Long-term trajectories of the human appropriation of net primary production: Lessons from six national case studies.

The 'human appropriation of net primary production' (HANPP) is an integrated socio-ecological indicator measuring effects of land use on ecological biomass flows. Based on published data for Austria, Hungary, the Philippines, South Africa, Spain and the UK, this paper investigates long-term trends in aboveground HANPP and discusses the relations between population, economic growth, changes in biomass use and land-use intensity and their influences on national HANPP trajectories. During early stages of industrialization, population growth and increasing demand for biomass drive land-cover change, often resulting in deforestation, which raises HANPP. During later stages, industrialization of agriculture boosts agricultural yields often faster than biomass demand grows, resulting in stable or even declining HANPP. Technological change improves agricultural area-efficiency (biomass provision per unit area), thereby decoupling population and economic growth from HANPP. However, these efficiency gains require large inputs of fossil fuels and agrochemicals resulting in pressures on ecosystems and emissions. Our findings corroborate the argument that HANPP alone cannot - as sometimes suggested - be used as a simple measure of carrying capacity. Nevertheless, analyses of long-term HANPP trajectories in combination with accounts of material and energy flows can provide important insights into the sustainability of land use, thereby helping to understand limits to growth.

Introduction

There is a long discussion about the carrying capacity of the earth for humans; that is, on the question how many people the earth can support (e.g., Cohen, 1995; Martinez-Alier, 1987; Pfaundler, 1902). The ‘human appropriation of net primary production’ or HANPP measures the combined effect of land use and biomass harvest on the availability of trophic energy in ecosystems, thereby providing a measure of the scale of human activities as compared to ecological processes in terrestrial ecosystems (Daly, 1992). Following the influential study of Vitousek et al. (1986), who found that humans globally appropriate almost 40% of terrestrial NPP, HANPP has often been cited by ecological economists as a particularly striking example for the limits imposed by environmental constraints on further population or economic growth (Costanza et al., 1998; Daly, 1992; Meadows et al., 1992).

The basic idea was simple: Humans compete with all other heterotrophic organisms for NPP as their source of trophic energy (Vitousek et al., 1986). So if humans use 40% of the NPP today, the consequences would be dire if that number were to grow to 80 or even 100%, which would soon be the case, given the short doubling times resulting from current rates of population and GDP growth (Costanza et al., 1998; Meadows et al., 1992). This notion has lost credit, however, largely due to the recognition that the links between population and economic growth are a lot less straightforward (see Sagoff, 1995 and Davidson, 2000 for a critical discussion). Nevertheless, interest in HANPP has remained vivid. In particular, HANPP has recently gained attention as an indicator capable of linking natural to socioeconomic processes and of generating an integrated picture of socio-ecological conditions (Haberl, 1997; Haberl et al., 2007; Imhoff et al., 2004; Krausmann et al., 2009; Wright, 1990) — a major goal of sustainability science (Kates et al., 2001; Parris and Kates, 2003).4

During the last decades, the concept of HANPP has been advanced and proposals for a standardization of definitions and methods have been made (Erb et al., 2009b; Haberl et al., 2007; Imhoff et al., 2004). A considerable number of empirical case studies on global and regional patterns of HANPP have been published (e.g., Erb et al., 2009b; Haberl et al., 2007; O'Neill et al., 2007; Vackár and Orlitová, in press), as well as several long-term (decadal to centennial) national time series of HANPP (Kastner, 2009; Kohlheb and Krausmann, 2009; Krausmann, 2001; Musel, 2009; Niedertscheider et al., in press; Schwarzlmüller, 2009).

This paper uses existing case studies for a comparative discussion of long-term changes in the aboveground HANPP in six countries: Austria, Hungary, the Philippines, South Africa, Spain and the UK. We aim to better understand the processes that drive long-term changes in HANPP at the national level and to contribute to a better understanding of how population and economic growth, changes in biomass use and land-use intensity are related, how they shape the magnitude and spatial pattern of HANPP, and what can be learned about these interactions and about ecological limits. We address a number of highly policy-relevant issues, in particular related to the suitability of HANPP to serve as an indicator of sustainability and to the question of providing sufficient food, feed, fiber and fuel for the growing number of humans on earth in a sustainable manner (Evans, 1998; Godfray et al., 2010).

In the next section, we briefly define HANPP and give an overview of the six national case studies that provided the database for our comparative analysis. The following section presents the development of a number of aggregate indicators derived from HANPP data in the case studies in a comparative manner. This is followed by a Discussion section that analyzes drivers of the observed trends, focusing on the significance of land-cover change, land-use intensity, biomass use, biomass trade and the ecological costs of improving land-use efficiency. The paper ends with conclusions on the potentials and risks of further improvements in the HANPP intensity of biomass production and an outlook at possible future global developments.

