Organic Nitrogen Nutrition Does Not Increase Glucosinolate Concentrations in Broccoli (Brassica oleracea L. var. italica)

Abstract

Concentrations of specific secondary metabolites can be higher in organically grown crops. This may be linked to organic nitrogen (N) nutrition that provides a gradual supply of N to crops over the growing season. This study examined whether organic N nutrition influenced the concentration of glucosinolates in broccoli crops. Nitrogen release patterns were determined from three synthetic (Rustica, 12% N; calcium nitrate, 15.5% N; urea, 46% N) and two organic fertilizers in an incubation experiment. Broccoli seedlings were then grown in two N dose response pot trials with different N source or application timing treatments to investigate growth and glucosinolate responses. Synthetic fertilizers released 84 to 89% of total N after 28 days, while chicken manure pellets and composted cow manure had only released 52% and 13% of total N, respectively, after 91 days. Broccoli yield and N content were generally higher in synthetic fertilizer treatments. Glucosinolate concentrations were generally higher in the synthetic fertilizer treatments, and only sinigrin and glucoiberin concentrations in the 800 kg ha⁻¹ N application rate of organic fertilizer matched those in the corresponding synthetic fertilizer treatment. Broccoli head weight was reduced when N was applied fortnightly compared to basal and weekly N applications, but glucosinolate concentrations were not significantly different. Overall, there was no evidence that organic (chicken manure) N nutrition, or the rate of N supply to broccoli plants, affect glucosinolate concentrations.

1. Introduction

The demand for certified organic food continues to grow, with the global organic food market in 2019 exceeding EUR 106 billion [1]. This growth is largely based on a high level of trust in certified organic produce by consumers [2]. Organic farming prioritizes natural inputs and practices to improve soil health, protect biodiversity, and reduce environmental impact compared to conventional farming, which often depends on synthetic pesticides and fertilizers. Organic farming emphasizes the use of natural fertilizers (including animal manures), composts, crop diversity, and rotation to meet crop nutritional requirements and for pest control [3]. However, as synthetic fertilizers and pesticides are prohibited from certified organic farming systems, crop yields are often lower compared to conventional farming systems [4].

Increasing consumer demand for certified organic food is in part due to a perception that organic produce is more nutritious than conventionally grown produce and free of pesticides [5]. A recent meta-analysis study indicated that food produced under organic systems contained significantly higher concentrations of some antioxidants, including phenolic acids, flavonols, flavanones, stilbenes, flavones, and anthocyanins, when compared to conventionally grown foods [6]. Studies isolating the effects of organic crop nutrition inputs suggest that the slow-release nature of organic fertilizers may be responsible for the observed increased plant secondary metabolite concentrations [7]. In particular, nitrogen (N) nutrition may play a role since increased N fertilizer rates have been linked to a reduction in some secondary metabolites in a range of horticultural crops including lettuces (Lactuca sativa L.) [8], tomatoes (Solanum lycopersicum) [9], and carrots (Daucus carota L.) [10,11], among others. However, reducing synthetic N fertilizer applications can also lead to a decline in yield [12], and it is therefore important to distinguish between increased content of secondary metabolites in organic crops versus increased tissue secondary metabolite concentrations due to yield reductions arising from nutrient deficiency.

Broccoli (Brassica oleracea L. var. italica Plenck) is a common vegetable crop, and like vegetable crops in the Brassicaceae family, is a good source of dietary glucosinolates that have been linked to a range of health benefits in humans [13]. Glucosinolates are sulfur (S)-rich compounds and are broadly classified into indole, aliphatic, or aromatic glucosinolates based on their precursor amino acid molecules [14]. The major glucosinolates found in broccoli include aliphatic-glucosinolates derived from amino acid methionine (glucoraphanin, glucoiberin, glucoalyssin, sinigrin, gluconapin, glucobrassicanapin, and progoitrin), indole glucosinolates derived from amino acid tryptophan (glucobrassicin, 4-methoxyglucobrassicin, and neoglucobrassicin), and the aromatic-glucosinolate derived from amino acid phenylalanine or tyrosine (gluconasturtiin) [15]. Among all glucosinolates found in broccoli, glucoraphanin has been widely studied for the anticarcinogenic properties of the glucoraphanin-derived isothiocyanate sulforaphane [16]. As with other brassica crops, previous studies have indicated that the accumulation and variation of glucosinolate profile in broccoli is highly influenced by genetic loci governing glucosinolate biosynthesis [17,18]. Moreover, tissue-specific concentrations of key glucosinolates in broccoli are strongly affected by N and S nutrition [19]. Schonhof et al. [20] reported total glucosinolate concentrations were low under insufficient S supply with optimal N supply, but high under insufficient N supply regardless of S supply. In addition to influencing total glucosinolate concentrations, N and S nutrition impact glucosinolate composition. Increases in applied N up to 150 kg N ha⁻¹ in the absence of applied S led to a decline in the concentrations of the aliphatic glucosinolate glucoraphanin and an increase in the indole glucosinolate glucobrassicin, where application of 50 kg S ha⁻¹ led to increased aliphatic glucosinolate concentrations and no change in indole glucosinolate concentrations [21].

