Broadbalk Crop Nutrient Content

The Broadbalk experiment was started in 1843 to investigate the relative importance of the plant nutrients nitrogen, phosphorus, potassium, magnesium and sodium (N, P, K, Mg and Na) on grain yield of winter wheat. From the very start of the experiment, samples of wheat grain and straw were kept from each plot, for chemical analysis, though some early samples were lost because the containers were damaged.

Early analysis

Lawes and Gilbert carried out analyses of wheat grain and straw for N, P, K, Ca, Mg and Na content on proportionally bulked samples from 10 plots, representing four ten-year periods (1852-1861, 1862-1871, 1872-1881 and 1882-1891). The results for crops grown in 1852-1871 were published in detail (Lawes & Gilbert, 1884). However, only the results for P, K, and Na content for four of the 10 plots for the second period (1872-1891) were published (Gilbert, 1895). Johnston (1969) summarises these early analyses.

Johnston (1969) also measured N, P, K, Ca, Mg and Na in grain and straw of the last two crops of Squarehead's Master grown in 1966 and 1967, on samples taken from all plots in the continuous wheat section, last fallowed in 1951, and from all sections carrying the first wheat after fallow. These were compared with samples from 1852-91 (mean of 1966-67 only, not individual values).

1968 to present

In 1968 Broadbalk was divided into 10 sections. It was decided that it was no longer practical to keep samples for chemical analysis from all plots (10 sections x 19 plots). Each year only wheat grain and straw were kept from each continuous wheat plot on Section 1 (19 plots) and from plots where wheat followed potatoes and beans (Sections 2, 4 and 7, 1968-1978). Potato tubers and bean grain and straw samples were also kept. Dyke et al (1983) summarises plant nutrients in crops grown between 1968-1978. All samples from 1970-75 were analysed for %N, P, K, Ca, Mg and Na. Samples from 1976-78 were analysed for %N only.

Thorne et al (1988) measured crop dry matter yield and %N, %P and %K from anthesis to maturity, from 1969-1984 on selected plots on Broadbalk: plots 07, 08, 15, 16, 17, 18, 21 and 22. Data was from first and second wheats in rotation with potatoes and beans. They also measured many other plant properties, including LAI, grain dry weight, number of ears, number of grains per ear, and total above-ground dry weight. Harvest index, date of anthesis and date of senescence were calculated.

From 1979-1985 %N was measured in grain and straw from the continuous wheat (Section 1) and the first wheat crop of the rotation, and in the potato tubers. From 1986-1995 %P, K, Ca, Mg and Na were also measured.

From 1996 %S was also measured (except in 2002). Nutrients were also measured in more continuous wheat sections (Sections 0 and 9 every year, and Section 8 in some years), and more of the rotational wheat sections.

Data is presented in e-RA as the following datasets:

BKWHNUTRI: Broadbalk wheat grain and straw nutrient data, 1968-2017. See grain nutrient data and straw nutrient data for details of what data is available.

BKBEANNUTRI: Broadbalk bean nutrient data, 1968-1978. See bean nutrient data for details of what data is available

BKPOTSNUTRI: Broadbalk potato tuber nutrient data, 1968-1996. See potato nutrient data for details of what data is available

BKOATNUTRI: Broadbalk oats nutrient data, 1996-2017. See oats nutrient data for details of what data is available

BKMAIZENUTRI: Broadbalk forage maize nutrient data, 1997-2017. See maize nutrient data for details of what data is available

Methodology

Nitrogen (N)

1968-1995: nitrogen content was determined by Kjeldahl digest, the digest was then analysed colorimetrically using a Technicon segmented flow analyser. If nitrate-N was expected to be high in the sample, the salicylic acid modification was used (Bremner, 1965).

1996 onwards: nitrogen content is determined by combustion analyser, based on the Dumas method. Originally a Heraeus Combustion analyser, currently a LECO combustion system is used.

Phosphorus (P), Potassium (K), Calcium (Ca), Magnesium (Mg), Sodium (Na) and Sulphur (S)

1968-late 1980s: dry ashing techniques, as described by Piper (1942) were used to produce an acid extract. The extracts were analysed for P colorimetrically on a Technicon Auto Analyser, or later equivalents (Alpkem continuous flow system / Skalar SanPlus segmented flow system), using the modified Murphy & Riley (1962) molybdenum blue method developed from Fogg & Wilkinson (1958). K, Ca, Mg and Na were determined using automated atomic absorption methods. The SP90 atomic absorption flame photometer was used to analyse Ca, Mg and Na; K was analysed by the EEL flame photometer until 1973, K was then analysed with Ca, Mg and Na by the SP90. S was not routinely measured until 1996.

1980s onwards: since the late 1980s the open tube nitric-perchloric digestion (Zarcinas, et al, 1987) has been used to produce an acid extract. P, K, Ca, Mg and Na have been determined by ICP-OES (Inductively coupled plasma-optical emission Spectrometer) since 1982, and S since 1996.

Blank (control) values: Data for sodium is presented after the analysis blanks (controls) have been subtracted from the raw data. Sodium blank values tend to be relatively large, and can be greater than the raw data. If subtracting the blank resulted in a negative value, this is shown as zero. Thus the sodium data should be treated with some caution. Blank (control) readings for the other nutrients are very small, relative to the treatment values, and have not been subtracted.

• Bremner, J. M. (1965). Total nitrogen. In Methods of Soil Analysis. Part 2 (ed. C. A. Black), pp. 1149-1178. Madison: American Society of Agronomy.
• Fogg, D.N. and Wilkinson, N.T. (1958). The colorimetric determination of phosphorus. Analyst, London, 83:406-414.
• Piper, C.S. (1942) Soil and plant analysis. The University of Adelaide, Adelaide, Australia.
• Zarcinas, B.A., Cartwright, B., Spouncer, L.R., 1987. Nitric acid digestion and multi-element analysis of plant material by inductively coupled plasma spectrometry. Communications in Soil Science and Plant Analysis 18, 131-146.

ICP-OES

Nitric/perchloric acid digestion

LECO combustion analyser

Ripe wheat

Broadbalk grain 1945

Rothamsted sample archive

Micro-nutrients

Selenium (Se)

See Fan et al (2008) in Key References below, for details of changes in Se concentration in wheat grain and soil between 1843 and 2000, in plots 3, 7, 9, 10, 14, 15 and 22.

Copper (Cu), Zinc (Zn) and Iron (Fe)

See Fan et al (2008) in Key References below, for details of changes in concentration of Cu, Zn and Fe between 1843 and 2000 in plots 3, 7, 9, 10, 14, 15 and 22.

Further information and acknowledgements

For analytical techniques used prior to 1968 see Johnston (1969) in Key References below.

For information on current analytical methods used for the Rothamsted Long-term experiments, please contact the Rothamsted Research Analytical Chemistry Unit - Harpenden laboratory,

With thanks to Andy Macdonald, Paul Poulton, Steve Freeman, Ruth Skilton and Wendy Gregory for help with compiling the information, images and text.

Wheat straw

Wheat grain

Rothamsted sample archive

Key References

Broadbalk Diseases

The following wheat root and stem diseases have been assessed on selected plots regularly since the introduction of rotations in 1968:

• Take-all Gaeumannomyces graminis (Sacc) Arx & Oliver var.tritici
• Eyespot Oculimacula acuformis and Oculimacula yallundae
• Sharp eyespot Rhizoctonia cerealis
• Brown foot rot Fusarium spp.

