DID HUMANKIND PREVENT A HOLOCENE GLACIATION ? Comment on Ruddiman ’ s Hypothesis of a Pre-Historic Anthropocene

Recently, W.F. Ruddiman (2003, Climatic Change, Vol. 61, pp. 261–293) suggested that the anthropocene, the geological epoch of significant anthropospheric interference with the natural Earth system, has started much earlier than previously thought (P. I. Crutzen and E. F. Stoermer, 2000, IGBP Newsletter, Vol. 429, pp. 623–628). Ruddiman proposed that due to human land use, atmospheric concentrations of CO2 and CH4 began to deviate from their natural declining trends some 8000 and 5000 years ago, respectively. Furthermore, Ruddiman concluded that greenhouse gas concentrations grew anomalously thereby preventing natural large-scale glaciation of northern North America that should have occurred some 4000–5000 years ago without human interference. Here we would like to comment on (a) natural changes in atmospheric CO2 concentration during the Holocene and (b) on the possibility of a Holocene glacial inception. We substantiate our comments by modelling results which suggest that the last three interglacials are not a proper analogue for Holocene climate variations. In particular, we show that our model does not yield a glacial inception during the last several thousand years even if a declining trend in atmospheric CO2 was assumed. 1. Holocene CO2 Growth: No Need for Anthropogenic Forcing Ruddiman’s interesting and thought-provoking idea of an early, global-scale effect of land use on climate is based on the assumption that the increase in atmospheric CO2 concentrations during the Holocene was anomalous with respect to trends reconstructed for the previous three interglacials. To corroborate his assumption, Ruddiman critically reassessed a number of mechanisms which have been proposed to explain the observed Holocene trend in atmospheric CO2: changes in terrestrial biomass, changes in oceanic carbon compensation, and changes in ocean temperature and re-growth of coral reefs. With respect to natural land cover changes as a cause for Holocene CO2 trends, Ruddiman discussed results based on simulations with GENESIS and DEMETER models (Foley, 1994) and PMIP intercomparison (Harrison et al., 1998). Here, we would like to add a word of caution that important atmosphere-biosphere feedbacks were not considered in these model examples. For subtropical deserts and semideserts, like North Africa, this limiting assumption leads to an underestimation of mid-Holocene precipitation changes (e.g. Claussen, 1997; Ganopolski et al., 1998). Hence we would like to draw the reader’s attention to a study by Brovkin Climatic Change (2005) 69: 409–417 c © Springer 2005 410 MARTIN CLAUSSEN ET AL. et al. (2002) in which we used the fully coupled atmosphere-ocean-vegetation model of intermediate complexity, CLIMBER-2 (Claussen et al., 1999; Petoukhov et al., 2000; Ganopolski et al., 2001), including interactive models of terrestrial and marine biogeochemistry. Our model results suggest that the largest changes of terrestrial carbon storage at 8000 yr BP (years before present) are associated with moistening of northern subtropical regions, while the increase in boreal forest biomass is smaller. Despite of a reduction in mid-Holocene atmospheric CO2 concentrations by 20 ppmv compared to pre-industrial values, the terrestrial biosphere in this interactive simulation contains about 90 PgC more than seen in the pre-industrial control simulation. In the late Holocene, release of this carbon to the atmosphere could explain an increase in atmospheric CO2 concentration of some 10 ppmv. This value is too small in comparison with reconstructions of Holocene atmospheric CO2 concentration. Furthermore, simultaneous growth of peat storage in northern wetlands (Gajewski et al., 2001), which was not included in our model study, might have reduced or even reversed the trend in atmospheric CO2 concentrations. Hence natural changes in the terrestrial biosphere alone are presumably not sufficient to explain observed trends in Holocene atmospheric CO2 concentration – in this respect we agree with Ruddiman. An aspect not included in this discussion are biogeophysical effects of land cover changes. If early Holocene land use would have resulted in a strong deforestation, then changes in albedo associated with deforestation in snow-covered regions should have led to a considerable cooling (e.g., Bonan et al., 1992) some 8000 years ago which tends to compensate or, at high latitudes, could overcompensate global warming due to emission of greenhouse gases by deforestation (Claussen et al., 2001). Hence Ruddiman’s estimates of temperature changes arising from changes in land use should be revised. Broecker et al. (1999) suggested that the Holocene increase in atmospheric CO2 concentration was caused by carbonate compensation in the ocean in response to the glacial–interglacial changes in the terrestrial carbon storage. Ruddiman rejects this explanation for two reasons. Firstly, he questions that the CO2 ‘rebound’ from 8000 yr BP to pre-industrial conditions should have been four times the size of the early Holocene CO2 decrease. However accounting for the very long time scale of carbonate compensation, the CO2 growth might be caused by changes that started much earlier than 8000 yr BP. This view is supported by wide-scale evidence of excessive accumulation of carbonate in deep ocean sediments during the Holocene (Milliman, 1993). In CLIMBER-2 simulations in which oceanic carbonate sedimentation was assumed to be constrained by changes in carbonate ion concentration as reported by Broecker et al. (1999), the model simulated a growth in atmospheric CO2 concentration in line with Taylor Dome data (Brovkin et al., 2002). In addition, excessive accumulation of carbonate presumably occurred due to the growth of coral reefs which were slowly adjusting to glacial–interglacial changes in sea level (Kleypas, 1997; Ridgwell et al., 2003). This process is independent of the CO2 ‘rebound’ at 8000 yr BP. DID HUMANKIND PREVENT A HOLOCENE GLACIATION? 411 As a second reason for rejection of the oceanic hypothesis, Ruddiman argues that atmospheric CO2 variations between Holocene and previous interglacials differ. We cannot challenge this view, because we have not undertaken any simulation of the fully coupled climate system for previous interglacials. On the other hand, the Rudiman’s hypothesis is also not free of problems. Two of its weaknesses – inconsistency both with Holocene inventory of terrestrial biomass and with observed changes in δ13C – are discussed by Joos et al. (2004). We would like to point at a further problem related to Ruddiman’s interpretation of paleo-climatic data. Ruddiman’s hypothesis of (declining) atmospheric CO2 variations due to land use is based on the extrapolation of the observed CO2 trend during the early Holocene to the entire Holocene. Ruddiman did not explain why such decline should take place. However others (e.g. Indermühle et al., 1999; Broecker et al., 1999; Brovkin et al., 2002) have attributed the drop in atmospheric CO2 concentration during early Holocene to the re-growth of the terrestrial biomass after the retreat of ice sheets. Such process is limited in time and is likely to not last during the whole Holocene. Furthermore, if we consider the Eemian interglacial as an example of natural dynamics of atmospheric CO2 concentrations, then the pre-industrial Holocene CO2 concentration deviates from the ‘natural’ one by only 10 ppm, which completely changes the starting point of Ruddiman’s discussion (see Figure 1b, thick full line and dashed line). Meanwhile variations of atmospheric CO2 concentrations were recorded for the entire interglacial of MIS 11 (EPICA community member, 2004). It is clearly seen that atmospheric CO2 changes little or tends to increase during the first ten thousand years of MIS 11 (see figure 3 in EPICA community members 2004, the age scale is given in the supplementary information). Also the Vostok data (Petit et al., 1999) reveal a just slightly decreasing atmospheric CO2 concentration over a long period of time during MIS 11. Hence atmospheric CO2 variations during MIS 11 seem to be at variance with those recorded for the last three interglacial. This fact alone would refute any argumentation based on a similarity of greenhouse gas variation during all interglacials. Regarding the question whether or not, anthropogenic land-cover change has led to an increase to atmospheric CO2 concentration, we agree to the statement that anthropogenic deforestation started as early as 8000 yr ago or even earlier. However, we surmise that its consequence for the terrestrial carbon cycle is hard to evaluate based on available terrestrial proxies. Even for well-documented present-day conditions, the uncertainty in CO2 emissions associated with land use is uncomfortably high (Prentice et al., 2001). It is even more difficult to assess the anthropogenic influence on the atmospheric CO2 in the past, many thousand years ago. Besides, as pointed by Joos et al. (2004), a deforestation of the magnitude proposed by Ruddiman should be accompanied by much stronger changes in δCO2 than observed in the Taylor Dome record (Indermühle et al., 1999). Hence we conclude that the case of atmospheric CO2 variations in the Holocene is still open; Ruddiman’s hypothesis and reassessment of earlier theories is not free 412 MARTIN CLAUSSEN ET AL. Figure 1. (a) Changes in maximum insolation at 65◦N computed according to Berger (1978) for the Holocene (full line, corresponding time axis at the bottom of the figure) and for the Eemian/early Weichselian transition (dashed line, time axis at the top of the figure). (b) Atmospheric CO2 concentration for the Eemian and last glacial inception following Barnola et al. (1987) (dashed line, time axis at the top of the figure), atmospheric CO2 concentration for the last 10,000 year from Indermühle et al. (1999) (thick line, time axis at the bottom of the figure) and scenarios (full thin line) of a declining trend in atmos