In conventional Australian vegetable production systems, N is typically supplied as a basal mineral N fertilizer with additional N top-dressed at specific growth stages as either calcium nitrate or urea [22]. In organic production, N is typically supplied as a basal application in the form of manures or composts that comply with the National Organic Standard [23]. The specific effect of organic (slow release) N nutrition on the growth and glucosinolate yields in broccoli is not known but may be significant given reports that organic N nutrition may improve secondary metabolite concentrations in other crops [3].

The present study was undertaken to investigate the role of organic N nutrition compared to conventional N nutrition on yield and glucosinolate concentrations in broccoli. It was hypothesized that when other key nutrients were balanced between treatments, organic N nutrition (i.e., a slow release of N) would result in higher glucosinolate concentrations in broccoli at similar N fertilization rates.

2. Materials and Methods

Three experiments were conducted to test the hypothesis that the slow release of N from organic fertilizers would result in higher glucosinolate concentrations than mineral N nutrition at similar produce yields. The first experiment characterized N release patterns from two organic amendments compared to three synthetic N fertilizers in an incubation study. The second study was an N dose–response experiment using the most appropriate organic amendment from experiment 1 as well as a mineral N fertilizer to investigate broccoli yield and glucosinolate responses to applied N fertilizer. The third experiment was an N dose response experiment with mineral N fertilizers applied weekly or fortnightly to mimic a slow-release N product compared to standard industry fertilizer practice, to investigate specifically whether N supply pattern influences glucosinolate concentrations in broccoli.

Soil for the experiments was collected from the 0–100 mm layer of a rhodic Ferralsol derived from a tertiary basalt located at the Wollongbar Primary Industries Institute (31°9′ S, 150°59′ E, elevation 140 m) as described in Weng et al. [24]. The field had previously grown hemp and two crops of maize. Selected physiochemical properties of the soil are summarized in

Table 1. Physiochemical properties of the 0–100 mm layer of the unamended Ferralsol, chicken manure pellets, and composted cow manure (mean ± SD, n = 2).

Property	Ferralsol	Chicken Manure Pellets	Composted Cow Manure
pH (1:5 water)	5.5 ± 0.0	6.5 ± 0.0	7.8 ± 0.0
Electrolytic Conductivity (dS m−1)	0.0 ± 0.0	15 ± 0.3	2.7 ± 0.1
Bray 1 Phosphorus (mg kg−1)	4.4 ± 0.1
Colwell Phosphorus (mg kg−1)	39 ± 1.4
Nitrate-N (mg kg−1)	7.1 ± 0.1
Ammonium-N (mg kg−1)	11 ± 0.0
Exchangeable Calcium (cmol kg−1)	3.3 ± 0.0
Exchangeable Magnesium (cmol kg−1)	0.7 ± 0.0
Exchangeable Potassium (cmol kg−1)	0.5 ± 0.0
Exchangeable Sodium (cmol kg−1)	0.1 ± 0.0
Exchangeable Aluminum (cmol kg−1)	0.3 ± 0.0
Effective CEC (cmol kg−1)	5.2 ± 0.1
DTPA-extractable Zinc (mg kg−1)	1.5 ± 0.1
DTPA-extractable Manganese (mg kg−1)	13 ± 0.0
DTPA-extractable Iron (mg kg−1)	93 ± 23
DTPA-extractable Copper (mg kg−1)	0.6 ± 0.0
Total Carbon%	3.9	36 ± 0.1	7.3 ± 0.1
Total Nitrogen%	0.3	3.7 ± 0.0	0.6 ± 0.0
Total Calcium%		3.6 ± 0.1	1.8 ± 0.1
Total Magnesium%		0.8 ± 0.0	0.3 ± 0.0
Total Potassium%		2.6 ± 0.2	0.6 ± 0.0
Total Sodium%		0.6 ± 0.0	0.1 ± 0.0
Total Sulphur%		1.3 ± 0.0	0.2 ± 0.0
Total Phosphorus%		1.6 ± 0.1	0.4 ± 0.0
Total Zinc (mg kg−1)		479 ± 8.5	145 ± 6.4
Total Manganese (mg kg−1)		512 ± 8.5	614 ± 41
Total Iron (mg kg−1)		6348 ± 450	21,705 ± 1571
Total Copper (mg kg−1)		208 ± 2.1	38 ± 1.3
Total Boron (mg kg−1)		45.6 ± 2.7	20 ± 0.3

