Fertilizing Methods and Nutrient Balance at the End of Traditional Organic Agriculture in the Mediterranean Bioregion: Catalonia (Spain) in the 1860s

By reconstructing the nutrient balance of a Catalan village circa 1861–65 we examine the sustainability of organic agricultural systems in the northwest Mediterranean bioregion prior to the green revolution and the question of whether the nutrients extracted from the soil were replenished. With a population density of 59 inhabitants per square km, similar to other northern European rural areas at that time, and a lower livestock density per cropland unit, this village experienced a manure shortage. The gap was filled by other labour-intensive ways of transferring nutrients from uncultivated areas into the cropland. Key elements in this agricultural system were vineyards because they have few nutrient requirements, and woodland and scrublands as sources of relevant amounts of nutrients collected in several ways.


Introduction
This work is part of a larger project that seeks to clarify the reasons for the abandonment of traditional organic management in Mediterranean agriculture. We wished to determine how sustainable these systems were with respect to nutrient replenishment into the soil and whether our results could contribute to improve contemporary organic farming practices in a region such as Catalonia (Spain). In an earlier study in which we reconstructed the energy balance in the same area for 1860 we found a positive return on energy investment of around 1.41 or 1.67 depending on the boundaries of the area under study (Cussó et al. 2006a, b;Tello et al. 2006Tello et al. , 2008. In this study we complete this socio-metabolic investigation by estimating the nutrient balance and assessing the maintenance of soil fertility.

Agrological and Socioeconomic Features of the Area Under Study
The municipality of Sentmenat is located in the Catalan Vallès county, some 35 km northeast of Barcelona, with a total area of 2,750 ha, of which 59 % were cultivated in 1861 (Fig. 1). The village was settled during the tenth century AD in a small plain located in a tectonic basin between Catalonia's littoral and pre-littoral mountain ranges. It has an average slope of 9.7 % and an annual rainfall of 643 mm. The heliothermic Huglin index of 2,168 is good enough for winegrowing-it has a minimum requirement of 1,500 and reaches a maximum municipal score of 2,778 in Catalonia (Badia-Miró et al. 2010). Rainfall and temperature allow for reasonable yields in cereal crops, at least in flatlands with a higher water retention capacity.
In 1860, 354 families and 1,713 people were registered in Sentmenat, a population density of 59 inhabitants per square km., allowing 1.7 ha (including the municipal area) or 1.4 of cropland per inhabitant. Seventy per cent of labour capacity was devoted to agriculture and 21 % to industrial activities. As many as 208 out of the 241 agricultural families were "peasants" or "landowners", while 21 worked as ploughmen tenants and 12 as daily labourers. Moreover, 187 out of the 208 landowners were so-called autonomous peasants who primarily worked their land with family labour, only hiring labour in peak seasons. Many landless labourers had kinship ties with peasant owners . Despite being far from egalitarian, this rural society enjoyed a broad degree of access to the land and can be basically seen as a peasant community (Netting 1993;Ploeg 2008).
The Gini coefficient of inequality in owned land distribution was 0.58 in 1859, or 0.51 if only cropland is taken into account. In 1735 this had been 0.77 and 0.67 respectively, and rose again to 0.76 or 0.70 in 1918 following the Phylloxera plague that killed all the old vines in the 1880s (Badia-Miró et al. 2010). The reduction in landownership inequality between 1735 and 1859 was driven by vineyard specialization (Garrabou et al. 2009). Many landowners and some peasant owners leased poor sloping soils previously covered by scrub and pastureland to an increasing number of non-heir relatives or landless immigrants who built terraces and planted vineyards . The use of the Catalan sharecropping contract called rabassa morta, which stayed in force until the death of the vines planted, was widespread, and led to lower levels of inequality recorded, reflecting a reduction in land-access and income inequality rather than in landownership distribution as such (Tello and Badia-Miró 2011).

