Introduction

Selenium is a fundamental dietary component for ensuring human health as there are twenty-five selenoproteins involved in numerous biochemical functions (Bellinger et al. 2009; Rayman 2012). Deficiency and toxicity levels of selenium are narrowly separated with respect to intake rates (COMA 1991; FAO/WHO 2004). Selenium deficiency can result in increased susceptibility of cardiovascular disease, certain types of cancers in addition to reduced fertility and immunity (Bellinger et al. 2009; Rayman 2012). Additionally, reports have suggested that selenium deficiency can result in inadequate thyroid function (FAO/WHO 2004). However, elevated dietary levels of selenium can cause conditions such as selenosis; clinical symptoms include structural changes in hair and nails, irritability, fatigue, skin lesion and abnormalities to the nervous system (FAO/WHO 2004). Evidence has suggested dietary intakes of selenium is insufficient for populations in Ireland and the UK (Bates et al. 2011; Murphy et al. 2002; Rayman 2000). In the UK and Ireland, selenium deficiency is becoming more prevalent, and may be partially explained by a reduction in wheat imports from selenium-rich areas (Adams et al. 2002; Broadley et al. 2006).

Selenium concentration in soil is partially dependent on geological parent material (Levender 1977). However, environmental and soil characteristics, including atmospheric deposition, soil pH, soil organic content, selenium species and the presence of competitive ions, create a dynamic in which a modest difference in key soil characteristics can influence soil selenium concentrations and subsequent transfer to plant material (Alloway et al. 2013; El-Ramady et al. 2015; Fordyce 2013). There is a paucity of research focused on understanding the transfer of selenium from the terrestrial environment to plants and subsequent translocation to higher trophic levels in the agronomic food-chain.

Owing to the dietary consumption patterns in Ireland and the UK, primary dietary sources of selenium include dairy products (Rose et al. 2010; Van Dokkum 1995). In recent years, dairy cows have been housed both during winter and all-year, increasing the intensity of the dairy industry (DAERA 2019). This shift in farm management practices increases reliance on silage and supplementation such as total mixed rations (TMR) where essential vitamins and minerals are regularly incorporated to supplement silage, for which selenium is also under investigated (Wichtel et al. 1996; DAERA 2019). Numerous studies have shown that milk's selenium concentration is sensitive to selenium concentrations provided in fodder to lactating ruminants (Alfthan et al. 2015; Sun et al. 2021; Azorín et al., 2020). Here, a comprehensive analysis of selenium in the dairy agroecosystem was undertaken to establish the extent environmental factors, coupled with husbandry practice, to determine important components in the transfer of selenium into dairy produce. The study includes the six counties of Northern Ireland, to ensure wide geographical coverall and underlying geology. Sampling took place during winter and summer months to include summer pasture verses winter housing husbandry.

Materials and Methods

Sample Collection

To investigate the flow of selenium through dairy agroecosystems, three separate sample collections were completed between January 2016 and December 2017. Two geographical studies were completed in winter (study 1) and summer (study 2), coupled with a detailed study to assess specific husbandry practices (study 3). The first sample collection was conducted on seventy-one dairy farms across Northern Ireland between January and February 2016. From each individual farm, six (50 ml) homogenised milks samples were collected from randomly selected cows. Additionally, a sample of soil and grass were collected from a grazing paddock, and a sample of silage that was provided to the dairy herd at the time of collection. A second survey was conducted in July 2017, where a subset of forty-eight farms were selected. During this sample collection, six (50 ml) homogenised milks were collected from randomly selected cows, in addition to a sample of grass from a grazing paddock. The third sampling campaign was conducted to investigate the milk selenium concentrations of herds housed all year-round. Five farms were sampled every two months for one year from January 2017 to December 2017. This study did not collect or analyse water samples, since reported water selenium concentrations are 10 g/L, therefore unlikely to influence milk selenium levels (WHO, 2011). Further details of the sample collection approach have already been reported in McKernan et al. (2020).

