Defining the beginning of the Anthropocene as a formal geologic unit of time requires the location of a global marker of an event in stratigraphic material, such as rock, sediment, or glacier ice, known as a Global Stratotype Section and Point (GSSP), plus other auxiliary stratigraphic markers indicating changes to the Earth system. Alternatively, after a survey of the stratigraphic evidence, a date can be agreed by committee, known as a Global Standard Stratigraphic Age (GSSA). GSSPs, known as ‘golden spikes’, are the preferred boundary markers10 (see Box 1).

Generally, geologists have used temporally distant changes in multiple stratigraphic records to delimit major changes in the Earth system and thereby geological time units, for example, the appearance of new species as fossils within rocks, coupled with other temporally coincident changes. Perhaps the most useful GSSP example when considering a possible Anthropocene GSSP is that marking the beginning of the most recent epoch, the Holocene38, because some similar choices and difficulties were faced. These include: not relying on solid aggregate mineral deposits (‘rock’) for the boundary; an event horizon largely lacking fossils (although fossils are used to recognize Holocene deposits); the need for very precise GSSP dating of events in the recent past; and how to formalize a time unit that extends to the present and thereby implicitly includes a view of the future.

Depending on the parameter considered, the current interglacial took decades to millennia to unfold, as global climate, atmospheric chemistry and the distribution of plant and animal species all altered. From these changes a single dated level within a single stratigraphic record was required to be chosen as a GSSP primary marker (Box 1; Fig. 2). Thus, formally, the Holocene is marked by an abrupt shift in deuterium (2H) excess values at a depth of 1,492.25 m in the NorthGRIP Greenland ice core, dated 11,650 ± 99 yr bp (before present, where ‘present’ is defined to be 1950)38. This corresponds to the first signs of predominantly Northern Hemisphere climatic warming at the end of the Younger Dryas/ Greenland Stadial 1 cold period38 (Fig. 2). Five further auxiliary stratotypes (four lakes and one marine sediment) showing clear correlated changes across the boundary complement the GSSP, consistent with the occurrence of global changes to the Earth system38. The requirements for a formal definition of the start of the Anthropocene are similar: a clear, datable marker documenting a global change that is recognizable in the stratigraphic record, coupled with auxiliary stratotypes documenting long-term changes to the Earth system.

Figure 2: Defining the beginning of the Anthropocene. a, Current GTS2012 GSSP boundary between the Pleistocene and Holocene38 (dashed line), with global temperature anomalies (relative to the early Holocene average over the period 11,500 bp to 6,500 bp)112 (blue), and atmospheric carbon dioxide composite113 on the AICC2012 timescale114 (red). b, Early Anthropogenic Hypothesis GSSP suggested boundary (dashed line), which posits that early extensive farming impacts caused global environmental changes, defined here by the inflection and lowest level of atmospheric methane (in parts per billion, p.p.b.) from the GRIP ice core59 (green), with global temperature anomalies (relative to the average over the period 1961 to 1990)115 (blue), and atmospheric carbon dioxide113 (red). c, Orbis GSSP suggested boundary (dashed line), representing the collision of the Old and New World peoples and homogenization of once distinct biotas, and defined by the pronounced dip in atmospheric carbon dioxide (dashed line) from the Law Dome ice core75,76 (blue), with global temperature data anomalies (relative to the average over the period 1961 to 1990)115 (red). d, Bomb GSSP suggested boundary (dashed line), characterized by the peak in atmospheric radiocarbon from annual tree-rings (black)103 (the Δ14C value is the relative difference between the absolute international standard (base year 1950) and sample activity corrected for the time of collection and δ13C), with atmospheric carbon dioxide from Mauna Loa, Hawaii, post-1958116, and ice core records pre-195875,76 (red), and global temperature anomalies (relative to the average over the period 1961 to 1990)116 (blue). Full size image Download PowerPoint slide

Defining the Anthropocene presents a further challenge. Changes to the Earth system are not instantaneous. However, even spatially heterogeneous and diachronous (producing similar stratigraphic material varying in age) changes appear near-instantaneous when viewed millions of years after the event, especially as time-lags often fall within the error range of the dating techniques. In contrast, Anthropocene deposits are commonly dated on decadal or annual scales, so that all changes will appear diachronous, to some extent, from today’s perspective (but not from far in the future)11,41. Judgement will be required to assess whether the time-lags following events and their significant global impacts are too long to be of use when defining any Anthropocene GSSP.

