The Antarctic Centennial Oscillation: A Natural Paleoclimate Cycle in the Southern Hemisphere That Influences Global Temperature

1. Introduction

Global temperature has oscillated at least since the beginning of the Phanerozoic Eon 542 million years (My) ago, manifest as long climate cycles that repeat with a period of 135–150 My [1,2,3]. Faster and smaller temperature oscillations are superimposed on the cooling phase of the most recent long cycle beginning at least ~55 million years before present (Mybp) [4], forming the earliest Marine Isotope Stages (MISs; [5]), or global glacial cycles (“ice ages”). MISs recurred initially every ~41 thousand years (Ky) [5], the same repetition period as the integrated Obliquity/Precessional Orbital Cycle (OPOC; [6,7]), and grew larger by more than an order of magnitude over millions of years as the Earth cooled an additional ~1.5 °C [5]. MISs shifted to a quasi-regular period of ~80–120 Ky and doubled in amplitude at the Mid-Pleistocene Climate Transition 1.25–0.75 My ago [5,8,9,10,11,12,13,14]. This temperature-oscillation pattern continues to dominate multi-millennial climate variability up to the present, inducing quasi-regular global glaciations separated by shorter warm interstadials. The contemporary climate constitutes a prolonged warm interstadial comparable to those that have alternated in the past with cold stadials corresponding to global glacial maxima.

Superimposed upon these multi-millennial climate cycles are numerous shorter global and regional climate cycles ranging in period from several millennia down to a few weeks. Included among these faster oscillations are millennial-scale cycles, particularly the Bond cycle [15,16,17,18,19,20,21,22,23], and centennial-scale cycles, notably the Antarctic Oscillation (AAO) known also as the Southern Annular Mode (SAM) and tracked quantitatively by means of the Southern Oscillation Index (SOI) [24,25,26,27,28,29,30,31]. These interdependent Southern Hemisphere (SH) temperature-proxy oscillations exhibit both centennial [24,25,26,27] and decadal [28,29,30,31] frequency components. Similar periodicity appears in independent reconstructions of more contemporary temperature proxies from James Ross Island [27] and snow accumulation in stacked records from snowpits at Vostok [32].

These and other centennial-scale temperature oscillations recurred with regularity in the past and are, therefore, likely to continue to oscillate at similar periods to affect future climate [33]. Understanding these cycles can therefore improve the empirical basis for climate projection across time scales most immediately pertinent to human affairs, decades to centuries [34,35]. Analysis of oscillatory climate regimens and their interactions on diverse time scales also promises to improve our understanding of regional and global climate dynamics, including especially the balance between human and natural forcing of present and future climate change, which are among today's most compelling environmental and policy issues.

Toward these ends we report here a previously-unexplored centennial-scale temperature-proxy oscillation, the Antarctic Centennial Oscillation (ACO), recorded in stable isotopes frozen in ice cores at Vostok, Antarctica [36,37] and three additional Antarctic drill sites distributed widely on the East Antarctic Plateau (EAP), namely, EPICA Dronning Maud Land, EPICA Dome C and Talos Dome. Paleoclimate records from these four drill sites have been synchronized on the most precise ice-core chronology available, the Antarctic Ice Core Chronology of 2012, or AICC2012 [38,39], enabling the most accurate possible estimates of the relative timing of ACO occurrence at different drill sites.

Centennial-scale temperature oscillations in the Vostok paleoclimate record were recognized qualitatively nearly two decades ago [40], but their nature and spatial extent have not been characterized previously, “precluding a definitive conclusion about their meaning” [40] (p. 111). Here we characterize the ACO quantitatively, document its occurrence at the aforementioned four ice-core drill sites on the EAP, explore the ACO contribution to Dansgaard-Oeschger (D-O) oscillations in the Northern Hemisphere (NH), draw parallels with contemporary climate cycles including particularly the Antarctic Oscillation (AAO), and evaluate possible implications for the current and future climate.

In particular, the contemporary global warming increase of ~0.8 °C recorded since 1850 has been attributed widely to anthropogenic emissions of carbon dioxide (CO₂) into the atmosphere. Recent research has shown, however, that the concentration of CO₂ in the atmosphere has been decoupled from global temperature for the last 425 million years [3] owing to well-established diminishing returns in marginal radiative forcing (ΔRF) as atmospheric CO₂ concentration increases. Marginal forcing of temperature from increasing CO₂ emissions declined by half from 1850 to 1980, and by nearly two-thirds from 1850 to 1999 [3]. Changes in atmospheric CO₂ therefore affect global temperature weakly at most. The anthropogenic global warming (AGW) hypothesis has been embraced partly because “...there is no convincing alternative explanation…” [41] (p. 12). The ACO provides a possible alternative explanation in the form of a natural climate cycle that arises in Antarctica, propagates northward to influence global temperature, and peaks on a predictable centennial timetable.

2. Methods

Open-access databases containing stable isotope records [36,37,38] and computed planetary orbital (Milankovitch) parameters and consequent surface insolation [6,7] were analyzed using conventional methods detailed in the Supplementary Materials (SM). The cycle nomenclature proposed by Wunsch [42] was modified for the centennial-scale Antarctic cycles identified here. The start time for this analysis, 226,408 years before 1950 (yb1950), is the oldest for which the sampling resolution of the Vostok temperature-proxy record is sufficient to detect centennial-scale events in compliance with the Nyquist-Shannon sampling-frequency criterion of two samples per cycle (SM). Oxygen and hydrogen isotope ratios are used widely and interchangeably as paleoclimate temperature proxies [43,44,45,46,47], although the former contains a proxy component of ice volume and other environmental signals not present in the latter [43,48]. We use both proxies in this study as is conventional.

Three definitions were used to describe temperature-proxy oscillations comprising ACOs: oscillations larger than 0.25 °C in amplitude (~90% of 546 cycles identified over 226 millennia), oscillations from 0.045 to 0.25 °C in amplitude (~10% of 546 cycles), and small oscillations characterized by a positive inflection of the temperature-proxy record followed by a negative inflection on a rising or falling background temperature-proxy curve (an additional 104 cycles creating a maximum sample size n = 650). These smallest ACOs are accompanied by larger homologs in climate records from other drill sites, showing that they are genuine temperature-proxy oscillations and not noise. The third definition was used only to compute cycle period; excluding it increased mean computed period by ~15%. All other properties of ACO cycles were computed using only the first and second ACO cycle definitions. ACO properties evaluated include period (trough-to-trough and peak-to-peak), amplitude (half the excursion of the warming phase plus half the excursion of the cooling phase), and warming and cooling duration and rate, as detailed in the SM. Reduction of temperature-proxy and atmospheric carbon dioxide (CO₂) concentration data to analytic format and color-coding of centennial-scale climate cycles depicting the above three definitions appear in SM Table S1, enabling replication of this study and supporting further independent analysis.

This study is limited to East Antarctic temperature-proxy records from ice cores extracted from four drill stations on the EAP synchronized on the AICC2012 chronology: Vostok, EPICA Dome C (EDC), Talos Dome (TD) and EPICA Dronning Maud Land (EDML) [38]. Meaningful comparison of ACO cycles across these diverse climate records required equalizing corresponding sampling resolutions as closely as possible to Vostok by averaging (filtering). The averaging protocols for non-Vostok records were established using a simple arithmetic protocol (SM) and are required to replicate the results of our study. Averaging protocols are therefore reported in figure legends as the bin width over which averages were computed from the original raw temperature-proxy data followed in parentheses by the start time of the averaging. Quantitative analysis of the timing of homologous cycles in different climate records was restricted to the most recognizable cycles associated with well-established climate events. These unambiguous “signpost” ACO cycles can be matched objectively based on their position relative to known climate events, following the methodology established by numerous previous climate investigators [38,40,49].

