Factors influencing fungal and aflatoxin levels in Turkish hazelnuts (Corylus avellana L.) during growth, harvest, drying and storage: A 3-year study

Abstract

The levels aflatoxins in Turkish hazelnuts have been monitored over a 3-years period (2002–2004). Periodical sampling was made in 72 different orchards at different locations representative of the hazelnut-growing areas and post-harvest applications. Various parameters (aflatoxins, water activity, moulds) were analysed and environmental conditions (temperature and relative humidity) recorded during growing and at different stages of harvest and post-harvest processing, involving three different harvesting methods (collection in nets, from the ground, etc.) and four drying techniques (traditional sun-drying, mechanical drying, etc.). Fungal and aflatoxin analyses (HPLC) showed no significant difference except between samples which had been in contact with the ground and those which had not (at 95% confidence level). Aflatoxins levels from the orchard recorded a maximum of 0.77 ± 0.08 ng g⁻¹ from a total of 1624 samples. Regarding harvesting and post-harvest processes, the only application where aflatoxins were detected was in samples which had been in direct contact with the ground (max. 3.18 ± 0.03 ng g⁻¹). Aflatoxin formation was low during storage (max. 0.34 ± 0.003 ng g⁻¹). As a result of mycological studies, a total of 5546 Aspergillus flavus (89%) and A. parasiticus (11%) species were isolated and identified from samples. The results indicated that harvesting hazelnuts into a canvas by shaking the trees, manual harvesting of mature hazelnuts where possible, use of jute instead of nylon sacks and mechanical drying technique would minimize aflatoxin levels in hazelnuts. These recommendations have been implemented and about 4000 people in the hazelnut industry have been trained in these practices.

Keywords:

• Hazelnut
• aflatoxin
• moulds
• harvesting
• drying
• storage
• prevention

Introduction

Aflatoxin contamination is an important issue in the areas of food safety and international trade since they are potent carcinogens and teratogens in humans and farm animals (Moreu 1979; Bullerman 1986; Pitt 2000; Campbell et al. 2003; Magan and Olsen 2004; European Commission 2006). The total aflatoxin action threshold level for nuts is set at 20.0 ng g⁻¹ by the US Food and Drug Administration for domestic foods. The European Union (and Japan) has set their threshold levels for imported nuts, intended for direct human consumption or use as an ingredient in foodstuffs, at least five times lower–at 4.0 ng g⁻¹ for total aflatoxins (European Commission 2006). However, the low threshold levels for aflatoxin have significantly increased the probability of the rejection of tree-nut shipments by the major importing nations of the EU and Japan. Therefore, the determination of aflatoxin risk and preventive measures need to be implemented to minimize food safety concerns and for purely economic reasons.

In Turkey, hazelnuts are traditionally sun-dried and may be subject to mould growth and, like other nuts, subsequent aflatoxin formation due to prolonged drying under humid or rainy conditions (Simsek et al. 2002). The global production of hazelnut is increasing by more than 5% annually; Turkey being the most important producer (75% market share), followed by Italy (15%), Spain (3%) and the US (3%) (Annual Statistic 2006).

Aspergillus flavus and A. parasiticus are known to be the primary aflatoxin-producing species, and are closely associated with agricultural environments and crops, including tree nuts (Magan and Olsen 2004; Bayman et al. 2002; Denizel et al. 1976a). A. flavus populations are influenced by agricultural and processing practices but, in many cases, the mechanism and reason are unclear. Several environmental factors are known to influence aflatoxin production, but temperature and relative humidity (RH) are critical (Northolt et al. 1976; Chiou et al. 1984; Denizel et al. 1976b). The hard shell of nuts is a good barrier against bacterial and fungal contamination (Bayman et al. 2002; Campbell et al. 2003); nevertheless, the rate and degree of contamination are dependent on temperature, humidity, soil and storage conditions. Prevention, particularly by excluding or reducing toxigenic mould growth and toxin production in susceptible food crops, is the most effective way to restrict aflatoxin contamination (Magan and Olsen 2004; Barung et al. 2006). Additional factors, such as substrate composition (Sakai et al. 1984), storage time, insect damage and presence/absence of a shell, also influence fungal growth and aflatoxin production (Schatzki and Ong 2001; Campbell et al. 2003). It is also important to recognize, however, that the interaction of all these factors may provide varying results as regards fungal growth and mycotoxin production, even on identical substrates.

