Full text
In many parts of the world, nitrogen‐fixing legumes are grown along with cereals to form a large part of a nutritionally balanced diet and stable agroecosystem (WHO/FAO, ). In contrast, legumes have declined in European cropping systems where they typically occupy a few percent of the arable surface (Zander et al., ). The decline has been attributed to various factors but notably the competitive advantage afforded to duty‐free imports of protein and oilseed crops from the 1960s, in the wake of which Europe came to rely on external sources for most of its needs in legume products (Hausling, ; Kuhlman, Helming, & Oudentag, ; Westhoek et al., ). Legume production has also been limited by poor policy support, unstable yields, and a widespread “lock‐in” preventing the uptake of legumes among the main cereal‐based cropping systems (Cernay, Ben‐Ari, Pelzer, Meynard, & Makowski, ; Magrini et al., ; Stagnari, Maggio, Galiena, & Pisante, ). Increasingly, however, the value in legume production is being recognized and stimulated globally through the International Year of Pulses () and in Europe by a range of policy initiatives (Bues et al., ; European Parliament, ; Hausling, ), in consequence of which the areas cropped with legumes have increased in some EU countries (De Cicco, ).
The positive case for legumes is based on a range of factors, not the least of which is food security (EU, ). The heavy reliance on imported food and feed protein would threaten food supply in conditions of blockade or war, high tariffs, environment disasters and increasing global demand for legumes. Strong environmental arguments for legumes center on their ability to fix their own nitrogen from the air, leave some of this as residue for the next crop and hence reduce the total mineral nitrogen input (Jensen, Peoples, & Hauggaard‐Nielsen, ; Kopke & Nemecek, ; Panday, Li, Askegaard, & Olesen, ). Legumes have many other beneficial ramifications for ecological sustainability, including enhanced soil structure and habitat diversity and introducing breaks in the sequence of cereal crops to allow management of pests (Kopke & Nemecek, ; Stagnari et al., ). Nutritional and health benefits of legumes arise from various properties of the grain, including plant protein of low glycaemic index and, when mixed with meat products, reducing the energy density of food intake to counter cardiovascular disease (Aleksandrowics, Green, Joy, Smit, & Haines, ; Rebello, Greenway, & Finley, ; Schneider, ). Finally, an ethical awareness has risen among consumers and governments that home‐grown legumes offer routes to a more environmentally conscious agriculture and to food systems that are less exploitative of natural resources and ecosystems in other parts of the world (Aleksandrowics et al., ; Schneider, ; Stagnari et al., ).
The current position, however, is that the increased need for nitrogen (N) during the intensification of crops and grass in the mid to late 20th century (Erisman, Sutton, Galloway, Kilmont, & Winiwarter, ; Rudel et al., ) continues as an embedded reliance on mineral N fertilizer and imported legume protein (Westhoek et al., ). High inputs of mineral N have caused a range of environmental problems related to pollution as greenhouse gases (CCC, ), losses to water, and reductions in arable biodiversity, the long‐term impacts of which remain major concerns (Ascott et al., ; Hawes, Begg, Squire, & Iannetta, ). Attempts to reduce nitrogen usage in agriculture from a peak in the late 1980s have met with partial success. EU practices resulting from set‐aside and nitrate directives (EC, ; EEC, ) followed by national action led to a lowering of input in some instances, but most effectively in managed grass rather than arable (e.g., in the UK: Defra, , ; Fertiliser Practice, ). Further regulation of N use in cereal crops will be limited simply because of the relatively high N concentration needed in grain.
While increasing legume inclusion is relevant to the whole of Europe and other parts of the world, the methods by which it can be achieved will need to be adapted to both the local cropping systems and the legume species and products that have potential for expansion (Revoyron et al., ). For example, demands to expand legume production will be fruitless if there is limited land available or if several legume products are competing for the same space. Therefore to establish a methodology, a case study was conducted within a well‐defined, maritime, agro‐climatic region of eastern Scotland (Squire, ). Beans (Vicia faba L.) and peas (Pisum sativum L.) became established in Britain in prehistoric times (Treasure & Church, ). The role of legumes in regenerating soil fertility was recognized in the improvements era of 1700–1850 when they were reported to occupy variously 1 in 4 to 1 in 12 years of the arable‐grass sequences (e.g., Wight, ). However, no absolute assessment of area occupied by grain legumes was made before 1850. Subsequently, they suffered a decline to a low point in the 1930s. Following a minor resurgence, peas and beans have occupied as little as 1% of the arable surface in the last two decades (ERSA, ). Accordingly, their current land coverage brings limited savings to the mineral nitrogen budget and unavoidable reliance on imports of legume‐based products for food and animal feed (Schneider, ). In this and any case study, however, several uncertainties need to be resolved as a prelude to designing a program of legume expansion.
- Potential for legume expansion in postintensification cropping systems. The reasons should be established whether the historical decline in legume cultivation has led to changes in the cropped ecosystem that might now limit expansion of legumes. For example, 20th‐century intensification, including the move to winter crops and associated increased fertilizer and pesticide usage, might have relegated legumes to unique localities or crop systems with specific requirements that no longer exist in the wider agroecosystem.
