The first revolution in energy input to the biosphere was the origin of photosynthesis. The earliest life forms were probably fuelled by chemical energy stored in compounds in their environment, but the supplies would have been small, except in unusual environments with concentrated volcanic/metamorphic activity such as deep sea vents near mid-ocean ridges (if plate tectonics started early on the Earth). Shortage of chemical energy on a global scale would thus have severely restricted the spread of chemolithoautotrophic life. The first truly global biosphere arose when early life began to harness the most abundant energy source on the planet – sunlight. Evidence for the photosynthetic fixation of carbon dioxide from the atmosphere is coincident with the first putative evidence for life on Earth > 3.7 Ga , and perhaps as early as 4.1 Ga . It takes the form of small particles of graphite carbon, which have a likely biogenic origin, and an isotopic signature consistent with carbon-fixation by the enzyme RuBisCO.

The first photosynthesis was not the familiar kind, which uses water as an electron donor and produces oxygen as a waste product. Instead, molecular phylogenies suggest that several forms of anoxygenic photosynthesis evolved independently, early in the history of life, long before oxygenic photosynthesis . This makes energetic sense as there are several donor compounds from which it is easier to extract electrons than water, requiring fewer or less energetic photons and simpler photosynthetic machinery. Hydrogen gas (H2) gives up its electrons the easiest and may thus have fuelled the first photosynthesis . Other potential electron donors include elemental sulphur (S0) derived from sulphur dioxide (SO2) gas, or ferrous iron (Fe2+) dissolved in the ancient oceans . The meagre supply of these compounds (relative to H2O) limited the energy input to the early biosphere. For example, the present-day flux of H2 emanating from volcanic processes can only support ∼0.1 EJ yr-1 (3 TgC yr-1) of anoxygenic photosynthetic net primary production (NPP) , over 4 orders of magnitude less than present marine biosphere (1800 EJ yr-1 or 48 PgC yr-1).

The challenge for the first photosynthetic biosphere would thus have been to evolve the means of recycling the scarce materials that it needed to metabolize, especially the electron donors for photosynthesis. The ease or difficulty of evolving recycling has been examined theoretically by simulating “virtual worlds” seeded with “artificial life” forms and leaving the resulting ecosystems to evolve . In these simulations, the closing of material recycling loops robustly emerges , even if they incur an energetic fitness cost . The empirical record of how and when recycling emerged in the early Earth system is sparse, but there is some evidence for biogenic methane production by 3.5 Ga . This would have recycled hydrogen (and carbon) back to the atmosphere. If the early biosphere was fuelled by anoxygenic photosynthesis based on H2, then recycling of hydrogen via methane production and photolysis could have boosted global NPP to 1.8 EJ yr-1 (48 TgC yr-1) or 0.1 % of the modern marine biosphere . If volcanic activity on the early Earth was elevated by an order of magnitude, a hydrogen-fuelled biosphere might have approached 1 % of modern marine NPP . Alternatively, if early anoxygenic photosynthesis used the supply of reduced iron upwelling in the ocean then its NPP, controlled by ocean circulation, might have reached 77–225 EJ yr-1 (2–6 PgC yr-1) or ∼10 % of modern marine NPP (Fig. ). A potential constraint on early biosphere productivity is provided by the carbon isotope record of marine carbonate rocks, which is conventionally interpreted as indicating that the proportion of carbon buried in organic form (rather than inorganic carbonates) was around 20 % even as early as 3.5 Ga. Given greater inputs of carbon from the mantle on the early Earth, this would imply a marine organic carbon burial flux in excess of the present 60 TgC yr-1, setting a lower limit on NPP at the time (assuming no heterotrophic recycling, i.e. all organic carbon produced was buried). This would likely preclude H2-based photosynthesis as the dominant source of carbon 3.5 Ga onwards, suggesting instead an iron-fuelled (or even oxygenic) biosphere. However, a more nuanced interpretation of the carbon isotope record allows for the possibility that little organic carbon was buried for large parts of the Archean Eon .