We used standard quantitative methodologies in this study, including Fourier (spectral) analysis, auto- and cross-correlation, and conventional t-tests. Our results show, however, that climate data including those evaluated here do not necessarily comprise a stationary time series, implying that modern advanced methods that assume neither linearity nor stationarity may be more appropriate. Such methodologies include, to various degrees, Wavelet Analysis (which is nonetheless based on the Fourier transform), Empirical Mode Decomposition, Hilbert-Huang transform, etc. We consider these more advanced analytic approaches unnecessary in this initial analysis of the ACO, but our findings nonetheless imply that their absence here constitutes a methodological limitation. As developed in the Discussion, we consider this limitation inconsequential because we mainly evaluated short time segments of the climate record that comprise relatively stationary time series as confirmed by auto- and cross-correlation results.

We evaluated the major potential sources of noise in the reconstructed climate records, including frequency aliasing, dating uncertainty, variance in cycle amplitude, cycle periodicity, sample resolution and stratigraphic noise from non-climate signals in the original ice-core records, and concluded that they did not affect the present interpretations significantly (SM). The SAS JMP software used here for spectral power analysis employs Fisher’s Kappa Test Statistic (κ) as a white-noise test (SM). The κ statistic returns the probability that the distribution analyzed is generated by Gaussian (random) white noise against the alternative that the distribution is nonrandom (SM) and is reported in the figure legends. To compute confidence intervals of spectral density peaks, Fourier coefficients were estimated using forward stepwise regression (SM). Such stepwise selection can exaggerate the precision of the alpha estimates by as much as an order of magnitude [50], however, and we therefore conservatively set the confidence limit threshold for spectral peaks at an order of magnitude higher (0.005, 99.5%) than usual (0.05, 95%).

3. Results

3.1. Cycle Nomenclature

We used the cycle nomenclature developed by Wunsch [42] as modified here (SM). By this nomenclature, the ACO documented in the present study is designated as the TO_C350V cycle (Temperature-proxy Oscillation_{Centennial−scale 350 year mean period Variable}) based on its mean repetition period of 352 years over the time period evaluated, 226.4 to 0.149 Kyb1950, and based on the observed systematic variation of ACO period and amplitude in the time series as documented below. The acronyms “ACO” and “TO_C350V” are used interchangeably throughout this paper. A second, millennial-scale oscillation, termed the TO_K1500V cycle using the modified Wunsch nomenclature, appears homologous with the previously-identified Bond cycle in the NH [17,18,19,20,21] and is also evaluated here using temperature-proxy data from the SH.

3.2. Cycle Form

The TO_C350V cycles analyzed here are evident in raw temperature-proxy records as separate, individually-identifiable oscillations, each composed of several sampling data points (Figure 1). These centennial temperature-proxy cycles are superimposed on ascending temperature phases of MISs such as the Last Glacial Termination (LGT; Figure 1a) and recovery from the Antarctic Cold Reversal (ACR, Figure 1b), on descending phases of the temperature-proxy record such as the cooling phase of the ACR (Figure 1b), and on relatively flat regions of the Vostok temperature-proxy record including the Last Glacial Maximum (LGM) and the Holocene (Figure 1a,d, respectively). All 650 TO_C350V cycles identified in this study over the 226.4 Ky evaluated using the three definitions described in the Methods are labeled starting with the most recent cycle at 149 yb1950 on the Vostok GT4 glaciological chronology using alphanumeric symbols, following the convention adopted for MISs [5] (Figure 1). These time series show qualitatively an apparent progressive decline in cycle period (an increase in frequency) combined with a proportional increase in cycle amplitude, both of which are confirmed quantitatively below.

3.3. Cycle Periodicity

Spectral analysis of Antarctic paleoclimate records was done previously for the Holocene for the time period from 11,500 ybp to the present [51]. We extend this analysis here using the more accurate data synchronized on the AICC2012 chronology and including climate records from Vostok, EDC, EDML and TD; [38,39]. We evaluate these climate records over broader time spans than previously done, including the Holocene, and over older and longer time periods, to enable detection of a broader complement of cycle periodicities. The resulting spectral power density periodograms show peaks at decadal, centennial, millennial and multi-millennial periodicities (Figure 2). Spectral peaks occur at similar frequencies in ice-core records from different drill sites (cf. Figure 2a,b), consistent with the hypothesis that the same temperature cycles manifest at different Antarctic drill sites located on the EAP.

Broadband periodograms covering several hundred millennia exhibit the primary MIS frequency peak at a period of 106–108 Ky, as well as secondary and tertiary peaks of Milankovitch cycles at the orbital obliquity (41 Ky) and precession (21 Ky) cycle periods (Figure 2a,b), replicating previous results [36] (cf. our Figure 2a with Figure 4a in [36] (p. 432)). Spectral analysis of shorter records shows possible representations of the obliquity orbital cycles in the Talos Dome climate record at 38.3 Ky (Figure 2c), but corresponding peaks at EDML differ in period by up to 25% and hence their etiology is correspondingly less certain. The spectral periodograms constructed over shorter time periods (Figure 2c,d) also lack the longer MIS cycle (Figure 2c,d) as expected, since the time period evaluated is too short to detect it.

Spectral analysis of temperature-proxy records over shorter time periods (narrower frequency bandwidths) were done for the period from ~12,500 to 0 Kyb1950, confirming the results of previous studies [51]. The Vostok periodogram (Figure 3) shows six discernible (p < 0.005) centennial-scale spectral power density peaks ranging from 193 years, near the average period of the ACO across the Holocene, to 1096 years, near the reported millennial period of the Bond cycle in the NH [17,18,19,20,21]. Relaxation of confidence limits to 95% (p < 0.05) adds three discernible spectral peaks at 282, 448 and 580 years which however we do not consider further here owing to the possible exaggeration of precision in the mathematical method used to compute the confidence limits (Methods). These findings confirm previous evidence of multiple centennial peaks over the Holocene [51] (cf. Figure 3 here with their Figure 6, p. 356) and they are reminiscent of multiple discernible spectral peaks recorded in climate records from the NH, e.g., [23].

Periodograms of the remaining three AICC2012 climate records during the Holocene are similar to the periodogram of the Vostok record (Figure 4). All are bounded near the low end by a peak corresponding approximately to the mean period of the TO_C350V cycles and near the high end by a peak corresponding to the Bond cycle in the NH and ranging from 825 to 1027 years. Between these extremes lie at least four additional centennial-scale peaks in all AICC2012 climate records evaluated.

The periods of these peaks coincide closely with the six discernible (p < 0.005) spectral peaks identified in the Vostok climate record (adjacent arrows in Figure 4 demarcated by ovals). The difference between the closest corresponding peaks in other records was computed in both relative terms (positive and negative signs included) and, more conservatively, in absolute terms (signs omitted). The mean (n = 6) relative and absolute differences, respectively, between the frequency of Vostok and non-Vostok spectral peaks (Figure 4) are 1.0% and 5.8% (Vostok versus EDC; Figure 4a), −1.5% and 3.8% (Vostok versus TD; Figure 4b), and −0.7% and 4.7% (Vostok versus EDML; Figure 4c). Homologous spectral peaks at different drill sites therefore match each other on average within ±0.2% (relative difference) or ±2.4% (absolute difference).