- Conflict for land area between legume species and products. The potential for conflict should be assessed between different legume species or products and the crops they might replace. Expansion might not be feasible, for example, if different commercial opportunities for legume products all competed for a limited agricultural space and if current high‐value crops would be at risk of displacement.
- Environmental benefits of legume expansion. The consequent benefits should be quantified, especially through savings of mineral nitrogen fertilizer, but also more widely for example in reduction of pesticide usage, rehabilitation of soil, and restoration of habitat.
Crucial to resolving the above uncertainties is new data that have recently become available through the EU's Integrated Administration and Control System (IACS), from which it should be possible to quantify where legume crops are grown, the crops they are grown with, and hence the crop systems in which they occur and could be expanded. Analysis using IACS data here examines the above three questions within this case study region. To focus the arguments, the paper considers the scope for enhancing three current production options—beans for aquaculture feed, pulses (peas but also beans) for human consumption, and legume alternatives to grass for farm animal feed (mainly peas).
The region lies in the east of Scotland between latitudes 54° and 61°N and mostly within 40 km of the sea. The first crop‐grass census was taken in 1854 when summaries of crop areas were given for the administrative regions (old counties) current at that time (Thorburn, ). Census was repeated in the four decades after that (Porter, ); then, a comprehensive annual census was undertaken from 1902 to the present (Agricultural Statistics Scotland, ; ERSA, ). Land managed for agricultural production (excluding rough grazing) covers around 1.9 million hectares and, averaged over the previous 15 years, consists of arable crops (602,000 ha), short‐term or rotational grass (358,000 ha), and permanent grass (5 years and over, 936,000 ha). Fields do not remain in the same category indefinitely, many switching over time, particularly between crops and short‐term grass. The public census records allow legumes to be identified in three periods: beans and peas grown for harvesting as grain crops, categories that persisted from 1854 up to the 1950s; forages grown as crops, rather than as crop‐grass swards, and variously designated as vetches, tares, and cereal–grass mixtures, recorded in the middle decades of the century; and several grain legume types that emerged after 1940, starting with legumes as vegetables and extending to include “vining” peas for human consumption and peas and beans for combining (mostly for animal consumption). The first and third categories overlapped for a few years, when both were recorded, and parts of the third (e.g., for combining) must have been similar to the main components of the first but were designated separately in the census.
While all census yearbooks give areas of crops and grass, it is not possible from publicly available data to define crop sequences or crop systems. After 2000, however, information per field has become available through the EU's Integrated Administration and Control System (IACS) via Scottish Government. IACS allows definition of consecutive crops grown in spatially referenced land units (fields). The raw data were subject to quality control, which led to some fields being excluded because data are incomplete. Quality control was considered effective for 5 years, 2003–2007, which was also a period of relative stability in legume area. From the several hundred thousand fields registered in Scotland, a sample set was defined that was most likely to be targeted in a program of legume expansion. Fields were excluded if they were designated as “rough grazing” for the whole period, and hence very unlikely to be converted to arable, or contained grass for the whole period, designated “grass 5 years and older.” The remaining fields—all containing arable only or a mix of arable and grass in sequence—were mostly in the lowland cropped ecosystems of eastern and southern Scotland (maps are given in Results3). For the estimation of nitrogen inputs from a separate government survey (see below), fields were also excluded if data on specific crop types were incomplete or uncertain. The final set for the full analysis here consisted of >53,000 fields (or >265,000 crop‐years), each defined by a location, area, and sequence of crops or grass. IACS allows definition of four types of legume in this region: beans for animal consumption (1,123 crops within the defined overall sample set), beans for human consumption (104 crops), peas for animal consumption (727 crops), and peas for human consumption (592 crops). Together, the legumes constitute around 0.9% of all crop‐years in the final sample set, a value consistent with their percentage occurrence in the government crop census that includes all fields.
Trends in fertilizer nitrogen inputs were taken from Fertiliser Practice (, and all previous yearbooks to 1988) which gives nitrogen application in kg per hectare to the 1960s, augmented by earlier records (Church & Lewis, ; Squire, ). The period of the late 1800s and early 1900s is described by agronomic accounts that imply a consistent input of nitrogen to the main crops as sulfate of ammonia or nitrate of lime (Paterson, ), from which N in kg per hectare is here calculated.
For estimation of N inputs for each field in the IACS set, a nominal nitrogen input was estimated based on data in Fertiliser Practice which give area treated, average rate of nitrogen in kg/ha to fields where it was applied, and a corresponding overall application rate (OAR) averaged over all fields including those to which it was not applied. For the main cereals, these two rates are very similar, since fertilizer is applied to almost all fields, but they differ for grass since fertilizer is not applied to all fields in all years. For the period studied, OAR was 89% of the field rate for grass under 5 years and 79% for grass of 5 years and over. OAR is used here as it more correctly describes the mean application to grass over a period of years. Animal manures were applied to part of the crop‐grass area, ranging from 10% of the area of high‐input crops to 28% of that of grass. However, the overall contribution of manures to nitrogen additions remains uncertain from present survey protocols (Fertiliser Practice, ).