Methods and Data

The results of HANPP calculations strongly depend on the respective definition used, and definitions vary considerably between studies (Erb et al., 2009b; Haberl et al., 2004, 2007). We here use data from six studies that defined HANPP as the difference between the NPP of potential vegetation (NPP0) – the vegetation that would prevail in the absence of land use – and NPPt; that is, the fraction of the NPP that remains in ecosystems after harvest (NPPh). NPPact denotes the NPP of the currently prevailing vegetation (Haberl et al., 2004, 2007). The difference between NPPact and NPP0, that is the NPP change resulting from land conversion, is denoted as ΔNPPLC. Accordingly HANPP can be defined as follows:

HANPP = NPP0 − NPPt with

NPPt = NPPact − NPPh and

ΔNPPLC = NPP0 − NPPact

Two processes contribute to HANPP: (1) the change in NPP resulting from land conversion (ΔNPPLC) and (2) withdrawal or destruction of biomass during harvest (NPPh). We here discuss not only HANPP, but also its components, in particular NPPact and NPPh. We only refer to data for aboveground NPP, as data on belowground NPP were not available for all underlying studies.

In addition to HANPP and its components, we are also interested in changes in yields and area-efficiency; that is, in the amount of biomass gained per unit area and year. We define HANPP intensity (HANPPi) as the HANPP per unit of harvest5:

HANPPi = HANPP / NPPh.

The inverse of HANPP intensity has been interpreted as a measure of land use efficiency: If ΔNPPLC is low (it may even become negative), most or all HANPP results from harvesting biomass (NPPh) which means that little or no productivity potential is foregone due to land management (Krausmann and Haberl, 2007).

We use data from six published case studies which provide comparable data on the development of aboveground HANPP and its components over decadal to centennial periods of time. Table 1 gives an overview of the case studies.

Table 1

Overview of the national HANPP studies used in this paper and socio-economic and bio-geographic characteristics of the six countries. Sources: Population growth and population density are based on Maddison, 2008; GDP per capita in constant 2005 USD and average annual precipitation are from World Bank, 2011; annual temperature means were calculated from Hijmans et al., 2005.

Country	Observed period	Reference	GDP/cap (PPP) 2005 [USD/cap/yr]	Population density 2005 [cap/km2]	Population growth 1910–2005 [%]	Precipitation [mm/yr]	Temperature [°C]	HANPP in the year 2000 [% of NPP0]
Austria	1830–1995	Krausmann, 2001	33.377	101	23%	1110	5.6	51%
United Kingdom (UK)	1800–2005	Musel, 2009	32.731	249	35%	1220	8.4	68%
Philippines	1910–2003	Kastner, 2009	2.927	295	890%	2348	25.4	62%
Spain	1955–2003	Schwarzlmüller, 2009	27.377	80	103%	636	13.1	62%
Hungary	1961–2005	Kohlheb and Krausmann, 2009	16.955	108	26%	589	10.4	71%
South Africa (RSA)	1961–2006	Niedertscheider et al., in press	8.597	36	616%	495	17.0	21%

The six countries are quite different with respect to their bio-geographic and socio-economic conditions (Table 1): Austria and Hungary are two neighboring central European countries with temperate climate. Austria is dominated by the Alps and has a high share of woodlands. In contrast, Hungary is characterized by fertile plains used for crop production but a more continental climate with lower average precipitation. The United Kingdom (UK) is situated in north-west Europe. It also has a temperate climate which is favorable for crop production in its southern part, while large areas in the north are only suitable for extensive grazing. Spain in south-western Europe is characterized by a Mediterranean climate; average annual precipitation is low and large parts of the country are considered semi-arid and feature a high share of irrigated crop production.

In addition to the industrialized European countries, two non-European countries with much lower income were included. The Philippines are an archipelago comprising of over 7000 islands. They are located in the Western Pacific Ocean and have a hot and humid tropical maritime climate. A large share of the land has been deforested and is used for agriculture. South Africa (abbreviated RSA for ‘Republic of South Africa’) is the southern-most country of Africa. It is mostly characterized by a subtropical semiarid climate. Only a small percentage of the land is used for crop production.

In contrast to Austria, the United Kingdom and Hungary, where population grew only modestly in the 20th century, the Philippines, South Africa and Spain experienced high population growth (Table 1). At the turn of the 21st century, only South Africa with a population density of 36 inhabitants per km² can be considered sparsely populated; all other countries are densely populated with population densities above 80 cap/km². South Africa is also the only country with a comparatively low HANPP. The aboveground HANPP of all other countries is far above the global average level of 29% in the year 2000 (see also Haberl et al., 2007; Krausmann et al., 2009).