Sample Analysis

Sample preparation and analysis follows that of McKernan et al. (2020) in an associated study on iodine. All samples, milk, soil, silage, and grass were stored in a freezer (− 20 °C) during sample collection period. Milk samples were freeze-dried using a Christ Alpha1-4 LD Plus freeze dryer (Christ, Osterode, Germany) and soil, grass and silage samples were oven dried 70 °C overnight (N6S, Genlab Widnes, England). To obtain a homogenous and representative powder, oven dried silage and grass samples were placed in the Planetary Ball Mill (RERSH, PM100 Ball Mill, Germany) at 500 rpm for 5 min. Oven dried soil was ground using a mortar and pestle, then passed through a 2 mm sieve. The removal of fat from freeze-dried milk samples was necessary for ICP-MS quantification of selenium, freeze-dried milks were then fat extracted by homogenisation in a 2:1 petroleum ether/diethyl ether (BDH Dorset, Poole), creating a powdered milk.

Powdered milk, soil grass, and silage samples were accurately weighed (~ 100 mg) to four significant places in 50 ml polypropylene centrifuge tube. Additionally, relevant Certified Reference Materials (CRM) triplicate for each instrument run accurately weighed to (~ 100 mg), CRM for skimmed milk powder, ERM BD151; soil, NCS ZC73007; and citrus leaf, NCS ZC73018. Then 1 ml of 5% (w/v) tetramethylammonium hydroxide (TMAH) solution (Sigma-Aldrich, USA) was added to each sample and three empty tubes for assigned blanks and allowed to soak overnight. The following day, 4.5 ml of deionised water > 18.2MΩ.cm (at 25 °C) was added and placed in a microwave for a 50-min cycle to assist sample digestion (Mars 6 240,150, CEM Microwave Technology) To the cooled digestate, 10 µl of internal standard rhodium was added (Fluka Analytical, Sigma-Aldrich, USA), and volume was made up to 10 ml using standardised deionised water. The Limit of Detection (LoD) was calculated as the average concentration in the blank + 3 times the standard deviation for element mix. The milks digestate were centrifuged and 3 ml if the digestate was filtered using a cellulose steripop TM 0.45 µm Millipore express filter unit (Millex, Millipore, USA), then 3 mL of 0.5% (w/v) was added to the filtered digestate to create a one in two dilution. Similarly, a 1 in 10 dilution of the soil, silage and grass samples were prepared by centrifuging samples, adding 1 mL of sample digestate to 9 mL 0.5% TMAH. Selenium concentration in the digests was quantified by inductively coupled plasma-mass spectrometry (ICP-MS) (Thermo Scientific Icap Q, Thermo Scientific, USA).

To determine the organic matter of soil, loss of ignition (LoI) was utilised, where 5 g (± 1 g) was weighed into a crucible and weight recorded. The weighed soil sample was then placed into a muffle furnace for 3 h at 500 °C (Carbolite, England). The ashed sample was allowed to cool in a desiccator and the ashed sample and crucible was reweighed and recorded to determine organic matter content. To determine the pH content of the soils, 2 g of soil was weighted into a 50 ml polypropylene tube. Following this, 10 ml of standard deionised water was added to the tube and samples were vortexed at 1,500 rpm for 15 min (VWR, DVX-2500 Multitube Vortexes), pH was determined using VWR, Phenomenal pH 100H instrument.

Statistics

All data were analysed using GraphPad Prism (version 8.11, USA), with a p-value p ≤ 0.05 considered to be significant. Preliminary statistical checks found that the data set was not normally distributed, therefore , non-parametric statistical tests were run on winter, summer and husbandry sampling collection including, Mann Whitney, Kruskal Wallis, One-way ANOVAs and Two-way ANOVAS. Additionally, all data sets were log-transformed to run linear regression analysis on variables.