Pesticide inputs are used along with N inputs to characterize the degree of agronomic intensity among crops and crop systems. Of various metrics produced by government pesticide surveys (PUS, ), “formulations” is used here rather than mass of pesticide (which can be misleading since mass has decreased over time as new active substances have become more effective). A formulation consists of one or more active substances designed for a specific target. The survey records the number of different formulations applied to the main crops and represents the result as an area of application (previously termed spray‐hectares). Here, formulations are summed for main pesticide groups—fungicide, herbicide, and insecticide with molluscicide—but exclude growth regulators. Dividing this treated area by the area sown with the crop gives an indicator of pesticide use intensity for each crop, representing the number of times that a crop is treated with formulations during a season. Grass of both types received very little pesticide, spring cereals 5–6 formulations per season, winter cereals 9–11, and potato >15. The grain legumes grown in the region generally receive fewer formulations than spring cereals but more than grass. Examples of the approach are given by Squire, Hawes, Valentine, and Young ().
The main designations for crops and short‐term grass (with abbreviations as necessary) are as follows: spring barley, SB; winter barley, WB; winter wheat, WW; spring wheat, SW; winter (WO) and spring oats (SO) usually combined as “oats”; winter oilseed rape, WOSR; turnips and swedes; set‐aside; grass 5 years and older, grass>5; and grass of less than 5 years, grass<5. The nominal agronomic intensity is estimated for each field in the IACS sample set by averaging the N and pesticide formulations over the crops in the sequence. At this scale, no allowance is made for adjustments due to residual N left by previous crops, but while these vary with soil, weather, and local practice they would be averaged out over several years in the census.
Trends in areas occupied by grain legumes and in mineral nitrogen input appear unrelated (Figure ). Areas sown with beans and peas treated as grain crops declined after the first census in the 1850s reaching a low in the late 1930s, when together they occupied 0.23% of the area sown with arable crops. After a temporary rise in the 1940s, they decreased and were unrecorded after 1960. Commentators attributed the decline in the late 1800s to reliance on potato in the subsistence economy and a long‐term trend to supplant pulses with legume‐grass forages as the main source of biologically fixed nitrogen (Findlay, ; Porter, ). Legume crop forages, harvested mostly for grain (dashed line) similarly reached a low point in the 1930s, increased temporarily in the 1940s, mainly through a rise in cereal–legumes mixtures, but then also receded. The combined area of peas, beans, and forage crops in 1930 was 0.92% of arable crops and 0.47% of crops and rotational grass (combined as “Arable Land” at that time). From the 1940s, several new categories of grain legumes appeared in the census, notably legumes classed as vegetables, “vining peas,” and peas and beans for stockfeed, which are here combined as the gray line. The peak in the 1980s was due to a temporary rise in peas. In the terminology employed by Revoyron et al. (), the region as a whole has passed through a series of coherence phases, when the type of legume cultivation and area were more or less stable, and transition phases when the species, products, and areas changed. Given the previous variation, however, it is unclear whether the rise after 2000 represents a durable upward trend. Between 2000 and 2015, grain legumes, comprising peas and beans for combining and vining peas for human consumption, persisted in a coherence phase, in which their area varied between 0.7% and 1.3% of the area of crops and short‐term grass. The mean from the government agricultural census over the period 2003–2007 was 0.87%, similar to that in the final IACS sample over the same period used in the analysis here (see Materials and Methods).
Enlarge this image.Trends in (a) legumes grown as crops, separated into areas sown with beans and peas treated as grain crops (heavy line to the left, 1854–1960), forage legumes (dashed), and later census categories of beans and peas after 1939 (marked “all types”) including for grain, freezing and fresh products (gray line to the right), from various sources, mainly Agricultural Statistics Scotland () (Materials and Methods); and (b) mineral nitrogen application to arable (tillage) crops and grass from Fertiliser Practice (), supplemented with projected trend from 1952 to 1971 estimated from earlier sources (e.g., Church & Lewis, ; Squire, ), and the horizontal dashed line indicating typical mineral N inputs preintensification (Paterson, )
Mineral fertilizer was applied to some crops in the late 1800s and early 1900s, but there are too few systematic records to allow detailed analysis. Accordingly, mineral N applications over this early period are represented by the level dashed line up to 1950 using figures cited by Paterson (). Mineral N increased steeply as part of agricultural intensification after 1950, reaching a peak around 1990. Trends before 1972 were estimated from Church and Lewis () and after 1972 from Fertiliser Practice (). Nitrogen inputs were then regulated downwards as a result of various EU directives and set‐aside, but the more so for grass than crops. Despite fluctuations due to these and other factors, Squire () estimated than total N inputs to crops were reduced to around 80% of those at the 1990 peak. Therefore, grain legumes began their decline in the 1800s, well before the intensification‐driven, steep rise in mineral N applications between 1950 and 1990; and the recent rise of grain legumes occurred during the period of level or regulated N input to arable crops. Trends in legume cultivation of the previous 150 years have therefore occurred independently of trends in total mineral nitrogen inputs.