The take-all fungus infects winter wheat roots in the autumn causing black necrotic lesions on the roots. Early infections can lead to uneven growth in the spring and occasionally plant death. If severe infection occurs in June/early July above ground symptons often show as patches of premature ripening plants with reduced yields and grain quality. Comparisons of yields and of differences in amounts of take-all between continuous wheat on Broadbalk and wheat in other fields growing shorter sequences of cereals culminated in the development of the hypothesis of 'take-all decline'. This phenomenon is now widely recognised and has been shown to occur in all the susceptible cereals.

Large take-all patches on wheat

Severe take-all on wheat

Eyespot was first identified in the UK in 1935 on Broadbalk. It is common on intensively cultivated cereals in heavy soils and is favoured by long wet and cold periods in winter and spring. The fungus can survive on infected stubble and in the soil for 2-4 years. Early symptoms are difficult to assess and can range from a brown smudge to a typical eye shaped lesion. Careful removal of the outer leaf sheath can reveal a small black dot, the penetrating stroma. It penetrates the leaf sheath eventually infecting the stem. Severe infection softens the stem, often showing as white heads and can lead to lodging of the crop as it matures.

Eyespot lesions on wheat

Lodging of wheat caused by eyespot

Severe eyespot on straw

Sharp eyespot causes numerous stem and stem base lesions. The fungus overwinters primarily as mycellium on infected stubble but has a large host range which can also act as a source of inoculum. As a result, the fungus is not easily controlled by rotation. Early symptoms show as well defined lesions on the outer leaf sheaths and frequently have interveinal tissue shredding within the lesion. Infection can occur at any time during the growing season. Late infections often remain on the leaf sheaths but early infections can penetrate the straw often causing multiple lesions that can be observed as far up the stem as the 4th node. Severe infection can cause white heads and make the straw brittle. The disease is favoured by cold and dry conditions and is less prevalent in intensively cultivated cereals.

Sharp eyespot lesions with interveinal tissue shredding

(Galium tricornutum) in Broadbalk

Sharp eyespot on straw extending ouver several nodes

Brown foot rot is caused by Fusarium spp. and is both soil and seed-borne. Infection often begins at the base of the leaf sheath and spreads up the leaf. The fungus continues to spread eventually reaching the nodes and straw causing two distinct symptoms. An overall general browning, mainly attributed to Fusarium culmorum, and streaky brown lesions caused by Microdochium nivale. Severe Fusarium infection can result in whiteheads and occasionally lodging of the crop.

Brown foot rot lesions caused by Microdochium nivale

Severe brown foot rot caused by Fusarium culmorun

Measurements

Take-all, eyespot, sharp eyespot and brown foot rot have been routinely assessed on Broadbalk on Sections 9 and the 1st, 2nd and 3rd wheats of the rotational sections since 1968. Several plots were assessed each year and from 1985 mainly plots 7 (N2PKMg), 10 (N2), 11 (N2P), 13 (N2PK), 15 (N5PKMg) and 21 (FYMN2). Disease assessments were generally carried out in late June/early July, at growth stage 69-77. Data in e-RA is currently available from 1968 - 2009, and 2016.

No disease assessments were done from 1982-1984. This was a transition period during which the rotation period changed from 3 to 5 years.

The percentage of plants with slight, moderate and severe take-all infection is assessed as follows: 0 = no infection (healthy); 1 = < 25% of the root system infected (slight take-all); 2 = 25-75% infected (moderate take-all); 3 = > 75% roots infected (severe take-all). A take-all rating (TAR) is determined (Gutteridge et al., 2003). It is calculated as follows: (1x % plants with slight infection) + (2 x % plants with moderate infection) + (3 x % plants with severe infection). The TAR is a measure of take-all intensity with a range from 0 (no take-all) to 300 (severe infection on all plants).

The severity of infection by eyespot is assessed as described by Scott & Hollins (1974). Plants with slight eyespot have small lesions occupying less than half of the circumference of the straw. Plants with moderate eyespot have lesions occupying more than half of the circumference of the straw. In plants with severe eyespot the straw is completely girdled by lesions and/or tissue softened.

Symptoms of sharp eyespot are often very superficial and a distinction is made only between symptoms that are slight, small lesions occupying less than half of the circumference of the straw or severe where lesions occupy more than half of the circumference of the straw and usually extend over more than one internode (Goulds & Polley, 1990).

Brown foot rot is classified as slight or severe (Goulds & Polley, 1990). Plants with slight brown foot rot have either a general light browning occupying most of the circumference of the straw (typically caused by Fusarium culmorum) or fewer than 5 dark brown-black narrow streaks c.1mm wide (typically caused by F. nivale). Severe symptoms are defined as a general dark brown discoloration of the straw usually extending over more than one internode (F. culmorum) or more than 5 brown-black streaks greater than 1mm wide (F. nivale).

With thanks to Richard Gutteridge for providing the photographs and helping to compile the text.

For more information, refer to the Rothamsted Guide to the Classical Experiments 2018 page 17

Key References

Broadbalk Earthworm measurements

Earthworm populations were assessed on Broadbalk in 1920-21, 1979, 2014 and 2015. Details are given below. The results are not directly comparable, as different methods were used for each assessment.

Feb 1920-Jan 1921 (Morris, 1922)

Morris recorded insect and other invertebrate fauna, including earthworms, on Broadbalk between February 1920 and January 1921. Samples were taken from two plots: FYM (Plot 2), receiving 35t/ha FYM annually since 1843 and the unfertilized plot (Plot 3) - no fertiliser or manure since 1843.

Samples were taken from the Western end of the plots, using a metal box 23x23cm with a total of five depths: 0-2.5cm, 2.5-7.6cm, 7.6-12.7cm, 12.7-17.8cm and 17.8-23cm. 23 samples were taken from each plot, approximately every 12 days, between Februrary 1920 and January 1921. Soil was not sampled on rainy days, due to the difficulty of examining wet soil. FYM was applied to Plot 2 and the Plots were ploughed on October 13th 1920. The plough depth was around 14-16cm, i.e. the fourth layer measured.

Earthworms were defined as belonging to the sub-order Terricolae in the order Oligochaeta, including Lumbricus sp.

Results:

Earthworms were recorded in all 12 months, and in all five depths (0-23cm)
Earthworms were most common in the second soil layer sampled (2.5-7.6cm deep)
Over twice as many earthworms were present in the plot given FYM since 1843 as in the plot never given fertiliser or manure
A total of 2.50 million earthworms/hectare (250/m2) were recorded in the FYM plot from 23 samples
A total of 1.13 million earthworms/hectare (113/m2) were recorded in the unfertilized plot from 23 samples
Earthworms were thought to occur at deeper depths than 23cm in both plots

September 1979 (Edwards & Lofty, 1982)

Edwards & Lofty measured earthworm numbers and total biomass in various Broadbalk plots growing continuous winter wheat, to investigate the effects of nitrogen (N) fertilizers. Earthworm populations were sampled in September 1979 by pouring dilute formalin on to 16 0.25 m ² quadrats and collecting the worms that were brought to the surface.

Results:

All species of earthworm were more numerous in plots treated with organic fertilizers
There was a strong positive correlation between the amounts of inorganic N fertilizer applied and populations of earthworms
Plots receiving both inorganic and organic N had the largest populations of earthworms

See September 1979 earthworm data (pdf) for summary of the Edwards & Lofty (1982) data from Broadbalk.

Spring 2014 (Sizmur et al, 2017)

Earthworm surveys were carried out on four parts of Broadbalk in spring 2014, receiving the following treatments each year:

A 1m x 14m area on the northern edge of each plot was used, divided into four equal sub-plots. Two earthworm surveys were carried out in each sub-plot, a total of 8 surveys per treatment.

Earthworm surveys were conducted by excavating a 20 x 20 x 20cm cube of soil, which was brought back to the laboratory and sorted to find and identify all the earthworms. Deep burrowing (anecic) earthworms were extracted by pouring a 5 L aqueous solution containing 6g/l of mustard flour (see Sizmur et al 2017 for full details).