Results

The selenium concentration in the CRMs and samples were above the LoD. Furthermore, milk, soil, grass and silages recovered for CRMs was 91%, 88% and 94%, respectively (Table 1). A ten-fold variation in soil selenium concentration was observed, ranging from 0.15 to 1.33 mg/kg DM, with a median value of 0.46 mg/kg DM. Linear regression with location variables, latitude, and longitude, indicated soil selenium concentrations were not influenced by proximity to the marine environment. Soil selenium concentrations generally correlated with those provided in the Tellus map (Fig. 1), the soil selenium concentrations Northern Irish Tellus Project range between < 0.2–7.8 mg/kg−1 DM and a median soil selenium concentration 0.7 mg/kg−1DM (BGS, 2019). Moreover, high soil selenium concentrations are found in mountainous regions (Fig. 1). Similar findings were found in this study, the highest soil selenium concentrations are predominantly reported in coastal mountainous region in County Antrim with a median soil selenium concentration 0.69 mg/kg DM ranging from 0.29–1.33 mg/kg DM, and County Down with a median soil selenium concentration 0.66 mg/kg DM ranging from 0.16–1.15 mg/kg DM (Figs. 2 and 3). In Count Armagh, despite the Tellus map identifying this as a region with scattered high selenium concentrations (BGS, 2019), in this study the soil selenium concentrations were lower ranging from 0.15 to 0.79 mg/kg DM, with a median concentration 0.34 mg/kg DM.

Table 1 Selenium recoveries from Certified Reference Material (CRM)
Full size table
Fig. 1
figure 1

Selenium distribution for Northern Irish from the concentration data in the BGS database (BGS, 2019), overlaid with sample locations for the environmental sampling campaign. Circles with cross are for summer pasture and winter housed, filled triangles are for winter

Full size image
Fig. 2
figure 2

Linear regression scatter graphs to demonstrate statistically significant relationships between various variables for pastures

Full size image
Fig. 3
figure 3

Selenium concentration (mg/kg DM) of soil (hexagonal) (n = 71), silage (diamonds) (n = 71), grass (triangles winter, inverted triangles following summer) (n = 71), winter milk just winter housed (open squares) (n = 71), winter milk all-year housed (closed dark blue squares) (n = 30), summer pasture-fed milk (open circles) (n = 48)

Full size image

In this study, the LoI of the soil ranged between 6–34%, linear regression analysis indicated that soil LoI has a statistically significant positive relationship (P = 0.004) with soil selenium (Fig. 2). Additionally, soil pH concentrations ranged from 4.2- 6.9, linear regression analysis indicated a negative statistically significant relationship (P = 0.0117) observed between soil selenium and soil pH (Fig. 2). Linear regression analysis indicated positive significant correlation between soils and winter grass (P = 0.003) (Fig. 2), however, no relationship was observed for summer grass. The selenium concentration in winter grass ranged from 0.01 to 0.16 mg/kg DM with a median of 0.04 mg/kg DM. Furthermore, selenium concentrations in winter grass were ~ tenfold lower than correlating soils. The selenium concentrations of summer grass were ~ 2-times higher than winter grass and ranged between 0.05 to 0.30 mg/kg with a median value 0.07 mg/kg DM. A Mann–Whitney test found that the selenium concentrations in summer and winter grass to be statistically different (P < 0.0001) (Fig. 3). Further linear regression analysis did not find a statistically significant relationship between summer or winter grass. The selenium concentrations in silage ranged from 0.01 to 0.39 mg/kg DM with a median of 0.03 mg/kg DM. Mann–Whitney analysis indicated that selenium concentrations in silage were significantly lower in comparison to summer and winter grass selenium concentrations (p- < 0.0001 and p-0.002), respectively (Fig. 3). Additional linear regression analysis found no statistically significant relationships between silage and summer grass, winter grass and soil selenium concentrations.