Four categories are discernible in the IACS records (see Materials and Methods), namely peas and beans for both animal feed and human consumption (abbreviated to beans‐animal or B‐A, etc.). Peas and beans grown for animal feed have the widest distributions and may be considered equivalent to the peas and beans grown for grain in early census records (Figure ). Accordingly, the present distributions of these two crops (Figure ), each dot representing a field, are superimposed on the proportional distributions from the 1854 census based on administrative counties at that time. In 1854, beans were to a degree concentrated in the center and southeast whereas peas covered a wider area including the northeast. The distributions of beans‐animal and peas‐animal today, while broadly similar to those in 1854, indicate a loss of cropped area to the west and northwest of the current concentrations, implying the distributions have retreated eastwards. Since areas grown with beans and peas are only a little less now than in the 1850s, the concentration of these crops has increased in more easterly regions. Notably, peas‐human for freezing and canning, which was not grown at the time of the earlier census, is concentrated in a narrow band near the center‐east of the latitudinal range, while beans‐human is a minor crop located in two small areas. At least part of the reason for the restricted distribution of peas‐human is the need for proximity to processing plant. However, both beans and peas occur today broadly within the historical areas in which grain legumes have been grown.
Enlarge this image.Recent geographical distribution based on IACS data (Materials and Methods) of four grain legumes (a) beans for animal consumption (beans‐animal), (b) beans‐human, (c) peas‐animal, and (d) peas‐human, fields shown by dots with, superimposed on (a) and (c) distribution of comparable crops in the 1854 census, each circle representing relative area of crop recorded for each administrative county of that time
The distributions of the four pulse categories suggest they might be grown with different combinations of other crops. For example, the areas occupied to the northeast by peas‐animal are known to have a low proportion of winter crops. The proportions of crops grown over the period were estimated from IACS records for each field that contained a legume (Figure ). The legume occupied 20% of crop‐years (1 year in five), except in a few fields that grew more than one legume during the accounting period. In each case, the crops most commonly grown in the same field as the legume are spring barley (SB) and winter wheat (WW), but distinct differences are apparent between the legume types. For example, beans‐animal had a high proportion of WW, while peas‐animal had less of the cereals and about 20% grass<5 years and both peas‐human and beans‐human had a higher proportion of potato than the others. In the whole of the area designated crops and fallow, the census records show that SB occupies 2.5 times as much area as WW. The reduction of SB to an area much more similar to that of WW in Figure implies that SB is the crop most commonly displaced when one of the legumes is grown in a crop sequence.
Enlarge this image.Proportions of crops grown with the four categories of grain legume: (a) beans‐animal, (b) beans‐human, (c) peas‐animal, and (d) peas‐human derived from IACS data (see text)
Since the main crops differ in agronomic input, the crop profiles in Figure suggest the legumes are not confined to fields of a specific or limited range of intensity. Fields containing a legume were therefore assigned nominal nitrogen and pesticide inputs based on the national means in respective government surveys (see Materials and Methods and examples in Squire et al., ). If for example, a field supported WW, WB, SB, WOSR, and a legume, then the nominal inputs would be the means of these crops from the respective survey. The resulting nominal inputs from all fields including a legume were assembled into a grid defined by ranges of nitrogen in intervals of 20 kg/ha and ranges of pesticide in intervals of one formulation (Figure ). The relative number of fields falling in each grid square (relative to the total for each legume) is indicated by the size of the circles, each positioned at the upper limits of N and pesticide for each grid square. The line from the origin to the maxima is provided for visual comparison. Fields that occupy circles to the right along the N axis contained high proportions of winter crops, while those to the center of the N axis were mainly spring crops and grass. Fields that occupy circles to the top of the pesticide axis contained mainly winter crops and potato, those to the center mainly spring cereals and those at the bottom mainly grass. The presence of the legume in all cases would have reduced the nominal nitrogen intensity compared to similar sequences without the legume.
Enlarge this image.Intensity of inputs estimated for fields growing a grain legume (a) beans‐animal, (b) beans‐human, (c) peas‐animal, and (d) peas‐human; annual mean nitrogen and pesticide inputs averaged across crops in 5 years based on national fertilizer and pesticide surveys (Materials and Methods), crops then grouped in intervals of fertilizer and pesticide, size of circles representing the proportion of fields in each nitrogen‐pesticide grid square, the line from zero to the absolute maxima providing visual comparison
While each of the four legumes was grown across a wide spread of input intensity, both beans‐human and peas‐human were grown in a large proportion of sites located above the line in Figure , indicating they were grown in crop combinations of very high input. In contrast, beans‐animal occurs on both sides of the line but mostly below it, and peas‐animal occurs mostly below the line but notably at a concentration of sites of moderate N (60–80 and 80–100 kg/ha) but low pesticide, typical of fields containing a high proportion of grass in the crop sequence. The results in Figures and confirm that legumes have not become relegated to, for example, low‐input fields, but instead occur in, and offer the potential for expansion across, a wide range of crops and grass.