Most but not all matching spectral peaks at EDC, TD and EDML are as distinct as the corresponding peaks in the Vostok record (Figure 4). For example, peak 6 at Vostok with a period of 1096 years, which may be analogous with the Bond cycle in the NH (the TO_K1500V cycle), is matched with a comparable peak at 1050 years in the TD record. We did not assess the statistical discernibility of non-Vostok spectral peaks and have no explanation for the occasional apparent differences in relative spectral density energy of peaks that are otherwise matched in period. The findings nonetheless support overall the hypothesis that the same temperature-proxy cycles manifest at geographically-distributed EAP drill sites.

Spectral power periodograms of non-stationary time series can be ambiguous in the frequency domain inasmuch as the changing frequency of any variable climate cycle can broaden (“smear”) a spectral power peak, rendering it less distinct. Additionally, in the event of discrete frequency shifts in the underlying reference signal, power periodograms can manifest distinct spectral peaks even when they originate from the same, variable (non-stationary) climate cycle. We therefore conservatively sought additional confirmatory evidence for nonrandom periodicity in the time domain of Antarctic climate records in the form of lagged or serial autocorrelation correlelograms. We generated correlelograms over relatively short time frames of the Vostok record dated on the GT4 chronology, including from 75 to 55 Kyb1950, 20 to ten Kyb1950, and ten to 0.149 Kyb1950. The GT4 chronology for Vostok differs only slightly from the AICC2012 chronology over time periods analyzed here, as demonstrated by the near-identity of the respective age models over the time period studied here (SM Figure S3).

The null hypothesis that TO_C350V cycles comprise random variation in cycle structure was tested by means of cyclic autocorrelation coefficients. We find that autocorrelation coefficients alternate between positive and negative at the same periodicity as the corresponding TO_C350V cycle frequency (Figure 5). Near peaks and troughs, nearly all of these autocorrelation coefficients are discernibly different from zero at low alpha levels (at least at p < 0.05). These autocorrelation results supplement and extend spectral periodograms to confirm that TO_C350V cycles comprise nonrandom periodic sequences. Such positive autocorrelation results would not be possible unless the short time series evaluated represent relatively stationary time series over the time periods evaluated.

3.4. Geographic Distribution of ACOs

The demonstration of similar spectral density periodograms at different drill sites on the East Antarctic Plateau suggests that TO_C350V cycles are common to all four ice-core records evaluated here. We tested this hypothesis more directly by examining individual TO_C350V cycles using the identical approach pioneered by previous investigators, namely evaluation of cycle waveform and sequence (Methods). Three of the four AICC2012-synchronized Antarctic climate records over the period 70 to 63 Kyb1950 were sampled originally at a high enough frequency to detect centennial-scale temperature-proxy oscillations in compliance with the Nyquist-Shannon sampling-frequency criterion: Vostok, the nearby (~550 km) and slightly lower-elevation (~248 m) EDC, and the more distant (~2300 km) and still lower-elevation (~596 m) EDML (Figure 6) (distances between drill sites and elevations are from [51]).

The time period from 70 to 63 Kyb1950 encompasses two well-studied climate events, Antarctic Isotope Maximum (AIM) # 19 and AIM # 18 [38], which were formerly designated as AIM # 5 and AIM # 4 [40], respectively. These AIMs precede D-O events # 19 and # 18, respectively, in Greenland ice cores from the NH, with south-to-north propagation latencies of 1.5 to 3.0 millennia [38,40,49,52]. Examination of individual TO_C350V cycles shows that AIM # 19 is composed of multiple TO_C350V cycles, identified as ACO 195a, 195 and 194, while AIM # 18 is composed of TO_C350V cycles 186–181. These identified TO_C350V cycles are recognizable across the three climate records compared (grey dashed connector lines in Figure 6), consistent with the hypothesis that TO_C350V cycles are distributed geographically across the EAP to at least these three drill sites.

The identified TO_C350V cycles that compose AIM # 19 and AIM # 18 do not occur at the same time in different climate records, i.e., the homologous cycles at one drill site are shifted forward or backward in time relative to other drill sites when displayed on a common chronology (Figure 6). For example, mean peak-to-peak latencies from Vostok to EDC over ten homologous cycles that encompass AIM # 19 (cycles 195a-190) was computed as −158.5 years. The negative sign implies that Vostok peaks lag homologous peaks at EDC. Similarly, peaks of the ten TO_C350V cycles associated with AIM # 18, 186a-178 are asynchronous, occurring at a mean time latency from Vostok to EDC of −211.2 years (Vostok lags). In both of these examples, however, the mean latency is smaller than the reported dating uncertainty for this portion of the respective climate records [38] and therefore in these examples the observed difference in latencies cannot be distinguished from chronological noise or dating uncertainty.

In contrast to these examples, the mean latency between the peaks of homologous TO_C350V cycles from Vostok to EDML for cycles 195a-190 is −335.1 years (Vostok lags EDML), while the mean latency between homologous cycles 186a-178 is −557.7 years (Figure 6a). These means are discernibly greater than the corresponding mean latencies between Vostok and EDC (t-test, p < 0.0006, n = 10). The Vostok-to-EDML mean latency also exceeds the sum of the decadal variance that arises from averaging and the estimated dating uncertainty for this glacial period, which is 500 years [38] (p. 1737). These latency differences from Vostok to EDML cannot, therefore, be attributed to dating uncertainty. Instead, the differences in latency between homologous cycles at different drill sites are genuine (non-artifactual) climate signals.

The corresponding mean peak-to-peak latency from EDC to EDML between homologous cycles 195a-190 is −176.6 years (EDC lags), discernibly smaller than the mean latency from Vostok to EDML of −335.1 years (t-test, p < 0.0009, n = 9). For cycles 186a-178, the comparable mean homologous peak-to-peak latencies are −322.6 years (EDC to EML) and −557.6 years (Vostok to EDML), discernibly different from each other (t-test, p < 0.0006, n = 10). Chronological uncertainty reported for this period of the AICC2012 chronology is from 20 to 100 years [38] (p. 1737). The measured differences in mean latencies are therefore not only discernible with high probability but again exceed the dating uncertainty of the AICC2012 chronology, in these examples by a factor up to >30. Comparing Vostok-to-EDC latency with EDC-to-EDML latency yields non-discernible differences for cycles 195a-190 (p > 0.66) and cycles 186a-178 (p > 0.13). In both cases the corresponding mean latencies nonetheless exceed reported chronological uncertainty.

The century-scale and discernible differences in peak time latencies across different climate records therefore signify that within the limits of chronological uncertainty, homologous TO_C350V climate cycles arrive at different times at the four drill sites analyzed. Homologous cycles appear first at the lowest drill site nearest the ocean (EDML) and centuries later on the high EAP at Vostok and EDC. The latency between sites exceeds the duration of individual cycles and yet the characteristics and sequences of TO_C350V cycles are retained across climate records as they propagate from drill stations close to the ocean to more distant sites high on the EAP. Although the expansion and contraction of the probably-homologous Antarctic Oscillation (AAO; see below) comprises a standing wave, its cyclic variation imparts apparent movement that is reflected here as changes in latencies between homologous ACO cycles at different drill sites. Retention of the internal architecture of ACO cycles and their sequences as the cycles propagate over time and space reflects a “memory” in the Antarctic climate system at least a few centuries in duration. The differences in latencies of TO_C350V cycles identified here parallel and corroborate the timing differences reported between major climate events in Antarctica identified from methane-synchronized Antarctic climate records [53,54].