Results:

Plots given FYM contained a significantly greater biomass and number of earthworms than the plots never given FYM
Total earthworm biomass was 109 g/m2 in plots given FYM, compared to 6 g/m2 in plots never given FYM
Total earthworm numbers were 400/m2 in plots given FYM, compared to 70/m2 in plots never given FYM
28 years of straw incorporation had no significant effect on the earthworm populations
There was no significant interaction between straw and FYM on earthworm abundance or biomass

Broadbalk total earthworm biomass and abundance, spring 2014 (Sizmur et al, 2017)

+/- standard errors of the mean, based on four replicate sub-plots and two surveys per sub-plot.

For further details of the different earthworm species recorded in the survey, see Spring 2014 biomass (pdf) and Spring 2014 abundance (pdf).

Earthworm biomass and abundance were also measured in spring 2014 on the long-term straw incorporation experiment at Rothamsted, which had four rates of cereal straw incorporated annually for 28 years. The highest rate of straw (20t/ha) resulted in significantly greater earthworm abundance and biomass. See Sizmur et al (2017) for more details.

Autumn 2015 (Stroud et al, 2016)

Stroud et al carried out midden counts on two Broadbalk plots in September 2015, to assess populations of anecic, deep burrowing Lumbricus terrestris earthworms. The plots received the following treatments each year:

Midden assays (5 m2 per plot) were carried out with 1 m2 mustard validation assays. 1.5l of mustard solution containing 20g mustard powder in water were poured within a 0.25 m2 square quadrat in a random location within the plot. Earthworms (L. terrestris) were collected for species analysis and then released. The results are not directly comparable with earlier assessments, as different methods were used.

Results:

The plot ammended with FYM + N contained 0.3 middens per m2
The plot ammended with NPK contained 0.13 middens per m2

Middens counts were also carried out on three other arable sites at Rothamsted and Woburn, with and without organic ammendments (straw, biosolids, and organic wastes). Organic matter applications enhanced L. terrestris populations, however these populations were very low, never exceeding 4.6 per m2. See Stroud et al (2016) for more details.

Key References

Broadbalk grain quality data

The Broadbalk experiment was started in 1843 to investigate the relative importance of different fertilizers and manures on the grain yield of winter wheat. Since 1974, the wheat grain has been regularly analysed for standard grain quality characteristics.

Data available

BKGR_QUALITY: Broadbalk wheat grain quality, 1974-2018, on selected plots, containing:

• Thousand grain weight (TGW) since 1974
• Hagberg Falling Number (HFN) since 1999
• Hectolitre weight (HLWT) since 1999
• Grain size classes (% grain between 1-3.5mm) since 1999
• Wheat variety
• Previous crop (for the Sections of Broadbalk in rotation)
• Number of years since last non-wheat crop, which indicates when the section was last fallowed, for the Sections in continuous wheat

See grain quality data available for details of what measurements are available, and from which plots and sections of Broadbalk. Data is available for Section 1 (continuous wheat) and the Section in the 1st wheat of the rotation every year, with other sections in other years.

Other data:

• Wheat Hectolitre weight (HLWT) derived from Bushel weights, 1844 - 1955.

BKOATS : Broadbalk oats grain quality, 1996-2018, for all plots each year, containing:

• Thousand grain weight (TGW) since 1996
• Hectolitre weight (HLWT) since 1999
• Grain size classes (% grain between 1-3.5mm and % 1-2.2mm) 1999-2004

Methodology and background information

TGWs: The thousand grain weight is the weight in grams of 1000 cereal kernels. It is determined using an automatic grain counter. After counting, the grain is dried overnight at 105 degrees C. TWGs are useful for determining the evenness of grain size (by comparing several TGWs). The TGW is also important when calculating the drilling rate for sowing cereals.

HFN: The Hagberg Falling Number is a recognised international test used by millers to assess the level of germination/sprouting within a batch of wheat intended for bread making. Developed in the 1950s, the Hagberg apparatus measures the time taken for a plunger to fall through a standardised hot slurry of milled wheat flour. As wheat begins to germinate it produces the enzyme alpha-amylase, which reduces the bread making potential of the wheat, and reduces the HFN. A HFN of 300 seconds or more is generally accepted as the cut-off point for bread making. A low HFN would reduce the value of the crop and could mean that the grain could only be sold for animal feed. The process includes 60 seconds of mixing, when the plunger is automatically rapidly lifted up and down by the mechanism. At 60 seconds, the plunger is released at its highest point so it can fall through the mixture. Thus the minimum possible reading is 60. THe HFN is determined on fresh grain. From 2016 the HFN is measured with a Perten Instruments FN1000, and is the mean of two readings.

HFNs in 2007 and 2010 were very low. This is probably because it was very wet around harvest, and the grain had started to sprout before it could be harvested.

HLWT: The hectolitre weight is the weight in kilograms of 100 litres of grain. It is calculated from specific weight of grain. Specific weight of grain is measured using a chondrometer and is equal to the weight in grams of a one litre volume of grain. HLWT is calculated from specific weight (grams per litre) as g/l x 0.10033 + 0.42119 = kg/hl. HLWT is determined using fresh grain, and the sample is weighed fresh.

HLWT is used to quantify the size of grains and the proportion of broken or thin, shrivelled grain. Millers use HLW to help determine the quality of grain and its potential end use as bread, biscuit or pasta flours.

Size classes: Size classes of grain are determined using a Winnower sieve, using fresh grain. The % of grain between 1-3.5mm is recorded. This is divided into the % between 1-2.25mm and the % between 2.25 and 3.5mm. The proportion of grain between 2.25 and 3.5mm is what would normally be used by millers for flour. This is calculated by subtracting the proportion of grain between 1-2.25mm from the proportion of grain between 1-3.5mm. A weighed sample is passed through slotted sieves. The same sample is sieved twice, using different sized sieves in turn. Screenings are undersized, broken or shrivelled grain between 1 and 2.25mm. Admixture comprises impurities, such as straw, chaff, weed seeds and earth, which must be removed before milling marketable flour. Screenings and admixture represent a loss to the miller, so a maximum of 2% is normally allowed.

In 2014, a few of the plots had a measurable amount of grain greater than 3.5mm. If you would like this data, please contact the e-RA Curators.

wheat grain

Grain counter

Hagberg falling number apparatus

Winnower sieve

Half-litre chondrometer

Litre chondrometer

Early studies

Hectolitre weights (HLWT), derived from bushel weights, are available for all plots and sections on Broadbalk from 1844 - 1955, for dressed grain measured at threshing. Bushel weight is the weight of a bushel of grain (one bushel = 36.369 litres). Bushel weights (measured in lbs/bushel) have been converted to HLWT (kg/hl), (one lb = 0.4536 kg). From 1844 to 1901 Broadbalk was harvested by hand, and then by binder until 1956. The sheaves were 'stooked' on the plot, then carted to a barn and stored until threshing in winter, when bushel weight was determined by weighing a bushel measure filled with grain (see Atkinson et al, 2008 in Key References below). The measurement of bushel weights was discontinued in 1955 (Johnston and Garner, 1969). If you would like access to this early data, please contact the e-RA Curators.

Related studies

Godfrey et al (2010) assessed wheat grain protein compostion and dough properties on selected Broadbalk plots, 2005-2007. See Key References below.

Atkinson et al (2008) looked at the relationship between grain specific weight/hectolitre weight and the winter North Atlantic Oscillation for plot 22 (FYM) Section 1 of Broadbalk, from 1844-2001. Grain specific weight was derived from the bushel weight of dressed grain, 1844-1955 (see Early studies, above). Specific weight was measured in archived samples of clean grain from 1956-2001. For further details, see Key References below.