It is estimated that 5–10% of food and other agricultural crops are unfit for human/animal consumption due to fungal damage, at a cost of nearly 16 billion USD per year (Pitt and Hocking 1997; Moreu 1979). The total cost of tree nut sales lost to aflatoxin contamination averages around $50 million per year (Cardwell et al. 2001). The impact of potential aflatoxin contamination on hazelnuts as regards food safety and international trade has created the impetus to develop methods and strategies for reducing aflatoxins in pre- and post-harvest hazelnut products. Therefore, the aim of this study was to determine the level of aflatoxin contamination in Turkish hazelnuts from orchard to storage and implement preventive measures in pre- and post-harvest applications.

Materials and methods

Chemicals & instruments

All chemical reagents and standards were obtained from Sigma-Aldrich-Fluka Co. Ltd. (Taufkirchen, Germany), unless otherwise stated. The aflatoxin standards were obtained from R-Biopharm Rhone Ltd. (West of Scotland Science Park, Glasgow, UK). HPLC analyses were done using a Shimadzu Class-Vp 5.0, Shimadzu RF-10AXL (florescence detector); column: ACE C18, 250 × 4.6 mm I.D., 5 µm particle size (Advanced Chromatography Technologies, Aberdeen, Scotland).

Methods of analysis

Mould enumeration

An assessment of the level of mould contamination was carried out by suspending 25 g from each hazelnut sample in 225 ml of sterile dilution solution. Homogenised samples were diluted and inoculated on Dichloran Rose Bengal Agar (DRBC-Oxoid) by the pour plate method for enumeration. The plates were incubated at 25 ± 1°C for 5–7 days. Fungal counts from plates, having between 15 and 150 colony-forming units (cfu), were used in calculating the total mould count per g in each sample (Pitt and Hocking 1997; Samson et al. 1995).

Isolation and identification of Aspergillus flavus group

Hazelnut samples (50 kernels) were surface disinfected by a 2 min immersion in 70% ethanol followed by 2 min in 0.4% chlorine; then, 40 kernels were plated directly (2 kernels per plate) onto Aspergillus Flavus and Parasiticus Agar (AFPA; Oxoid). Plates were incubated at 25 ± 1°C for 3–5 days and examined visually under a stereomicroscope. Fungal growth was recorded after incubation and colonies having a yellow/orange colour on the underside were isolated as possible A. flavus/parasiticus growth. Any colony suspected of belonging to the A. flavus group was sub-cultured on Czapek (CZ; Merck) and Malt Extract Agar (MA; Merck) (Pitt and Hocking 1997; Samson et al. 1995). Morphological characteristics were observed during growth and identification of A. flavus and A. parasiticus was made according to the classification given by Pitt and Hocking (1997) and Samson et al. (1995).

Detection of aflatoxin production ability of fungi

Pure isolates of A. flavus and A. parasiticus were transferred to YES (Yeast Extract Sucrose) Agar and incubation for 14 days at 25°C for the production of secondary metabolites (Samson et al. 1995). After the incubation period, the YES medium was mixed with chloroform in a stomacher shaker for 1–2 min. The chloroform extracts of culture filtrates of A. flavus and A. parasiticus were qualitatively analysed on thin-layer chromatography (TLC) plates under UV light. The TLC plates, spotted with ∼20 µl of chloroform extract, one spot of standard aflatoxins (mixture of B₁, B₂, G₁ and G₂; R-Biopharm Rhone Ltd.) and one spot of chloroform extract plus standard aflatoxins were developed in a chloroform/acetone solvent system (90:10, v/v). The plates were observed under UV light (at 366 nm) for detection of various aflatoxins by comparison with a standard after 5–6 min air drying of the plates. Chemical confirmation of aflatoxins was carried out by spraying 25% H₂SO₄ (Reddy et al. 1970; Stack and Pohland 1975).

Determination of aflatoxin (B₁, B₂, G₁ and G₂)

Qualitative aflatoxin tests for detection of aflatoxin-producing fungi were performed using TLC plates and quantitative test were carried out by HPLC. After each pre-harvest, harvest and post-harvest steps, samples were taken and sent to the laboratory under cold-chain conditions (4°C) to conduct aflatoxin B₁ and total aflatoxin analyses. The samples were homogenized with a Waring blender using water (1:1, v:v). Aflatoxins were quantified using HPLC–IAC with post-column derivatization from AOAC (999.07; AOAC 2000) (limit of detection (LOD): B₁ > 0.04 ng g⁻¹; total > 0.1 ng g⁻¹).