To enable a more direct visual comparison of relative waveforms, amplitudes and timing of homologous ACOs, the corresponding temperature-proxy records were shifted in time by the mean latency between peaks so that they appear in artificial temporal register (Figure 6b–d). Waveforms and relative amplitudes of homologous TO_C350V cycles appear qualitatively most similar between Vostok and nearby EDC, although comparisons across drill sites separated by greater distances and elevations also often show similar waveforms and relative amplitudes. Smaller-amplitude cycles in one climate record invariably manifest as larger-amplitude cycles in one or more other records, e.g., cycles 193a and 184 at Vostok and cycles 189a and 183 at EDC (Figure 6c,d, respectively). This finding, repeated across all time frames evaluated in this study (not shown), establishes the requirement to include the smallest temperature oscillations under cycle definition 3 (Methods). Waveform matching is 1:1 not only for homologous signpost cycles (Figure 6a,c) but also for homologous non-signpost cycles (Figure 6b).

Spectral analysis of the time period from 70 to 63 Kyb1950 across a narrower (centennial) bandwidth reveals the same spectral density structure seen above in similar records over comparable time periods of the Holocene. That is, the periodogram is bounded near the high-frequency (short-period) end of the spectrum by a centennial-scale peak of approximately the same period as mean TO_C350V cycles and on the low-frequency (long-period) end by a long-period cycle of 1471 years (Figure 7). Both of these periods are larger than the corresponding values in more recent time periods, however (cf. Figure 7 with Figure 3 and Figure 4), suggesting that the period of the millennial cycles identified here declines over time, as confirmed more directly below from individual ACO cycles in the time domain. The spectral peak at 1471 y shows nearly the same periodicity as the reported 1470-year period of Bond events in the NH ([17,18,19,20,21], but see [22,23]). This cycle is designated here as the TO_K1500V cycle using the climate-cycle nomenclature modified here from Wunsch [42].

Decomposition of the millennial-scale cycle from the original temperature-proxy time series data (Figure 6a) reveals a characteristic temporal pattern in the sequence of TO_C350V cycles that underlies a stereotypic architecture of the TO_K1500V cycle. This pattern consists of a largely-monotonic increase over several centuries in the amplitude and frequency of successive TO_C350V cycles culminating at the end of each such sequence in a single extreme warming peak followed by a single extreme cooling trough. This stereotypical pattern of TO_C350V cycle trains is evident in the internal temperature-proxy structure of AIM # 19 and AIM # 18 (Figure 6a) and throughout the 226.4 Ky period studied here (not shown) including the Holocene (see below). These observations suggest that AIMs are formed by the periodic (millennial-scale) algebraic summation of sequences of high-frequency TO_C350V cycles of which they are composed.

The hypothesis that TO_C350V and TO_K1500V cycles at Vostok comprise more widely-dispersed Antarctic climate signals was evaluated more directly by tracking single, identifiable signpost centennial cycles across the four AICC2012-synchronized climate records over the last 21 millennia (Figure 8). During the LGM and the start of the LGT (Figure 8a), TO_C350V signpost cycles include those that occur at the beginning of termination from 18,677 yb1950 (TD) to 18,364 yb1950 (EDC), namely, cycles 74–69. These signpost cycles appear near the time of the onset of the LGT at 18,934 yb1950, as estimated independently from observed sea level minima (19,000 ybp ± 250) [55], and are matched one-to-one across all four AICC2012-synchronized climate records (gray dashed connector lines in Figure 8a). Additional TO_C350V signpost cycles associated with the ACR from 14,872 (TD) to 14,456 (Vostok) yb1950, including cycles # 64–62 at the onset of the ACR (Figure 8a), are among the most recognizable TO_C350V signpost cycles identified in this study.

Peak-to-peak latencies between homologous cycles at different drill sites from 21 to 14 Kyb1950 (Figure 8a) follow a different pattern from those described above for the older time period of 70 to 63 Kyb1950 (Figure 6a and Figure 8a). Prior to the LGT (Figure 8a), latencies from TO_C350V cycle peaks at Vostok to homologous peaks at EDC are near zero or weakly positive. Latencies from Vostok ACOs to homologs at EDML and especially TD are strongly negative, i.e., Vostok lags. After the LGT begins, however, latencies from Vostok to EDC remain near zero or weakly negative while latencies from Vostok to EDML and especially from Vostok to TD shift toward positive (Figure 8a), i.e., Vostok tends to lead. These latency shifts manifest within a few centuries following the onset of glacial termination. The latency shifts are not ascribable to dating uncertainties across these records, which they exceed by up to more than an order of magnitude as noted above. TO_C350V cycle latencies between different drill sites are therefore variable and reversible on relatively short geologic time scales. These results suggest that glacial termination and the associated increase in temperature transiently synchronized TO_C350V peaks across the four AICC2012-synchronized climate records. Cycle coherency, i.e., the percent of identified centennial cycles that are matched 1:1 with homologs in this time frame across all cycles, exceeds 99%. Systematic changes in the latencies between homologous cycles at different drill sites are significant in part because they may bear upon the method of teleconnection between the different drill sites (see below).

Similar analysis of the time period from 15 to 9 Kyb1950 (Figure 8b) shows that signpost TO_C350V cycles are matched 1:1 across all four climate records. The most recognizable signpost TO_C350V cycles in this time period occur at the beginning and end of the ACR, namely TO_C350V cycles # 63-61a and cycles # 56–51, respectively. TO_C350V cycle # 55 demarcates the end of the LGT and the start of the Holocene across all four climate records evaluated and is the single most recognizable TO_C350V signpost cycle identified in this study. All remaining non-signpost TO_C350V cycles in the Vostok record can also be matched with putative homologs in the other records. Cycle coherency across all cycles in all records is 98.3%, i.e., most centennial-scale cycles in the Vostok record, including both signpost and non-signpost cycles, are matched 1:1 with homologous cycles in the three AICC2012 records, and conversely.

The relative timing of TO_C350V cycle peaks over this time period (Figure 8b) shows that Vostok typically lags the other three drill sites and that peak-to-peak latencies generally exceed dating uncertainty. For example, TO_C350V signpost cycle # 62 at Vostok lags its homolog at EDC by 139 years, EDML by 230 years, and TD by 427 years. All three latencies exceed the ten-to-200-year dating uncertainty reported for this time frame [38] (p. 1737) and are larger by up to two orders of magnitude than the decadal variance associated with averaging. Therefore TO_C350V cycles # 62 and # 55 occur earlier at TD and EDML than the homologous TO_C350V cycles at Vostok and this time difference is genuine rather than artifactual.

The time lag between homologous TO_C350V cycle peaks from TD to Vostok is equivalent to at least two TO_C350V cycles, implying retention in the climate system of information about the relative amplitude, waveform and sequence of TO_C350V cycle # 62 as this climate signal propagates from TD to Vostok over hundreds of years and thousands of kilometers. These latency differences between the four AICC2012 sites, particularly TD and Vostok, persist throughout this six-Ky period and increase to 750 years, equivalent to from three to four complete TO_C350V cycles at the cycle frequency that prevails over this time period. This latency exceeds the corresponding 200-year dating uncertainty and exceeds the uncertainty estimated for comparisons between TD and Vostok of 20–100 years [38] (p. 1737) by a factor of up to 37.5. These findings therefore further exemplify the aforementioned “climate memory”, which is expressed as the conservation of TO_C350V sequences and cycle parameters across multiple TO_C350V cycles and across distances of thousands of kilometers.

