hokkaido and Aomori Prefecture, but the strong interest in reconstructing the Middle Jomon pottery-sequence of the Kanto area might have generated a higher density of 14 C dates during this interval (see for instance ). This http://www.hookupswipe.com/asian-dating-app/ might indeed be the reason why we observe a strong positive deviation around 5,000 cal BP in Kanto. However, the fact that we observe similar peaks in other proxies (e.g. counts of residential units) seems to support our argument, which the SPD is genuinely reflecting an increase in population size. g. exceptionally large settlements) might potentially be affecting some of the pattern we observe in the SPD, albeit the broader similarity to other proxies does not seem to suggest this.
The results of our SPD analysis might be a consequence not only of changes in the underlying population but also the result of variations in the site-to-population ratio. Sites might in fact vary in their function (e.g. settlements vs. field camps), size (i.e. number of residential features), duration of occupation, and archaeological visibility as a result of changes in the subsistence-settlement patterns. Using counts of residential features as basic unit of analysis does not necessarily solve this issue. In fact, one should consider also variations in the size of residential units, and more crucially the patterns of residential mobility. For instance, a change from year-round settlements to seasonal shifts would increase the number of sites (and hence lead to a higher density of 14 C dates), even if the population size remains unchanged. Similarly, if the seasonal shifts involved fission-fusion of residential units, the average size of the seasonal residential bases could be smaller than the nucleated residential bases of fully sedentary hunter-gatherers.
Our case studies do have evidence of potential changes in the subsistence-settlement pattern during the temporal scope investigated here. In the case of Aomori Prefecture and Kanto, the archaeological evidence from the middle of the Middle Jomon period is characterized by an abundance of extremely large nucleated settlements such as the Sannai a site in Aomori Prefecture [3,64] and the Miharada site in Gunma Prefecture in Kanto [65,66]. Towards the end of the Middle Jomon period, however, scholars have reported an increase in the number of smaller settlements [17,67]. Similarly, settlement analyses in the Eastern Tokyo Bay area have shown continuous fluctuations in the shape of the site-size distribution between the Early and Late Jomon periods . For an earlier phase, lithic and settlement data seem to indicate that the transitional period between the Early and Middle Jomon in some parts of Kanto was characterised by a temporary shift from a collector to a forager strategy . It is an issue open for debate whether under the assumption of a constant population size these changes in settlement pattern would translate into tangible patterns in the proxies we examined here. If archaeological visibility is higher for all the residential bases, we might expect an increase from a collector to a forager system in terms of site number (forager systems would imply a higher number of residential move for the same interval), and hence also a higher number of 14 C dates. If the sampling is unbiased (for example, if the sampling is more representative to the full spectrum of other site types such as caches and stations), this might not be the case. Further studies aiming to establish differences in the site-to-population ratio for different settlement patterns is thus a key aspect for the development of SPD analysis.