The next question concerns the extent of geographical space available for expansion of legumes. The distribution of fields in the intensity matrix in Figure suggests that certain legume types might be associated with specific crop‐grass systems. The analysis proceeded in two stages: a set of crop‐grass systems in which the legumes were grown was first defined; then the areas which these systems occupied in the whole region were quantified. The first stage considers only the 2,500 fields growing legumes (about 1% of the area, see Materials and Methods), while the second stage considers all >53,000 fields in the IACS sample that could potentially grow legumes. The reasoning is that if a legume type was presently grown in a defined combination of crops, then the legume could be most readily expanded into a system that was as similar as possible to that combination, but presently without a legume.
The fields containing legumes were partitioned sequentially into groupings depending on simple rules defining minimum combinations of grass and cereals. The first split was into sites that contained legumes but had only grass as the other crop. The second split was into sites that contained both grass and spring cereals but no winter cereals. The third was into sites that contained both spring and winter cereals but no grass, and so on. In all, six crop‐grass systems of increasing input intensity were identified in which legumes were grown (Table ).
Six crop‐grass systems identified for legume expansion from a sample of >53,000 fields covering 331,000 ha, identified by numerals I to VI (column 1), short name (column 2) and simple inclusion rules (column 3)| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 |
| Crop‐grass ID | Crop‐grass type | Inclusion rules | % Area | Main crops (% of the respective crop‐grass type) | Nominal mean N input (kg/ha) | Nominal mean pesticide formulations | % All legumes among categories | Main legumes | Legume replacing | N at 20% legume (% initial) |
| I | Grass with arable break | 80% grass, other crop | 14.5 | Grass<5 66.4%, grass>5 13.6%, SB 14.5%, SO 1.6%, turnip 1.5% | 102.8 | 0.93 | 3.6% | P‐A | Grass<5 15%, grass>5 5% | 80.1 |
| II | Grass‐arable (spring cereal) | 20%–60% grass, spring crop; no winter cereal | 19.4 | SB 43.2%, grass<5 42.3%, grass>5 3.9%, oats 3.7% | 101.9 | 2.5 | 6.6% | P‐A | Grass<5 10%, SB 10% | 77.0 |
| III | Arable spring cereal | Spring cereal; no grass or winter cereal | 15.5 | SB 88.6%, set‐aside 4.3%, oats 4.0%, WOSR 0.9% | 100.2 | 4.9 | 8.9% |
P‐A B‐A P‐H |
SB or other spring cereal 20% | 79.2 |
| IV | Arable mixed cereal | At least 20% winter and 20% spring cereal, no potato | 20.7 | SB 40.4%, WW 22.6%, WB 16.8%, WOSR 10.3%, oats 3.9%, set‐aside 3.8%, | 144.3 | 7.4 | 39.5% |
P‐A B‐A P‐H |
SB 15%, WB 5% | 83.2 |
| V | Arable winter cereal | At least 20% winter cereal, no spring cereal, no potato | 8.5 | WB 36.9%, WW 34.7%, WOSR 19.7%, set‐aside 3.8%, WO 3.1% | 177.7 | 9.2 | 18.3% |
P‐A B‐A B‐H P‐H |
WB 10%, WOSR 10% | 78.6 |
| VI | High‐input cereal plus potato | Winter cereal, potato | 8.5 | WW 28.9%, SB 26.8%, potato 20.2%, WOSR 8.6%, WB 8.3%, set‐aside 1.6% | 149.4 | 9.9 | 10.4% |
B‐H P‐H |
SB 20% | 86.1 |
Data on current status of the crop‐grass systems are shown in columns 4 to 7: percentage of area covered, proportions of main crops, and estimates of intensity provided by nominal mean nitrogen and pesticide inputs based on government survey. Columns 8 and 9 show percentage of current legumes grown in each system and the most prevalent legume types (abbreviations P for peas and B for beans, A for animal and H for human consumption). Columns 10 and 11 show indicative target crops for replacement by the legumes in column 9 and estimated effect of 20% legume inclusion on N inputs for each crop‐grass type.
The sample set of >53,000 fields was then searched by similar rules, which were adapted to account for the fact that the fields did not contain a legume (Table ). For example, category I in those fields that contained a legume was defined as consisting of 80% grass and 20% legume; but the rule was changed to 80% grass and 20% “any other crop” for the fields that did not contain a legume. The six crop‐grass systems are listed as I to VI in Table , each with a general descriptor (column 2), the rules that defined the system (column 3) and a range of metrics quantifying the proportions of crops and legumes together with indicators of agronomic intensity, the mean nominal nitrogen and pesticide inputs. Between them, fields in categories I to VI covered almost 90% of the total fields in the sample. The remaining fields fell between or outside these crop‐grass types as defined by the rules.
Legumes are currently spread across all systems, but “arable mixed cereals” (system IV) had the highest concentration at 39.5%. Notably, the high‐input systems V and VI also had sizeable proportions. However, the legume types were not evenly represented across the systems: peas‐animal was by far the main legume at the lower end of the range in systems I and II, whereas peas‐human and beans‐human were the main legumes at the high end in system V (consistent with Figure ). All four legumes occurred in the intermediate cereal‐based systems III, IV, and V.