The time period from ten to five Kyb1950 (Figure 8c) encompassing the early Holocene shows a different pattern from the earlier time periods analyzed. During the early Holocene, the latencies between homologous TO_C350V peaks approach zero across the four AICC2012 climate records. The peak-to-peak latency from TD to Vostok reverses for several cycles starting at approximately eight Kybp, when TO_C350V peaks at TD follow corresponding peaks at Vostok. This reversal entails changes in peak-to-peak latencies of up to a few hundred years, which exceed by more than an order of magnitude the summed variance of dating uncertainty (ten to 100 years) and averaging noise (decadal). During this period, the Holocene Climate Optimum, temperature is greater than at any other time analyzed in this study and peak latencies across sites shifts from negative to positive (Vostok leads). These results illustrate that the latencies at different drill sites shift from synchrony to reversal during the warmest climate studied here, the early Holocene. Cycle coherency over the time frame from ten to five Kyb1950 including both signpost and non-signpost cycles is 98.0% (Figure 8c).

The most recent Holocene (Figure 8d) is characterized by four millennial-scale (TO_K1500V) cycles culminating in temperature maxima followed immediately by temperature minima (e.g., TO_C350V cycles # 24, 18, 13 and 4). These cycles are evident to variable degrees across all four of the AICC2012-synchronized climate records (Figure 8d). In contrast to the pre-Holocene and early Holocene, however, the TO_C350V cycle peaks occur nearly simultaneously at different drill sites, i.e., the mean peak-to-peak latencies between identified TO_C350V cycles at different drill sites approach zero. The reversal of latencies seen in the early Holocene (i.e., the shift from Vostok-lags to Vostok-leads) is absent, particularly for the most recent two millennia of the Holocene. Corresponding peak-to-peak latencies are less than the summed variance of averaging and chronological uncertainty. The warmest portion of the paleoclimate record analyzed here is, therefore, also characterized by the closest synchrony of homologous TO_C350V cycles, which may provide clues about the mechanism of regional propagation of the ACO as discussed below. Cycle coherency over the recent Holocene is 100% across all cycles, signpost and non-signpost (Figure 8d).

3.5. ACO Period

Mean TO_C350V cycle period computed from Vostok temperature-proxy data is 352 years over the 226.4 Ky time period evaluated, ranging from 63 to 1174 years (

Table 1. Descriptive statistics for TO_C350V climate cycles computed from 226.4 to 0.149 Kybp.

Statistic	Period (Years)	Amplitude (°C)	Warming Duration (Years)	Cooling Duration (Years)	Warming Amplitude (°C)	Cooling Amplitude (°C)	Warming Rate (°C/Century)	Cooling Rate (°C/Century)
Mean	352	0.67	184	189	0.66	0.68	0.43	0.44
Min	63	0.05	19	28	0.01	0.01	0.01	0.01
Max	1174	3.22	713	861	3.09	3.46	5.00	4.52
σ	175	0.44	109	114	0.53	0.52	0.46	0.45
CV	0.50	0.66	0.59	0.60	0.79	0.77	1.07	1.03
SS (n)	646	546	546	546	546	546	546	546

TO_C350V cycle statistics computed from the data contained in SM Table S1 for the Vostok ice-core record (GT4 glaciological chronology) [36,37]. Period was computed using all three definitions of a TO_C350V cycle. Remaining statistics were computed using the more restrictive first two definitions (SM, 2. Methods). Abbreviations: Min, minimum; Max, maximum; σ, standard deviation of the mean; CV, coefficient of variation (standard deviation divided by the mean); SS (n), sample for same column.

). The decrease in cycle period over time shown qualitatively in the time series of Figure 1 is confirmed quantitatively in Figure 9. Over the 226.4 Ky analyzed, TO_C350V cycle period declined by ~100% on average from approximately 500 to 250 years (Figure 9a). By the end of the Holocene TO_C350V cycle period declined on average to less than 100 years. Mean ACO period for the 325 cycles from 88,196 to 0.149 Kyb1950 (mean 295.9 years) is discernibly smaller than the mean ACO period for the 325 cycles from 226.4 to 88.196 Kyb1950 (mean 423.7 years) (t-test, p < 0.00001). This long-term change in TO_C350V cycle frequency is not attributable to variance from frequency aliasing, which is estimated as <0.5% (SM). Neither is this long-term change in TO_C350V cycle frequency attributable to variable sample resolution, which changes only slightly over long portions of the record (SM, Figure S2) while ACO frequency increases both systematically and episodically over the same time periods (Figure 9a).

The ACO is locked in phase with the integrated 41 Ky orbital Obliquity/Precession Orbital Cycle or OPOC [6,7] (Figure 9a). Peaks in TO_C350V cycle period occur in approximate phase with insolation minima at 65° N computed from the OPOC. Plotting the time series of cycle period for every TO_C350V cycle (Figure 9b) shows large, short-term or episodic fluctuations in TO_C350V period. High-frequency (short-period) bursts of TO_C350V cycles occur during the warm interstadials identified by odd-numbered MISs (Figure 9b), while low-frequency (long-period) TO_C350V cycles occur during the intervening cold stadials. Differences between stadial and interstadial TO_C350V cycle period for individual cycles approach 800%. These findings are consistent with the hypothesis that ACOs exhibit relatively high sensitivity to ambient temperature.

TO_C350V cycle period declines to near its lowest sustained level by the Holocene, a trend that continues and accelerates up to and throughout the CE. High-frequency TO_C350V cycles continue up to the CE, where TO_C350V cycles oscillate at a higher frequency than at any time during the previous 226.4 millennia (Figure 1 and Figure 9b). These observations confirm directly the decrease in cycle period (increase in cycle frequency) that is evident visually in climate record time series (e.g., Figure 1) and is shown by spectral analysis of different time periods of the same climate records (Figure 2, Figure 3, Figure 4 and Figure 7). These findings also demonstrate that the climate signals corresponding to ACOs cannot be attributed to “noise” except under the circular reductio ad absurdum that MIS cycles and corresponding orbital forces modulate noise in climate records to yield statistically-discernible variations, in which case they do not conform to the definition of noise.

3.6. ACO Amplitude

The mean amplitude of TO_C350V cycles computed from Vostok data is 0.67 °C over the 226.4 time period evaluated, ranging from 0.05–3.22 °C (Table 1). The increase in TO_C350V cycle amplitude suggested qualitatively by the time series of the last 20 Ky (Figure 1) is confirmed quantitatively for the entire 226.4 period of this analysis (Figure 10). Over this time period TO_C350V cycle amplitude increased by approximately two-thirds, from 0.52 to 0.82 °C (Figure 10a) in parallel with the decline in TO_C350V cycle period described above (Figure 9). By the end of the Holocene, centennial cycle amplitude routinely approached 2–3 °C. The average cycle amplitude of the 273 cycles that occurred from 226.4 to 88.196 Kyb1950 (0.62 °C) is discernibly smaller than the average cycle amplitude of the 273 cycles that occurred from 88.196 to 0.149 Kyb1950 (0.72 °C) (t-test, p < 0.000025), i.e., amplitude increased discernibly across the two time periods. The increase in TO_C350V cycle amplitude is not attributable to changes in sample resolution, which remained relatively stable as cycle amplitude increased over the same time periods (cf. Figure 10a with SM Figure S2).