Land-uses, Livestock Densities and Manure
Vineyard specialization developed during the nineteenth century whereby some land, usually the best, was devoted to grain, legume and vegetable polyculture. In 1861, the Fig. 1 Location of the study area: the municipality of Sentmenat and neighbouring townships in the province of Barcelona and Catalonia (Spain) extreme scarcity of natural pastures (12.4 of the total) seriously constrained livestock production. The majority of cropland consisted of vineyards or olive groves that extracted less nitrogen while pruning supplied a useful byproduct contributing nutrients to the soil. At the same time, thanks to the increase of arboriculture, the ratio of uncultivated area to land sown with herbaceous crops could be maintained as high as 2.4, and the ratio of permanent landcovers to annually sown land was as high as 5.1 (Table 1). All these features were typical of the Mediterranean-type of "intensive organic agriculture" (Sieferle 2001;Wrigley 2004) that went into a steep decline during the economic globalization at the end of the nineteenth century leading to World War One (Tello et al. 2006Marull et al. 2008).
A crucial component of this form of pre-industrial organic agriculture was the number of cattle grazed on uncultivated pastures and foraged crop waste in order to provide enough manure to sustain the land sown with cereals (Krausmann 2004): in 1865, only five head per square km in Sentmenat (seven including donkeys)-a live weight density of only 12 livestock units (LU) of a standardised weight of 500 kg (LU500) per cropland square km. (Table 2). In comparison, Cunfer and Krausmann (2009) found 24 LU500 per square km of agricultural area in the intensively cropped Austrian village of Theyern in 1829, and 4-13 LU500 in Finley Township (Kansas) in the very extensive land-use American Great Plains between 1895 to 1915. This density of livestock would provide only 1.5 tonnes of fresh manure per cropland hectare, a figure corresponding to the 1.37 tonnes recorded in 1919 in the first statistical survey of fertilizers in the province of Barcelona. The input to sustain a highly intensive regime of organic agriculture recommended by agronomists of the timewas 10 tonnes per cropland hectare or almost ten times these amounts (Aguilera 1906;Cascón 1918;Slicher van Bath 1963).
Nevertheless, these average figures do not account for marked differences between crops. No manure was used for growing vines, and only very small quantities in olive groves. This explains the role played by vineyard specialization in reducing the ratio between land sown with cereals and uncultivated land (Table 1). If we assume that all manure was applied only to growing grains, livestock densities would rise to 46 LU500 per square km of cropland and average inputs to 5.6 tonnes of fresh manure per sownland hectare, which corresponds to the 6-7 tonnes per hectare attributed by other sources to the rain-fed cultivation of cereals in the province of Barcelona during the second half of the nineteenth century-including applications ranging from 22-32 tonnes per hectare on irrigated lands. These would be double the inputs of between 2.5-5 tonnes per hectare applied in the United States at that time (Cunfer 2004(Cunfer , 2005Burke et al. 2002), and matched the average of 4 to 5 tonnes per hectare in England and Wales from the mid-nineteenth century to World War Two (Brassley 2000).

How the Nutrients Gap Was Closed
Even assuming woody crops received no manure, there remains a significant gap between available livestock densities and fertilization required. Hence we conclude that either other organic inputs were used or unsustainable soil mining was occurring until chemical fertilizers came to be used. Cunfer and Krausmann (2009) conclude that thanks to high livestock densities Austrian farmers were able to return over 90 % of nitrogen (N) extracted to cropland, although they produced little marketable crop surplus. In contrast, farmers on the American Great Plains produced plenty of exports but used few animals to exploit rich grassland soils, thus returning less than half of N extracted. After depleting soil fertility for over six decades, they faced a steep decline in crop yields from 1880 to 1940, when chemical fertilizers were introduced (ibid).
To compare these cases with Western Mediterranean agriculture we reconstruct a complete nutrient balance for our case study. Nutrient outputs and inputs in crops and seeds have been estimated, taking into account both the harvest index and the reuse of by-products (Table 3). Some 40 kg N per hectare were removed annually from irrigated lands and vegetable gardens, three times more than the average and 5.6 times the N taken up by vineyards. Rain-fed intensive rotations of grains sown without fallow extracted 39 % of all N in 22.6 % of cropland, about 22 kg N per hectare. Vineyards drew 7 kg N per hectare, including grapes and pruning-shoots. Although occupying two-thirds of cropland, vineyards removed only 38 % of N, 28 % of P and 18 % of K.