Enlarge this image.Frequency histograms of nominal mean nitrogen input in three representative crop‐grass systems V, IV, and III (labeled on the figure and defined in Table ), based on averaging inputs to component crops (bars) and then after substitution of one of the crops by a grain legume assumed to receive no fertilizer (lines)
The procedure for identifying crops for substitution is partly subjective, and one approach is given here for illustration. The main criteria were that the most profitable crops namely potato and winter wheat (ERSA, ) were not substituted, no crop was entirely removed by substitution, and otherwise the most prevalent crop was substituted. The crops to be replaced by the legume were therefore mostly SB in the lower input categories (consistent with the proportions in Figure ) and then either WOSR, WB, or SB in the high‐input categories. In all categories, one crop or combinations of one or more were available in sufficient number for substitution. Problems would have arisen, for example, if one of the high‐input categories consisted solely of WW and potato, but such was the varied nature of these cropping systems, all categories had a range of potentially substitutable crops. Moreover, more than one legume could in principle be expanded in each of the categories except the extremes I, II, and VI where only one could be expanded.
One further possible limitation was considered. Expansion of peas‐animal and beans‐animal would be feasible throughout the wide ranges occupied, respectively, by peas and beans historically (Figure ), but peas and beans for human consumption might be constrained by nonbiotic factors such as proximity to processing facilities. However, 85% of crop‐grass type VI, which would be the most suitable for expansion of peas‐human, exists within the current latitudinal range of that legume in Figure .
Finally, legumes were introduced to fields by replacing one or more of the crops indicated in Table . At an indicative 20% legume substitution, values ranged around 20% reduction in N, but were greater where high N crops such as WOSR were substituted. If legumes were increased to 20% in all crop‐grass categories, then total N input weighted by area would be reduced to 82.7%. However, the crop‐grass systems differed markedly in the shape of the frequency distribution of N among fields. In Figure , the initial distribution is shown by the vertical columns and the recalculation after substitution by the lines. The distribution in the crop‐grass system of highest N input (winter cereals, system V, Figure a) was highly concentrated, 60% of fields occurring in the 180–200 kg/ha band while the other fields were distributed in immediately lower bands. Substitution reduced the peak to the 140–160 kg/ha band. Of the remaining five systems, two (IV, mixed cereals and VI, mixed cereals with potato) showed a much broader distribution with a similar peak and spread, that for IV being shown (Figure b). The peak N input of 140–160 kg/ha was moved down to 120–140 kg/ha after substitution. The remaining three systems (III, II, and I) also had a highly concentrated distribution but peaked at a much lower N input. In system III (spring cereals), 80% of sites occurred in the 100–120 kg/ha band which was shifted to 80–100 kg/ha after substitution. The profiles of N input in crops systems I and II (not shown) were similar to that in III, consistent with the similar N inputs to spring cereals and grass<5.
In addition to the N not applied to the legume, the legume will also leave a residual nitrogen that contributes to the pools of N in the soil. The next crop may take up some of this soil nitrogen (not necessarily just or all that from the legume), and hence, careful N budgeting can lower the applied N requirement of the next crop. While there are very few direct measurements of residuals in fields in this region, the evidence from elsewhere in Europe is that legumes in high intensity systems will still fix nitrogen and leave a residual, but that the quantity of residual and the proportion of that taken up by the next crop are highly variable and sometimes small (Iannetta et al., ; Jensen et al., ; Panday et al., ). Applying meaningful residuals to the whole set of fields consistently is therefore not feasible.
However, the potential is illustrated for the extremes of the range in Figure . Typical crop sequences are shown for crop‐grass systems V (Figure a, winter cereals) and III (Figure b, spring cereals). Accumulation of applied mineral N (based on nominal fertilizer inputs) is shown as step changes at the beginning of each year through a sequence of years. The number shown next to the end point of the accumulation is the annual mean, in this case starting with 188 kg/ha in system V. Analysis of the IACS data shows that the legume, in this instance beans‐animal, most commonly precedes WW, so should be introduced in place of WB or WOSR where indicated by the arrows. The legume itself is here assumed to receive a “starter” nitrogen application of 20 kg/ha (which is reported in the fertilizer estimates in some years). Replacing WB brings the annual mean to 87% (not shown), and replacing WOSR brings it down to 83% (shown). Beans‐animal occurred more than once in a small proportion of fields (3%), and by replacing both WOSR and WB with two legumes crops, the mean annual N is reduced to 132 kg/ha (70%). Finally, a residual nitrogen is assumed to be left by the legume and used by the next crop, a value of 50 kg/ha being well within estimated limits, bringing the annual mean to 116 kg/ha (62%, not shown). Further reductions would be possible to 100 kg/ha (53%) by applying no fertilizer to the legume and assuming a larger but still realistic residual of 75 kg/ha. A comparable analysis for a sequence of all SB in system III shows that similar changes reduce annual mean N from 104 to 44 kg/ha (44%).