The increase in TO_C350V cycle amplitude over 226.4 millennia is not uniform. Peaks in cycle amplitude occur in approximate antiphase with surface insolation minima at 65° N that result from planetary orbital cycles (Figure 10a; orbital parameters are from [6,7]). Plotting the time series of cycle amplitude over every TO_C350V cycle (Figure 10b) rather than in five Ky averages (Figure 10a) reveals large, short-term episodic fluctuations in TO_C350V amplitude. Peaks in TO_C350V cycle amplitude occur during warm interstadials associated with odd-numbered MISs (Figure 10b), when cycle period is smallest (Figure 9). Troughs in centennial-cycle amplitude occur during cold stadials corresponding to even-numbered MISs (Figure 10b), when cycle period is largest (Figure 9). The ratio of interstadial-to-stadial TO_C350V amplitude exceeds 60, suggesting that TO_C350V cycle amplitude is not only more variable than cycle frequency, but also several times more sensitive to temperature. TO_C350V cycle amplitude increases to its highest sustained level, 2–3 °C, during the Holocene and particular during the last millennium (Figure 1 and Figure 10b). These findings confirm directly the increase in cycle amplitude suggested in time series (e.g., Figure 1).

3.7. Orbital Influences

The period and amplitude of TO_C350V cycles appears qualitatively to be synchronized with variations in surface insolation induced by the 41 Ky integrated OPOC (Figure 9a and Figure 10a respectively). To test this hypothesis quantitatively, the time series of cycle period and amplitude were plotted on the same time scale as the time series of the OPOC [6,7] for the threshold glacial ablation insolation of τ = 325 W/m² [6,7]. ACO cycle period maximizes at the lowest NH insolation energies and minimizes during or near the highest NH insolation energies (Figure 11a). The lagged cross-correlation coefficients between insolation and TO_C350V cycle period (Figure 11b) alternate between positive and negative at the OPOC frequency and are discernible, in peaks and troughs of the autocorrelelogram, at probabilities from p < 0.05 to p < 10⁻¹⁰. TO_C350V cycle period is therefore locked in approximate antiphase with the surface insolation cycle induced by the integrated OPOC, i.e., less insolation is associated with lower ACO repetition frequency. As anticipated from the relation between cycle period and amplitude (Figure 9 and Figure 10), cycle amplitude also oscillates in approximate antiphase phase with surface insolation (Figure 12).

4. Discussion and Conclusions

Centennial-scale climate cycles reported previously by several investigators [33,40,52,56,57] and in this paper are significant in at least three contexts. First, centennial-scale climate cycles demonstrate “an important role of natural multicentennial variability that is likely to continue” [33] (p. 614). When both the mean and variance of any centennial-scale climate cycle are known, as is the case for the TO_C350V cycles documented here (Table 1), then the future behavior of such cycles can be projected within well-defined confidence limits. Understanding centennial-scale temperature cycles can therefore contribute to precise climate projections over timelines that are most pertinent to human and civilizational life cycles, decades to centuries. This approach to the projection of future climate change has been pioneered by Liu and colleagues [56] based on analysis of tree ring data from the Tibetan Plateau. From past centennial-scale temperature oscillations, they project a steep decline of temperature on the Tibetan Plateau of ~3 °C between 2006 and 2068, followed by a weaker warming trend and continuing on a cyclic basis into the future [56] (their Figure 5, p. 2292).

Second, centennial-scale paleoclimate cycles comprise a “natural” source of temperature forcing, i.e., one that is free from anthropogenic influences. Human impact on global climate from agriculture and land clearing may have begun as early as the mid-Holocene [58,59], but earlier climate change was presumably devoid of anthropogenic influences. Characterizing past cycles of temperature fluctuation can therefore help inform the distinction between natural (non-anthropogenic) and anthropogenic forcing of climate in the present, as discussed further below.

Third, Antarctic temperature fluctuations on several time scales are reflected worldwide and in the NH after a delay of 0.5 to 3.0 millennia [40,49,52,53,60,61,62,63,64,65,66,67,68,69,70,71]. These delays were measured for older time periods, however, generally before the LGT, and may be shorter for more recent climate events in a warmer environment (see below). Given the close association between AIMs in the Antarctic and D-O events in the NH, as demonstrated repeatedly by previous investigators [38,40,49,52], the discovery here that AIMs are composed of summated TO_C350V cycles constitutes strong evidence that ACOs manifest globally. The centennial-scale climate cycles identified in the NH [33,56,57] may be northerly manifestations of the Antarctic TO_C350V climate cycle documented here, a hypothesis that remains to be tested. In the meantime, the present findings demonstrate that the ACO and its potential modern counterpart (the AAO; see below) influence the temperature of the NH. This finding suggests a potentially-fruitful research direction aimed at assessing the impact of the contemporary AAO on global climate and weather. Our study raises the possibility that the ACO/AAO entrains global temperature and serves as the primary pacemaker of centennial fluctuations in temperature in both hemispheres while simultaneously modulating shorter cycles.

4.1. Geographic Distribution of ACO Cycles

Comparison of TO_C350V cycles across the four AICC2012-synchronized climate records evaluated here supports four conclusions and interpretations. First, spectral density peaks from the Vostok paleoclimate record since the LGM are matched (±2.4%) with peaks from other EAP climate records synchronized on the AICC2012 chronology. In addition, 100% of signpost cycles and more than 98% of all single, individual TO_C350V cycles evaluated at Vostok match one-to-one with homologs at the three other drill sites evaluated here and conversely. Therefore, TO_C350V cycles occur at all four AICC2012-synchronized drill sites on the EAP despite the separation of the sites by up to thousands of km and elevation differences of up to more than a kilometer [51].

Second, the time latencies between the peaks of individual, identified TO_C350V signpost cycles in different climate records (“homologs”) varied systematically by up to several centuries. These latencies exceed reported variance arising from dating uncertainty in the AICC2012 chronology by up to nearly 40-fold, and therefore cannot be ascribed to chronological uncertainty or “noise”. Prior to the LGT, identified TO_C350V cycles are detectable first at TD and EDML and appear centuries later at EDC and Vostok. In this limited sample of four drill sites, therefore, the locations where TO_C350V cycles can first be detected (TD and EDML) are closer to the ocean, lower in elevation, warmer and wetter than those where homologous TO_C350V cycles occur centuries later on the higher, drier and colder EAP (EDC and Vostok). This timing sequence encourages future research directions on the origin of TO_C350V cycles in which the sample is expanded to the eleven best-studied Antarctic drill sites to evaluate propagation latencies and directions using a larger sample size. Backward extrapolation of these cycle parameters may pinpoint the geographic origin of the ACO cycle.

Third, the measured latencies between peaks of homologous TO_C350V cycles at different drill sites vary and reverse on short (centennial) time scales. Before the LGT, TO_C350V cycle homologs across the climate records evaluated here were asynchronous, but after the LGT and during the warmer Holocene, homologous TO_C350V cycles became synchronous, i.e., the latency of homologous ACOs at different sites approached zero. Some of the reduction in latency between homologous cycles at different drill sites is expected from the observed increase in cycle frequency over time and the consequent compression of individual cycles. Most of the change in latency, however, arises from genuine (non-artifactual) differences in the arrival time of homologous TO_C350V cycle peaks at different drill sites. As noted, this difference in latencies typically exceeds chronological uncertainty in the AICC2012 climate records studied here by more than an order of magnitude. This finding supports the hypothesis that the propagation velocity of ACOs increases with temperature, which in turn bears upon the mechanism(s) of regional cycle teleconnection that remain to be investigated.