Overall, this distribution reveals the rationale behind the priority given to the scarce manure: it was first applied to irrigated land, and then to rain-fed cereals rotated with Nfixing leguminous crops or green manures. Vineyards were not fertilized with manure except at planting, and only received small amounts of other organic fertilizers such as leaf litter and branches buried in ditches dug between rows of vines, or burning and ploughing into the soil the hormigueros (formiguers in Catalan). These resembled small charcoal-kilns made with piles of dried vegetation that were burnt under a soil cover to generate slow and incomplete combustion. The material obtained was used as fertilizer or soil conditioner (Olarieta et al. 2011;Figs. 2 and 3).
Some 20,195 kg of N were annually removed from the 1,618 ha of ploughed land in Sentmenat circa 1860-65, equivalent to 12.5 kg N per hectare. All locally produced manure contained only about 12,164 kg N. Considering that at least 50 % was lost in the dung pile, the N available would be reduced to 6,082 kg, or a maximum of 3.8 kg N per hectare a year (Cascón 1918; Tisdale and Nelson 1956;Johnston 1991), thus requiring alternative sources of nutrients and agricultural fertilization practices to fill this gap. Five different possibilities are considered: 1) human sewage and garbage; 2) symbiotic bacterial fixation through leguminous crops; 3) green manures; 4) burying fresh biomass into the soil; and 5) material generated by hormigueros.
One of the most difficult components of any organic nutrient balance to measure is the value adopted for a For sheep and goats maintained in grasslands 50 % of manure has been discounted considering that it could not be recovered by locking the herd at night in a pen or taking it to stall. b Estimated by us from the available feed and assuming the existence of five chickens or rabbits per household. c d See Table 7. e Rain-fed and irrigated herbaceous crops and vegetable gardens Source: our own estimate made from the livestock census of 1865 in the district, the data provided by contemporary literature and the assumptions made in the energy balance published by Cussó et al. (2006b). The following references have also been taken into account: Bouldin et al. (1984), Loomis and Connor (1992), Sørensen et al. (1994), Tisdale and Nelson (1956), Archer (1985) atmospheric N fixation made by symbiotic bacteria. Even today, the scientific literature presents bewildering variation in the figures of N fixed by leguminous plants. This can be largely explained by the circumstantial nature of the symbiosis between legumes and Rhizobium bacteria whereby the presence of high doses of mineral N in the soil suppresses  Tisdale and Nelson (1956), Loomis and Connor (1992), and Angás et al. (2006) bacterial fixation. Moreover, only a part of the N content of a leguminous plant comes from the atmosphere. Before the Rhizobium nodulation develops in the roots, the plant needs to uptake mineral N from the soil and therefore not all the N absorbed before the flowering and maturation of the grain can be attributed to the Rhizobium nodules. The lower energy cost of directing the available carbon towards their own growth rather than to Rhizobium colonies, which may remain inactive, explains why legumes break symbiotic N fixation when there is enough mineral N in the soil. This flexibility has a lot to do with the crucial role legumes played in the millennial development of organic agriculture, in which the mineral N was practically always lacking in the soil (McNeill and Winiwarter 2006). Unfortunately, this creates considerable uncertainty about the actual symbiotic fixation in each particular circumstance. Values ranging from 10 kg to over 300 kg N per hectare a year have been estimated ( Herridge and Bergersen 1988). There are examples and opinions that reduce N symbiotic fixation to very low values, or even assume a net negative outcome if the grain is removed and plant residues are not incorporated into the soil. The only safe rule is that symbiotic and free fixation are greater the poorer the mineral N content of the soil. Therefore, the N mobilized by leguminous crops from the atmosphere would have been higher in past organic agricultural systems, a hypothesis that contemporary organic farming may well help to  corroborate (Obersom et al. 2007). Despite these uncertainties, we arrived at the preliminary estimates shown in Table 4.
Green manure provided another important source of leguminous N-fixing properties. We have found sufficient historical sources to conclude that green manures were used in the province of Barcelona during the second half of the nineteenth century, and were widely endorsed by agronomists of that period. However, we do not have precise data for the average area sown, the species used or the amount of atmospheric N fixed. As a very preliminary rough estimate, and assuming that 3.6 % of herbaceous cropland was sown annually with green manure, about 165,900 kg of aerial biomass may have been buried into the soil. We assume that the atmospheric N fixed was the only net input flow from green manure that must be included in the calculation, since the rest of the nutrients are simply recycled into the soil.