Enlarge this image.Examples of estimated reduction in cumulative nitrogen (N) application in crop sequences following substitution of existing crops with a grain legume, numbers to the right showing initial N input (kg/ha) averaged over the 6 years then downwards, N input and percentage of initial following increasing substitution in: (a) winter cereals, system V, initially with 188 kg/ha then replacing WOSR with legume including 50 kg/ha residual (83%), replacing WOSR and WB with legumes and same residual (70%), and as previous but omitting N fertilizer to legumes and assuming substitutions plus residual N of 75 kg/ha (50%); and (b) the same procedure for spring cereals, system III, initially with 104 kg/ha and replacing one SB (66%), two SB (73%), and finally two SB with 50 kg/h residual (43%)
Several routes are open to a region or country that wishes to increase the inclusion of nitrogen‐fixing legumes in its agriculture. Choices need to be made between familiarity and innovation. Traditionally, beans and peas have been widely grown in this case study region. Other grain and many forage legumes, including cereal–legume mixtures, have been grown in the past and may find opportunities in modern cropping systems following new research. The current prevalence of the once most widespread form of legume cropping, the clover–grass mixture (Findlay, ), is highly uncertain due to the absence of appropriate records in any agricultural census. Until the status of clover–grass mixtures is better understood, the most immediate route to greater legume inclusion for better health and environment is therefore through expanding the area sown with existing grain legumes.
The question was raised in the Introduction as to whether 20th‐century intensification, particularly its use of high inputs of mineral nitrogen and a shift to winter crops, had excluded grain legumes from parts of the arable‐grass surface and therefore restricted opportunities for expansion. Unequivocally however, the combination of crop census with modern IACS records showed that intensification in this region had not specifically excluded legumes, which had declined before intensification, but rather had created opportunity for their inclusion in high intensity crop sequences. At the same time, combinations of low‐input systems, notably grain legumes with spring cereals and rotational grass, had been retained within extensive areas of grazing land and feed crops for livestock. There is now a wider range of crop systems in which the existing grain legumes could expand than there was in the early 1900s before intensification. The six crop‐grass systems defined here are the result of a local and particular historical legacy and a current diversity of land use, but a comparable analysis would form an appropriate baseline for any program of legume expansion in any region.
The interrogation of IACS data allowed resolution of further uncertainties—specifically whether legume inclusion would remove existing profitable crops and whether the different legume types would be in competition for the land available for expansion. On the first of these questions, the sown area of the crop most likely to be substituted in this region, spring barley, has been highly flexible over the past century, increasing or decreasing by more than 20% on many previous occasions to accommodate trends in other crops and changes in policy. For example, it increased from being a secondary crop to replace oats in the mid‐20th century, decreased by over 20% to accommodate winter barley and winter oilseed rape in the late 20th century, shifted in response to the coming and going of set‐aside in the 1990s and typically varied between 10% and 20% in response to years of extreme weather (Agricultural Statistics Scotland, ; ERSA, ). There are ample physical space and a likely flexibility in this crop for legume expansion in present systems therefore, without having to design and implement totally new systems or lose the area of the most profitable crops. The only possible exception is winter cereals (system V) where in any case, winter barley or oilseed rape is available in sufficient quantity to be substituted.
The savings of mineral N by legume expansion were only partly estimated due to incomplete data. At its peak in the late 1980s, mineral N applications were similar for “tillage crops” (mainly arable) and grass at 130 kg/ha (Fertiliser Practice, ). Over two subsequent decades, that to grass had fallen to around 70 kg/ha or 54% but that to tillage to only 120 kg/ha or 92% (Fertiliser Practice, ; Squire, ). The analysis here allows estimates of further reductions following legume inclusion to 20%, bringing the mean for tillage crops to around 100 kg/ha or 77% of the peak in the late 1980s. Further major reductions to grain‐based agriculture will be difficult since crops need a minimum N concentration in grain. Without reducing cereal yield itself, there are few options for further reduction in N other than by maximizing the use of the biologically fixed residual from legume crops, yet the quantity of this residual in present crop‐grass systems is highly uncertain. The fixation rate by grain legumes in the field is well documented to lie typically between 100 and 350 kg/ha (Kopke & Nemecek, ; Panday et al., ), but the proportion of this left as residual N has shown great variability between locations and cropping systems. Comprehensive field studies are now needed to confirm the calculations in Figure that suggest mineral N could be taken to 50% of current inputs by increasing legume inclusion to one third and assuming realistic residuals are fully used.
Broader questions of conflict or competition between the various economic and environmental benefits of legumes can also be resolved through analysis of the type presented here. Potential options are summarized in Figure , which shows the six crop‐grass systems, aligned I to VI vertically in boxes. Each of the four legume types is indicated by P‐H, P‐A, B‐H, B‐A (where for example, P‐A represents peas for animal consumption, etc.), and beneath these letters, vertical lines cover the systems in which the stated legume type is most likely to expand (adapted from Table ). The target of increased production or environmental benefit is given in the box to the right of each subfigure. Whatever the target, management has to operate through one or more crop‐grass systems, not simply the legume crop in question. The legumes types (P‐A, etc.) that any target has to affect are shown in bold type in each subfigure.