. The soil was air-dried and sieved to 4 mm prior to use.

2.1. Experiment 1—Nitrogen Release Patterns from Organic and Mineral Fertilizers

An incubation experiment was established to determine the N release rates from three synthetic N fertilizers, Campbells Fertilisers Australasia NPK “Rustica” (12% N), calcium nitrate (15.5% N), and urea (46% N), and two organic amendments, cow manure-based compost (0.6% N) and pelletized chicken manure (3.7% N). A nil-amendment control treatment (soil only) was also included. The incubation method was based on Eldridge et al. [25] with a few modifications. A two-level plywood and steel rod frame was used to house 18 50 mm-diameter × 200 mm-high PVC columns. Caps were placed at both ends with a hole for the sample port drilled in the bottom cap, which was subsequently sealed with a Shimadzu rubber septum. A 20 mm-high layer of glass wool (4 g) was compressed into the bottom of each column, before 210 g of the air-dried soil, along with the treatment amendments were added to create a 100 mm-high soil layer at the bottom of the column.

The experiment was run for 91 days to represent the typical duration of a broccoli crop (90–120 days). Three replicates were established for each of the six treatments. Fertilizer N was applied at the equivalent (pot surface area basis) of 560 kg N ha⁻¹, which is the approximate amount of N applied at the permitted maximum rate of compost applied to Australian organic farms (20 t ha⁻¹, 3.8% N) [26]. Selected characteristics of the composted cow manure and pelletized chicken manure are summarized in Table 1. Each of the N fertilizers was added to the 210 g of air-dried soil and thoroughly mixed before being added to each designated PVC column. The soil in the columns was then wet to the pre-determined 90% field capacity and subsequently incubated in a Sanyo CFC Free incubator set at 35 °C in the laboratory at Southern Cross University (Lismore, Australia). Solute samples were collected at 0, 1, 3, 7, 11, 16, 21, 28, 35, 42, 49, 56, 63, 70, 77, 84, and 91 days after establishment.

On each allocated sample day, the rubber septum and column cap were initially removed before 100 mL 0.01 M calcium chloride was applied at ̴9:30 am to leach dissolved N into solution from each column. This solute was then collected in a new 135 mL tube placed underneath the column. At 11:30 am, the bottom of the column was resealed with a rubber septum as described by Eldridge et al. [25] and 25 mL of (N-free) nutrient solution (Yoshida et al. [27]) was added to each sampled column to replace nutrients other than N that may have been leached out of the columns with the collected samples. Columns were re-capped at 3 pm and placed back in the incubator until the next sampling day. The top caps were removed 3 days after the samples were taken for 1 h on each occasion to refresh the column headspace air [25].