According to many local contemporary sources, crop byproducts and forest biomass were directly applied to the soils as fertilizers, besides being used as compost matter in the manure pile. Two procedures were employed: 1) a direct burial of fresh vegetal matter in ditches dug between rows of vines; 2) ploughing into the soil ashes, charcoal and topsoil burnt in the hormigueros (Miret 2004).
In order to estimate the local biomass potential, the ratio between land sown with grains, land devoted to arboriculture and the available biomass that could be removed from woodland or scrubland was analysed. The amount of nutrients added to the soil by the burial of fresh biomass is easy to infer from its N-P-K content (although only the organic N is taken into account, disregarding any possible loss by mineralization). The amount of nutrients supplied by each hormiguero has been taken from Olarieta et al. (2011). It seems that any net N contribution would have been negligible but the hormigueros would have added some amounts of P and K, which could also result in a significant yield increase of legumes intended to supply N (Johnston 1991).
However, there remain some unknown aspects of the impact this method may have had to the biotic component of soil fertility. According to the agronomist Cristobal Mestre and the chemist Antonio Mestres (1949), the rise in temperature experienced by the topsoil covering the hormiguero caused a variation in the populations of soil microorganisms that may help to explain the harvest increases obtained in experimental fields fertilized in this way compared with control plots-for example, by increasing free atmospheric N fixation (see Table 5 for our own preliminary estimate).
We assume that the burial of biomass and the hormigueros played a role in filling the remaining gaps in the nutrient balance. They appear in our balance sheet as a minor component because the estimated number of hormigueros is small due to the considerable uncertainties that still prevail about the size of each hormiguero and the amount of biomass burnt in them. Acknowledging that this issue deserves to be further studied, we have taken as a cautionary option an average figure of 13 hormigueros per cropland hectare per year (or 20 if only applied to vineyards), a figure adjusted to the locally available forest biomass-while figures up to 200 (Roca 2008) or even 700 per hectare per year (Barón de Avalat 1780) can be found. Taking into account the high labour inputs demanded by these techniques, it seems reasonable to assume that their use would depend on the relative scarcity of other fertilizers and the abundance of cheap labour. We came to a similar conclusion considering the task of removing fallen branches and dried biomass from the Mediterranean forests and scrub land, which usually become prone to wildfires (Pyne 1997; Grove and Rackham 2001).

An Organic Nutrient Balance Close to Equilibrium?
We matched the nutrients utilized by crops, or lost through other processes, with two different estimates of their replacement by various fertilizing methods: a) a maximum potential amount of N-P-K which the mass balance tells us should be somewhere in the local agro-ecosystem; and b) the fraction we believe was actually put into the soil discounting material losses by these fertilizing methods: manure piles, cesspools, Source: our own, based on the N-P-K composition per unit weight of the legumes used in our balance (Bassanino et al. (2007), Berry et al. (2003), Castellanos et al. (1996), Drinkwater et al. (1998), Domburg et al. (2000), Holland et al. (1999), LaRue and Patterson (1982), Loomis and Connor (1992), Obersom et al. (2007), Peoples and Craswell (1992), Phillips and DeJong (1984), Schmidtke et al. (2004), Tisdale and Nelson (1956), Wilson ed. (1988) and the other references given in Table 7 latrines, hormigueros, burial of fresh biomass, crop legumes or green manure (Table 6). This balance is not designed to assess accurately all nutrient flow transported by livestock, agricultural labour and natural processes. Some minor flows have been omitted, such as erosion losses which could be largely offset by the accumulation of sediments in other nearby lands-depending on the scale of analysis. Nor have we assigned values to the mineralization processes in the soil, or the possible increase obtained in atmospheric N fixation by stimulating free bacterial activity through piles of hormigueros. But even admitting a margin of error, which can only be reduced through future calibration and comparison with other balances, we believe that the usefulness of this assessment lies in its heuristic function. We think that this balance sheet helps us to reveal some basic features of the societal attempts made to close the flow of nutrients in highly intensive organic agriculture of a Mediterranean-type. Despite inaccuracies and uncertainties it allows us to formulate some results. First, the amount of nutrients available to sustain cropland fertility could have been almost large enough to replace the main macroelements taken from the soil by crops and natural processes, provided that the processing efficiency of animal manure and human sewage was not lower than 50 % in N, 90 % in P and 80 % in K. We suppose as well a high labour input allocated to make hormigueros or bury fresh biomass in order to import nutrients -mainly K-from uncultivated areas to cropland. Should these assumptions be changedfor example by considering a loss higher than 50 % of N content in manure management and reuse of sewages-the totality of nutrients extracted would not have been replenished (Fig. 4). On the other hand, we know that N losses in manure piles could only be reduced up to 30 % if the floor of livestock stall was paved and the compost process was accurately managed (Cascón 1918).