The waste products of early metabolisms would have altered the environment. The long-term burial of organic carbon, even if it was a small flux, would have removed carbon from the atmosphere (and ocean) tending to cool the planet. This cooling effect could have been profound given that today 15 ZgC are stored as organic carbon in sedimentary rocks, compared to 38 EgC in the ocean–atmosphere system. Somewhat counterbalancing the net removal of carbon to the crust, the conversion of atmospheric CO2 to methane would have increased radiative forcing, tending to warm the planet. As a crustal reservoir of reduced carbon accumulated in sedimentary rocks, some organic carbon would later be exposed on the continents, potentially supporting heterotrophic productivity there. Relatively low estimates of global productivity make it unlikely that the macro-nutrients nitrogen and phosphorus became limiting, making them under-tapped resources in the ocean environment.

The next major revolution in energy input to the biosphere was the origin of oxygenic photosynthesis, using water as an electron donor . To split water requires more energy (i.e. more high energy photons of sunlight) to be captured than in any of the earlier anoxygenic forms of photosynthesis. It was contingent on the prior origin of anoxygenic photosynthesis in that two existing photosystems – derived from anoxygenic photosynthetic ancestors – were wired together in the same cell . To be naturally selected, oxygenic photosynthesis required an environment – plausibly freshwater – where easier electron donors were absent or had been drawn down to limiting concentrations. The resulting cyanobacterial cell was the ancestor of all organisms performing oxygenic photosynthesis on the planet today. It took up to a billion years to evolve , with the first evidence of oxygen appearing 3.0–2.7 Ga .

Once oxygenic photosynthesis evolved, the productivity of the biosphere was no longer restricted by the supply of substrates for photosynthesis, as water and carbon dioxide were abundant. Instead, the availability of nutrients, notably nitrogen and phosphorus, would have become the major limiting factors on global productivity – as they still are today. Oxygenic photosynthesis would have flourished wherever nutrients were available and anoxygenic photosynthesis drew down its electron donors to limiting concentrations, or where oxygen removed those electron donors by oxidizing them. Anoxygenic photosynthesis might have flourished underneath oxygenic photosynthesis in parts of the surface ocean if and when anoxic waters bearing Fe2+ extended up into the sunlit photic zone , and this would have set up some competition for the nutrients nitrogen and phosphorus.

Constraints on nutrient concentrations in the early ocean are scarce . Nitrogen would initially have been in the form of ammonium (rather than nitrate), but the advent of an oxygen source plausibly triggered the onset of nitrification and denitrification . Nitrification could have produced small pools of nitrate in restricted surface ocean `oxygen oases' with nitrogen in the form of ammonium elsewhere. Whether denitrification could then have caused nitrogen scarcity , depends on whether nitrogen fixation had evolved and could counter-balance it . Iron and vanadium-based nitrogen fixation were plausibly already widespread , although molybdenum-based nitrogen fixation may have evolved later . Thus phosphorus was probably the ultimate limiting nutrient, as it is today. Lower terrestrial weathering fluxes of phosphorus (relative to present) have been predicted, due to a shift from terrestrial to seafloor weathering to balance the carbon cycle earlier in Earth history, and this would have tended to reduce ocean phosphorus concentration, because seafloor weathering is not a source of phosphorus . Initial work estimated only ∼10–25 % of today's phosphorus concentration in the Late Archean ocean , however subsequent studies have revised this upwards to ∼1–4 times present-day phosphorus concentration . Furthermore, nutrient recycling by the microbial loop within the surface ocean was conceivably more efficient than today because eukaryotic mechanisms of exporting organic matter out of the surface ocean were absent. One model suggests that marine NPP may have been ∼25 % of today's productivity (450 EJ yr-1 or 12 PgC yr-1) in the Late Archean ∼2.7 Ga .

With the advent of oxygenic photosynthesis there was thus an order of magnitude increase in organic carbon production (Fig. ). The extra flux of carbon sinking into the anoxic depths of the ocean would initially have fuelled methanogenesis (as sulphate was yet to build up significantly in the ocean, ). The resulting upward flux of methane could support widespread methanotrophy near the source of oxygen from oxygenic photosynthesis, consistent with very isotopically light organic carbon from ∼2.7 Ga . A large flux of methane, equivalent to around 60 % of the primary production sinking out of the surface layer of the ocean , would also escape to the atmosphere, warming the planet. However, if the CH4 : CO2 ratio in the atmosphere approached 0.1, photochemical production of an organic haze that scattered sunlight back to space would have triggered cooling . This process would be self-limiting, but might help explain the earliest glaciations ∼2.9 Ga .