Fourth, comparison of homologous TO_C350V cycles across four Antarctic climate records reveals a “climate memory” in Antarctica, reflected in the conservation of centennial cycle architecture between homologous cycles at different drill sites as TO_C350V cycles propagate across Antarctica. The transition from asynchrony to synchrony of TO_C350V cycle homologs during the most recent glacial termination and the Holocene suggests that this climate memory fades and disappears with rising temperature. The mechanism(s) of this climate memory are unknown, but a plausible candidate is hysteresis in the temperature/ice system, since these components of the Antarctic climate exhibit the prerequisite long time constants and temperature sensitivity [9].

4.2. Relation of the ACO to Contemporary Climate Cycles

The TO_C350V climate cycle documented in this study for the past 226.4 millennia presumably continues to oscillate into the present. The most likely contemporary candidate for such a comparison between past and present manifestations of the ACO is the interrelated AAO/SAM/SOI cycle (Introduction). The geographic scope and repetition period of this contemporary climate oscillation is similar to TO_C350V cycles over the recent Holocene, and to our knowledge there is no other tenable candidate for this comparison in the SH based on the geographic scope or the oscillatory period of known contemporary climate cycles.

This hypothesis receives support from the observation that the TO_C350V cycles identified here are similar in period and relative amplitude to comparable cycles of the AAO that have been identified independently from more contemporary tree-ring data [26,27] and from stacked temperature-proxy records from Vostok snow pits [32]. Similarly, westerly wind records in the SH (Falkland Islands) for the past 2600 years as assessed from records of pollen distribution, show discernible AAO spectral peaks at 242, 185, 150 and 95 years that average 168 years [72] (Figure 5a, p. 194), similar to the 193-year discernible peak in the Holocene spectral periodogram of the ACO (Figure 3) and the 162-year average period over the last 2,600 years (Figure 1d). We therefore conclude, subject to confirmatory research, that the ACO is the probable paleoclimate precursor of, and is continued into the present by, the contemporary Antarctic Oscillation (AAO/SAM/SOI).

4.3. Intensification of the Antarctic Climate

The demonstration here that TO_C350V cycles increased in frequency and amplitude over hundreds of millennia indicates that temperature fluctuations in the Antarctic have become more frequent, rapid, and extreme over geologic time, i.e., the ACO has strengthened during the 226-millenium time period encompassed by this study. Similarly, the comparison here of spectral periodograms corresponding to different time periods suggests that the TO_K1500V cycle declined in period by ~50% over the past 70 millennia, from about 1500 years to 800 years. This millennial cycle oscillates at the same frequency as the AIMs known to precede D-O events in the NH [23,38]. These findings support the hypothesis that the Antarctic climate is dynamic and has intensified by a factor of two over the past ~226 millennia. The strengthening of the Antarctic climate has implications for contemporary and future regional and global climate as discussed further below.

These findings show that contrary to past assumption, climate cycles such as the ACO and the millennial-scale TO_K1500V cycle that paces AIMs do not comprise stationary time series. This conclusion has implications both for analytic methodologies that can be brought to bear on climate data and also for the interpretation of natural versus anthropogenic influences as discussed below. In respect to the methodological issues, many analytic approaches to climate data are based on the assumption of linearity and a stationary time series. Spectral analysis, for example, which is based on Fourier transformation, is technically applicable only to stationary time series. Our study shows that climate data do not necessarily comprise a stationary time series, yet spectral analysis is used almost universally in climate science, including in most of the references we cite here (e.g., [2] (Figure 12, p. 123), [3] (Figures 7 and 10), [4] (Figure 4, p. 689), [5] (Figure 12, p. 14), [6] (Figure 2b,d,f, p. 509), etc. Our study suggests the need for a comprehensive expert review of methodological approaches to the analysis of climate data in view of potential non-stationarity.

The dilemma for the present work is that we used standard techniques of climate science in a study whose results show that these techniques may not be well-suited to climate data. This dilemma constitutes a limitation of this study. We believe that this limitation is minor because we evaluated primarily relatively short segments of the climate record, which comprise relatively stationary time series as demonstrated by strongly-positive auto- and cross-correlation results. At the same time, our study suggests that it would now be fruitful to apply the more reliable and modern methods of Wavelet Analysis, Empirical Mode Decomposition, Hilbert-Huang Transforms, etc., to the climate data analyzed here. These data are presented in full in the SM Table S1 for exactly such purposes.

The increase in frequency over time of both the TO_C350V and TO_K1500V cycles found here in the Antarctic may help resolve the debate over the oscillatory period of related Greenland Interstadials (GIs) including both Heinrich events and D-O oscillations [22,23]. In particular, the 1470-year cycle period of D-O events (GIs) in Greenland [20] has been characterized as “potentially an artifact of arithmetic averaging” [23] (p. 23). Our finding here that the repetition period of a comparable cycle in the Antarctic (the TO_K1500V cycle) declines from ~1500 to 800 years over the last 70 Ky suggests an alternative explanation, namely, that spectral peaks of climate oscillations in the NH represent discrete steps in a variable, non-stationary time series composed of several discernible spectral density peaks that nonetheless originate from a single, variable millennial-scale climate cycle. This proposal is consistent with the discovery of several statistically-discernible spectral peaks in Greenland ice core data ranging from 1500 to 800 years, the same range found here in the SH for the last 70 Ky, and which have also been attributed in part to a non-stationary time series [23].

4.4. Climate Stability and the Origin of D-O Oscillations

Rapid fluctuations in temperature-proxy records during the Holocene and earlier have long been interpreted as indicators of an unstable and chaotic global climate ([16,73,74,75,76] and references therein). In contrast, our study shows that such temperature-proxy fluctuations comprise a well-defined, predictable, centennial-scale climate oscillation with quantifiable descriptive statistics (Table 1). These cycles vary systematically with multi-millennial glacial cycles (MISs) and with related and presumably causal planetary orbital forces (the integrated OPOC), or Milankovitch cycles, and therefore cannot be attributed to “noise”. It appears, therefore, that the observed temperature-proxy fluctuations over the past hundreds of millennia do not imply instability of a chaotic global climate, but are indicative instead of exactly the opposite—a highly stable and predictable climate consisting of systematic changes in temperature that follow quantified oscillatory parameters, namely a centennial climate cycle that interacts with a millennial climate cycle as documented here.

Dansgaard-Oeschger (D-O) events in the NH, for example, recently termed Greenland Interstadials or GIs [23], have been portrayed as exemplars of climate instability [16] and “abrupt climate change” [77]. D-O events manifest in Greenland ice cores as a periodic sudden (decadal) and "runaway" (positive feedback) exponential warming of several °C followed by a longer and more linear cooling recovery. Computer simulations have been interpreted to imply that such a sudden change in climate “inherently limits [climate] predictability” [77] (p. 111). In contrast to this interpretation, the empirical analysis reported here shows that demonstrated precursors to D-O (GI) events in the Antarctic, namely AIMs, are composed of several TO_C350V cycles that recur in rapid succession at the highest observed TO_C350V repetition frequencies, driven by the TO_K1500V cycle. These TO_C350V cycles add together over time (summate) with millennial periodicity to form AIMs, manifesting the TO_K1500V cycle which was identified previously in the NH as the Bond cycle [17,18,19,20,21]. The only prerequisite for such “temporal summation” of TO_C350V cycles in the SH is their recurrence at a sufficiently high frequency to ensure that one TO_C350V cycle begins before the preceding cycle ends, so that successive ACOs “piggyback” on one another to create a larger AIM.