Selenium concentrations were reported in milk as DM, the selenium concentration of subsequent milks is reported in brackets as fresh weight (FW). There was no significant difference found between milk selenium concentrations and any other environmental variables (region, soil characteristics, grass and silage). Winter milk selenium concentrations ranged from 0.161–0.393 mg/kg DM (21 to 51 µg/kg FW), and median summer milk selenium concentrations ranged from 0.101–0.560 mg/kg DM (13 to 73 µg/kg FW). In addition, there was no significant correlation between winter and summer milk selenium concentrations. Mann–Whitney analysis found that the selenium concentrations of winter milks elevated ~ 20%, in comparison to summer milk samples collected from the same herds (p-0.0002), with median values reported 0.25 and 0.21 mg/kg DM (33 to 27 µg/kg FW), respectively (Fig. 3). Furthermore, the winter milks produced by herds that are housed all-year round and those that are housed only during winder show no statistical difference. One-way ANOVA of the individual milk’s selenium concentrations at individual farm level reported a significantly different milk selenium concentrations (p- < 0.0001), moreover the variation of milk selenium concentrations at individual farm levels was deduced via the relative standard deviation (RSD) as a percentage. The RSD in summer milk selenium concentrations ranged from 5 to 43% with a median of 13%, and the RSD in winter milks ranged from 4 to 39% median 10%, with a Mann–Whitney test reporting higher variation (RSD) in summer milk samples in comparison to winter (P = 0.0028). Further linear regression analysis found no statistical difference between the milk selenium concentrations at farm level. A two-way ANOVA reported that seasonal variation was significant, however, the individual farm milk selenium concentrations and the interactions between farm and season were not significantly associated. To understand the milk selenium concentrations from the same farms during summer and winter months, median milk selenium concentrations were ranked, and ranges were plotted for both sample collections (survey 1 and survey 2) (Fig. 4). This showed that for 14 farms with the lower median selenium concentrations reported, the corresponding median winter milk selenium concentrations varied substantially. Moreover, a greater milk selenium variation was observed for these farms, with limited or no overlap of range within the herd on the same farm. Furthermore, it is worth noting that in this section of the graph winter median milk selenium concentrations are consistently higher than summer milk median concentrations. However, farms with median milks selenium concentrations of 0.21 mg/kg (27 µg/kg FW) or above, the median milk selenium concentrations for winter and summer milk are more comparable, with ranges overlapping more frequently. On 12 occasions the summer milk selenium median was higher than corresponding winter milk selenium median. Interestingly, the farm that that the highest summer milk selenium median reported, had 11th lowest milk selenium median during winter from the same herd.

Fig. 4
figure 4

Selenium concentration (mg/kg DM) in winter and subsequent summer milk ranked in order of summer milk concentrations. Blue squares are the winter milk median per farm, and winter farm data range is illustrated by the black lines; green circles are the summer medians, while the grey shading outlines the summer range

Full size image

Discussion

The soil selenium concentrations reported here for Northern Irish pastures, having a range between 0.15 and 1.3 mg kg−1 DM and a median soil selenium concentration 0.5 mg kg−1 DM and were comparable to those reported by the Northern Irish Tellus Project having a range between < 0.2—7.8 mg kg−1 DM and a median soil selenium concentration 0.7 mg kg−1 DM, a baseline survey of NI topsoils at a sampling density of 2 km2 (BGS, 2019). Similar soil selenium concentrations were reported in Irish and UK studies, primarily reporting a range between 0.1 and 2.5 mg kg−1, with the exception of elevated soil selenium concentrations reported in seleniferous regions (Teagasc 2022; Broadley et al. 2006; Shand et al. 2012). Tan (1989) categorised soils in relation to total soil selenium concentrations, as an indicator of the abundance and status of selenium in the terrestrial environment, regarding this is an important consideration for selenium transfer through the food-chain. Selenium concentrations in soils were categorised as selenium: deficient; marginal; sufficient, rich and excessive. Comparing the selenium results of the current data observed in this study to these parameters, it can be said that at a minimum Northern Ireland can be described as a selenium sufficient environment, as > 90% of soils containing soil selenium levels within or above this range (> 0.175 mg kg−1). High soil selenium concentrations correspond to the mountainous regions of Northern Ireland (BGS, 2019), which are primarily high rainfall districts with organic soils, predominately peat, and therefore, atmospheric deposition through elevated rainfall reported in these regions may partially explain the selenium concentrations in these regions (Sweeney 2014; Sun et al. 2016).