The six case studies provide data on aboveground HANPP and related parameters for different periods of time in the 20th century. For Austria, the UK and the Philippines the data cover a period of time exceeding one century. The studies for Spain, Hungary and South Africa cover the second half of the 20th century. We here focus on changes in the 20th century.

All six studies used the basic definition of HANPP outlined above. However, there were some differences with respect to the inclusiveness of the definition of NPPh. Five of the six case studies (all except the Austrian one) used a comprehensive concept of NPPh that includes not only biomass harvested for further socioeconomic use (such as crops or timber), but also all biomass destroyed during the harvest process, even if the biomass is not further used by society. Examples of such by-flows are residues remaining in the field or bark and twigs of felled trees not removed from the forest. The methods and data sources used to quantify HANPP were also similar in all cases but the Austrian one. These five studies combined statistical data on land use/land cover and biomass harvest with information on actual and potential productivity derived from a dynamic global vegetation model (LPJ).6 The Austrian study, published much earlier than the other studies, applied a less inclusive definition of NPPh that only included the extraction of biomass used by society. Moreover, the NPP0 used in this study was based on a simple static productivity calculation based on temperature and precipitation as well as dominant plant species making up potential vegetation. Therefore, the comparability of the Austrian results is limited, as will be discussed below.

All data used in this article were taken from original publications (Table 1); in addition, auxiliary data were taken from databases and literature cited below. In order to enhance comparability, NPP data are presented as carbon flows per unit area and year (kg C/m2/yr).7 Original data given in dry matter were converted by assuming a C content of dry matter biomass of 50%. Data for Hungary and Austria were originally presented in Joule gross calorific value (GCV). These data were converted to kg C by assuming an average GCV of 18.5 MJ/kg and a carbon content of 50% per kg dry matter (Haberl et al., 2007). In the case of Hungary, the original HANPP calculation did not include unused crop residues as part of NPPh. To assure consistency with the other case studies and to enhance comparability, these flows were extrapolated from data on commercial harvest using region specific harvest factors from Haberl et al. (2007). This increased overall HANPP in Hungary by 15 to 40% compared to the published data.

Obviously, using only six case studies (four of which are European countries) in order to detect general temporal trends of HANPP cannot lead to comprehensive results valid at the global scale. However, to our knowledge these case studies are the only national long-term studies of HANPP which have been published so far, and a comparative discussion of their results sheds light on very generic temporal trends visible in all (or most) of the countries investigated. Secondly, despite the strong European bias, the case studies do cover a wide array of biogeographic and economic conditions, allowing for a comparison of very different framework conditions of HANPP development during the 20th century.

Comparison of National HANPP Trends

Fig. 1 shows the development of HANPP and key components of HANPP in the six countries. In the UK and the Philippines, HANPP increased considerably during the first half of the 20th century (Fig. 1a). It peaked in the 1960s and has since stabilized (Philippines) or even declined (UK). In Spain and Hungary, HANPP also declined in the second half of the 20th century. In these four countries, HANPP was high and amounted to 60–70% of NPP0 in the year 2000. In Austria, HANPP was somewhat lower and comparatively stable throughout the observed period. South Africa is characterized by a stable, low level of HANPP throughout the observed period.

Fig. 1

Development of HANPP and its components in Austria, Hungary, the Philippines, South Africa (RSA), Spain and the United Kingdom (UK). (a) HANPP in % of NPP of potential vegetation (NPP0), (b) NPP of the currently prevailing vegetation (NPPact) in kg C/m2/yr, (c) Harvested NPP (NPPh) in kg C/m2/yr and (d) HANPP intensity (HANPP/NPPh). Sources: Calculated from the studies referenced in Table 1.

Fig. 1b shows that the average productivity of the vegetation (NPPact per unit area and year) slightly declined in the UK and the Philippines during the first half of the 20th century. Around mid-century, the trend reversed and productivity began to increase. Spain, Hungary and Austria also show substantial increases in NPPact since the 1960s. In the four decades between 1960 and 2000, growth of NPPact ranged from 12% in Austria to 44% in the UK.

In parallel to NPPact, we find a significant growth in the amount of biomass harvested (NPPh). The growth of NPPh in the four decades since 1960 ranged from 28% in Austria to 54% in Spain. In Hungary, both NPPact and NPPh increased to a very high level in 1989. With the collapse of the agricultural production system of the planned economy in Hungary, agricultural productivity and harvest plummeted. In the decade after the regime change, both parameters showed strong oscillations. In South Africa, the pattern is quite different: The level of all flows is much lower than in the other six countries. NPPact declined until 1980 and shows a dramatic peak in 2000 that can be explained by climatic anomalies (see below). NPPh grew by 26% in the last 40 years.