Gutteridge et al (2003) investigated the relationship between take-all disease and wheat grain quality (TGW, HLWT, HFN and grain size) at other field sites at Rothamsted. For further details, see Key References below.

Grain protein content is an important property, determining suitability for breadmaking. A minimum protein content for breadmaking wheat in the UK is typically 13% dry basis. Protein % can be calculated as %N x 5.7. See Broadbalk Crop Nutrient Content for details of Broadbalk grain %N.

For more details, contact the e-RA Curators.

Acknowledgements

With thanks to Steve Freeman, Chris Hall, Andy Macdonald and Paul Poulton for help with compiling the information, images and text.

Key References

Broadbalk Yields - Other Crops and Fallow

Broadbalk Description of other crop yield data
Other arable crops and fallow have been included in the rotational sections of Broadbalk since 1968 (see Cropping plan for details). See below for further details of fertilizer applications and other treatments, including to the fallow sections . Yield data is available from the following datasets:

BKBEANS Spring bean grain and straw yields 1968 -1978; Winter bean grain and straw yields 2018 -
BKPOTATO Potato tuber yields 1968 -1996
BKMAIZE Forage maize whole crop yields 1997 - 2017
BKOATS Winter oat grain and straw yields 1996 - current year

Spring field beans (Vicia faba) were grown on the rotational sections of Broadbalk between 1968 and 1978, as part of the potato > beans > wheat rotation on sections 2, 4 and 7. For full details of the cropping sequence see Cropping on Broadbalk. The varieties were Maris Bead (1968-74), Minor (1975) and Minden (1976-78). The beans were sown in March and generally harvested in September. The crops received the same inorganic fertilizer (including N) and FYM applications as the wheat crops. The N is applied in the seedbed. Grain and straw yields at 85% DM, thousand grain weights, variety, sowing and harvest dates are available from dataset BKBEANS.

Winter field beans (Vicia faba) replaced forage maize on the rotational sections of Broadbalk from 2018 onwards, and the rotation was changed to wheat/wheat/oats/wheat/beans. For full details of the cropping sequence see Cropping on Broadbalk. The beans are sown in the autumn and harvested in August/September. No fertilizer N or FYM is applied; P, K and Mg are as for the wheat crops. Grain and straw yields at 85% DM and thousand grain weights are available, along with variety, sowing and harvest dates from dataset BKBEANS.

Potato (Solanum tuberosum) was grown on the rotational sections of Broadbalk between 1968 and 1996, as part of the potato > beans > wheat rotation of Sections 2, 4 and 7 (1968-78) and then as part of the potato > wheat > wheat > wheat > fallow rotation of Sections 2-5 and 7 (1979-96). For full details of the cropping sequence see Cropping on Broadbalk. The varieties grown were Majestic (1968-69), King Edward (1970-75), Pentland Crown (1976-93), Estima (1994-96). The crops received the same inorganic fertilizer and FYM applications as the wheat crops. Fertilizer N was applied in the spring, before rotary cultivation to prepare the seedbed. Chitted potato seed was generally planted in April and harvested in September; please apply to the e-RA Curators for a full list of dates since 1968. The potatoes were usually harvested earlier than normal to allow timely drilling of winter wheat, hence the yields are lower than might be expected.

Data available includes total tuber yield, tuber DM content and % ware (tubers that do not pass through a 3.81cm ‘riddle’ or grader). ‘Ware’ potatoes are those grown for human consumption, which need to be a minimum size, generally over 40mm. Potatoes are graded according to size, using a potato riddler.

Forage maize (Zea mays) was grown as a whole crop, for silage. It has been grown in the rotational sections 2-5 and 7, 1997-2017, as part of the wheat > wheat > wheat > oats > maize rotation. Maize is a C4 plant. As such, the carbon it contains has a different ¹³C "signature" than that in the C3 plants that have been grown previously on Broadbalk. Thus, we can distinguish maize-derived organic matter from that of organic matter already in the soil. For full details of the cropping sequence see Cropping on Broadbalk. The variety Hudson was grown 1997-2014, and the variety Severus 2015-2017. Forage maize is generally sown in May and harvested in September; please apply to the e-RA Curators for a full list of dates since 1997. Maize receives the usual inorganic fertilizer and FYM applications, the same as wheat. In 2013, forage maize yields may have been reduced, due to the accidental application of herbicide to the crop in June/July 2013. The last crop of forage maize was grown in 2017. A new rotation including winter beans instead of maize was started in 2018.

Winter oats (Avena sativa) has been grown on the rotational sections 2-5 and 7 since 1996, as part of the wheat/wheat/wheat/oats/maize rotation. For full details of the cropping sequence see Cropping on Broadbalk. The varieties grown were Image (1996-2000), Revisor (2001) and Gerald (2002-onwards). Revisor, a spring variety, was sown in 2001 due to poor autumn weather preventing a winter variety from being sown. Oats are generally sown in October and harvested in August; please apply to the e-RA Curators for a full list of dates since 1996. No fertilizer N or FYM was applied to the oats crops harvested 1996-2017; K and Mg were applied as usual. Thus, on plots were P and K is not limiting, any differences in yield between treatments were due to residues of inorganic N from previous applications or from differing amounts of N being mineralised from the soil organic matter.

Since 2018 the rotation was changed to wheat/wheat/oats/wheat/beans, and N is now applied to the oats at 1/2 the usual rate as a single application in mid-April (i.e. 24, 48, 72, 96, 120 and 144 kgN/ha). FYM is applied at the usual rate to the sections growing oats.

Bare fallow, 1968-1995 was included as part of the three year rotation (wheat > wheat > fallow) on sections 3, 5 and 6 from 1968-1979, and as part of the three or five year rotation on sections 2-5 and 7 from 1979-1995. The fallow sections received FYM and castor meal, and the autumn inorganic fertilizers (P, K, Na and Mg) at the same time as the cropped sections (in the autumn, before ploughing). However, no inorganic N was applied to the fallow sections. The fallow sections were cultivated several times, to remove weeds, and herbicides were also applied. The continous wheat sections (0, 1, 6, 8 and 9) are occasionally bare fallowed to control perennial weeds, most commonly Section 8 (which does not receive herbicides).

Bare fallow, 1926-1967. Regular bare fallowing was introduced in 1926, with cultivations to kill the weeds. No FYM, castor bean meal or inorganic fertilizer was applied to the fallow sections. This decreased the soil organic matter content, especially on the plots usually given FYM. No herbicides were applied to the fallow sections.

For more information, refer to the Rothamsted Guide to the Classical Experiments 2018 pages 7-12.

Broadbalk soil chemical properties background information

The Broadbalk wheat experiment, established in the autumn of 1843, has been cultivated since at least 1623, and probably much earlier (Avery & Bullock, 1969). Soil chemical properties have been measured at regular intervals since 1865.

• Soil Organic Carbon (%SOC) Open Access long-term changes in SOC from selected treatments, 1843-2015 (doi: 10.23637/KeyRefOABKsoc-02)
• Soil Total nitrogen (%N) Open Access long-term changes in soil N concentration from selected treatments, 1843-2010 (doi:10.23637/BK-oadata-soilN-01)
• Plant-available P (Olsen P) Open Access long-term changes in Olsen P from selected treatments, 1843-2010 (doi:10.23637/keyrefoabkolsenp)

Soil pH: The plough layer (0-23 m) is limed when necessary to maintain a minimum soil pH of 7.0 - 7.5. Broadbalk was first limed regularly from 1955-1967, with plots given the larger inputs of ammonium fertilizers receiving more lime than the controls. Liming stopped from 1968-1975, but began again in 1976-1992, with 3 or 4 sections being limed each year. From 2007 onward selected plots have been limed every 5-6 years based on soil pH measurements to maintain top-soil pH around 7.0-7.5. Selected plots were limed in autumn 2018.