Several approaches have been put forward to define when the Anthropocene began, including those focusing on the impact of fire42, pre-industrial farming43,44,45, sociometabolism46, and industrial technologies1,39,40,41,47, but the relative merits of the evidence for various starting dates have not been systematically assessed against the requirements of a golden spike. Below, we review the major events in human history and pre-history and their impact on stratigraphic records. We focus on continuous stratigraphic material that may yield markers consistent with a GSSP (lake and marine sediments, glacier ice) and on the types of chemical, climatic and biological changes used to denote other epoch boundaries further in the past. We proceed chronologically forward in time, presenting the reason why each event was originally proposed, evaluate the existence of stratigraphic markers, and assess whether the event provides a potential GSSP. The hypotheses and evidence are summarized in Table 1. Following the evidence review we briefly consider the relative merits of the differing events that probably fulfil the GSSP criteria, and assess related GSSA dates.

Table 1: Potential start dates for a formal Anthropocene Epoch Full size table

Pleistocene human impacts

The first major impacts of early humans on their environment was probably the use of fire. Fossil charcoal captures these events from the Early Pleistocene Epoch42,48. However, fires are inherently local events, so they do not provide a global GSSP. The next suggested candidate is the Megafauna Extinction between 50,000 and 10,000 years ago, given that other epoch boundaries have been defined on the basis of extinctions or on the resultant newly emerging species10. Overall, during the Megafauna Extinction about half of all large-bodied mammals worldwide, equivalent to 4% of all mammal species, were lost49. The losses were not evenly distributed: Africa lost 18%, Eurasia lost 36%, North America lost 72%, South America lost 83%, and Australia lost 88% of their large-bodied mammalian genera50,51. So the Megafauna Extinction was actually a series of events on differing continents at differing times and therefore lacks the required precision for an Anthropocene GSSP marker.

Origins and impacts of farming

The development of agriculture causes long-lasting anthropogenic environmental impacts as it replaces natural vegetation, and thereby increases species extinction rates, and alters biogeochemical cycles. Agriculture had multiple independent origins: first occurring about 11,000 years ago in southwest Asia, South America and north China; between 6,000–7,000 years ago in Yangtze China and Central America; and 4,000–5,000 years ago in the savanna regions of Africa, India, southeast Asia, and North America52. Thus, the increasing presence of fossil pollen from domesticated plants in sediment is too local and lacking in global synchrony to form a GSSP marker. Critically, for the Holocene GSSP, auxiliary markers within stratigraphic material did not include any human-derived markers38, illustrating the lack of anthropogenic impacts at that time. Long-lasting cultural evidence related to agriculture is similarly constrained. Although ceramics are datable and preserved in stratigraphic records (for example, the mineral mullite41), they appeared in Africa before agriculture, while early southwest Asian farming cultures did not produce ceramics. Similarly, anthropogenically formed soils, derived from intensive farmland management, have also been suggested as a marker of the Anthropocene53. Although these soils are widespread, like vegetation clearance, they are highly diachronous over about 2,000 yr, thus excluding their use as a GSSP marker54.

A series of Neolithic revolutions resulted in the majority of Homo sapiens becoming agriculturalists to some extent by around 8,000 yr bp, rising to a maximum of about 99% by about 500 yr bp46. The Early Anthropogenic Hypothesis posits that the current interglacial was similar to the previous seven interglacial periods until around 8,000 yr bp43,55. By comparison with the closest astronomical analogue of the current interglacial (795,000–780,000 yr bp)55, atmospheric CO 2 should have continued to decline after 8,000 yr bp, eventually reaching about 240 parts per million (p.p.m.), and the onset of glaciation should have begun43,55. However, by 6,000–8,000 yr bp, farmers’ conversion of high-carbon storage vegetation (forest, woodland, woody savanna) to crops and grazing lands, plus associated fire impacts, may have increased atmospheric CO 2 levels, and postponed this new glaciation43 (Fig. 2). Thus, the lowest level of CO 2 within an ice core record could, in principle, provide a golden spike, but the CO 2 record lacks a distinct inflection point at this time (Fig. 2). Furthermore, the evidence that human activity was responsible for the gradual increase in CO 2 after 6,000 yr bp is extensively debated43,56,57,58.