Determination of water activity and moisture

Water activity values of all samples were measured using a Novasina Novatron water activity analyser at 25^°C constant temperature. During harvesting and post-harvest investigations, a mobile laboratory was set-up in the field and moisture content and other physical experiments were conducted from this mobile unit. Moisture content of each sample was measured as previously described (AOAC 2000).

Statistical analysis

All results were analysed using analysis of variance followed by least-significant difference (LSD) with SPSS software ver.7.5.1 (SPSS Inc., 1999). Significance was determined at p < 0.05 for all analyses.

Sampling of hazelnuts for analysis

Pre-harvest studies

Hazelnut samples were collected from three different parts of the Black Sea region, a major hazelnut-growing area of Turkey (east, middle, and west) over consecutive 3 years (2002, 2003 and 2004). The names of the regions and number of orchards are given in Table I and displayed in Figure 1. The number of the orchards was determined according the hazelnut production capacity of the region. The orchards were selected homogeneously and located at altitudes of 0–250, 250–500 and 500–750 m in each region. However, no significant difference was found in total mould counts or aflatoxin incidence for altitude, which was then ignored during evaluation. Periodical sampling was done for 4 months (May–August) from flowering stage until harvest. In each year, the periods between each sampling and the starting date were set according to climate and regional differences. The number of samplings in each year varied between eight and nine, depending on maturation of the hazelnuts.

Figure 1. Locations of the 72 orchards (each point on the map indicates the location of the orchard from western (1–24), central (25–44) and eastern (45–72) areas of the Black Sea region).

Sampling was carried out in each orchard according to “Z” pattern (AOAC 2000) and 11 trees were marked at five points in an orchard. In total, the same 55 marked trees in an orchard were used for subsequent sampling throughout the 3-year period. The sampling was done at different maturation stages until the harvesting period. Hazelnut samples (minimum 3 kg for each case) were placed in a polystyrene box with a cooling gel (pre-frozen to −20°C) and transferred to the TÜBİTAK MRC Food Institute within the same day. Upon arrival, the temperature and weight of the samples were recorded and samples were subjected to total mould count, mould isolation, water activity and aflatoxin analyses. For aflatoxin tests, hazelnut samples from flowering stage to kernel formation were homogenized after removing the green hulls. After kernel formation, the hazelnuts were cracked and shells were removed manually, taking precautions against possible contamination.

Table I. Sampling data from three hazelnut-growing regions.

			Number of orchards
Region	Orchards no.	City	2002	2003	2004
West	1–24	Adapazari	12	12	12
		Düzce	12	12	12
Central	25–44	Samsun	9	9	9
		Ordu	12	12	12
East	45–72	Giresun	15	15	15
		Trabzon	12	12	12
Total			72	72	72

Harvesting and post-harvest studies

Harvesting

Tombul variety of hazelnuts (Corylus avellana L.) were harvested during the August 2002 and 2003 seasons in two provinces, located to the east and west of the Black Sea. Harvesting was done using three different techniques; early manual harvesting, harvesting to a canvas (shaking the branches to collect the hazelnuts from canvas laid under the branches) and harvesting from the ground (“windfall”). Early manual harvesting was done 1 week before the officially announced harvesting period to determine the effects of early harvesting on aflatoxin formation. Harvesting to a canvas was done within the normal harvesting period by shaking the branches to knock the hazelnuts. Harvesting from the ground was done by collecting the fallen hazelnuts (naturally matured) towards the end of the harvesting period. Hazelnuts were collected into jute or nylon sacks to determine the effect of different materials on aflatoxin formation. The harvesting scheme for each application is given in Table II. Application nos. 1, 3 and 4 were the applications carried out by some farmers under the poorest conditions. Application no. 2 is the post-harvest stage performed by most farmers carry out. Application nos. 5A and 5.B were those recommended, using meshed shelves and mechanical driers. Application nos. 2 and 5 were repeated for 2 years.

Table II. Applied harvesting techniques to determine their effects on aflatoxin formation in hazelnuts.

	Applications
Post-harvest stages	No. 1	No. 2	No. 3	No. 4	No. 5
	2003	2002/2003	2002	2003	2002/2003
(1) Harvesting	Manual/early harvesting	Ground*	Ground	Ground	Canvas
Sacks used	Nylon	Jute	Nylon	Nylon	Jute
(2) Withering	10 Days in nylon sacks	3 Days on concrete floor	3 Days on ground	7 Days on ground plus 3 days in nylon sack	3 Days on mesh shelves
(3) De-hulling
(4) Drying	Ground	Concrete floor	Ground	Ground	(A) Mesh shelves
					(B) Mechanical drying
*Ground application was preformed in direct contact with soil.