With the exceptions of the Philippines, all countries show a considerable decline in HANPP intensity (the ratio of HANPP over NPPh) – that is, a growing HANPP efficiency of biomass production. Improvements of aggregate HANPP intensity were largest in Spain, Hungary and the UK, where HANPP intensity in the year 2000 was around 40% lower than in 1960 (Fig. 1d). In the UK and Hungary, HANPP intensity even reached a level of one or below one, which means that NPPh roughly equals HANPP and ∆NPPLC is at or below zero. This is the case when land use does not reduce the land's productivity compared to the NPP of potential vegetation (NPP0). In the Philippines, HANPP intensity did not change significantly during the 20th century.

Discussion

Comparability of Six Cases

As outlined in the Methods and Data section, all case studies used the same general definition of aboveground HANPP, but there were some deviations in the Austrian case. For Austria, the definition of NPPh was less inclusive (i.e. only used biomass extraction was considered) and a different, static method was used to estimate NPP0 (and therefore also the productivity of forests and natural grasslands). This less inclusive definition of NPPh results in a substantial underestimation of NPPh and thus lowers HANPP values. However, we do not believe that this strongly affects the temporal trend of HANPP and its components. The effect of the use of a static approach to estimate NPP0, as opposed to the LPJ results that consider changes in climate and atmospheric CO2 concentration which are underlying the other case studies, is less straightforward. However, because NPP0 is not only used as a reference state for calculating aggregate HANPP, but also underlies the calculation of NPPact of forests and natural grasslands, we assume that the general trend should be largely valid.

Another issue that requires explicit consideration is the stark increase in NPPact in South Africa around the year 2000. This peak is related to unusually strong rainfall in South Africa's drylands during that period. After a period of precipitation events above average during the La Nina period in 1999/2000, biomass productivity in huge parts of the RSA increased considerably. It is estimated that in several regions of Southern Africa biomass production rose by around 40% (Anyamba et al., 2002). These patterns during La Nina events were most significant in the westerns parts of South Africa, the Karoo (Nicholson and Selato, 2000). In any case, the values for South Africa around the year 2000 have to be interpreted with care.

What Drives HANPP Trajectories?

At the beginning of the 21st century, the aboveground HANPP was very high in four of the six countries. Despite considerable differences in climate and land use, the aggregate level of HANPP in Spain, Hungary, the UK and the Philippines was similar; that is, within a range of 60% and 70%. Austria's HANPP was a bit lower, but at 50% still far above the global average. If a more inclusive definition of HANPP had been used, the Austrian level would have been closer to that of the other four densely populated countries. HANPP is much lower in the only sparsely populated country in our sample, South Africa. This is in line with a recent cross-country analysis which has shown that population density has a strong effect on a country's level of HANPP. Typically, sparsely populated countries have a low level of HANPP, whereas HANPP is high in countries with high population density (Krausmann et al., 2009).

The HANPP trends (Fig. 1) bear some noteworthy similarities that may suggest a general pattern: In two of the three cases for which centennial data are available, HANPP increased in the first half of the 20th century. This growth came to a halt or was even reversed in the second half of the 20th century. In all six countries, harvest (NPPh) increased substantially in the second half of the 20th century, but HANPP stabilized (Philippines, South Africa) or even declined (all European cases). A comparison of HANPP trends and the development of GDP per capita (Fig. 2) shows that economic growth is not related to increases in HANPP. GDP is growing in all six countries at an average annual rate between 1.5% (Hungary) and 3.4% (Philippines). In contrast, annual growth rates of HANPP are negative or small and range between − 0.1% (Austria, UK) and 0.8% (Philippines). Consequently, HANPP per unit of GDP is falling rapidly in all countries (Table 2). HANPP behaves similar with respect to population growth: While it has been shown that there is a strong positive correlation between population and biomass harvest as well as biomass use across countries and over time (Krausmann et al., 2009; Steinberger et al., 2010), our data indicate that HANPP is not growing in line with population (Fig. 2): Population numbers are increasing in all cases, but HANPP remains stable or is even declining in the observed time periods. The only exception is the Philippines, where HANPP is growing in the first half of the 20th century, but at a much slower pace compared to population. As a result, per capita HANPP is declining considerably in all countries (Table 2). Biomass harvest per capita, in contrast, is even increasing in the four European countries (Table 2). This indicates a decoupling of biomass harvest and HANPP. To understand the decoupling of HANPP from economic growth and population growth we need to discuss the changes in the land use and biomass production systems that underlie HANPP in more detail.