Methane provides a clearer inflection point, which may provide a possible GSSP at 5,020 yr bp, the date of the lowest methane value recorded in the GRIP ice core59 (Fig. 2). Archaeological evidence suggests that the inflection is caused by rice cultivation in Asia and the expansion of populations of domesticated ruminants. Comparisons of changes in atmospheric methane from the current and past interglacials43, and some methane δ13C value evidence60, also suggest a human cause. However, a model study suggests that orbital forcing altering methane emissions from tropical wetlands may be responsible61. Auxiliary markers could include stone axes and fossilized domesticated crop pollen and ruminant remains, but these do not provide temporally well-correlated markers that collectively document globally synchronous changes to the Earth system.

Collision of the Old and New Worlds

The arrival of Europeans in the Caribbean in 1492, and subsequent annexing of the Americas, led to the largest human population replacement in the past 13,000 years62, the first global trade networks linking Europe, China, Africa and the Americas63,64, and the resultant mixing of previously separate biotas, known as the Colombian Exchange63,64. One biological result of the exchange was the globalization of human foodstuffs. The New World crops maize/corn, potatoes and the tropical staple manioc/cassava were subsequently grown across Europe, Asia and Africa. Meanwhile, Old World crops such as sugarcane and wheat were planted in the New World. The cross-continental movement of dozens of other food species (such as the common bean, to the New World), domesticated animals (such as the horse, cow, goat and pig, all to the Americas) and human commensals (the black rat, to the Americas), plus accidental transfers (many species of earth worms, to North America; American mink to Europe) contributed to a swift, ongoing, radical reorganization of life on Earth without geological precedent.

In terms of stratigraphy, the appearance of New World plant species in Old World sediments—and vice versa—may provide a common marker of the Anthropocene across many deposits because pollen is often well preserved in marine and lake sediments. For example, pollen of New World native Zea mays (maize/corn), which preserves very well41, first appears in a European marine sediment core in 160065. The European Pollen Database lists a further 70 lake and marine sediment cores containing Zea mays after this date. Phytoliths can similarly record such range expansions66. Specifically, the transcontinental range extension of at least one Old World species into the New World (banana, as phytoliths in Central and tropical South America sediments) and a second species from the New World expanding into the Old World (maize/corn, as pollen preserved in sediments in Eurasia and Africa) together constitute a unique signature in the stratigraphic record. This transcontinental range expansion—stratigraphically marking before and after an event—is comparable to the use of the appearance of new species as boundary markers in other epoch transitions49,67.

Besides permanently and dramatically altering the diet of almost all of humanity, the arrival of Europeans in the Americas also led to a large decline in human numbers. Regional population estimates sum to a total of 54 million people in the Americas in 149268, with recent population modelling estimates of 61 million people58. Numbers rapidly declined to a minimum of about 6 million people by 1650 via exposure to diseases carried by Europeans, plus war, enslavement and famine58,63,68,69. The accompanying near-cessation of farming and reduction in fire use resulted in the regeneration of over 50 million hectares of forest, woody savanna and grassland with a carbon uptake by vegetation and soils estimated at 5–40 Pg within around 100 years58,70,71,72. The approximate magnitude and timing of carbon sequestration suggest that this event significantly contributed to the observed decline in atmospheric CO 2 of 7–10 p.p.m. (1 p.p.m. CO 2 = 2.1 Pg of carbon) between 1570 and 1620 documented in two high-resolution Antarctic ice core records73,74,75,76 (Fig. 2 and Box 2). This dip in atmospheric CO 2 is the most prominent feature, in terms of both rate of change and magnitude, in pre-industrial atmospheric CO 2 records over the past 2,000 years75 (Fig. 2).