Oxygen remained a trace gas, O2 < 10-5 PAL (present atmospheric level), until 2.45 Ga as indicated by the mass independent fractionation of sulphur isotopes (MIF-S), preserved in sediments older than this, which shows that the ozone layer was absent and high energy ultraviolet radiation reached the surface (creating the signal), and sulphate had yet to accumulate in the ocean (allowing the signal to be preserved) . Elevated concentrations of methane in such a reducing atmosphere would have supported an increased flux of hydrogen loss to space, causing the long-term oxidation of the surface Earth system . Stability broke down 2.45–2.3 Ga in the “Great Oxidation” event . The MIF-S signature disappeared indicating that oxygen rose > 10-5 PAL sufficient to form an ozone layer. Massive deposits of oxidized iron appeared in the form of the first sedimentary “red beds”, and oxidized iron also appeared in ancient soils, indicating that oxygen increased to > 10-2 PAL. Models suggest that once enough oxygen built up for the ozone layer to start to form, this shielded the atmosphere below from UV and slowed down the removal of oxygen by reaction with methane . This created a strong positive feedback explaining the abruptness of the Great Oxidation .

The Great Oxidation destabilized other environmental variables. As oxygen rose, atmospheric methane concentration declined , which could help explain the series of Huronian glaciations and the low-latitude Makganyene glaciation 2.32–2.22 Ga . The reaction of oxygen with sulphide in continental rocks plausibly produced sulphuric acid that dissolved phosphorus out of apatite inclusions in the rocks and fuelled marine productivity . The oxidizing power unleashed in the Great Oxidation could thus have made another limiting resource, phosphorus, more available, boosting energy input to the biosphere. One model estimates that marine NPP in the Proterozoic Eon after the Great Oxidation was ∼1300 EJ yr-1 (34 PgC yr-1) or ∼70 % of today's value . This would have supported increased organic carbon burial, which is inferred to have occurred during the “Lomagundi” carbon isotope excursion, 2.23–2.06 Ga , potentially triggering an “overshoot” of atmospheric oxygen . However, there were large crustal reduced sinks for oxygen at the time , and after ∼ 150 Myr, excess buried organic carbon was recycled through the crust back to the surface, consuming oxygen and deoxygenating the ocean . After this protracted interval of instability, the Earth entered an even longer period of stability, known as “the boring billion”.

Marine productivity during this protracted interval of the Proterozoic Eon is very uncertain. We know that the deep ocean remained largely anoxic and “ferruginous” (with Fe2+ in solution), with euxinic waters (SO4-reducing) at intermediate depths along some ocean margins, and surface waters largely oxygenated . Several authors have argued for very low productivity partly on the grounds of a sparsity of organic carbon rich shales, but largely based on theoretical arguments for low nutrient conditions. Phosphate supply to the ocean could have been reduced by scavenging onto iron oxides forming in freshwater and estuarine environments . Phosphate could also have been efficiently removed from ocean waters by the formation of mixed Fe2+ / Fe3+ compounds such as “green rust” . However, reducing deeper waters and sediments (especially euxinic ones) should have been effective at recycling phosphorus and shuttling it back to the surface ocean, consistent with high estimates of phosphate concentration at 1.7 Ga . Nitrogen limitation has been argued for on the grounds of a lack of molybdenum for nitrogen fixation , but the existence of alternative nitrogenases makes this unlikely . Instead, heterogeneous ocean redox conditions could have supported a mixed nitrogen cycle with ammonium in the predominantly reducing waters of the deep ocean and small reservoirs of nitrate in oxygenated waters. In such a system there would be large fluxes of denitrification along the extensive interfaces between oxygenated and anoxic waters, counterbalanced by large fluxes of nitrogen fixation in surface waters replenishing the nitrogen reservoirs. Indeed the fact that the Great Oxidation was never reversed sets a lower bound on Proterozoic productivity of ∼25 % of modern marine NPP in an existing model .