Drying of hazelnuts

Different sun-drying techniques used by farmers (drying on the ground/soil, concrete floor, etc.) were evaluated for levels of aflatoxin formation. Drying was also carried out using two layers of mesh shelves and a mechanical dryer operating at 40°C, which was designed specifically for hazelnut farmers. Details of the drying processes will be discussed in a future article.

Hazelnuts were dried until the moisture content decreased to 5%, which is the accepted, safe moisture level. Average ambient air temperature and relative humidity in the province was 23°C/78% in 2002 and 22°C/75% in 2003, but ambient temperature reached 27–29°C in the afternoon. During post-harvest treatment, the ambient relative humidity reached 96% on rainy days in both the 2002 and 2003 seasons.

Samples were also collected from harvest sites under adverse conditions, i.e. during times of high humidity (rainy days or just after rain) from sites which were in contact with soil. The hazelnuts were protected from rain by nylon canvas without any air circulation.

Storage of hazelnuts

Dried hazelnuts were stored under controlled (5 ± 2°C, 65 ± 5% RH) and uncontrolled conditions (storage on-site without any temperature and humidity control). Temperature and environmental conditions in the storage facility were recorded using AZ Data loggers.

Training and dissemination of knowledge

To disseminate information about aflatoxin and preventive measures, approximately 4000 people (growers, traders and exporters) were trained over the three years. Posters (20,000) and two series of brochures (37,000) were also published and distributed.

Results and discussion

Sampling from orchards

The sampling was done from the upper and lower branches of the same trees from each orchard on three consecutive years (2002, 2003 and 2004). The number, total mould count and water activities of samples from 72 orchards are given in Table III. Statistical analysis was carried out if there was a significant difference among altitudes, regions or sampling year (SPSS 10.1 version; t-test). However, there were no significant difference among altitude or regions over three consecutive years in terms of total mould counts (p > 0.05). The samples of different maturation stages were also analysed for their compositional characteristics and enzyme activities to investigate if any correlation existed with aflatoxin formation; these results are reported elsewhere (Seyhan et al. 2007). However, the number of aflatoxin-contaminated samples were insufficient for a correlation study.

Table III. Total mould count and water activities of samples from 72 orchards.

Years	No. of samples	Total mould count [cfu g^−1] (min–max)	A w (min–max)
2002	648	1.00 × 10^2–3.87 × 10^6	0.95–0.99
2003	640	1.22 × 10^3–2.09 × 10^5	0.96–0.99
2004	336	1.80 × 10^1–3.40 × 10^5	0.95–0.98

In Turkish hazelnuts, the mould count ranged between 1.8 × 10¹ and 3.8 × 10⁶ cfu g⁻¹ from flowering stages to harvest. Arrus et al. (2005) investigated the mould count of Brazil nuts from various regions and found differences between the regions in the range 10⁶–10⁷ cfu g⁻¹. In 109 freshly harvested maize samples, the total mould counts ranged 1.9 × 10⁴–3.5 × 10⁶ cfu g⁻¹, and ranged 1.8 × 10²–1.6 × 10⁴ (average: 3.4 × 10³) in 32 cashew nut samples from Nigerian markets (Adebajo and Diyaolu 2003; Ono et al. 2006). Total mould count of 10³–10⁴ cfu g⁻¹ have been reported in 143 freshly harvested pistachios from Turkey and the mould count increased to 10⁵–10⁶ cfu g⁻¹ during storage (Heperkan et al. 1994).

Aflatoxin was not detected (LOD for B₁ < 0.04 ng g⁻¹; total aflatoxin <0.1 ng g⁻¹) in 640 and 336 samples from orchards, from the flowering stage to harvest, for 2003 and 2004, respectively. Aflatoxin was detected in seven orchards in 2002 prior to harvest at low levels and the maximum level was 0.77 ± 0.08 ng g⁻¹. The percentage of aflatoxin positive samples were only 1.54% of 648 samples in 2002.