Studies in Norway and Sweden reported proximity to the marine environment influenced soil selenium concentrations, suggesting elevated soil selenium concentrations in these regions could be attributable to atmospheric deposition from sea (Shand et al. 2010; Wu and Lag. 1988). This contrasted with the data in this study as soil selenium concentrations did not correlate with proximity to the marine environment (P > 0.05), however coastal mountainous regions in counties Antrim and Down exhibited a higher median selenium concentration in comparison the other counties. Moreover, even though the Tellus map identifying County Armagh having high soil selenium concentrations, in this study the soil selenium concentrations were lower in this region. This may be attributable to period of sample collection during the winter months, where selenium is lost from soil via leaching (Li et al. 2017), or simply due to the fact that Tellus surveyed all soils based on a sampling grid, while the current study only focused on pasture soils. However, Northern Ireland is a generic maritime environment with prevailing winds derived from the Atlantic Ocean.

For this study, soil selenium concentration correlates with soil organic content matter, agreeing with literature studies (Supriatin et al. 2015; Tolu et al. 2014). Previous investigations have demonstrated that that selenium bound to organic material can immobilise selenium in the terrestrial environment, consequently, reducing the bioavailability of selenium for soil–plant transfer (Johnson et al. 1991; Li et al. 2017; Wang et al. 2013). This current study found that the selenium concentration in soil decreased with increasing soil pH. This negative correlation between total soil selenium concentration and pH was also observed in other studies (Xing et al. 2015; Dhillon and Dhillon 1999; Xu et al. 2018). Xing et al. (2015) reported that selenium bioavailable fractions (water soluble and exchangeable fractions) in the soil positively correlated significant with soil pH, despite the reduction in total soil selenium. Literature studies have indicated that the selenium species selenate dominates under oxidising conditions, elevating bioavailability, and mobility (Shaheen et al. 2017; DeLaune and Seo 2011; He et al. 2010). Selenium in soil can be lost from the terrestrial environment via the volatisation into the atmosphere or leached due to rainfall, due to the elevated abundance of water-soluble selenate in well aerated soils in comparison to selenite (Li et al. 2017; Xing et al. 2015; Xu et al. 2018).

The selenium concentrations in forage here were comparable with other non-seleniferous Irish pastures which ranged from 0.02–0.50 mg kg−1 (Teagasc 2022). The results of the current study indicated that increased soil and winter grass selenium concentration are positively correlated. This agrees with other studies (Fordyce et al. 2010; Fordyce 2013). The uptake of selenium to plants can depend on various factors including, the species of selenium, soil characteristics affecting selenium bioavailability/motility, and competitive ions (Gupta and Gupta 2000; Liu et al. 2015; Haug et al. 2007; Zhu et al. 2009). Selenate present in soil has a greater affiliation to plant material in comparison to selenite (Gupta and Gupta 2000; Liu et al. 2015), probably attributable to research observing selenate as the dominant species of selenium in soluble fractions in the environment. Here, the selenium concentrations in summer grasses were significantly higher in comparison to winter grasses. This is likely explained by the extended dry periods during summer months, increasing the presence of soluble selenium forms selenate, reduced loss via leaching from persistent rainfall coupled with increased aeration from field management and ploughing during these months (Teagasc 2022). Whereas reduced selenium concentration in winter grasses may be due to poorer availability.

Selenium transfer to herbage is not only important to understand the transmission to dairy products, but also to maintain the health of the herd (Fordyce 2013). Studies recommended minimum forage selenium concentrations to be above 0.1 mg kg−1 DM to mitigate symptoms deficiency occurring livestock (Fordyce 2013; Levender 1977, Yara 2022; Teagasc 2022). The selenium concentrations for the majority (> 80%) forage samples in this study were considered deficient, with similar figures reported in other studies (YARA 2022; Teagasc 2022). This insufficient selenium concentration found in forage is an interesting finding, not meeting the requirements for grazing livestock despite the selenium concentrations in the soil being classified as sufficient with respect to selenium concentrations. Thus, feed supplementation, as observed here, is an obvious way of improving herd health.