Two nutrient samples were collected from the solute extracted from each column for analysis. Solute for each nutrient sample was collected using a syringe, before transferring through a syringe filter (0.45 µm) into two 10 mL nutrient vials—one for mineral N (nitrate (NO₃⁻) and ammonium (NH₄⁺)) and one for total N—and subsequently stored at −20 °C until analysis. Analysis for NO₃⁻, NH₄⁺, and total N was carried out via colorimetry following methods of Rayment and Lyons [28] using a QuikChem 8500 Series 2 Flow Injection Analyzer (Lachat Instruments, Loveland, CO, USA). Total N samples were first digested in persulfate–sodium hydroxide and then analyzed using the Flow Injection Analyzer. Cumulative N release (%) for each fertilizer was calculated using the equation:

Cumulative N Release (%) = ((TN_f − TN_c)/N_i) × 100

where TN_f is the total measured nitrogen recovered from the treatment column, TN_c is the total measured N recovered from the control (soil only) treatment column, and N_i is the initial N application rate for the fertilizer treatments (111 mg N column⁻¹).

2.2. Experiment 2—Yields and Glucosinolate Concentrations in Broccoli Grown with Organic and Mineral N Sources

Broccoli was grown using two N source treatments, organic N and mineral N. The organic N source was chicken manure pellets applied as a single basal dressing while the mineral N treatment involved an initial N dose (50% of total N dose) of Rustica followed by two doses of calcium nitrate (25% of N dose for each application) at four leaf and budding stages as per industry practice. Seven N doses (0, 25, 50, 100, 200, 400, and 800 kg N ha⁻¹) were used for the two N sources, with three replicate pots per N source × N dose combination. Pots were placed randomly (completely randomized design) in a temperature-controlled glasshouse at NSW Department of Primary Industries Alstonville Research Station. Temperatures in the glasshouse ranged from 8 to 30 °C and humidity ranged from 40 to 90% throughout the experiment.

The soil used for the experiment was the same Ferralsol used in experiment 1. The soil was air-dried and sieved to 4 mm. Free draining plastic pots (250 mm diameter, 10 l volume) were then filled with 7 kg of air-dried soil, and the appropriate treatment fertilizer was added and thoroughly mixed. Basal nutrients were added to ensure that all treatments had the same amount of phosphorus, potassium, and S using combinations of single superphosphate (guano for organic treatments with 0% N), potassium sulfate and elemental S. To ensure no deficiencies of calcium, magnesium, zinc, or boron occurred, basal applications of dolomite (9.8 g pot⁻¹), zinc sulfate (0.05 g pot⁻¹), and Granubor (14.3% B, 0.06 g pot⁻¹) were applied to all pots.

2.3. Experiment 3—Yields and Glucosinolate Concentrations in Broccoli Grown with Mineral N Applied Weekly, Fortnightly, or as Industry Practice (Three Split Applications)

The experiment comprised of three N treatments—urea applied as a basal dressing plus in-crop urea applications at week 4 and budding (industry practice), urea applied weekly, and urea applied fortnightly—and seven N doses (0, 11.5, 23, 46, 92, 184, and 368 kg N ha⁻¹). Each N treatment and N dose combination was replicated three times, with the pots initially placed randomly in the same temperature-controlled glasshouse as experiment 1. Equal amounts of other basal nutrients were applied to each pot to ensure that all treatments had the same amount of P, K, and S using combinations of single superphosphate, potassium sulfate, and elemental S, with rates based on the same amounts of P, K, and S by the Rustica fertilizer in experiment 2 at the 400 kg N ha⁻¹ dose rate. Basal applications of dolomite and Granubor were also the same rates as experiment 2, while zinc sulfate was applied at 0.04 g pot⁻¹.

2.4. Experiments 2 and 3—Broccoli Cultivation and Measurements

Broccoli cv. Bejo Belstar seedlings, purchased from Seedlings Organics, Bangalow, SW, Australia, were used for the pot trial experiments. Evenly sized seedlings were transplanted into pots at one seedling per pot on 26 July 2021. Upon planting, soil was immediately wet to 85% field capacity, which was pre-determined in a lab prior to planting. Pots were re-randomized and re-wet to 85% of field capacity three times per week for the duration of the trials. All in-crop fertilizer applications were made by applying fertilizer granules to the soil surface immediately before watering. All pots were sprayed weekly with DiPel^® Bacillus thuringiensis (5 g l⁻¹) and Eco-Oil (5 mL l⁻¹) to prevent any build-up of Heliothis spp. and Chrysodeixis eriosoma numbers, with Eco-Oil used as an organic wetting agent. Both products were sprayed late in the day for each application to prevent UV degradation and improve efficacy.