Data available

Soil chemical properties have been measured at regular intervals on Broadbalk since 1865, in topsoils (0-23cm) and subsoils. See soil measurements 1843-1944 and soil measurements 1966-2017 for details of what data is available. Not all plots, sections or soil depths have been measured every year. The following soil chemical properties have been measured:

• %N : total soil Nitrogen %
• %SOC : soil organic Carbon %
• Olsen P : plant-available phosphorus, by extraction with a solution of 0.5M NaHCO₃, buffered at pH 8.5
• pH : soil pH in water (1:2.5 soil : solution)
• Exchangeable cations : exchangeable Ca, K, Mg and Na by extraction with 1M ammonium acetate solution
• % inorganic carbon (CaCO₃-C) : by calcimetry
• soil weights : see soil physical properties for details

Soil Sampling

Broadbalk soil has been sampled on many occasions over the years. However, because the method of sampling has changed and the experiment has been divided, first into two halves, then into five sections and finally into 10 sections, it is not always advisable to directly compare one sampling with another without careful thought. This table shows, where possible, how samples taken over time relate to each other. Selected plots and depths have been sampled on other occasions; most samples still exist in the Sample Archive.

Since 1992 a systemic sampling plan has been adopted. In 1992, 1997, 2005, 2010 and 2015 all five continuous wheat sections were sampled (0, 1, 6, 8 and 9). In the intervening years the remaining sections in rotation were sampled, one per year, so that all sections were sampled every five years. All sections were sampled in 2000 prior to treatment changes being introduced. See table for full details.

In autumn 2004 it was apparent that parts of the field had been ploughed slightly deeper than 23cm as sub-soil clay was visible in random patches across the field. Thus in 2005 and 2006 all plots were resampled to create a new baseline, if necessary, for soil chemical properties. The sections were then sampled systematically from 2008 onwards.

In 2000-2004 archived soil samples from selected plots sampled in 1865, 1881, 1893, 1914, 1936 and 1944 were re-analysed for soil pH, Olsen-P, exchangeable cations, Total %N, %SOC and CaCO₃-C. Values for %N and %SOC from the re-analysis of the 1865 samples were very different to the original data and the 1881 and 1893 re-analysed soils, so the original data was used.

Data before 1926 (when the experiment was divided into sections) are given for whole plots only, from the re-analysis in 2000/2004. More recent data are available for individual plots within each section.

1936, 1944 and 1966 data is included with later data, although the experiment was not divided into 10 sections until 1968.

1936 data: 10-20 cores taken with a narrow auger from each of the five Old Sections were bulked within each plot. Data for Old Section I is used for current sections 0 and 1, Old Section II for current sections 2 and 3, etc. (see soil sampling plan for more details). All the data is from the re-analysis of old samples in 2001-2002.

1944 data: four holes were taken from each of the (then) five Old Sections; holes 1 & 2 were on current Section 0, holes 3 & 4 on current Section 1 etc, so the data can confidently be allocated to the modern sections. Data is presented as the mean of the two holes for the topsoil. Subsoil data is not available for each modern section, as the samples were bulked from four-sub-samples for each of the five old sections. Data for Old Section I is used for current sections 0 and 1, Old Section II for Sections 2 and 3, etc. (see soil sampling plan for more details). The soil was re-analysed in 2003-5 from selected plots from all Sections and depths. All data is from the re-analysis except %CaCO₃ which was analysed in 1944. Soil pH and Exchangeable Na were not measured in the 1944 subsoil samples.

1966 data: Section 0 and 1 use data from Old Sections Ia and Ib respectively; Sections 2 and 3 use data from Old Section II; Sections 4 and 5 use data from Old Section III; Sections 6 & 7 use data from Old Section IV; Sections 8 and 9 use data from Old Sections Va and Vb respectively. Bulked Va and Vb samples (ie Sections 8 and 9) were used to determine %SOC (plots 2.1 and 2.2 only) and for all plots for Olsen P, %CaCO3 and Exchangeable K. There was no plot 1 in 1966, this was created in 1968. The soil was sampled in September 1966 but no day is given. It is shown as 15/09/1966 in the database.

Data from the samplings in 2001-2004 is not included as this only covered sections 2, 4, 5 and 7 and there was a comprehensive sampling of all sections in 2000 and then in 2005/6.

Soil Sampling Methods

Samples between 1865 and 1914 were taken with an open-ended metal box, 9 inches (23cm) deep and usually 6 x 6 inches (15 x 15cm) across. There were between three and eight sample positions on each plot which were bulked together for each depth on each plot. In 1944 a spade was used to sample the 0-23cm layer, and the subsoil (23-46cm) was sampled with a semi-cylindrical auger.

In 1936 and from 1966 onwards samples were taken with a semi-cylindrical auger. 10-20 cores were taken from the different soil layers for each individual plot within each section and bulked together for each plot. Small diameter cores taken by semi-cylindrical augers cannot be used to determine soil weights, but provided enough are taken the sample better represents the proportions of SOC, N, P, K etc in the soil than a few large box samples. See soil physical properties for details of soil weights.

Samples were taken in the autumn, after the crop had been removed, but before ploughing, except for Section 3, 1996 which was sampled in March.

Soil sampling Broadbalk 1943

Soil sampling Broadbalk 1943

Broadbalk ploughing in 2013

Analysis methods for soil chemical properties

All soil samples are air-dried and sieved <2mm. Data are given for air-dried soil (approximately 98% dry matter). When calculating total amounts in the soil (e.g. kg N ha^-1 ) you may wish to convert to oven-dry soil (i.e. 100% dry matter). Broadbalk standard soil weights are given for oven-dry soil.

For information on current analytical methods used for the Rothamsted Long-term experiments, please contact the Rothamsted Research Analytical Chemistry Unit - Harpenden laboratory,

Total soil % nitrogen (%N)

1865: Original soda lime analysis for total N (Johnston, 1969b, table 5.10) multiplied by a factor derived from the comparison of soda lime and LECO analysis values for 1881 and 1893 samples. Soda lime analysis by the method of Will and Varrentrapp (Watt, 1863). See Johnston (1969a, p 50) for more details.

1881-1944: Selected samples re-analysed in 2001-4 by combustion analysis, based on the Dumas method, using a LECO combustion system. Measured on air-dried, finely ground soil (to pass a 355 micron or 44 mesh sieve).

1966, 1987-8: Kjeldahl digest method for total N (Bremner, 1965). The digest was then analysed colorimetrically using a Technicon continuous flow analyser. Measured on air-dried soil, finely ground to pass a 0.5mm sieve.

1992 onwards: Combustion analysis, based on the Dumas method, using a LECO combustion system. Measured on air-dried, finely ground soil (to pass a 355 micron or 44 mesh sieve).

Soil % organic carbon (%SOC)

1865: Derived from original soda lime analysis for total N and C:N ratios for 1893 for organic carbon (Dyer, 1902).

1881-1944: Selected samples re-analysed in 2001-4 by combustion analysis, based on the Dumas method, using a LECO combustion system to measure total carbon. SOC determined as total C minus CaCO₃-C, measured by a calcimeter (see below). Measured on air-dried, finely ground soil (to pass a 355 micron or 44 mesh sieve).

1966: Chromic acid titration method (Walkley and Black, 1934). Correction factor of W-B x 1.3 used, which is equivalent to organic C by Tinsley or total C by combustion minus CaCO₃-C. But see also Johnston (1969b, p 97). Measured on air-dried soil, ground to pass a 0.5mm sieve.