The Great Oxidation increased energy consumption by the biosphere, even with no change in energy input, because respiring organic matter with oxygen (2870 kJ mol-1) yields an order of magnitude more energy than breaking it down anaerobically (e.g. 232 kJ mol-1 for alcohol fermentation). This greater energy source facilitated the evolution of new levels of biological organization, in the form of eukaryotes. The ancestral (heterotrophic) eukaryote is thought to have had mitochondria performing aerobic respiration. The timing of eukaryote origins is deeply uncertain, but with putative biomarker evidence 2.7 Ga now rejected , and molecular clocks suggesting a last common ancestor 1.8–1.7 Ga , they may post date the Great Oxidation. Mitochondrial respiration in turn allows eukaryotes to support a much larger genome than prokaryotes, giving them the capacity to create more complex life forms with multiple cell types , the first evidence for which appears ∼1.2 Ga .

The next revolution in energy input to the biosphere involved encapsulating an existing metabolism – oxygenic photosynthesis – in progressively more complex, eukaryotic organisms and symbioses – algae, lichens and land plants (with mycorrhizal fungi). This energy revolution involved increasing the supply and utilization of limiting nutrient resources needed to perform photosynthesis and increasing the area over which it occurred.

The lineage containing all extant photosynthetic eukaryotes arose 1.7–1.4 Ga , but eukaryotic algae only became ecologically significant relative to cyanobacteria ∼740 Ma, when biomarkers of algae become more prevalent in ocean sediments and the diversity of eukaryote fossils starts to increase . Larger eukaryote cells are better at exploiting excess nutrients in polar surface oceans, but would also have removed carbon and nutrients from the surface ocean more efficiently, thus reducing recycling, with uncertain overall effects on productivity . More efficient carbon export to sediments plausibly increased phosphorus removal from the ocean, lowering global productivity and tending to oxygenate the deep oceans , and contributing to CO2 drawdown and global cooling . CO2 drawdown by silicate weathering might have been enhanced by the arrival of eukaryotes (fungi and algae) in microbial ecosystems on the land . Estimates of the productivity of global microbial mats, based on a simple area-scaling of modern desert crust , suggests only 3–11 % of today's terrestrial NPP, comparable to today's cryptogamic cover, which achieves 1–6 % of terrestrial NPP . However, deserts are unproductive environments and modern cryptogamic cover is living in a world dominated by vascular plants. Taking the ecophysiological model of cryptogamic cover and considering higher atmospheric CO2 and lack of competition from vascular plants, putative Neoproterozoic-early Paleozoic land biota might have achieved ∼25 % of today's terrestrial NPP. Whatever the cause(s), the Earth experienced two low-latitude “snowball Earth” glaciations – the Sturtian (starting 715 Ma) and Marinoan (ending 635 Ma) – amidst a protracted interval of instability in the global carbon cycle. Iron formations deposited during these events apparently record high concentrations of phosphate in the ocean , which could be explained by a shutdown of biological uptake and removal to sediments. In the aftermath of glaciations, high productivity could have been fuelled – at least temporarily – by elevated phosphorus concentrations. There is evidence of at least partial oxygenation of the deep ocean, and complex eukaryotic life including animals began to flourish in the oceans. However, by the early Phanerozoic, phosphorus concentrations were broadly comparable to today , implying comparable levels of marine NPP.

The key change in energy input to the biosphere and material cycling came later with the rise of plants on land, starting around 470 Ma and culminating in the first global forests by 370 Ma . This roughly doubled global NPP, increasing it by an order of magnitude on land and potentially indirectly in the ocean. Terrestrial NPP is estimated to have exceeded today's value (∼2100 EJ yr-1 or 56 PgC yr-1) on average during the Phanerozoic (Fig. ), with peaks potentially exceeding twice the present value . To colonize the land new nutrient acquisition mechanisms were required, achieved through symbioses with mycorrhizal fungi and nitrogen-fixing bacteria. Plants and their associated mycorrhizal fungi accelerated the chemical weathering of the land surface in search of rock-bound nutrients, notably phosphorus. Ultimately, stunningly effective recycling developed, such that the average terrestrial ecosystem today recycles phosphorus ∼50 times through primary production before it is lost to freshwaters .