There was a decreasing trend in water activity of samples prior to harvesting but the level varied between 0.95 and 0.99. The optimum water activity for growth of A. flavus is 0.98–0.99 (ICMSF 1996; Pitt and Miscamble 1995) and the range for growth of A. flavus and A. parasiticus is between 0.99 and 0.80 (Bresler 1998; Holmquist 1983; Nesci 2003). Studies on hazelnuts and pistachios suggest that optimum temperature and RH for aflatoxin production is 25–30°C and 97–99%, respectively (Diener and Davis 1967; Schindler et al. 1967; Northolt et al. 1976; Simsek et al. 2002). Although the recorded water activities of the samples fits the optimum conditions for aflatoxin formation, in this study aflatoxin was not detected during 2003 and 2004.

Incidence of A. flavus and A. parasiticus

Hazelnut kernels infected with A. flavus and A. parasiticus (AFP) were positive according to the yellow/orange colour of the underside of colonies. The% incidence for AFP in 2002, 2003 and 2004 is given in Table IV for the three regions listed in Table I. Sampling was carried out in hulls until the start of kernel formation. After kernel formation, sampling involved removing the green hulls surrounding the kernel.

Table IV. Incidence for A. flavus/parasiticus (AFP%) in hazelnut samples during 2002, 2003 and 2004.

		AFP% for sampling period
		2002				2003				2004
Region	Orchard no.	1	2	3	4	1	2	3	4	1	2	3	4
East	24	0	6	24	68	1	8	3	56	10	53	26	47
Central	21	3	5	18	73	1	6	0	63	15	36	27	61
West	27	0	7	27	79	3	3	1	63	4	40	39	54
*Sampling periods were selected to coincide with the same period in the three consecutive years, i.e. 1: 5–30 July (flowering stage); 2: 3–19 June (kernel development); 3: 20–30 June (development stage); 4: 4–20 August (harvesting stage).

The% incidence of A. flavus/parasiticus (AFP%) in developing hazelnut samples differed over the three consecutive years, possibly caused by mycoflora alterations and stress/activating factors affecting fungal growth via environmental changes (Choudhary and Sinha 1993; Bayman et al. 2002; Campbell et al. 2003; Magan and Olsen 2004). In general, although the AFP% fluctuates through the sampling period there is an increasing trend from the start of kernel formation over the three consecutive years. The incidence for AFP fluctuates due to the changes in surface flora affected by alterations in ecological and climate conditions (Prado 1991). Temperature and relative humidity in the region was obtained from the Turkish State Meteorological Services General Directorate and average, minimum and maximum values are given in Table V.

Table V. Air temperature and relative humidity in the region (Turkish State Meteorological Services General Directorate).

		Average		Min		Max
Region	Province	RH (%)	Temp (°C)	RH (%)	Temp (°C)	RH (%)	Temp (°C)
West	Düzce	68	13.7	28	−7	95	28.5
	Akcakoca	83	13.2	37	−2	96	27.4
Central	Samsun	77	14.8	31	−2	96	27.2
	Ordu	74	14.9	28	−2	95	28.4
East	Giresun	70	15.1	19	−1	92	27.6
	Trabzon	70	15.2	22	0	95	27.8

Choudhary and Sinha (1993) studied the relationship between A. flavus and competing moulds in maize, reporting that toxigenic A. flavus was mostly isolated in monsoon (63%) and winter periods (52%). Relative humidity of the environment and moisture of the substrate were significantly correlated with A. flavus in maize (Choudhary and Sinha 1993). A negative correlation was observed between A. flavus growth and rainfall. Significant competition was demonstrated between A. flavus, Penicillium spp. and A. niger, and a significant positive correlation was recorded in seasons having higher temperatures (Choudhary and Sinha 1993).

A. flavus and A. parasiticus were identified at the genera level in accordance with morphological characteristics and taxonomic criteria (Samson et al. 1995; Pitt and Hocking 1997). The percentage of A. flavus and A. parasiticus differed in each year (see Table VI) but the percentage of A. flavus was greater than A. parasiticus–89 and 11%, respectively, from 5546 identified samples averaged over 3 years. Systematic studies on the microflora of fresh nuts are limited, but it appears that A. flavus and A. parasiticus have a particular affinity for nuts and oil seeds (Pitt 2000). A. flavus was the dominant fungi isolated from foodstuffs, defined as the primer producer of aflatoxin with a close relationship to the agricultural environment and crop (Hill et al. 1983; Abdel-Hafez and Saber 1993; Abdel-Gawad and Zohri 1993; Bayman et al. 2002; Şimşek et al. 2002; Adebajo and Diyaolu 2003; Arrus et al. 2005b).

Table VI. Number of isolates and % of A. flavus and A. parasiticus in three consecutive years.