1987-8: Dichromate digestion, modified Tinsley (Kalembasa and Jenkinson, 1973) to measure organic C. Measured on air-dried soil, finely ground to pass a 0.5mm sieve.

1992 onwards: Combustion analysis, based on the Dumas method, using a LECO combustion system to measure total C. Measured on air-dried, finely ground soil (to pass a 355 micron or 44 mesh sieve). SOC determined by subtraction of CaCO₃-C, measured by a calcimeter (see below).

Inorganic carbon (IC) also known as calcium carbonate-C or CaCO₃-C

Soils sampled in 1865-1936 were re-analysed in 2000-4. Most of the 1944 data was from the 1944 analysis, except for a few plots (9, 18 and 19) that were analysed in 2001. IC was not measured in 1987-88, as %SOC was determined directly by Tinsley analysis. In other years IC is subtracted from total carbon to give %SOC.

All samples up to 2012 were analysed by a calcimeter. CO₂ is liberated from CaCO₃ in the soil sample by treating with hydrochloric acid (HCl) in a closed system. The amount of CaCO₃ is calculated by comparing the pressure produced by the sample against the pressure produced by known weights of CaCO₃, using a mercury filled manometer. %CaCO₃-C or %IC is derived from %CaCO₃ by dividing by 8.333. Since 2014 inorganic C has been measured by an automated Skalar Primacs inorganic carbon analyser.

Soil pH

Soil pH in water, with a 1:2.5 soil:water suspension, mean of two readings. Soils from selected treatments sampled in 1865-1944 were re-analysed in 2000-4. Measured on air-dried soil, sieved < 2mm.

Olsen-P (plant-available phosphorus; also known as bicarbonate soluble-P or NaHCO₃-soluble P)

Soils from selected treatments sampled in 1865-1944 were re-analysed in 2000-4. All samples were analysed by the Olsen method, in which soil is extracted with a solution of 0.5M NaHCO₃, buffered at pH 8.5 (Olsen et al, 1954). The extract is then analysed by continuous segmented colorimetric flow analysis. Measured on air-dried soil, sieved < 2mm.

See Blake et al (2000, 2003) for a discussion of the P balance on Broadbalk and changes in soil P fractions over time (Key References, below).

Exchangeable cations - Calcium (Ca), magnesium (Mg), potassium (K) and sodium (Na)

Selected soils sampled in 1865-1944 were re-analysed in 2000-6, and those sampled in 1966 were re-analysed in 2019. All soils were extracted with 1M ammonium acetate (NH₄CH₃CO₂) solution, after the method of Metson (1956). The data is expressed as mg kg^-1 in dry matter, with 5mg of soil in 100ml of leachate. Since 1983, the extracts have been analysed by ICP-OES (Inductively Coupled Plasma - Optical Emission Spectrometer). Measured on air-dried soil, sieved < 2mm.

Blank (control) values: Analysis blanks were subtracted from exchangeable Na, and other cations if necessary (e.g. K in 1987).

Samples taken in 1966 were re-analysed in 2019. The exchangeable K values differ from those published by Johnson (1969b), when a slightly different technique was used for measuring exchangeable cations, which involved swirling the samples in successive amounts of ammonium acetate and decanting off the solution once settled. In 1966 the extracts were measured with a flame photometer. Bolton (1972) reports on exchangeable Ca and Mg in Broadbalk soils from 1856-1966 for some sections, from the western (top) end of the field, but the 1966 data has not been found. It is recommended that the re-analysed data from 2019 is used for the 1966 samples.

See Bolton (1972) for changes in exchangeable Ca and Mg, 1856-1966 and Blake et al (1999) for a study of soil K content, crop K uptake and K balance in Broadbalk (Key References, below).

Methods References:

Further information and acknowledgements

See Goulding et al (2000) for measurements of nitrate leaching from the Broadbalk wheat experiment from 1990-1999 (Key References below).

With thanks to Andy Macdonald, Paul Poulton and Steve Freeman for help with compiling the data and text.

Key References

Broadbalk site and soil physical properties

The Broadbalk wheat experiment, established in the autumn of 1843, has been cultivated since at least 1623, and probably much earlier (Avery & Bullock, 1969). In his first Rothamsted paper, published in 1847, Lawes described the soil as "a heavy loam resting upon chalk, capable of producing good wheat when well manured" . Here, details are given of the site, the soil description and standard soil weights.

Site details

• Location: Rothamsted Research, Harpenden, Herts, AL5 2JQ, UK
• Latitude: 51 degrees 48 mins 36 secs North
• Longitude: 0 degrees 22 mins 30 secs West
• GB Grid Reference: TL123 136
• Gradient: A slope of 1 degree, West to East
• Irrigation: There is no irrigation

Plot area

When the experiment was established in 1843 most of the plots were very large. Most comprised an ‘a’ and ‘b’ half (each 3.77m wide) and were 320m long (the length of the field). Plots 21 and 22 are a little narrower. In 1894 the two halves were combined, giving a total plot area of 0.24ha. As the experiment progressed these large plots have been subdivided into different Sections (see Plans and treatments for more details), with corresponding changes in the area harvested. The current plot lengths vary depending on the Section and are between 15.24m (Section 0) and 28.04m (Section 1). The plots are now 6m wide (except plots 21 and 22 which are 4m wide) with 48 rows at 12.5 cm spacing. The harvested area is 2.1m wide. The harvested area is shown in most of the datasets.

Soil moisture and drainage

Soil Moisture Characteristics: With nearly 200 plots in the experiment, there isn't water retention information for all treatments. Water retention characteristics for contrasting plots can be found in Salter & Williams (1969). For more recent information, see Gregory et al, 2010, where retention characteristics have been fitted to the van Genuchten model.

Soil Drainage: Lawes described the soil of Broadbalk as having good natural drainage. However, as it was an experimental field, it was decided to improve the drainage, to allow greater access. Tile drains were installed under each strip in autumn 1849 (except plot 20). They were installed in the centre of the strip, under the furrow separating the 'a' and 'b' halves (see Plan 1852-1925). Tiles drains were installed under strip 2.1 (2A) in autumn 1884 before the first application of FYM. Tile drains were 5cm in diameter and 60-76 cm deep, and they discharged into a 10cm diameter main drain at the east side of the field.

Lawes and Gilbert realised that the drains could be used to measure losses of plant nutrients from the different fertilizer treatments. Small outlet pits were dug at the intersection of each strip drain and the main drain in December 1866 to sample the water draining from strips 2 to 16. This was not ideal, as there was the risk of the samples being contaminated. The drains from strips 17-19 were opened in November 1878. In spring 1879 the collection of drainage water was improved, with the drains from each strip discharging into their own pit, which overflowed into a separate, deepened main drain which was kept open. In 1897/98 this main drainage ditch was enlarged, the base concreted and the sides bricked. (Details from Johnston & Garner, 1969).

Map of 1623 showing field which became the Broadbalk experiment under arable cultivation

Broadbalk spreading chalk in 1954, Section V

Broadbalk 1935, tractor plough

Soil description

• FAO Classification: Chromic Luvisol (or Alisol)
• U.S. Soil Taxonomy: Aquic (or Typic) Paleudalf
• Soil Survey of England & Wales Group: Stagnogleyic paleo-argillic brown earth (Avery, 1980)
• Soil Survey of England & Wales Series: Predominately Batcombe Series (Avery & Catt, 1995 - see Soil Map link on left hand side).

For more details of the Batcombe and other soil series, see Cranfield University 2018 Soils Guide.

Soil texture class: Clay loam to silty clay loam over clay-with flints. The soils contain a large number of flints and are slightly calcareous. Below about 2m depth the soil becomes chalk. The experiment is under-drained and the site is free draining.