In any event, we are not assuming that actual fertilization always balanced crop extractions in each farm or plot. A very important issue that is masked in average figures is to how social inequality affected the availability of livestock manure, woodland or scrubland cuts, and latrines. In spite of the fact that the maximum potential of fertilizers available was probably enough to maintain soil fertility, we believe that poorer winegrowing tenants may have worked at a deficit level. Commoner (1971) considered a basic principle of an ecosystem's functioning to be "everything goes somewhere." Our balance shows, for example, that a portion of K was obtained from burying or burning biomass in hormigueros. Thus, any remaining K gap could probably have been closed by increasing labour and biomass allocated to make them. Another important issue that requires comment is that the proportion of cropland devoted to feed and fodder to support livestock could be kept relatively low due to the role played by agricultural recycling and natural pastures (Figs 4 and 5). This material eco-efficiency required careful management of cropland, uncultivated land and livestock breeding-which was also a key to the corresponding high degree of energy efficiency (Cussó et al. 2006a, b).

Discussion
These results help to explain the high incidence of winegrowing in Sentmenat circa 1860-65. Two-thirds of the cropland acreage devoted to vineyards brought about a significant saving of N and P. The importation of 1,556 Hl a year of wheat, together with some amounts of salted fish and rice, meant an annual gain of 2,561 kg N, 433 kg P and 459 kg K which accumulated in sewage. While the N content in the wine exported was negligible, the P taken yearly from wine was around 414 kg and the K around 2,070 kg. As a consequence, the nutrient trade balance led to a net annual gain of some 2,561 kg N and 433 kg P, together with a net annual loss of 1,611 kg of K (Tello et al. 2006Garrabou et al. 2009Garrabou et al. , 2010Badia-Miró et al. 2010).
However, the ability to access the full potential of nutrients available in the local agro-ecosystem is not the same as the ability to collect and reintroduce them into a Mulch, grasses, acorns, branches or bushes that could also be partly used to burn in«hormigueros», along with pruning and other by-products of crops. We have assumed that only a quarter of the available biomass in woodland and scrubland was used in this way. b We have considered the maximum potential number of«hormigueros»according to the available biomass. Source: our own from Cussó et al. (2006b), and results of fieldwork and analysis performed by José Ramon Olarieta croplands. Most of our uncertainties arise over the difference between potential and actual nutrient availability. Bearing in mind the processing losses of animal manure and human sewage, the actual availability of animal manure and human wastes would cover only 33 % of N, 84 % of P and 62 % of K required to replace extraction by crops.
-Each«hormiguero»is made with an average of 68 kg of woody biomass. -As a result of the combustion we have 2.5 kg of char and 2.5 of ashes. -The composition of the ashes from the« hormiguero»is the same as if the same type of woody biomass were burnt elsewhere.
-They are made in equal parts of pruning and woodland or scrub cuts. A. Average natural losses Drinkwater et al. (1998), Galloway et al. (2004)), Jambert et al. (1997), Kosmas et al. (1997), Parton et al. (1996), Rana and Mastrorilli (1998), Rosswall and Paustian (1984), Tisdale and Nelson (1956), Torrent et al. (2007) Leaching: 5.5 kg N/ha Denitrification: 1.5 kg N/ha irrigated Ammonia volatilization: 5 % green manure N inputs B. NPK composition of nutrients extracted by crops Soroa (1934), CESNID (2003), Mataix (2002). Moreiras-Varela et al. (1997) Source: our own based on the previous tables. (Item number corresponds with the numbers in Table 6) have covered about 14 % of the K required in order to balance the local agro-ecosystem in 1860-65. In other words, while the agronomists of the day were correct in noting the inadequacy of local livestock densities, other options were available for Mediterranean-type intensive organic agriculture. Nevertheless, these alternatives were highly labour-intensive. Hence we come to a third conclusion: the main limiting factor regarding organic nutrients was not biophysical, but technical and economic. Rather than the maximum potential of N-P-K available in the agro-ecosystem, what mattered most was the actual capacity to combine and recycle them as fertilizer taking into account the chain of losses experienced in dung piles, latrines, cesspools, sewers or hormigueros. A key limiting factor was the amount of human and animal labour needed for that purpose.