Increased silicate weathering lowered atmospheric CO2 levels, plausibly triggering the Late Ordovician glaciations , although others question the magnitude of early plant effects on the carbon cycle . A more established view is that the weathering effects of later plants, notably the first deep-rooting trees forming forests, caused the later Permian-Carboniferous glaciations. Increased phosphorus weathering supplied nutrient to the oceans, increasing marine productivity and plausibly triggering oceanic anoxic events . The increase in organic carbon burial with the rise of plants also increased atmospheric oxygen, as revealed in the charcoal record . Although ignition sources (lightning, volcanoes) have always existed on Earth, there was little to burn before land plants arose, and experiments show that O2 > 15 % of the atmosphere is required for biomass combustion to be sustained . The first charcoal evidence for natural fires coincides with the appearance of vascular plants on drier land ∼420 Ma . Plants in turn provided a new source of organic carbon for burial in sediments, especially new structural carbon polymers (including lignin), which are hard to biodegrade. Fungi evolved to recycle these, but a delay may have caused atmospheric O2 to peak in the Carboniferous . The continuous charcoal record indicates O2 persistently > 15 % of the atmosphere since 370 Ma .

The rise in atmospheric oxygen and increase in food supply brought about by land plants has allowed a flourishing of animal complexity from aerobic pathways – including the emergence of us humans. Today, the total global energy flux through heterotrophic biomass, based on a 10 % conversion efficiency of 100 PgC yr-1 with energy density 40 kJ gC-1, is ∼400 EJ yr-1, roughly half on land and half in the ocean. Natural fires additionally consume ∼55 EJ yr-1 (1.4 PgC yr-1) , and human-induced fires ∼45 EJ yr-1 (1.1 PgC yr-1) , giving a total biomass burning flux today of ∼100 EJ yr-1 (∼2.5 PgC yr-1) , or ∼2.5 % of the energy and carbon captured in photosynthesis.

During most of their existence humans lived as foraging societies in an uncontrolled solar-energy system , simply tapping into the existing energy and material cycles of the biosphere, without deliberately controlling them by systematic land management, and without introducing new biogeochemical pathways. The first human revolution in energy input was the intentional use of fire, which set humans apart from all other species. With it humans extended their energy utilization beyond their biological metabolism towards areas outside the human body. This marked the beginning of a social metabolism – a collectively organized extension of energy and material use by human societies .

There is robust evidence that Homo erectus could control fire from 790 ka in Africa and from 400 ka in Europe . The ability to cook, which implies the control of fire, may date as far back as 1.5 Ma . Cooking provided higher food energy, higher food diversity through detoxification, and a selective force to develop social abilities and large brains, thus playing a key role in human evolution. The use of fire may also have facilitated humans occupying colder climates , and developing increased abilities to cooperate , a decisive element of their evolutionary success.

Use of fire for cooking increased energy input to approximately 7–15 GJ cap-1 yr-1, i.e. a factor of 2–4 above the average physiological energy demand of 3.5 GJ cap-1 yr-1 . Assuming a population of 2–4 million at the beginning of the Neolithic , overall energy capture by humans amounted to roughly 14–60 PJ yr-1, a factor of ∼1000 below the global human energy input in 1850 and ∼10 000 below today's (Fig. ). In foraging societies, biomass accounts for more than 99 % of material input. Materials are used predominantly for energetic purposes, as fire wood or food. Thus the energetic and the material social metabolism were practically identical.

Based on their direct energy and material inputs, foraging societies had a negligible impact on the global environment. However, the intentional use of fire for hunting, clearing land and other purposes could have caused significant environmental impacts – accepting that the empirical evidence regarding frequency, scale and age for applying those intentional burning techniques is highly contested. Potential impacts include extinction of large Pleistocene land animals and ecosystem tipping events, including shift of vegetation to desert shrub triggering a weak monsoon in Australia , rapid landscape transformations in the mesic environments of New Zealand , the wet tropical forests of the pre-Columbian Amazon , and across the savannas and woodlands of Africa .