Box 2: Origins of the 1610 decrease in atmospheric CO 2 Is the CO 2 decline real? Two independent high-resolution Antarctic ice core records from the Law Dome and the Western Antarctic Ice Sheet show a reduction in atmospheric CO 2 of 7–10 p.p.m. between 1570 and 162073,74,75 (Fig. 2). A smaller CO 2 decrease is also observed in less highly resolved Antarctic cores117,118. The decline exceeds the measurement error of the cores, 1–2 p.p.m., and experiments suggest that it does not result from in situ changes within the ice core119. Did human activity cause the decline? The arrival of Europeans in the Americas led to a catastrophic decline in human numbers, with about 50 million deaths between 1492 and 1650, according to several independent sources58,63,68,69. Contemporary field observations of soil120 and vegetation121 carbon dynamics following agriculture abandonment suggest that about 65 million hectares (that is, 50 million people × 1.3 hectares per person) would sequester 7–14 Pg of carbon over 100 years (that is, 100–200 Mg of carbon per hectare total uptake, above- and below-ground). Reduction in fire use for land management would additionally increase carbon uptake outside farmed areas. Studies using a variety of methods report broadly consistent estimates58,70,71,72 of carbon uptake by vegetation of 5–40 Pg (2.1 Pg of carbon = 1 p.p.m. atmospheric CO 2 over shorter timescales, lessening over time127). Given that maximum human mortality rates were not reached for some decades after 149262,63, and maximum carbon uptake would take place 20–50 yr after farming abandonment, peak carbon sequestration would occur approximately between 1550 and 1650. Some model studies spanning thousands of years find a net land surface carbon uptake spanning 1500–1650 across the Americas58, while others do not122. However, in general, evidence from such studies weakly constrain the problem because Holocene carbon cycle modelling is designed to investigate changes associated with long-acting slow processes (carbon uptake by peat or coral reefs) and feedback mechanisms (oceanic outgassing, oceanic uptake and CO 2 fertilization of vegetation), and probably poorly represent the short period of the CO 2 dip (for example, ref. 57). For example, a study calculating a net zero impact of the cessation of farming in the Americas122 included a large soil carbon flux to the atmosphere, which contradicts field evidence120,123, and had the effect of offsetting the uptake from growing trees122. Carbon cycle models with robust representations of land-use change and subsequent vegetation regeneration following the Americas population catastrophe will be required to improve estimates of carbon uptake compared with carbon accounting studies. The approximate magnitude and timing of carbon sequestration make the population decline in the Americas the most likely cause of the observed decline in atmospheric CO 2 . Atmospheric74,124,125 and tropical marine δ13C analyses126 also support uptake of CO 2 by vegetation rather than oceanic uptake. The 1600 Huaynaputina eruption in Peru78,79 probably exacerbated the CO 2 minima, and a lagged oceanic outgassing in response to the land carbon uptake probably contributed to the fast rebound of atmospheric CO 2 after 1610127. In addition, multi-proxy reconstructions of temperature indicate that, after accounting for both solar and volcanic radiative forcing, additional terrestrial carbon uptake is required to explain temperature declines over the 1550–1650 period107. This is consistent with uptake by vegetation following the population crash in the Americas107.

On the basis of the movement of species, atmospheric CO 2 decline and the resulting climate-related changes within various stratigraphic records, we propose that the 7–10 p.p.m. dip in atmospheric CO 2 to a low point of 271.8 p.p.m. at 285.2 m depth of the Law Dome ice core75, dated 1610 (±15 yr; refs 75, 76), is an appropriate GSSP marker (Fig. 2). Auxiliary stratotypes could include: the first occurrence of a cross-ocean range extension in the fossil record (Zea mays, in 160065) plus a range of deposits showing distinct changes at that time, including tephra77,78 and other signatures from the 1600 Huaynaputina eruption detected at both poles and in the tropics77,78,79; charcoal reductions in deposits in the Americas71 and globally80; decreases in atmospheric methane, enrichment of methane δ13C, and decreases in carbon monoxide in Antarctic ice cores60,81,82,83,84; pollen in lacustrine sediments showing vegetation regeneration85; proxies indicating anomalous Arctic sea-ice extent86; changing δ18O derived from speleothems from caves in China and Peru14 and other studies noting changes coincident with 1600 and the coolest part of the Little Ice Age (1594–1677; ref. 87), a relatively synchronous global event noted in geologic deposits worldwide87.

The impacts of the meeting of Old and New World human populations—including the geologically unprecedented homogenization of Earth’s biota63,64—may serve to mark the beginning of the Anthropocene. Although it represents a major event in world history62,63,64,88, the collision of the Old and New Worlds has not been proposed previously, to our knowledge, as a possible GSSP. We suggest naming the dip in atmospheric CO 2 the ‘Orbis spike’ and the suite of changes marking 1610 as the beginning of the Anthropocene the ‘Orbis hypothesis’, from the Latin for world, because post-1492 humans on the two hemispheres were connected, trade became global, and some prominent social scientists refer to this time as the beginning of the modern ‘world-system’89.