The milk selenium concentrations in this study ranged from 13 to 73 µg/kg FW. Previous Spanish, German, Australian and Northern Irish studies reported similar selenium concentrations in unpasteurised to range from 6.79–47.6 µg kg−1 FW. (Lomeback 1978; Rodriguez et al. 2005; Tinggi et al. 2001; O’Kane et al. 2018). In the current study milk selenium concentrations did not significantly correlate with any of the other parameters measured. However, seasonal variation was found, with significantly higher milks selenium levels reported in winter in comparison to summer months. Similar differences were observed in Norwegian and Dutch studies, with elevated selenium concentration in winter months in comparison to summer milk selenium concentrations 17 and 11 µg kg−1, 16.5 and 10.3 µg kg−1, respectively (Frøslie et al. 1985; Koops et al. 1989). This seasonal variation is likely attributable to the agricultural practices in Europe, where herds are generally housed indoors during winter as grass yield is diminished during these months. Thus, the herds exposure to dietary supplementation in the form TMR routinely added to silages and meal is greater than in summer months where a herd is out on pasture (March et al. 2014). Feed supplementation during summer, pasture feeding months, may benefit both herd heath and the utility of dairy produced as it will be enhanced in selenium.

Numerous studies suggested that milk selenium concentrations were a sensitive indicator of selenium application to agricultural land (Alfthan et al. 2015; Sun et al., 2021; Azorín et al., 2020). More recently studies have found that supplementation with antioxidants, selenium, vitamin and sunflower oil can significantly improve the nutritional profile of milk, specifically milk selenium and improve cow health (Salles et al., 2022a, b). Furthermore, numerous studies investigated the effect of selenium supplementation on milk selenium concentrations in controlled conditions. Investigations of this nature demonstrate that supplementation in dairy herds can increase selenium concentration in a dose-dependent manner and highlights the profound influence of different selenium supplementation to the consequential derived food produced, in particular for milk selenium concentrations (Givens et al. 2004; Juniper et al, 2006; Grace et al. 2001; Heard et al. 2007). However, housed conditions contrast with natural pasture managed herds were with external variables, such as fluctuating weather conditions, which may profoundly influence operating procedures. Variable milk selenium concentrations were evident in this study as one-way ANOVA, on individual milks selenium concentration at farm level reported that the selenium concentrations varied significantly, and the RSD ranged from 4 to 43%. This is an interesting finding as the lactating cows within a dairy herd are generally exposed to the same environment and feed, therefore, theoretically should produce consistent milk selenium concentrations. Therefore, this suggests that farmers within Northern Ireland employ variable feeding practices. The significant variability of milk selenium concentrations may be attributable to farm management practices, including inconsistent mixing of TMR, change in brand/source of TMR due to competitive pricing and or availability resulting in some cow’s receiving a more concentrated or diluted TMR ration than other members of the herd. A study by McCarthy et al., (2021) reported variable mineral intake levels among cattle, with higher selenium levels in the livers of ‘high-intake’ cows compared to ‘low intake’ cows. Thus, milk selenium concentrations could be influenced by cow feeding behaviour. These hypothesised reasons for inconsistent TMR could mean a single herd may be exposed to several variants of supplementation at any time, and this would be reflected in milk selenium concentrations as previously discussed. The fluctuating milk selenium concentrations during summer months, may be attributable to adverse weather conditions, where herds are brought inside for periods of time therefore fodder would vary between TMR and grazed. More uniform management practise with respect to selenium supplementation could be used to enhance milk selenium and decrease the variation observed.

Milk is a considered a primary source of selenium, yet the flows of selenium through agroecosystems is not well understood. To the authors knowledge, this novel study is the first to investigate the flows of selenium through dairy agroecosystems to establish the extent environmental factors, together with husbandry practice, influence the transfer of selenium into dairy produce. Results from this study report soil selenium concentrations are influenced by soil characteristics and reflect in the grass selenium concentration. Despite the strong environmental relationship, a disconnect in relation to milk selenium concentration and the agricultural land, similar findings were reported investigating iodine in agronomic environment (McKernan et al. 2020). The variable milk selenium concentrations reported in this study is attributable to inconsistent farm management practices where the fodder provided to herds is variable. Understanding selenium flows through the food -chain is multifaceted and complex. Therefore, to alleviate variable milk selenium concentrations, consistent feeding practices should be encouraged to ensure milk selenium concentrations occur between the narrow range existing for deficiency and toxicity. This would also improve herd health as well as leading to a higher quality, i.e. enhanced fortification with selenium, of milk produced.