Years	Total no. of isolates	No. of A. flavus	%	No. of A. parasiticus	%
2002	2497	2105	84	392	16
2003	1999	1907	95	92	5
2004	1050	943	90	107	10
Total/Average	5546	4955	89	591	11

Aflatoxin production capacity of isolated fungi

Isolates of A. flavus and A. parasiticus were monitored for their toxigenic potential. The percentage of toxigenic fungi differed according to year; however, no significant difference was found among regions (p > 0.05). In 2002 and 2003, toxigenic fungi represented 4% of 1920 identified fungi but was 48% of 1028 identified fungi in 2004. Toxigenic fungi totalled 566 (19%) of the 2948 identified fungi for the three consecutive years.

Toxin-formation potential of some fungi isolated from various foodstuffs is given in Table VII, with the range varying between 5.7 and 55% (Jayaraman and Kalyanasundaram 1990; Mishra and Daradhiyar 1991; Kamphuis et al. 1992; Munimbazi and Bullerman 1996; Vazquez Belda et al. 1996; Freitas Costa and Scussel 2002; Gatti et al. 2003). Kamphuis et al. (1992) determined the percentage of the toxigenic fungi at 5.7% in 35 examined dried maize samples. Munimbazi and Bullerman (1996), in a study of 95 isolates from foodstuffs in Burundi, found that 39% of A. flavus and 100% of A. paraciticus were toxigenic. Similarly, Doster et al. (1994) isolated 11 Aspergillus spp. from Californian pistachios and reported that 65% were toxigenic. The number of isolates (2948) in this study is larger than those listed in Table VII and established that the percentage of toxigenic fungi varied with year–ranging 4-48% over the three consecutive years.

Table VII. Percentage incidence of some toxigenic fungi isolated from various foods.

Food	No. of samples	% Incidence	Reference
Foodstuff from Burundi	95 isolates	39% A. flavus	Munimbazi et al. (1996)
		100% A. paraciticus
Cheese production from Arzua	120 Aspergillus sp.	19%	Vazquez et al. (1996)
California pistachios	11 Aspergillus sp.	65%	Doster et al. (1994)
Brazilian coffee		25% mycotoxin	Freitas Costa et al. (2002)
		13% aflatoxin
Rice bran	34 samples	55% toxigenic	Jayaraman et al. (1990)
Maize	35 samples	5.7%	Kamphuis et al. (1992)

Harvesting techniques

Harvesting methods used in this study were based on the worst conditions operated by farmers and the recommended applications to prevent aflatoxin formation. Hazelnut orchards are located on very steep sites in the Black Sea region of Turkey, between 0 and 1000 m above sea level. Mature hazelnuts naturally drop to the ground and, thus, this harvesting technique is easy and occasionally used in the region; labour costs are low and it is possible to collect all hazelnuts at the same time. However, there is a risk of mould contamination on the surface of hazelnut hulls. In the early manual harvesting technique, labour costs are high and all hazelnuts cannot be collect due to differences in maturity; thus, a number of picking times are requires. Producers sometimes use this technique to sell hazelnuts earlier; however, both mature and immature nuts (low weight/small size hazelnuts) are collected. Schatzki and Pan (1997) studied the distribution of aflatoxin in small pistachios where low weight (small size) might be an indicator of pre-harvest weakness or damage. Smaller nuts showed greater aflatoxin content, but the size dependence was not significant. Harvesting to canvas by shaking is recommended to prevent contact of hazelnuts with the ground; however, it was not practical in very steep orchards due to the confined space. Farmers were recommended to use manual harvesting of mature nuts into small plastic or wooden baskets.

Drying of hazelnuts

In Turkey, hazelnuts are generally dried under the sun by small producers. The problem is that the weather in the Black Sea region is very humid and the harvesting season can be wet. Drying may extent to 2–3 weeks during rain, which increases the risk of mould contamination and consequent formation of aflatoxins. Some farmers collect hazelnuts 1 week earlier in the harvesting period for economic considerations and prefer to use nylon sacks due to ease of availability and cheaper price. However, jute sacks are recommended owing to their high air permeability, but are more expensive than nylon sacks. The hulls of the hazelnuts rot and heat is produced when the sacks are over-packed and kept for a few days before laying under sun to dry. This application is simulated in application no. 1 (see Table II). Due to respiration and reduced air circulation, hazelnuts are become wet and the risk of mould contamination increases. In 2002, jute sacks were used during the applications but aflatoxin levels were not determined in the samples. In 2003, nylon sacks were used instead of jute sacks and although samples were left in them for 10 days (Table II), moisture decreased only 2.27% (from 29.85 to 27.58%).