Foraging societies need large areas. Although the energy density of natural vegetation ranges over 0.1–1 W m-2 (NPP of 3.16–31.6 MJ m-2 yr-1) , the bulk biomass of the most abundant plants, grasses and trees, is not edible for humans. The very small share of human-edible natural biomass restricts the population density of foraging societies to no larger than ∼0.02–0.2 cap km-2 . Such low population densities and the necessity to stay mobile prevent the accumulation of artifacts and the development of complex institutions, e.g. institutions to deal with conflict are prohibitively costly as long as moving away is an attainable alternative. Therefore foraging societies are typically portrayed as small egalitarian groups of low internal complexity, based largely on a few extant foraging societies who have been pushed aside to marginalized environments. In more favourable environments higher resource intensities could have supported higher population densities and significantly more complex social structures, including settlements, handcraft, trade and social stratification .

By the beginning of the Holocene, 11 700 BP, humans had successfully inhabited all continents. Then, within a few thousand years a fundamentally new socio-metabolic energy regime emerged on all continents except Australia, involving the domestication of wild plant and animal species and the control of their reproduction via husbandry. Agriculturalists greatly enhanced the area productivity of edible species at the expense of non-edible species and of food competitors. In contrast to pre-agricultural societies they lived in a controlled solar energy system .

Agriculture had multiple independent origins; in the Near East (∼ 10 000 BP), Peru (∼ 10 000 BP), South China (8500 BP), North China (7800 BP), Mexico (4800 BP), East North America (4500 BP), and possibly sub-Saharan Africa (4000 BP) . Archaeological evidence from several sites at the shoreline of the Persian Gulf has revealed a rapid colonization of this area by advanced agricultural and urban societies at around 7500 BP. As sea-level rise from the last glacial low stand was only completed in the Persian Gulf 7000–8000 BP, there could be even older agricultural sites in areas that are now beneath the Indian Ocean . Explaining the relatively rapid transition to agriculture is one of the most controversial topics in universal history. The puzzle is that early agriculture, especially farming, was not obviously superior to foraging. Ethnological studies have shown that early farmers spent more hours to exploit their food base, relied on a less diverse and less stable diet, were more prone to diseases, and even had lower productivity in terms of calorific return on labour investment . The Neolithic revolution therefore tends to be explained as a necessity driven transition, fostered by population pressure , deterioration of resources (extinction of Pleistocene megafauna), or climate change. Whatever the reasons for switching from foraging to farming, it creates a lock-in once population densities exceed the natural carrying capacity of the surrounding ecosystem. Then reverting to foraging cannot occur without substantially reducing population numbers.

After ∼7000 BP complex agrarian civilization emerged . Extant biomass was still the energy source for almost all energy uses: food, fodder, heat, mechanical power and chemical transformation (metallurgy). Wind and water power used by agrarian civilizations (sailing ships and mills) were locally important but contributed only marginally to the energy input. Despite huge variations in agrarian land use systems, a defining condition is that energy supply is tightly coupled to productive land and (human and animal) labour working on the land. Without any external energy subsidies in the form of mechanical power and synthetic fertilizers, a larger usable energy output from extant biomass typically requires more land or more labour input on existing land, thus putting relatively strict limits to the possibility of increasing energy supply per capita . Higher yields could be achieved by various improvements in agricultural technology but those improvements were typically population driven and lead to absolute growth in energy capture per area of land while per capita energy availability stagnated or even declined . Estimates of global average energy input to agrarian societies are 45–75 GJ cap-1 yr-1 roughly a factor of 5 greater than in foraging societies . With the estimated population rising to ∼450 million in AD 1500 (when the agrarian mode of subsistence dominated the global population), overall energy capture by humans may have reached ∼20 EJ yr-1 , a factor of 300 above the foraging regime, but 30 below today. When the industrial revolution took off around 1850 human population was ∼1.3 billion and energy capture had reached ∼60 EJ yr-1 (Fig. ).