Harvest was conducted when all of the treatments had set florets and were fully developed (day 91). On the day of harvest, subsamples of the inner and outer florets were taken and weighed fresh, then freeze-dried under vacuum at 0.63 mBar and −70 °C using a Christ Alpha RVC vacuum centrifuge (Christ™, Osterode am Harz, Germany) for glucosinolate analysis. All remaining leaf, head, and stem material was dried at 40 °C in a drying room for 7 days before a subsample was collected and ground to <0.5 mm using a ring mill. Approximately 0.2 g of the ground sample was used for total N analysis using a TruMac LECO CNS analyzer (model 630–300–400, Leco Corporation, St Joseph, MI, USA).

Concentrations of major glucosinolates in freeze-dried florets were quantified using methods described in Borpatragohain et al. [14], where each ground sample (~15 mg dry weight) was extracted with 1.5 mL of 70% aqueous methanol. Samples were then homogenized using a Qiagen Retsch MM 301 TissueLyser II (Qiagen Retsch, Hilden, Germany) at 30 Hz for 45 s and then centrifuged for 15 min at 15,000 rpm at 7 °C using a Sigma laboratory tabletop centrifuge (Osterode am, Harz, Germany). An aliquot of 200 μL of supernatant was transferred into a 2 mL Agilent HPLC screw-cap vial (Agilent Technologies Australia, Mulgrave, Victoria, Australia). The samples were then dried using a Martin Christ Alpha RVC (Osterode am Harz, Germany) at successively reduced pressures of 180, 120, 80, 50, 20, and 5 mbar each at 1 h intervals. The pressure was kept at 5 mbar overnight. The dried samples were resuspended in 1.5 mL water containing 1.17 μmol l⁻¹ glucotropaeolin (a glucosinolate not found in brassicas) as an internal standard. The tubes were mixed by inverting several times.

Eight glucosinolates (sinigrin, gluconapin, progroitrin, epi-progoitrin, glucoiberin, glucoraphanin, glucobrassicin, and gluconasturtiin) were quantified using High-Performance Liquid–Chromatography Mass-Spectrometry (HPLC-MS; Agilent 1260 Infinity II, Agilent Technologies, Palo Alto, CA, USA) with the temperature set to 30 °C. The instrument possessed a binary pump and Agilent diode array detector (DAD, 1260) together with an Agilent InfinityLab liquid-chromatography mass-spectrometry detector XT single quadrupole mass analyzer (Agilent Technologies, Palo Alto, CA, USA). Solvents used for the analysis were 0.01% trifluoroacetic acid in MillliQ water as solvent A and 0.005% trifluoroacetic acid in acetonitrile as solvent B. The 18 min run consisted of 0% B (4 min), 25% B (10 min), 100% B (18 min), and 0% B (10 min). Single-ion monitoring mode then detected the signals for the glucosinolates with sinigrin (at m/z ratio of 358 at 0–8 min), glucotropaeolin (m/z ratio of 408 at 8–18 min), progroitrin and epi-progoitrin (m/z ratio of 422 at 1–10 min), gluconasturtiin (m/z ratio of 422 at 10–18 min), and gluconapin (m/z ratio of 372 at 0–18 min).

2.5. Statistical Analyses

All statistical analyses were conducted in R version 2.4.1. For both pot experiments, separate two-way analyses of variance (ANOVAs) (2 [fertilizer type] × 6 [fertilizer rate]) were conducted to test the effects of the N fertilizer dose and type on broccoli head weight, shoot weight, plant tissue N concentration, and glucoraphanin, glucoiberin, sinigrin, and glucobrassicin concentrations. Since there was no head formation (broccoli) in the control treatment, the control treatment was not included in the analysis. Data were log-transformed prior to analysis to satisfy assumptions of normality, and back-transformed means are presented. Post hoc pairwise differences between means (for all response variables) were calculated using the Estimated Marginal Means (emmeans) [29] and “multcomp” packages [30]. All uncertainty is indicated with the standard error unless indicated otherwise.