Soil texture, 0-23cm (from Gregory et al, 2010)

• Sand (2000 - 63 µm) : 25 %
• Silt (63 - 2 µm) : 50 %
• Clay (<2 µm) : 25 %
• Particle density g cm ^-3 : 2.56 %

There is considerable variation in soil texture across the site, with clay contents ranging from 19 - 39% (Watts et al, 2006). The mean clay content for Sections 0 (straw incorporated since 1986) and 1 (straw removed, both continuous wheat) is 28.3% (Watts et al, 2006). Clay content increases with soil depth in the Batcombe Series (from Jenkinson et al, 2008):

• 23-46 cm : 30 % clay
• 46-69 cm : 50 % clay
• 69-92 cm : 49 % clay

Soil pH: The plough layer (0-23 m) is limed when necessary to maintain a minimum soil pH of 7.0 - 7.5. Broadbalk was first limed regularly from 1955-1967, with plots given the larger inputs of ammonium fertilizers receiving more lime than the controls. Liming stopped from 1968-1975, but began again in 1976-1992, with 3 or 4 sections being limed each year, with 2.9t/ha of chalk being applied each autumn to all plots of each section. From 2007 onward selected plots have been limed every 5-6 years based on soil pH measurements to maintain top-soil pH around 7.0-7.5. Selected plots were limed in autumn 2018.

• For information on soil organic carbon, see Soil Organic Carbon (%SOC) Open Access long-term changes in SOC from selected treatments, 1843-2015 (doi: 10.23637/KeyRefOABKsoc-02)
• For information on soil concentration of Olsen P (plant-available P), see Plant-available P (Olsen P) Open Access long-term changes in Olsen P from selected treatments, 1843-2010 (doi:10.23637/keyrefoabkolsenp).

Soil weights, Mkg/ha

The following standard soil weights should be used for Broadbalk soil, both continuous wheat and rotational sections. All weights are in 10⁶ kg/ha of oven-dry fine soil. To convert to g/cm3 divide by depth in cm (eg 23) and multiply by 10. Data prepared by A J Macdonald and P R Poulton, February 2014.

Broadbalk standard soil weights 10⁶ kg/ha, 0-23cm

Broadbalk standard soil weights 10⁶ kg/ha, 23-46cm

Broadbalk standard soil weights 10⁶ kg/ha, 46-69cm and 69-91cm

Topsoil data is the mean of comprehensive measurements of soil weight made in 1865, 1881, 1893, 1914 and 2000. Figures in brackets are assumed. * indicates no samples taken. 23-46cm is the mean of all plots measured in 1865, 1881, 1893 and 1914. 46-69cm and 69-91cm is the mean of all plots measured in 1865, 1881 and 1893. See Dyer, 1902 for 1865, 1881 and 1893 data. See Watts et al, 2006 for 2000 data. 1914 data is previously unpublished.

Notes:

^a Plots 03-20, receiving inorganic fertilizer only, no manure (FYM). This also includes plot 03, given no fertilizer, and plot 19, recieving rape cake/castor meal.

^b Assume no further change in bulk density/soil weight on plots 2.1 and 2.2 after 2000. However, BD may increase slightly on plots in rotation as FYM is NOT applied to the oats.

^c BD will increase on plot 01 as applications of FYM stopped in 2000.The increase is assumed to be at the same rate as the decrease in BD between 1968-2000.

For deeper soil layers, use the following weights, taken from Dyer, 1902:

Acknowledgements

With thanks to Andy Macdonald and Paul Poulton for help with compiling the information and text.

Key References

Broadbalk Weeds

The Broadbalk experiment was started in 1843 to investigate the relative importance of different plant nutrients (N, P, K, Na, Mg) on grain yield of winter wheat. Weeds were controlled initially by hand hoeing and fallowing, but since 1964, herbicides have been applied to the whole experiment with the exception of Section 8. This is one of the few arable sites in the country where herbicides have never been applied. Weed surveys have been carried out annually in two phases, the first between 1933-1979 and the second from 1991-present day. There are earlier records too, with the first plot-by-plot surveys of weed species done in 1869 recording the presence of 23 species in stubble in September. Since then, approximately 130 weed species have been recorded on Broadbalk Section 8, but many of these only occur sporadically and about 30 of these are currently recorded annually (Species list 1869-2018). This site also provides an invaluable reserve for seven nationally rare or uncommon species including corn cleavers (Galium tricornutum), corn buttercup (Ranunculus arvensis), shepherd’s needle (Scandix pecten-veneris) and prickly poppy (Papaver argemone). This is now the only location in the UK where corn cleavers is known to occur naturally (see below). This resource and associated data enables various weed investigations including weed population ecology, studies on the effects of fallowing on the weed seed bank, seed dormancy and persistence, agroecology, and population dynamics of individual weed species. Recently, molecular approaches have been used to study the genetic diversity of weeds found on Section 8, this rare herbicide-free arable plot.

Section 8 was fallowed in 2015 and 2016 so there is no data for these years.

Current Survey 1991 to present

The current weed survey was started on Section 8 in 1991 and has continued annually ever since - although not on years when that section is in fallow i.e. 1994, 2001, 2008, 2015 & 2016. It was fallowed in 2015 and 2016 in order to reduce an infestation of Rumex obtusifolius in particular (see table below). Section 8 (called section VA 1958-1967) was created in 1968 when the experiment was divided into its present layout (Broadbalk plan today). The current assessment method records the presence of individual weed species in 25 random quadrats (0.1m2) per plot. Each year all 18 plots are surveyed meaning 450 quadrats are assessed per year, usually in June. Frequencies refer to the total number of quadrats in which a weed is recorded, the maximum being 25 per plot and 450 across all plots in the section. This method is more appropriate for detecting long-term trends in weed frequencies and population differences between plots than the earlier surveys (below) and provides a comprehensive set of 20 years of data for weed studies.

Corn cleavers

(Galium tricornutum) in Broadbalk

Weeds species in Broadbalk

non-herbicide Section 8 in early summer

Elevated view of Broadbalk

Section 8 weeds, mid-view, in late summer

Corn buttercup

Ranunculus arvensis

Shepherd's needle

Scandix pecten-veneris

Field poppy

Papever rhoeas

Data is presented in e-RA as the following datasets:

BKWEEDS_SUM: A summary of annual total frequencies for section 8 1991-present.

BKWEEDS_PLOT: Annual frequencies of each species per plot for section 8 1991-present.

There have been 53 species recorded in total since 1991 [species list 1991-2018] but on average there are 31 species found per year (with a maximum of 38 in 1993 & 1995 and a minimum of 26 in 2009) and there has been a gradual slight decline over the period. The long dataset allows general, overall trends to be observed and weed species are seen to differ greatly in their response to a given set of conditions, some declining, some increasing and others fluctuating - for example a decline in Papaver rhoeas between 1999 and 2003, and subsequent recovery (for which there is no obvious explanation). Of the species currently recorded annually ten species are locally common on many plots: Blackgrass (Alopecurus myosuroides), field poppy (Papaver rhoeas), common vetch (Vicia sativa), parsley-piert (Aphanes arvensis), scentless mayweed (Tripleurospermum inodorum), shepherd's needle (Scandix pectin-veneris), chickweed (Stellaria media), venus's looking glass (Legousia hybrida), creeping thistle (Cirsium arvense) and black medic (Medicago lupulina). A principal components analysis of the 1991-2002 survey data for 15 species showed clearly the influence of inorganic N fertiliser levels on the frequency of individual species. The frequency of one species (common chickweed (Stellaria media)) was greatly favoured by increasing amounts of nitrogen fertiliser from 0 to 288 kg N ha-1, others were strongly disadvantaged (e.g., black medic (Medicago lupulina) and horsetail (Equisetum arvense)), some were slightly disadvantaged (e.g., common vetch (Vicia sativa) and parsley-piert (Aphanes arvensis)), and some showed little response to differing N rates (e.g., blackgrass (A. myosuroides) and poppy (P. rhoeas)) (Moss et. al 2004).