Fig. 2

Development of HANPP in relation to GDP and population. Indexed (1961 = 1) development of GDP (1990 intl. Geary Khamis $ per capita and year), Population, HANPP (%), HANPP per capita and year and HANPP per $ GDP and year.

Table 2

Changes of HANPP per capita and year, HANPP per unit of GDP and year and Harvest (NPPh) per capita and year in the periods 1910–1961 and 1961 to 2005. Sources: own calculations based on Maddison (2008) for GDP and population.

	HANPP per GDP		HANPP per capita		Harvest per capita
	1910–1961	1961–2005	1910–1961	1961–2005	1910–1961	1961–2005
Austria	− 51%	− 71%	− 51%	− 15%	20%	12%
Philippines	− 71%	− 71%	− 71%	− 54%	− 50%	− 54%
UK	− 56%	− 64%	− 56%	− 15%	− 16%	32%
Spain		− 84%		− 29%		19%
Hungary		− 48%		− 2%		22%
RSA		− 65%		− 55%		− 49%

An explanation of these perhaps counter-intuitive findings needs to start with the recognition that HANPP measures the combined impact of land-use change and biomass harvest on trophic energy available in terrestrial ecosystems (see Methods and Data section). These two factors directly determine HANPP, but the interrelations between the two factors and their combined effect on HANPP are complex (Erb et al., 2009b; Krausmann et al., 2009). In order to explain the development of HANPP in the six case studies, we start with a closer look at the underlying changes in land cover. Table 3 shows the development of the share of forests and cropland of each country's total land area.

Table 3

Development of forest land and cropland in the period 1910 to 2000. Data were derived from HANPP studies referenced in Table 1.

	1910	1930	1960	1980	2000	∆1910–1960	∆1960–2000
	[% of total national territory]					[% during period]
(a) Share of forest land
Austria	40%	41%	42%	44%	47%	5%	13%
Hungary	n.d.	n.d.	14%	17%	19%		33%
Philippines	62%	56%	40%	28%	23%	− 35%	− 42%
RSA	n.d.	n.d.	8%	8%	8%		5%
Spain	n.d.	n.d.	20%	21%	22%		13%
UK	4%	5%	7%	9%	11%	76%	64%
(b) Share of cropland
Austria	23%	21%	18%	17%	14%	− 22%	− 21%
Hungary	n.d.	n.d.	60%	58%	52%		− 15%
Philippines	12%	19%	26%	33%	32%	114%	23%
RSA	n.d.	n.d.	11%	11%	9%		− 14%
Spain	n.d.	n.d.	43%	41%	36%		− 15%
UK	25%	24%	30%	29%	27%	19%	− 10%

n.d. … no data.

As Table 3 shows, forest area increased everywhere except in the Philippines throughout the observed period. Cropland areas declined in Austria, continuously increased in the Philippines, and increased in the first half of the 20th century in the UK, while they decreased in the second half of the 20th century in all countries except the Philippines. The Philippines are the only country with a massive deforestation over the 20th century: forest cover fell from 62% in 1910 to only 23% in the year 2000. The shift from agricultural land back to forests in recent decades has been observed in many now industrialized countries and is often referred to as ‘forest transition’ (Kauppi et al., 2006; Mather and Fairbairn, 1990; Mather and Needle, 1998; Meyfroidt et al., 2010).

Declining farmland areas and growing forest areas are likely to result in a decline of HANPP because HANPP per unit area is much higher on croplands, where aboveground HANPP is mostly above 90%, than in forests where HANPP levels are usually far below 40% (Haberl et al., 2007). Moreover, on cropland, the HANPP level per unit area and year is more or less independent of the yield level as NPPact and NPPh usually grow in parallel; that is, increases in plant growth are largely matched by increases in NPPh (see below). The general rule of thumb according to which increases in forest land result in decreases of HANPP whereas increases in farmland drive HANPP upwards may not hold, however, in arid regions. In these regions, irrigation may lead to a very large increase of NPPact over NPP0 and HANPP can even become negative, see for example the maps in Haberl et al. (2007).

When comparing the development in the Philippines to that in the European countries, we also need to consider their differences in terms of their respective stage in the socio-ecological transition from agrarian to industrial society (Fischer-Kowalski and Haberl, 2007), as well as the extremely strong population growth observed in the Philippines (see Table 1). While the European countries have all completed their agrarian–industrial transitions, the transition is still on-going in the Philippines. High levels of HANPP in the European cases in the early points in time were a legacy of deforestation processes that had mostly occurred before our observation periods started (e.g., Bork et al., 2001).