3. Results

3.1. Experiment 1

All synthetic fertilizers rapidly released either nitrate-N, ammonium-N, or organic-N during the first 1–7 days of incubation. The highest initial N releases were observed from the calcium nitrate and urea columns (Figure 1), where 77 mg nitrate-N column⁻¹ and 51 mg organic N were leached from the respective treatments on the day of fertilizer application (Figure 1A,E). In the Rustica treatment, 22–23 mg ammonium-N column⁻¹ was initially leached on day 0 and day 1 before steadily declining to approximately < 5 mg N/column from day 30 onwards (Figure 1C). From day 3 onwards, negligible amounts of nitrate-N or organic N were leached from any treatment, while small amounts of ammonium-N continued to be released from the fertilizer-treated columns until day 91 (Figure 1C). Nitrogen release from the organic amendments was slower but more prolonged compared to the synthetic fertilizers. For example, the composted chicken manure ammonium-N release peaked at around 8.6 mg column⁻¹ by day 3 but then steadily released 2–7 mg column⁻¹ from days 11 to 91 (Figure 1D). The composted cow manure had the lowest rates of N release across the incubation, with negligible concentrations of organic N and nitrate-N measured in leachate. Only low rates of ammonium-N leaching were observed from the cow manure, peaking at around 4.6 mg column⁻¹ by day 28 before declining to around 1.2 mg column⁻¹ by day 91. The unamended Ferralsol control only released 2–4 mg column⁻¹ nitrate-N during the first 3 days, with negligible nitrate-N leached thereafter (Figure 1A). Ammonium-N release was also low from the unamended Ferralsol, peaking at around 3 mg/column by day 7 and declining thereafter.

Total N release from all synthetic fertilizer treatments was faster and higher than both organic fertilizer treatments (Figure 2). Cumulative total N release from the three synthetic fertilizer treatments was from 160 to 195% higher than the best-performing organic treatment (chicken manure ̴ 52% released by day 91). Calcium nitrate and Rustica treatments had released approximately all total N by the end of the incubation (102 and 99%, respectively). Total N release from the cow manure was particularly low, with only 13% of the total N released by day 91. Total N release also occurred more rapidly from the synthetic fertilizers, with over 87% of total N in the calcium nitrate fertilizer released by day 1, while 82% of total N in Rustica and 77% of total N in urea had been released by day 7 (Figure 2). Comparatively, only 33% of total N from the chicken manure pellets had been released by day 28, at which point only 1% of total N had been released by the composted cow manure.

3.2. Experiment 2

There were significant effects on broccoli head weight and shoot N content from fertilizer source (chicken manure vs. synthetic Rustica), fertilizer dose rate, and their interaction (

Table 2. Significance of fertilizer source, fertilizer rate, and their interaction on key broccoli traits from two-way ANVOA statistical testing for the pot trial experiments.

Experiment 2	Fert. Source	Fert. Rate	Interaction
Head weight	**	***	*
Shoot dry weight	ns	*	ns
Shoot N content	***	***	***
Glucoraphanin	**	***	*
Glucoiberin	**	*	ns
Sinigrin	***	***	***
Glucobrassicin	***	***	ns
Experiment 3	Fert. timing	Fert. rate	Interaction
Head weight	**	***	ns
Shoot dry weight	**	***	*
Shoot N content	**	***	ns
Glucoraphanin	ns	***	ns
Glucoiberin	ns	***	ns
Sinigrin	**	***	ns
Glucobrassicin	ns	***	ns

ns = not significant at p ≤ 0.05, p ≤ 0.05 (*); p ≤ 0.01 (**); p ≤ 0.001 (***).

). Broccoli heads were present at 100 kg N ha⁻¹ and beyond for synthetic N fertilizer but were only present from 400 to the maximal 800 kg N ha⁻¹ dose rate in the organic N fertilizer treatments (Figure 3A). Head weights were around 5 g pot⁻¹ in both N source treatments at the 800 kg N ha dose rate (Figure 3A). There was an increasing trend towards higher total biomass yields up to 400 kg N ha⁻¹ observed for both organic and synthetic N treatments (Figure 3B). However, there were no significant differences observed within treatment rate between the organic or synthetic N source. Shoot N content was significantly higher in the synthetic fertilizer treatments compared to the organic fertilizer treatments from 200–800 kg N ha⁻¹ (Figure 3C).