Enlarge this image.Opportunities for legume expansion to satisfy six targets, indicated in rounded boxes, in three commercial products (upper subfigures a, b, and c) and three environmental benefits (lower subfigures d, e, and f). Squared boxes aligned vertically represent the six crop‐grass systems in increasing order of input intensity, I to VI in Table . The four legume types are shown as P‐A, peas for animal consumption, P‐H peas for human, B‐A beans for animal, and B‐H beans for human. Each legume type occurs in a range of crop‐grass systems (Table ) indicated by the vertical bars, solid where most of the legume is grown, and dashed to show minor occurrence. To achieve each economic target, one or more legume types must be activated for expansion, shown by the arrows from the target box passing through one or more of the systems
For illustration, options for three products—beans for aquaculture, peas and beans for human consumption, and peas to augment grass in stockfeed systems—are represented by the upper three parts of Figure (a, b, and c). Beans for animal feed (and hence aquaculture) are currently grown mainly in crop systems IV and V, and some also in III, while peas for animal feed are grown mainly in crop systems I and II. In each case, there is enough area to allow substitution to 20% (Table ). Moreover, these two requirements for legume product would not be in conflict since they operate through different crop‐grass systems. Expansion of the third use—grain legumes for human consumption—would occur in two of the systems, IV and V, which would be also activated for aquaculture. There is a potential conflict for space therefore between needs for aquaculture and home‐grown pulse food. Whether conflict actually arises depends on the degree of legume expansion predicted by the respective industry. For example, the demand for legume protein by aquaculture in Scotland could (given necessary developments in processing plant) be provided from 36,000 ha of home‐grown faba bean at a mean yield of 4.5 t/ha (Beans4Feeds, ). Currently, in “mixed cereals” and “winter cereals” (systems IV and V in Table ) far more than that area is grown with the most likely crops to be substituted. Similar projections now need to be made for a targeted expansion of other products to assess whether conflict for space would be a reality.
The desired environmental benefits, in addition to nitrogen reduction, were also found not to be in conflict with increased legume production for these three products with certain exceptions (Figure d,e,f). The expansion of beans for aquaculture and peas and beans for human consumption would be directly compatible with a targeted aim of reducing inputs and restoring soil condition in the high‐input crop systems V and VI (Figure d) which are associated with poor and possibly declining soil quality (Squire et al., ). Opportunities to enhance the landscape mosaic by growing more legume crops would accrue from expansion of all three products, but benefit to pollinators specifically (as an example of a topical group) would be more restricted (Figure e). Peas, most of which are self‐pollinated, are known to be a poor resource for pollinators, while beans, though open pollinated, are still only a moderate resource (Baude et al., ). An effective program of pollinator support would have to be achieved in addition to rather than primarily by grain legume expansion through enhancement of other food sources such as forage legumes (Baude et al., ) and the broadleaf weed flora (Hawes et al., ). The final target chosen as an example is pesticide reduction (Figure f). Applications are highest in crop‐grass systems IV, V, and VI (Table ). All current legumes are typically treated with fewer formulations than most other crops in these systems, and their expansion would reduce the overall pesticide load. Legumes would also reduce pesticide slightly in system III (spring cereals) but have no effect or even cause a small increase in systems II and I which receive very little pesticide.
In summary, many opportunities were identified for legume expansion in a region of long‐established agriculture in which legumes declined to a low point in the mid‐20th century. The analysis has confirmed that grain legumes had not been excluded by 20th‐century intensification but neither had they been encouraged to expand. While intensification into mixed cereals and winter cereals had provided more opportunities for legumes and their products, they persisted as a minor contributor to the evolving crop‐grass systems. Legumes could expand in principle throughout the region in a variety of systems, but expansion would need to be driven primarily by an increased demand from retailers and consumers. That such demand is growing is clear (Westhoek et al., ; Zander et al., ). Legume expansion would bring about a range of environmental benefits, notably through reduced mineral nitrogen inputs. Further research is needed, however, on the optimum positioning and management of legumes in cropping sequences to achieve the greatest reduction in N usage with least (if any) trade‐off for yield and economic output. However, charting the course toward further reductions in nitrogen, for example to contribute to greenhouse gas emission targets (CCC, ), will need action in addition to grain legume expansion. Much greater precision is needed in the available national data for nitrogen applied in manures, in supplements (mainly from legume protein grown elsewhere) and through nitrogen fixation in grass containing forage legumes, which are more resource‐efficient than grass alone (Suter et al., ). Models of the nitrogen balance in crop systems have been developed for this region and could be employed to quantify options (Young, Mullins, & Squire, ). More generally, IACS data coupled with national census could provide assessment of land and food security in any region. The methodology demonstrated here can now be extended to a wide range of questions in national and regional planning.
The work in this paper was funded by the Scottish Government Strategic Research Programme and the European Union's Horizon 2020 Research and Innovation Sustainable Food Systems (SFS) Programme—TRansition paths to sUstainable legume‐based systems in Europe (TRUE), Grant Agreement 727973. The authors thank two anonymous reviewers for helpful suggestions.
None declared.
© 2019. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Quesada, Nora 1 ; Begg, Graham S 1 ; Iannetta, Pietro P M 1 1 James Hutton Institute, Dundee, UK