Industrialization

The beginning of the Industrial Revolution has often been suggested as the beginning of the Anthropocene, because accelerating fossil fuel use and coupled rapid societal changes herald something important and unique in human history1,2,3,4,39. Yet humans have long been engaging in industrial-type production, such as metal utilization from around 8,000 yr bp onwards, with attendant pollution90. Elevated mercury records are documented at around 3,400 yr bp in the Peruvian Andes91, while the impacts of Roman Empire copper smelting are detectable in a Greenland ice core at around 2,000 yr bp92. This metal pollution, like other examples predating the Industrial Revolution, is too local and diachronous to provide a golden spike.

Definitions of the Industrial Revolution give an onset date anywhere between 1760 and 1880, beginning as an event local to northwest Europe88. Given the initial slow spread of coal use, ice core records show little impact on global atmospheric CO 2 concentration until the nineteenth century, and then they show a relatively smooth increase rather than an abrupt change, precluding this as a GSSP marker (Fig. 2). Similarly, other associated changes, including methane and nitrate15, products of fossil fuel burning (including spherical carbonaceous particles93 and magnetic fly ash94) plus resultant changes in lake sediments95,96 alter slowly as the use of fossil fuels increased over many decades. Lead, which was once routinely added to vehicle fuels, has been proposed as a possible marker, because leaded fuel was almost globally used and is now banned97. However, peak lead isotope ratio values from this source in sediments and other deposits vary from 1940 to after 1980, limiting the utility of this marker. The Industrial Revolution thus provides a number of markers spreading from northwest Europe to North America and expanding worldwide since about 1800, although none provides a clear global GSSP primary marker.

The Great Acceleration

Since the 1950s the influence of human activity on the Earth system has increased markedly. This ‘Great Acceleration’ is marked by a major expansion in human population, large changes in natural processes3,12,98, and the development of novel materials from minerals to plastics to persistent organic pollutants and inorganic compounds41,47,97. Among these many changes the global fallout from nuclear bomb tests has been proposed as a global event horizon marker41,47. The first detonation was in 1945, with a peak in atmospheric testing from the late 1950s to early 1960s, followed by a rapid decline following the Partial Test Ban Treaty in 1963 and later agreements, such that only low test levels continue to the present day (Fig. 2). A resulting distinct peak in radioactivity is recorded in high-resolution ice cores, lake and salt marsh sediments, corals, speleothems and tree-rings from the early 1950s onwards, declining in the late 1960s15,99. The clearest signal is from atmospheric 14C, seen in direct air measurements and captured by tree-rings and glacier ice, which reaches a maximum in the mid- to high-latitude Northern Hemisphere at 1963–64 and a year later in the tropics100. Although 14C has a relatively short half-life (5,730 years), elevated levels will persist long enough to be useable for several generations of geologists in the future.

While recognizing that many apparently novel industrially produced chemicals are occasionally produced in small quantities naturally101, chemical signatures from long-lived well-mixed gases in glacier ice or sediments may also meet GSSP criteria. Potential long-lived gases are the halogenated gases, such as SF 6 , C 2 F 6 , CF 4 (with half-lives of 3,000 yr, 10,000 yr and 50,000 yr, respectively). Most were first manufactured industrially in the 1950s, and many are measurable in firn air102, and with large enough samples could be measured in ice cores15. But although they are measurable, distinct peaks are very recent and sometimes absent because major declines in industrial production are occurring after the negotiation and ratification of the 1989 Montreal and 2005 Kyoto protocols.

Of the various possible mid- to late-twentieth-century markers of the Great Acceleration, the global 14C peak provides an unambiguously global change in a number of stratigraphic deposits. We suggest that an unequivocally annual record is the optimal choice to reflect the 14C peak, thereby giving a dating accuracy of one year. We propose that the GSSP marker should be the 14C peak, at 1964, within dated annual rings of a pine tree (Pinus sylvestris) from King Castle, Niepołomice, 25 km east of Kraków, Poland103 (Fig. 2). Secondary correlated markers would include plutonium isotope ratios (240Pu/239Pu) in sediments indicating bomb testing104, (fast-decaying) 137-Caesium97, alongside the presence of peaks in very long-lived iodine isotopes (129I, with half-life 15.7 million years) found in marine sediments105 and soils106.