Glucosinolates were only detected at 100 kg N ha⁻¹ dose rates and above from the synthetic N treatment and 400 and 800 kg N ha⁻¹ dose rates from the organic fertilizer treatments (Figure 4). Glucoraphanin and sinigrin were the dominant glucosinolates detected with respective concentration ranges of 0.3–0.5 µm g⁻¹ and 0.17–0.22 µm g⁻¹ across the 200–800 kg N ha⁻¹ dose rates in the synthetic N treatments (Figure 4A,C). Glucosinolate concentrations were generally higher in the treatments receiving the synthetic fertilizer, and they were also observed at lower dose rates (Figure 4), especially with glucobrassicin where concentration tended to increase from 100–400 kg N ha⁻¹ in the synthetic N fertilizer and 400–800 kg N ha⁻¹ from the organic N fertilizer treatment (Figure 4D). Overall, there were significant effects of fertilizer source (organic vs. mineral) and fertilizer dose rate on concentrations of all glucosinolates in broccoli florets, and significant interactions for sinigrin (p < 0.001) and glucoraphanin (p < 0.05) concentrations (Table 2).

3.3. Experiment 3

There were significant effects of fertilizer timing (basal vs. fortnightly vs. weekly) and fertilizer rate on head weight, shoot dry weight, and shoot N content and a significant fertilizer timing × rate interaction for shoot dry weight (p < 0.05) (Table 2). Fortnightly nitrogen application timing appeared to negatively impact broccoli head weights compared to the basal and weekly applications (Figure 5). Head weights in all treatments increased significantly from 92 to 184 kg N ha⁻¹, but there was no significant increase from 184 to 368 kg N ha⁻¹, and head weights in the fortnightly application were the lowest in each N rate group (Figure 5A). Similarly, there was no significant increase in shoot biomass beyond 184 kg N ha⁻¹, with maximum dry biomass across all N timing treatments of around 90 g pot⁻¹ (Figure 5B). In contrast to shoot weight, shoot N content continued to increase up to the maximal 368 kg N ha⁻¹ rate where it was approximately 3 g pot⁻¹ (Figure 5C).

The only glucosinolate to respond to the different fertilizer application timing was sinigrin (p < 0.01), where lower concentrations in the fortnightly fertilizer applications were generally observed (Figure 6C). In contrast to application timing, fertilizer dose rate had a significant (p < 0.001) effect on the concentrations of all four detected glucosinolates (Table 2), with glucosinolate concentrations generally increasing with dose rates (Figure 6). For the two dominant glucosinolates, glucoraphanin and sinigrin, there was no significant increase in floret glucosinolate concentrations beyond 92 kg N ha⁻¹ (Figure 6A,C).

4. Discussion

A number of market surveys and field studies have shown higher levels of specific glucosinolates in leaves and heads of organically grown broccoli compared to conventionally grown broccoli [31,32]. This study tested the hypothesis that slow-release N in organic nutrition regimes contributes to higher concentrations of glucosinolate in broccoli crops.

Initial characterization of N release patterns from mineral and organic N sources indicated a rapid release of N from urea, calcium nitrate, and Rustica, which is typical of synthetic N fertilizers [33]. Two of the synthetic N fertilizers (calcium citrate and Rustica) released all of their N by the end of the incubation; indeed, more N was eventually recovered from the calcium citrate treatment than was added with the fertilizer (102% N release). The greater than 100% recovery is only slightly more N than what was added and may have simply been a result of a propagated error. Alternatively, the continual release of N to above 100% of N applied in the calcium nitrate fertilizer may have been due to a priming effect, where additional N affects the rate of soil organic matter decomposition [34]. In contrast to the mineral N fertilizers, N release rates from the organic amendments were slow, with only 13% of total N released from cow manure compost and 52% of total N released from chicken manure pellets in the Ferralsol after 91 days. High variation in N mineralization rates from various composts and manures has been reported previously and is generally attributed to incubation temperature, incubation length, and amendment C:N ratio [35]. Overall, our results for the organic fertilizers are comparable to N release rates reported in other studies on manures and composts over 3- to 4-month periods [36,37]. Owing to the ultra-slow release of N from the cow manure compost, the chicken manure pellets were chosen for use in the subsequent experiment examining the effect of organic N nutrition on glucosinolate concentrations in broccoli.