A common application in the region is the harvesting of hazelnuts from the ground and drying on a concrete floor (Table II); the latter being generally preferred in the region due to the shorter drying time. The concrete floor is heated by the sun which helps dry the hazelnuts. Picking hazelnuts takes a long time in large orchards and hazelnuts may lie on the ground, which are favourable conditions for mould contamination and aflatoxin formation. In 2002, hazelnuts lay on the ground for 3 days; in 2003, they were 7 days on the ground and 3 days in nylon sacks (see Table II). These conditions were designed to observe aflatoxin formation. The hazelnuts, which lay for 7 days on the ground plus 3 days in nylon sacks, partially dried and it appears these hazelnuts had a short drying time, but 10 days is the general holding period before normal drying. The hazelnuts lost some moisture during this period depending on the weather conditions.

Mechanical drying time was significantly less (33 h) than sun-drying (≅110 h) and was independent of weather conditions. Although it is time-saving and has benefits for improved safety of the final product, it is not generally practiced by small farmers in the region due to energy costs compared to traditional sun-drying.

Aflatoxins in hazelnuts samples from post-harvest stages

Duplicate harvesting and post-harvest applications were performed in eastern and western sites of the Black Sea region and the results of post-harvest aflatoxin analyses are given in Table VIII. No aflatoxin was detected in the samples from application nos. 2 and 5 during 2002 and 2003, but contamination was recorded from applications in which hazelnuts were in direct contact with soil. Results showed that hazelnuts harvested from the ground (1.02 ng g⁻¹ total aflatoxin) and lying on the ground for 3 days after the de-hulling process contained aflatoxin (2.18 ± 0.02 ng g⁻¹ B₁ and 3.18 ± 0.03 ng g⁻¹ total aflatoxin). The maximum detected level of total aflatoxin was 3.18 ± 0.03 ng g⁻¹ and for B₁ was 2.18 ± 0.02 ng g⁻¹, which is very close to the EC limits (2.0 ng g⁻¹ for B₁ aflatoxin and 4.0 ng g⁻¹ for total aflatoxin). Direct contact with ground is crucial and any harvesting and drying techniques should prevent direct contact of hazelnuts with the ground.

Table VIII. Aflatoxin analyses during post-harvest.

	Aflatoxin (ng g^−1)
Sample	B1	Total
Application no. 1 (harvest stage 4)	0.42 ± 0.004	0.60 ± 0.01
Application no. 3 (harvest stage 1)	1.02 ± 0.01	1.02 ± 0.01
Application no. 3 (harvest stage 2)	2.18 ± 0.02	3.18 ± 0.03
Application no. 4 (harvest stage 4)	0.26 ± 0.002	0.44 ± 0.004

Early manual-harvested hazelnuts, held in nylon sacks for 10 days and dried on the ground, had a total aflatoxin level of 0.6 ± 0.01 ng g⁻¹. Similarly, hazelnuts harvested after 7 days on the ground, subsequently held in nylon sacks for 3 days and dried on the ground had a total aflatoxin level of 0.44 ± 0.004 ng g⁻¹. Nylon sacks are not recommended as a storage bag due to low air circulation and respiration; the hazelnuts become wet, which enhance mould growth and, consequently, aflatoxin formation. For aflatoxin formation, the effect of direct contact with the ground is clear. On the other hand, the effect of early harvest on aflatoxin formation is uncertain because counting the number of immature nuts in the 200-kg of harvested hazelnuts was not possible and the distribution of immature hazelnuts was not known.

Orchard samples under poorest conditions

In addition to post-harvest applications, sampling under adverse conditions was also carried out. Samples were taken randomly during the drying stage under conditions of high relative humidity (just after or during rainfall) in 2003 (29 samples) and 2004 (50 samples). In 2003, the percentage of aflatoxin-positive samples was 43% with a maximum aflatoxin level of 1.02 ± 0.01 ng g⁻¹. There were no samples over the EC limit (4.0 ng g⁻¹) in 2003. In 2004, the percentage of aflatoxin-positive samples was 16% with a maximum aflatoxin level of 24.36 ± 0.24 ng g⁻¹; 6% of the contaminated samples had levels >4.0 ng g⁻¹ aflatoxins and 10% had <4.0 ng g⁻¹. Water activities of these samples were higher than 0.80 due to the extended drying periods during rain, which increased the risk for aflatoxin formation. The optimum relative humidity conditions for aflatoxin formation are 97–99% (at equilibrium conditions, which correspond to 0.97–0.99 water activity) (Denizel 1976b). It has also been reported that aflatoxin can be formed at water activities of ∼0.82 (Denizel 1976a). FAO (1995) recommends that nuts should be dried to 0.70 water activities in a short period of time and storage conditions should not exceed 70% relative humidity. Therefore, the risk for aflatoxin formation increases during rain and high humidity conditions.