There are, of course, some ultimate agro-ecological limits inherent in any organic-based agrarian economy aiming to increase yields without overshooting the renewable resources available. Before reaching these limits it was possible to increase leguminous crops, which in 1860-65 covered just one quarter of cropland, and to use them as green manure. Here again the limiting factors appear to be more economic than agro-ecological. The water stress typical of the Mediterranean region was dealt with to some extent through increasing the water retention capacity of soils by increasing their organic matter content, or with temporary and permanent irrigation. Another option was specialization in arboriculture, which requires less water and extracts fewer nutrients from the soil. However, all these alternatives needed land improvements and labour investments, and these in turn had opportunity costs according to the relative market profitability of their alternative uses.
Fourth, the scope for increasing agricultural yields through more intensive organic fertilization was very limited unless land-uses were changed, as recommended by agronomists, by increasing the land sown with leguminous crops and using them as green manure or by increasing forage, livestock and manure. To a degree, either of these land-use changes were constrained either by the rainfall levels of the Mediterranean environment, or by actual market opportunities to reallocate land towards commercial woody crops (González de Molina 2002; Guzmán Casado and GonzálezDeMolina 2008;González de Molina et al. 2010;Vanwalleghem et al. 2011).
Finally, it should be emphasised that in Sentmenat circa 1860-65 the maintenance of cropland fertility was only possible through a permanent transfer of nutrients from uncultivated areas of woodland, scrub and pasture. This was of course an overriding feature of any past organicbased agricultural system. What draws most attention in this case study is the key role played by human labour in cropping legumes and green manure and transferring nutrients from woodland or scrub by means of hormigueros burnt and biomass buried into cropland as compared to the less significant role of livestock in that transfer. This was a key feature of Mediterranean organic agriculture that contrasted with other European bioregions (Fig. 5). Thus we come to our fifth and last conclusion: organic fertilizers rather than animal manure played a key rolealbeit small in absolute terms-in transferring nutrients from uncultivated areas into cropland. Besides being highly labour-intensive, these transfers imposed a relevant nutrient tribute on woodland or scrubland, mainly in terms of K, which added to the simultaneous extraction of timber, firewood or charcoal. The maintenance of cropland fertility was closely related to the sustainability of this multiple-use of forests, which up to a point might have been overexploited. Photographs taken during the first third of the twentieth century show diminished forest cover. At that time woodlands were reduced to a minimum in Catalonia, and even more in Spain: forest land occupied only 15 % of the country area in 1915 ( Tello and Sudrià 2010), and about 20 % in 1955 (Schwarzlmüller 2009).
Acknowledgments This work has been developed in the project HAR2009-13748-C03-01HIST on Environmental History of Mediterranean Agrarian Landscapes funded by the Spanish Ministry of Science and Innovation. A first version was presented in the session 3.5 on Sustainable agricultural systems: historical soil fertility and farm management at the First World Congress of Environmental History (WCEH2009). Following the advice of anonymous reviewers, all data were revised in depth by Elena Galán using the Manager of Energy and Nutrient Balances of Agricultural Systems (MENBAS). This accounting tool is now being developed at the University of Barcelona, and soon will be offered as an Open Access resource in our website: http://www.ub.edu/histeco/p2/eng/ index.php. After this major revision many details have been substantially changed, although the overall picture and general conclusions remain. Many references of historical sources have been suppressed for the sake of brevity, and may be found in Garrabou and González de Molina (2010). We thank to Joan Romanyà, Miriam Burriel, Mar Grasa and Marià Alemany at the University of Barcelona, Roberto García at the University of Jaén, and the two anonymous reviewers for their useful comments and corrections.