The increased population and energy flows due to farming increased the material inputs to, and waste products from, societies. The resulting environmental effects began early in the Holocene, but their scale is much debated . Irrigation began around 8000 BP in Egypt and Mesopotamia, leading to some localized salination and siltation of the land, reducing crop yields and encouraging a shift in agricultural crop from wheat to more salt-tolerant barley . The use of manure as fertilizer may have begun as early as 9000 BP in SW Asia and 7000 BP in Europe . The clearance of forests to create agricultural land and supply biomass energy and wood from 8000 BP onwards, reduced the carbon storage capacity of the land, transferring CO2 to the atmosphere . Cumulative carbon emissions may have approached 300 PgC by 500 BP contributing ∼20 ppm to atmospheric CO2 levels. The biogeophysical effects of forest clearance also affected the climate, regionally and remotely . Anthropogenic sources of methane started around 5000 BP with the irrigation of rice paddies and have contributed to changes in atmospheric CH4 concentration over the past ∼3000 years .

The energetic surplus generated by agrarian societies first allowed cities to become a widespread phenomenon ∼5000 years after the beginning of agriculture. This led to more complex social organization with increasing division of labour, technological innovations, social stratification and written language . This in turn requires re-integration via exchange, trade and redistribution creating mutual dependencies which increased the potential for conflict, prompting the inception of social institutions to deal with such conflicts (e.g. priests, judges). Additionally, stockpiling and concentration of resources in cities attracted predators stimulating institutions of defense (military). Social complexity has costs as well as benefits and a number of early complex societies collapsed when those costs became prohibitively high .

Agrarian societies are faced with relatively severe constraints regarding the energy surplus they can achieve. On average around 90 % of the population is required to work in agriculture. This limits the urban population engaged in non-food producing activities to no more than 10 % , although locally urbanization levels could be much higher. The outstanding role of bio-productive land as the main factor of production also explains the important political role of territory in agrarian societies. The intrinsic connection between social stratification and territory in agrarian societies can be illustrated by the role of land in medieval European feudalism, where the power of the nobility was strongly connected to the control over productive land. Economic growth was only possible through land expansion and increase in area productivity. Both have inherent practical limits and both require a growing population. Combined with hard constraints on transportation in pre-industrial societies, this leads to relatively fast local negative feedbacks in the energetic and material social metabolism and renders sustained material growth impossible on a per capita basis – making the distribution of material wealth a zero sum game.

Fossil fuels, especially coal and peat had been used for hundreds of years in China, Burma, The Netherlands and England . However, their contribution to the social metabolism always remained small. The key energy transformation of the industrial revolution came with the ability to massively scale-up fossil energy use . Unlike the Neolithic revolution, the Industrial revolution was a historical singularity. Its inception in 18th century England was followed by a worldwide expansion of the new energy regime, which is still ongoing. The fossil energy regime eventually surmounted the inherent thermodynamic constraints of agrarian societies that had existed for millennia by decoupling socially usable energy from bio-productive land and human labour . Within 150 years, from 1850 to 2000, global human energy use increased tenfold from 56 to 600 EJ yr-1 (estimates based on ), the world population went from 1.3 billion to 6 billion, and global GDP increased from 800 to 6600 intGK$. Thus by 2000 the annual global energy flux through human societies was one third of the global terrestrial NPP and one third above the total global energy flux through all non-human heterotrophic biomass (Fig. ).

Unlike the Neolithic revolution, the puzzle of the industrial revolution is not that it began, but that it continued . Similar innovation-driven growth periods in agrarian civilizations (e.g. the Dutch golden age) could not be sustained, because they were sooner or later counterbalanced by diminishing returns on energy investment in the agricultural sector. Even for the classical British economists Adam Smith, David Ricardo, Thomas Malthus, and John Stuart Mill, who witnessed England's industrial take-off, there was no doubt that diminishing marginal yields in the agricultural sector would eventually bring industrialization to a halt . A key challenge was to feed a growing industrial labour force with a controlled solar-energy based system of agriculture (given that the agricultural sector did not industrialize until the 1930s in the USA and the 1950s in Europe) . England was in a specially favoured position, because since the late 16th and early 17th century area yields, total agricultural production and labour productivity had been growing continuously . This allowed 18th century England to support a growing industrial labour force in the initial phase of the industrial revolution. When agricultural productivity gains eventually came to a halt around 1830 – while the population was still growing rapidly – England's hegemonic political position was instrumental to massively increase food imports .