The UK is a special case due to its low initial forest cover in the beginning of the 20th century. Despite very strong growth of forest area, HANPP increased in the UK in the first half of the 20th century, driven by an increase in cropland. In the UK, cropland increased at the expense of grasslands that were used relatively extensively in the beginning of the 20th century. The area of cropland grew due to the food production campaigns of the First, but most noticeably, the Second World War, when areas of grassland were plowed up for arable cultivation (Martin, 2000; Sieferle et al., 2006).

Perhaps unexpectedly, the expansion of settlement/infrastructure areas into agricultural land, observed in all six case studies (not shown), did not result in HANPP growth, sometimes rather in a reduction. This can be explained as follows: Soil sealing results in a HANPP of 100%, but settlement and infrastructure areas are usually accompanied by areas such as gardens and parks with an often quite high NPPact and mostly low NPPh. These areas are often irrigated and fertilized. As a result, HANPP on these areas is usually much lower than on intensively used croplands. Moreover, infrastructure areas mostly grow at the expense of cropland, and therefore their growth may even reduce HANPP, at least if the cropland is not shifted somewhere else, i.e. if cropland areas are shrinking, as observed in most countries.

As shown in Fig. 1c, NPPh grew massively in all six countries. In particular in the second half of the 20th century, these increases in harvest did not translate into further increases in HANPP. This was linked to a surge in average NPPact in the same period (Fig. 1b). As discussed above, the reforestation of agricultural areas with low productivity contributed to some extent to the increases in NPPact that counteracted further increases in HANPP. But the growth of NPPact on agricultural land was even more important, as it allowed to greatly increase harvests without increasing HANPP. NPPh is related to agricultural yields and can therefore be interpreted as an indicator of ‘output intensification’ (Lambin et al., 2000).

Agricultural intensification is therefore important for understanding HANPP trajectories.

Beginning after World War II, the industrialization of agriculture, and the so called ‘green revolution’ in the developing world (Evenson and Gollin, 2003; Tilman, 1998), helped to rapidly increase agricultural yields and biomass harvests. Agrochemical inputs, irrigation and a spatial reorganization of land use allowed for massive increases in the NPP of agricultural ecosystems. Fig. 3 shows that fertilizer use per unit of cropland multiplied in all countries in the two decades after 1960 and reached very high levels in the UK, Austria and Hungary. In these countries, fertilizer consumption declined since the 1980s, when pressures from environmental legislation and economic drivers triggered a more efficient use of agrochemicals (e.g. Krausmann et al., 2003). The massive decline in Hungary is a result of the collapse state planned agricultural production system and is also reflected in a drastic decline in NPPh and NPPact (Fig. 1).8

Fig. 3

Mineral fertilizer use (pure nutrient of nitrogen, phosphorus and potassium fertilizer) in the six case studies.

Fertilizer input per unit of agricultural land in the RSA reached a maximum in 1981. The decline in the 1980s was linked to the economic crisis of Apartheid (trade embargo, disinvestment), which considerably impeded agricultural modernization (visible also in a low NPPh on cropland after 1978). The slowdown of agricultural performance can be directly related to the removal of governmental subsidies as well as to the rising costs for fossil fuels and mineral fertilizers. In Spain and the Philippines, where agricultural industrialization progressed at a much slower pace, growth in fertilizer use continued to grow and is approaching Central European levels.9 In arid Spain, irrigation was also an important driver for increases in harvest and NPPact. According to data reported by the FAO (2011) the amount of irrigated cropland in Spain increased from 9 to 22% since 1961.

Fertilization and irrigation result in stark increases in NPPact and NPPh of agricultural areas but hardly affect HANPP on these areas, as the additional plant growth is subsequently harvested and the amount of NPP remaining in the ecosystem (NPPt) remains approximately at the same level. Moreover, the industrialization of agriculture accelerated the shift of agricultural land to forests discussed above: Capital intensive crop and livestock production systems were concentrated on the best farmland and land of marginal productivity was increasingly taken out of production and became available for reforestation, a process which has also termed agricultural adjustment to land quality (see Krausmann, 2006; Mather and Needle 1998). This contributed to the observed increases in average productivity in the four European countries. Yet another side effect of productivity growth was the implementation of political measures to prevent overproduction, e.g. subsidies for letting land lie fallow. This land is then not harvested and thus also contributed to a reduction of HANPP.