Corn cleavers (Galium tricornutum) is one of the rarest plants in the UK and this occurence on Broadbalk is the last known site in the UK. Numbers have increased from under five individuals in the 1990's to over 450 in 2011 and this has been a consequence of our active management strategy.

* 20 years of records taken over 23 years (not in those years when section 8 was fallow i.e. 1994, 2001 and 2008)

Meadow vetchling

Lathyrus pratensis

Horsetail

Equisetum arvense

Chickweed

Stellaria media

Dwarf spurge

Euphorbia exigua

Black medick

Medicago lupilina

Common vetch

Vicia sativa

Earlier surveys 1933-1967 and 1968-1979

Annual surveys were conducted from 1933 to 1979. Originally, there were no sections on Broadbalk, just long strips the length of the whole field (Broadbalk plan 1852-1926). In 1926 the field was divided into 5 sections (I-V) (Broadbalk plan 1926). The whole field length was sampled for weeds as no herbicide weed control took place, just fallowing every five years. This gave all plot and section combinations (between 90 and 129 plots per year). After 1968, it was divided in to ten sections (Broadbalk plan 1996-2017) and again the whole field was surveyed each year - enabling comparison of weeds with and without herbicides.

This appears to be an excellent resource, however, it has not been widely used, principally because of the inconsistent frequency categories used. Neither were quadrats used, rather the assessor walked in a zig-zag pattern along the plot noting all weeds within 45cm of the plot boundary. Consequently there is a limitation on the interpretation of the data for ecological studies. The codes used to indicate species presence and abundance (termed STATE in the database) include the following:

The following datasets are available:

BBKWEEDS_FAL for 1933-67 (FAL indicating fallow) - herbicides applied from 1964 to all sections except VA (which became section 8 in 1968). Data for all sections and all plots.

BBKWEEDS_ROT for 1968-79 (ROT indicating rotation with other crops, though not on section 8 which is rotated only with fallow): - herbicides applied throughout to all sections except 8. Data for all sections and all plots.

During these years, surveys were conducted twice yearly, usually in May and August (sometimes as early as April or as late as September). Supplementary surveys were done for special purposes such as for blackgrass which is not obvious in the early season surveys and become more obvious in the summer when flowering. It is recommended that you extract both STATE and SPEC_REMARK since the presence of a species may be indicated by a remark such as 'patch' even though there is no code for the state. In view of the large number of null values it is probably best to tick the checkbox for STATE and exclude 'null' and '-', meaning none. In these datasets, dates are termed START DAY rather than year, as there are two surveys each year.

There were 114 species names recorded during 1933-1967 [species list 1933-1967] and 113 species names during the 1968-79 surveys [species list 1968-79].

Earlier surveys - data not included in e-RA:

Originally, there were no sections on Broadbalk, just long strips the length of the whole field (Broadbalk plan 1852-1926). The first plot-by-plot list of weed species was done in 1869 and recorded the presence of 25 species in stubble in September [species list 1869] (Thurston, 1969). Hand weeding and hoeing was practiced but due to the shortage of labour during the 1914-18 war the field became very weedy. Between 1926-1929 the field was fallowed to eliminate weeds, three out of five sections fallowed for 2 years running (Broadbalk plan 1926-1966). From 1931 a regular cycle of fallowing 1 year in five was introduced and the effect of fallow on weed seeds was studied (Brenchley & Warington 1930). Routine plot-by-plot surveys were started in 1930 and made twice yearly; first in May after spring germinating weeds are large enough to identify without trampling the crop and secondly, after the crop is cut and harvested which showed late germinating species.

Further information and references

The review paper by Moss et al. (2004) provides the most recent summary of weed studies on Broadbalk. The Thurston (1969) report provides additional information on the earlier surveys. References in both papers provide comprehensive coverage of other studies. For more details, refer to the Rothamsted Guide to the Classical Experiments 2018 pages 15-16 or contact the e-RA Curators.

Acknowledgements: With thanks to Stephen Moss, Jon Storkey, Richard Hull and Graham Shephard (VCU) for help with compiling images and data.

Note on herbicides: Applied to section 1A from 1957; to VB from 1963; and to all other sections (except VA which became section 8 in 1968) from 1964.

Notes on species: These include records of volunteer plants such as potatoes and both names for species names which have changed.

List of Latin and common names of species in current Broadbalk Survey.

Groundsel

Senecio vulgaris

Plot 3 Section 8 Broadbalk

Plot 21 Section 8 Broadbalk

Key References

Broadbalk Yields

Wheat grain and straw yields have been recorded every year since the experiment began, with the first harvest in 1844 (the crop was sown in autumn 1843). Dried grain and straw samples have also been kept for chemical analysis since 1844; these are preserved in the Rothamsted Sample Archive.
Yield data is held in four datasets, which reflect the changes to the experimental layout and harvesting techniques:

BKYIELD Broadbalk wheat grain and straw yields 1844-1925
BKYIELD_F Broadbalk wheat grain and straw yields 1926-1953
BKYIELD_F85 Broadbalk wheat grain and straw yields 1954-1967
BKYIELD_R85 Broadbalk wheat grain and straw yields 1968-current year

See Plans and treatments for more details of changes to the experimental layout over time.

See List of key dates, including sowing and harvest dates.

Harvesting:
From 1844 to 1901 the wheat crop was cut by hand by scythes, from 1902 to 1956 a self-binder was used, originally horse drawn, and then powered by a tractor. Hand cutting with scythes was often necessary if the ground was very wet or the crop was badly lodged (flattened). After cutting the crop was bound into sheaves which were ‘stooked’ and left on the plot for about two weeks then ‘carted’ into barns where they were threshed over the winter. Cutting and carting may have been spread over several days. Thus the earlier datasets BBKYIELD and BBKYIELD_F include the dates of both cutting and carting the crop. From 1957 the plots have been harvested by a small plot combine harvester; only the central strip of each plot is taken for yield and samples. Before 1957 the plots were usually cut in early August, since 1957 combining has been in August or early September. Sowing and harvest dates are available from the e-RA Curators.

Yields from Section 8 (no herbicides):
Section 8 has never received herbicides, and many weed species are present. Many weed seeds and other weed debris are included in the 'grain' yield at harvest. The 'straw' will also include weed material. The FYM plots and plots given the higher N rates generally have more weed contamination than the other plots. A sub-sample of grain is cleaned (weed seeds and debris removed) after harvest, and an estimate of the cleaned grain yield made. Data currently in e-RA is of the uncleaned grain. It is not possible to correct the straw yield for weed contamination. For further details, and the cleaned grain data, please contact the e-RA Curators

Units:
Since 1954 dry matter (DM) content of the grain and straw has been measured at harvest, and all yields have been converted to 85% DM (datasets BKYIELD_F85 and BKYIELD_R85). Before 1954 DM content was not measured and the yields are expressed at the DM content at which they were harvested (around 85% DM).

Corrections:
2013 plot 17 Section 3, no straw yield was recorded. An estimated value has been used, based on the straw/grain ratio of plot 16, Section 3 (a plot with similar grain yield): (5.44/8.01) * 7.95 = 5.39 t/ha.

Broadbalk harvest 1880, hand cutting with scythes

Broadbalk harvest 1935, horse-drawn reaper-binder

Broadbalk harvest 1935, tractor binder