Central to this interpretation, and as noted above, AIMs in the SH have been established by several previous investigators as precursors of D-O (GI) events in the NH [38,40,49,52]. Each AIM precedes its corresponding D-O (GI) event in the NH by 1.5–3.0 millennia [40], linked by hemispheric teleconnection that has been attributed to the bipolar climate see-saw [40], although this teleconnection time may be smaller in the more recent Holocene (see below). The discovery here that AIMs are composed of identifiable TO_C350V cycles that maximize periodically, paced by the TO_K1500V cycle in the SH, provides a novel yet straightforward hypothesis for the origin of D-O (GI) oscillations in the NH. By this hypothesis, temporally-summated TO_C350V cycles generate AIMs in the SH with millennial periodicity, following the TO_K1500V cycle, and these AIMs in turn teleconnect by undemonstrated mechanisms to the NH, where they induce D-O (GI) events.

It follows that D-O (GI) events are neither attributable to nor diagnostic of climate “instability”, but are instead caused by systematic and predictable interactions between centennial-scale oscillatory climate cycles in Antarctica that are modulated on a variable millennial time scale in phase with the TO_K1500V cycle. This hypothesis implies that abrupt millennial-scale climate changes such as D-O (GI) events, and more gradual natural climate change on centennial scales, are inherently and precisely predictable. This conclusion in turn imparts new statistical power to projections of future climate, as exemplified by the work of Liu et al. [56]. This hypothesis can be tested potentially by examining counterparts to TO_C350V cycles in the NH, which is however beyond the scope of the present study.

4.5. Implications for Contemporary Global Warming

Recent changes in the AAO/SAM toward the most positive (meridionally-contracted) phase seen in several centuries have been attributed to increased atmospheric concentrations of anthropogenic greenhouse gases [26,27]. This conclusion assumes, however, that “natural” (non-anthropogenic) climate change comprises a stationary time series, enabling potential anthropogenic forcing signals to emerge from a flat “natural” baseline [78]. This assumption is falsified by the findings of the present study, which show that the frequency and the amplitude of TO_C350V cycles increased together over the last 226.4 millennia and also on a regular millennial schedule. Comparative spectral analysis of different time periods reported here suggests a similar conclusion for the TO_K1500V cycle.

The conclusion that Antarctic climate cycles increase in intensity over time and, therefore, do not comprise a stationary time series, has implications for understanding human impacts on the global climate. During most of the 226 millennia evaluated in the present study, humans exerted little or no influence on climate. The evolution of the SAM toward its most positive phase in several centuries during the last millennium may therefore result from the “natural” (non-anthropogenic) intensification of the Antarctic climate documented in this study, perhaps following the pattern identified here of millennial-scale increases in amplitude and frequency of ACOs terminated by a single large temperature peak and trough.

Consistent with this conclusion, it has been reported recently that natural variability in Antarctic temperature “overwhelms” any possible anthropogenic forcing of temperature for at least the past two centuries [79] (p. 917). In this case, and assuming that the ACO is the climatic precursor of the AAO, it may be concluded that the recent changes in the AAO toward the most positive phase seen in several centuries result not from increased atmospheric concentrations of anthropogenic greenhouse gases [26,27], but rather from the natural increases in amplitude and frequency of the ACO as documented in the present study.

The finding here of a natural climate cycle that oscillates on a centennial scale and impacts temperature in the NH has further implications for the debate about anthropogenic global warming (AGW). It has been considered widely that atmospheric carbon dioxide (CO₂) emitted by human activities is responsible for the contemporary global warming signal. Recent research shows, however, that neither atmospheric CO₂ nor marginal radiative forcing by CO₂ is generally correlated discernibly with global temperature over the past 425 million years [3]. Therefore, neither atmospheric CO₂ nor marginal radiative forcing by CO₂ caused temperature change in the ancient climate [3]. The same study showed that contemporary marginal radiative forcing by atmospheric CO₂ has declined by approximately two-thirds since the beginning of the Industrial Age owing to the well-known logarithmic relationship between radiative forcing and atmospheric CO₂ concentration [3]. Progressively larger concentrations of atmospheric CO₂ therefore yield progressively smaller global temperature change, minimizing any possible impact of CO₂ on global temperature.

A primary rationale for invoking anthropogenic emissions of CO₂ as the cause of contemporary global warming has been the perceived lack of alternative explanations, particularly the absence of a natural climate cycle that might account for the current global warming signal of ~0.8 °C since 1850. The 2017 US Global Change Research Program Report on climate [41] concluded, for example, that “For the warming over the last century, there is no convincing alternative explanation supported by the extent of the observational evidence” [41] (p. 12). The empirical documentation here of a natural paleoclimate climate cycle that impacts temperature in the NH and that is probably homologous with the contemporary AAO, namely, the ACO, provides such a possible alternative explanation. By this alternative explanation, the global temperature profile since 1850 is caused in whole or part by a natural climate cycle (the ACO/AAO) that is approaching or has reached its temperature peak. The peaking of this natural warming cycle would also explain the recorded deceleration of warming over the last two decades, termed the “warming hiatus” [80,81,82].

A key falsifying test of this “natural global warming” (NGW) hypothesis as a viable alternative to the AGW hypothesis revolves around timing. The last millennium at Vostok has encompassed eight ACO cycles, whose average period is, therefore, 125 years (Figure 1 and SM Table S1). The most recent recorded temperature peak at Vostok occurred in 1760. Incrementing this date by the average ACO period of 125 years over the past millennium projects ACO peaks at Vostok in 1885 and 2010. The latter date is consistent with the possible peaking of the contemporary global warming trajectory in the NH, assuming acceleration of ACO teleconnection velocities with increasing temperature as we suggest here for regional teleconnection within the SH. Pedro and colleagues have also reported that changes in Antarctic and Greenland climate occur with “little-to-no time lag between climate transitions” [53] (p. 671), consistent with the hypothesis that the ACO influences NH temperature with vanishingly-small latency in the recent paleoclimate record (the late Holocene). Similarly, “climate changes at Taylor Dome during the last glacial-interglacial transition were synchronous or near synchronous with changes in the North Atlantic region” [62] (p. 94).

This analysis of timing remains simplistic and preliminary in that it does not account for variance in ACO period and amplitude nor measured teleconnection times, but it nonetheless demonstrates to a first approximation that the ACO, in its proposed modern form of the AAO, is a plausible candidate for the source of contemporary global warming. This NGW hypothesis merits and requires further research as an alternative to the AGW hypothesis. As established by previous research [3], however, the possibility that contemporary global warming results from a natural climate cycle, rather than from anthropogenic emissions of CO₂, may not lessen the urgency to consider limiting anthropogenic CO₂ emissions. Peaks in atmospheric CO₂ concentration in the ancient climate [3] coincide closely with peaks in past mass extinctions [83], mediated perhaps by acidification of ocean water [84,85]. Therefore, irrespective of the source of the contemporary global warming signal, and subject to further research, limiting anthropogenic CO₂ emissions through existing and new international policy instruments such as the U.N. Framework Convention on Climate Change and the evolving Paris Climate Accords, respectively, may prove essential to preventing human-induced mass extinction of biodiversity.