The aggregate effect of the changes in land cover and in the intensity of land use was the observed stabilization or even a decline in HANPP in the second half of the 20th century in spite of massive increases in harvested NPP. Conveyed differently, the stabilization of HANPP can be seen as a result of considerable reductions of HANPP intensity: The industrialization of agriculture boosted NPPact on cultivated land and reduced ∆NPPLC. This allowed for reductions of the amount of HANPP per unit of harvested biomass. As shown in Fig. 1d, HANPP intensity declined considerably in the four industrialized countries throughout the second half of the 20th century. In Hungary and the UK, two countries with a very high share of cropland or intensively cultivated grassland, average NPPact even surpassed NPP0 and HANPP intensity reached values below one. Only in South Africa and the Philippines we find little evidence of significant reductions in aggregate HANPP intensity. While in South Africa this can be explained by extensive land use practices, for the Philippines it warrants a closer look. The apparent lack of improvements in HANPP intensity is largely due to the definition of HANPP intensity, which includes biomass destroyed/burnt without direct socioeconomic use into NPPh. Relating total HANPP to biomass extracted for further socioeconomic use would be a more comprehensive intensity measure. For the Philippines such a measure shows significant improvements in HANPP intensity: the value decreased from 8.7 in 1910, over 3.7 in 1960, to 2.7 in 2003. These efficiency gains can be mainly explained by the fact that compared to other ways of biomass appropriation, the use of fire for land clearing has lost significance during the 20th century (Kastner, 2009) and consequently HANPP per used biomass extraction declined while HANPP per total NPPh did not change much.

The improvements in HANPP intensity in the last decades came, however, at a considerable cost. Irrigation, fertilization and general intensification of land use not only boosted agricultural output, but the industrialization of agriculture increased the direct and indirect energy requirements of agriculture (e.g. mechanization and fertilizer use/Fig. 3). As a result, the energy return on investment (EROI) of agricultural production systems declined (Krausmann et al., 2003; Pimentel et al., 1990), while greenhouse gas emissions from agriculture increased (Smith et al., 2008).10 Intensive agricultural production also entailed a plethora of environmental pressures: Leaching of plant nutrients into ground and surface water, soil erosion, depletion of ground water reserves, release of toxic agrochemicals and many more (IAASTD, 2009).

Population growth and economic development drive the demand for biomass. More people need more food, and increasing income drives up the consumption of biomass-intensive animal products (Erb et al., 2009a). The substitution of fossil fuels for fuel wood is usually offset by an increase in timber demand during industrial development (Krausmann et al., 2009; Steinberger et al., 2010). Our six case studies nevertheless suggest that population and economic growth and HANPP are largely decoupled during industrialization (Haberl and Krausmann, 2001). The major underlying factor of this decoupling are the technological changes in agriculture discussed in the previous section. Another factor that needs to be taken into account when national HANPP trends are discussed, however, is international trade.

Biomass Trade and Embodied HANPP

The amount of biomass traded internationally is increasing rapidly. At the global scale, biomass exports grew by a factor of 6 from 1961 to 2008 and currently amount to 1.6 Gt (Gigatons, 1 Gt = 109 t = 1 Petagram or Pg) of fresh weight per year (FAO, 2011). It has been shown that biomass trade may considerably decouple domestic biomass consumption from domestic HANPP, because HANPP only considers effects on national territory and biomass trade may shift burdens abroad (Erb et al., 2009b). The ‘physical trade balance’ (defined as imports minus exports; Dittrich and Bringezu, 2010) of the six countries included in this study is shown in Fig. 4. According to these data, net biomass imports are large but slightly falling in the UK, increasing in Spain and in the Philippines, about balanced and more or less stable in Austria and South Africa, and falling in Hungary. These data suggest that growing imports may have contributed to the stabilization or decline of HANPP in Spain and the Philippines but rather not in the other four countries.

Fig. 4

Development of net biomass trade from 1961 to 2008 in kg C/m²/yr in the six countries. Negative net trade means net exports, positive net imports.

However, the physical trade balance of biomass does not provide a full picture in the context of HANPP. The reason is that the HANPP related to each unit of traded biomass is strongly dependent on its quality (e.g., grain vs. meat or cheese) as well as on regional differences in yields and conversion efficiencies (Erb et al., 2009b). The measure of embodied HANPP (eHANPP) has been introduced to correct for such distortions (Erb et al., 2009c; Haberl et al., 2009). eHANPP corrects national HANPP for the HANPP embodied in traded biomass products, similar to the ‘virtual water’ concept (Hoekstra and Hung, 2005). Unfortunately, no eHANPP time series are currently available — at present there exists only one global eHANPP dataset for the year 2000 (Erb et al., 2009c). Results for the six countries in our analysis are reported in Table 4.