Storage applications

Dried hazelnut samples were stored in controlled (5°C, 65 ± 5% RH) and uncontrolled storage conditions. Aflatoxin was not detected in the controlled storage conditions after 2 years of storage, although aflatoxin-producing fungi were recorded in the stored samples. Uncontrolled storage involved conditions in the region similar to those operated by farmers and traders. Temperature and humidity was recorded throughout the storage period; RH ranged 48–84% in uncontrolled conditions throughout the 2-year storage period.

Aflatoxin was detected in samples that were in direct contact with the ground (soil) during harvesting and drying. The level was 3.18 ± 0.03 ng g⁻¹ for hazelnuts harvested from the ground after 3 days (application no. 3 in Table II). Although increased aflatoxin contamination was expected during uncontrolled storage of samples from the no. 3 application, the level was low (max: 0.34 ng g⁻¹). The reason for the low level of aflatoxin formation in contaminated samples during uncontrolled conditions may be competitive environmental conditions.

Conclusions

Fungal contamination and subsequent production of aflatoxin can occur in hazelnuts in the orchard, at harvest, during post-harvest operations and in storage. Our results indicate that, although the risk of aflatoxin formation is present in the orchard, the most important stages to prevent aflatoxin formation are harvesting and post-harvest, including storage. Aflatoxins detected during sampling from orchards showed a maximum of 0.77 ± 0.08 ng g⁻¹ from 1624 samples (2002–2004). There was no significant difference between regions, altitudes or samples collected from upper and lower branches (p > 0.05). There was an increasing trend in AFP% as the hazelnuts reached maturity; thus, precautions during harvesting and post-harvest are essential in minimizing the aflatoxin contamination risk.

Three harvesting and four drying techniques were studied. Harvesting to a canvas was the recommended collection method as it prevented the contact with the ground. Due to the steepness of the sites, this is not always practical; then manual harvesting of mature hazelnuts into plastic or wooden baskets is recommended. It is important that contact of hazelnuts with the ground is prevented as there is increased risk of contamination for hazelnuts lying on the ground for long periods of time and exposed to humid weather or rain.

Mechanical drying of hazelnuts is the recommended technique to effectively prevent aflatoxin formation. It took ∼33 h to dry 300 kg of hazelnuts from ∼26 to 5% moisture content at 40°C. Sun-drying techniques are dependent on weather conditions and take longer than mechanical drying. Drying on mesh shelves is hygienic and safe but two layers are insufficient for an entire crop; it needs more layers and drying time, which took ∼4 days, weather depending. Drying on the ground is not recommended, as direct contact with the ground results in aflatoxin formation and prolonged drying times–3–3.5 days depending on weather conditions. The maximum detected level of total aflatoxin was 3.18 ng g⁻¹ in post-harvest stages.

The total number of the samples analysed in this study was 2113 over 3 years, including the sampling under orchards, harvesting and storage conditions. A total of 3.6% of the 2113 samples were aflatoxin-contaminated; 3.9% of these samples (max: 24.36 ± 0.24 ng g⁻¹) were over 4.0 ng g⁻¹ total aflatoxin and 96.1% were lower than the EC limit (4.0 ng g⁻¹). Hazelnuts harvested from the ground and allowed to remain on the ground posed the greatest danger; aflatoxin was also detected in hazelnuts kept in nylon sacks and those dried on the ground, but the levels again were lower than the EC limit.

In conclusion, pre-harvest contamination of hazelnuts by aflatoxins is a risk factor; thus, future sampling studies should focus on identifying the factors which increase aflatoxin contamination on the tree. Research into competing natural compounds/organisms for the prevention of aflatoxin contamination on the tree and during storage would also be valuable.

Acknowledgements

The authors gratefully acknowledge the Hazelnut Promotion Group (Turkey) for financial support, MRC researchers and research technicians for their assistance, Local Directorates of the Ministry of Agriculture and Rural Affairs and farmers for providing their orchards for sampling over the three years, cooperation of local authorities during field studies and Giresun Hazelnut Research Institute for the use of their facilities.