Regardless of N source or supply pattern, the concentrations of all glucosinolates detected in experiments 2 and 3 generally increased with N supply until around 100 to 200 kg N ha⁻¹, then plateaued at higher rates. While this has previously been reported in broccoli florets for glucobrassicin [38], not all glucosinolates appear to be limited in the same plateauing manner with increasing N supply. Glucoraphanin, the dominant glucosinolate in broccoli in our study and previous studies [20,31,38], has been reported to increase in concentration until optimal N supply is reached before declining at luxury N levels provided S nutrition is adequate [20,38], as was also reported in broccoli florets for glucobrassicin [38]. While it is possible that the highest N dose rates in experiment 2 and 3 were not sufficient to induce a reduction in glucoraphanin concentrations, the concentrations and composition of broccoli glucosinolates appear to differ by cultivar [31,39,40] and seasonal conditions [39]. For example, Wang et al. [41] reported glucoraphanin and glucoerucin were the major glucosinolates in 27 broccoli seed samples, while Valverde et al. [31] reported glucoraphanin and glucobrassicin were the dominant glucosinolates in florets with no detectable glucoiberin and inconsistent trace amounts of sinigrin. Thus, while broccoli florets in our study had higher concentrations of sinigrin and lower concentrations of glucobrassicin than other studies, these differences likely reflect differences in cultivar choice and the growing conditions. Cultivar choice and seasonal conditions may also have contributed to differences in the responses of glucosinolates to the luxury N supply used in our study.

Overall, available N appeared to be an important factor for biomass and glucosinolate concentrations, regardless of the source. While shoot N content was typically higher in the synthetic N treatment than the organic chicken manure treatment at N application rates above 100 kg N ha⁻¹, the 400 kg N ha⁻¹ synthetic N treatment and 800 kg N ha⁻¹ chicken manure treatment had similar shoot N content, biomass, and head yields. Examination of these two treatment combinations, where N supply pattern differed (slow-release organic N versus rapid synthetic N release) but crop N uptake was similar (around 1.6 g N pot⁻¹), indicates no significant differences in any measured glucosinolates concentrations in florets (Figure 4). Thus, the data do not support the hypothesis that organic N nutrition leads to higher glucosinolate concentrations, acknowledging that only one cultivar was examined in this study This is further supported by the results of experiment 3, where the application of N every week or fortnight in order to mimic a slow-release N source did not lead to greater concentrations of any key glucosinolates compared to synthetic N applied as per industry practice at any applied N dose (Figure 6).

That patterns of N supply had no significant effect on glucosinolate concentrations in broccoli when other key nutrients were balanced between treatments suggests other factors likely led to the higher glucosinolate concentrations in leaves and florets from organic broccoli reported in earlier field studies [31,42]. A lack of nutrient balancing between treatments is common in many agronomic comparisons of organic and mineral fertilizers (see review by Edmeades [43]), and even where nutrient rates are balanced, differences in nutrient form/species can influence loss rates in the field with consequences for crop nutrition. In our study, balancing key nutrients between treatments or applying them in excess of crop needs, and watering pots to weight to minimize nutrient losses, enabled a comparison of organic vs. mineral N supply in the absence of many confounding factors. Given the strong link between S and N nutrition in relation to glucosinolate levels in broccoli [20,38] and other brassica crops [44,45], it is possible that differences in S speciation may have contributed to previous results. For example, while more total S was applied to conventional broccoli crops in the study of Valverde et al. [31], it was applied in the form of sulphate in a field trial setting, and heavy rainfall documented in that study may have led to S leaching losses from the conventional treatment that did not occur in the organic treatment. Ultimately, the reasons for increased glucosinolate concentrations in organically grown broccoli compared to conventionally grown broccoli in previous studies [31,32,42] remain unknown but appear unlikely to be due to the slow-release supply of N in manure amendments.

5. Conclusions

Unlike a range of previous studies, our study did not find any increase in glucosinolate concentrations in broccoli florets from manure N vs. synthetic N and found no evidence that the supply pattern of N, as opposed to N dose, influences broccoli glucosinolate concentrations. While the results may have been influenced by the specific environmental conditions and manures tested, the study indicates that general conclusions regarding organic production techniques and secondary metabolite levels in crops cannot be reached and will likely be specific to the species of crop grown and the specific metabolites being investigated.