The availability of technologies to overcome bottlenecks in energy utilization also played a decisive role in the industrial revolution happening in England. Notably, the coincidence of a domestic endowment of coal with the emergence of a new technology complex consisting of the steam engine and coke-based iron smelting. With this technological complex energy constraints could be exceeded , which had previously limited coal extraction, steel production, and long-distance transportation.

The step increase in energy capture with industrialization is associated with fundamental changes in global material cycles. Material inputs to societies were transformed from biomass dominance to minerals dominance. Global average per capita material use increased from 3.4 to 10 t cap-1 yr-1 from 1870 to 2000, and with roughly constant biomass use of 3 t cap-1 yr-1, the average use of mineral and fossil materials increased from 0.4 to 7 t cap-1 yr-1 . In industrial economies ∼80 % per weight of the total annual outflow of materials is CO2, making the atmosphere the largest waste reservoir of the industrial metabolism . Between 1850 and 2000 global CO2 emissions from combustion of fossil fuels and materials processing increased 125-fold from 54 to 6750 TgC yr-1 and reached 9140 TgC yr-1 in 2010 .

Industrial societies require large physical stocks: buildings, transport infrastructure, energy, water and waste infrastructure, production facilities and durable consumer goods. For example, the material stock of industrializing Japan has increased by a factor of 40 between 1930 and 2005 reaching 38.7 billion tonnes or 310 tonnes per capita and the non-metallic minerals incorporated in residential buildings, roads and railways in the EU25 was 75 billion tonnes or 203 tonnes per capita in 2009 . In the USA the amount of iron incorporated in durable products and infrastructure increased from 100 million tonnes to ∼3200 million tonnes between 1900 and 2000 . Industrial societies also use a much larger diversity of minerals. Almost all metals are now commercially used in increasingly complex combinations . Overall recycling rates (measured as the global average of the content of secondary metal in the total input to metal production) of metals are uncertain. Recycling rates are above 50 % for only three metals (Nb, Ru, Pb), between 20–50 % for another 16, and below 20 %, often less than 1 %, for all the other ∼40 metals in wide industrial use . A recent study estimated that only 6 % of globally extracted materials are currently recycled within the socioeconomic system .

The global biogeochemical cycles of nutrients have also been transformed by industrialization. Between 1860 and 2005 anthropogenic creation of reactive nitrogen grew more than tenfold, from ∼15 to 187 TgN yr-1 . Furthermore, the creation of nitrogen oxides as a waste product of fossil fuel combustion increased from ∼0 to 25 TgN yr-1 . The excess reactive nitrogen was transferred to other environmental pools, partly denitrifying to atmospheric N2, but also contributing to eutrophication and acidification of terrestrial and coastal marine ecosystems, to global warming and to tropospheric ozone pollution. Analogous human-induced acceleration affected the P-cycle.

The industrial revolution also gave rise to entirely new metabolites. The CAS Registry currently includes 92 million unique chemical substances in commercial use of which only 320 000 are regulated in key markets. It is unknown how many of these substances represent entirely new chemicals and whether they are harmful to humans or the environment. With 15 000 new entries daily comprehensive in-vivo toxicity testing is practically impossible .

The industrial revolution expanded to the European continent and to the USA in the early 19th century, to Japan in the late 19th century and to large nations like China, India, and Brazil in the last decades of the 20th century. With the transition to an industrial mode of production the socio-economic power of the nobility (based on control over productive land) diminished and shifted towards the owners of the means of industrial production (which Karl Marx called capitalists). Large differences in consumption among countries persist until today (Fig. ; data from and the World Bank Income Classification, ). If we consider high income countries with an average energy use of 302 GJ cap-1 yr-1 as fully industrial, and upper middle and lower middle countries, with an average energy use of 140 and 74 GJ cap-1 yr-1 respectively as transitioning to an industrial energy regime, then ∼15 % of the world population lived in a mature industrial energy regime in 2000, ∼44 % were in transition, and the remaining ∼40 % still lived under largely agrarian conditions with average energy use amounting to 42 GJ cap-1 yr-1. The correlation between energy use and human development appears to be highly non-linear. At high levels of human development large increases in energy input have little or no effect on further increases in standards of living. However, at low levels of human development relatively small increases in energy input have large positive effects , for example supplying ∼3.5 kilowatt per person can greatly increase life expectancy .