Table 4

HANPP on national territory, HANPP embodied in traded biomass, eHANPP (the sum of HANPP on national territory and HANPP embodied in trade) and the ratio of eHANPP to HANPP for the six countries in the year 2000. Negative values of HANPP embodied in traded biomass indicate net exports. Data source: Erb et al. (2009c).

	HANPP on national territory [1000 t C/yr]	HANPP embodied in traded biomass [1000 t C/yr]	eHANPP [1000 t C/yr]	Ratio eHANPP/HANPP
Austria	22	7	29	1.34
Hungary	36	− 5	31	0.87
Philippines	166	16	181	1.09
RSA	184	− 15	168	0.92
Spain	113	65	177	1.57
UK	71	65	136	1.91

As Table 4 shows, eHANPP was almost twice as large as HANPP on national territory in the UK in the year 2000. This suggests that imports play a substantial role in supplying the UK with biomass-based products and are probably quite important for the decoupling between population, economic growth and HANPP. In the Austrian case, the eHANPP related to biomass imports is substantial, despite an almost balanced physical trade balance. By contrast, Hungary and South Africa are even ‘net exporters’ of eHANPP, i.e. the HANPP on their respective territories is larger than the eHANPP related to the products consumed in their national economies.

Conclusions and Outlook

Based on the evaluation of data for six case studies we assume that HANPP increases with population during early periods of industrialization (Fig. 5). When fertile land is abundant, growth in the demand for food and feed is met by the expansion of agricultural land. This results in deforestation and productive woodlands are replaced by less productive agricultural ecosystems. In this phase, population growth (or export production) outgrow any improvements in yields and drive an increase in HANPP, which can reach high levels of more than 70%. The industrialization of agriculture changes this trend (Fig. 5). In our case studies massive increases in agricultural yields resulted in increases in the NPPact of agricultural ecosystems. Consequently growing harvests could be achieved without further increasing HANPP, or could even go in parallel with reductions of HANPP. Increasing area productivity reduced the pressure to cultivate land of marginal productivity. Agricultural areas were increasingly taken out of production and reforested, which further contributed to reductions of HANPP. As a consequence, HANPP intensity, that is the amount of HANPP associated with each ton of biomass extraction, declined.

Fig. 5

Conceptualization of changes in HANPP and its components during the 20th century: During early periods of industrialization biomass harvest (NPPh) increased at the expense of NPP remaining in ecosystems after harvest (NPPt), mostly due to the expansion of cultivated areas. This drove increases in HANPP. The industrialization of agriculture allowed for increases in harvest by increasing the NPP of the prevailing vegetation. Additional harvest did not result in higher HANPP, but HANPP rather slight reductions in HANPP can be observed.

The analysis of six historical case studies has shown that biomass harvest can be decoupled from HANPP. This development may be seen as promising with respect to the forecasted growth of global population and the corresponding surge in the demand for food, feed and fuel. Indeed, data on national HANPP for 175 countries in the year 2000 suggest that a considerable share of countries have land use systems with a high HANPP per unit of biomass extraction (Haberl et al., 2007). In these countries, which are responsible for 40% of global HANPP but contribute only 26% of global biomass extraction, further improvements in HANPP intensity might be possible.

In spite of such potential efficiency gains, which could help to increase global biomass harvest without increasing HANPP, caveats are warranted. HANPP is not an all-encompassing indicator of ecological pressures resulting from land use and it is not sensible to important problem shifts associated with agricultural intensification (cf. Marull et al., 2008). The national case studies have shown that reductions in HANPP intensity result from increased land-use intensity. These increases are associated with a reduction of energy efficiency of biomass production and considerable ecological costs. The green revolution was based on the industrialization of agriculture and entailed a surge in fossil energy and agrochemical inputs. It is clear that further improvements in HANPP efficiency are likely to lead to a further increase in agricultural inputs and aggravate environmental pressures, at least if they are based on the currently predominant technologies.

Our analysis of long-term trends of HANPP from six national case studies indicates that the ecological concept of carrying capacity cannot be applied to human societies in a straightforward manner, at least since the Neolithic revolution. By means of technology, humans are able to drastically increase the capacity of terrestrial ecosystems to supply food and fuel for human use and to decouple human population and economic growth from HANPP. This does, however, not mean that there are no limits for growth. Expanding the capacity of the earth to provide human society with biomass comes at considerable ecological costs and the long term sustainability of the achieved gains in HANPP efficiency is questionable. An analysis of HANPP and related parameters can help to better understand the dynamic relation of population and resource use and the processes involved in these transitions.