How food matrices modulate folate bioaccessibility: A comprehensive overview of recent advances and challenges.

The incomplete absorption of dietary folate makes it crucial to understand how food matrices affect folate bioaccessibility. Bioavailability encompasses bioaccessibility, which depicts the proportion that is liberated from the food matrix during digestion and becomes available for absorption. Bioavailability studies are expensive and difficult to control, whereas bioaccessibility studies utilize in vitro digestion models to parameterize the complex digestion, allowing the evaluation of the effect of food matrices on bioaccessibility. This review covers the folate contents in various food matrices, the methods used to determine and the factors affecting folate bioaccessibility, and the advances and challenges in understanding how food matrices affect folate bioaccessibility. The methods for determining bioaccessibility have been improved in the last decade. Current research shows that food matrices modulate folate bioaccessibility by affecting the liberation and stability of folate during digestion but do not provide enough information about folate and food component interactions at the molecular level. In addition, information on folate interconversion and degradation during digestion is scant, hindering our understanding of the impact of food matrices on folate stability. Moreover, the role of conjugase inhibitors should not be neglected when evaluating the nutritional value of food folates. Due to the complexity of food digestion, holistic methods should be applied to investigate bioaccessibility. By synthesizing the current state of knowledge on this topic, this review highlights the lack of in-depth understanding of the mechanisms of how food matrices modulate folate bioaccessibility and provides insights into potential strategies for accurate evaluation of the nutritional value of dietary folate.

Flavor challenges in extruded plant-based meat alternatives: A review.

Demand for plant-based meat alternatives has increased in recent years due to concerns about health, ethics, the environment, and animal welfare. Nevertheless, the market share of plant-based meat alternatives must increase significantly if they are to support sustainable food production and consumption. Flavor is an important limiting factor of the acceptability and marketability of plant-based meat alternatives. Undesirable chemosensory perceptions, such as a beany flavor, bitter taste, and astringency, are often associated with plant proteins and products that use them. This study reviewed 276 articles to answer the following five research questions: (1) What are the volatile and nonvolatile compounds responsible for off-flavors? (2) What are the mechanisms by which these flavor compounds are generated? (3) What is the influence of thermal extrusion cooking (the primary structuring technique to transform plant proteins into fibrous products that resemble meat in texture) on the flavor characteristics of plant proteins? (4) What techniques are used in measuring the flavor properties of plant-based proteins and products? (5) What strategies can be used to reduce off-flavors and improve the sensory appeal of plant-based meat alternatives? This article comprehensively discusses, for the first time, the flavor issues of plant-based meat alternatives and the technologies available to improve flavor and, ultimately, acceptability.

B Vitamins in Legume Ingredients and Their Retention in High Moisture Extrusion.

Legumes have been recognised as healthy and environmentally friendly protein sources. Knowledge about the vitamin B contents in legume ingredients and extrudates is scarce. In this study, we investigated thiamin, riboflavin, niacin, and folate in various faba bean, lupin, and pea ingredients. Further, the retention of B vitamins in high moisture extrusion was studied. Prior to liquid chromatographic determinations of thiamin, riboflavin, niacin, and folate, vitamins were extracted by acid hydrolysis (niacin), enzymatic treatment (folate), or their combination (thiamin and riboflavin). The contents (on a dry matter basis) varied greatly among different ingredients: the thiamin content was 0.2-14.2 µg/g; riboflavin, 0.3-5.9 µg/g; niacin, 8.8-35.5 µg/g, and folate, 45-1453 ng/g. Generally, the highest levels were in flours and protein concentrates, whereas low levels were observed in isolates. The retention of B vitamins was excellent in high moisture extrusion, except for folate in faba bean, where the folate contents were 42-67% lower in the extrudates than in the respective ingredient mixtures. In terms of both vitamin B contents and their retention, extrudates containing substantial amounts of flour or protein concentrate are promising plant-based sources of thiamin, riboflavin, niacin, and folate.

The bioaccessibility of folate in breads and the stability of folate vitamers during in vitro digestion.

Both the liberation and stability of endogenous folate are relevant to the bioaccessibility of folate. Since folates are unstable, in addition to studying the natural folate content in foods, bioaccessibility should be considered. To understand folate changes during digestion, a mixture of standard folate compounds was subjected to a static in vitro gastrointestinal digestion assay. Next, different types of bread were analysed to study how food matrices influence folate bioaccessibility. Folates were identified and quantitated by a UHPLC-PDA/FL method. Folic acid and 10-formylfolic acid were stable throughout the digestion, and the conversions among formyl folates and 5,10-methenyltetrahydrofolate were triggered at the gastric phase. Tetrahydrofolate began to degrade during the oral phase and was lost completely during the gastric phase. During the intestinal phase, 5-methyltetrahydrofolate began to degrade and suffered a 60% loss. With bread matrices, folate conversions and the decrease of reduced folates were also common, but the extent of changes varied. Generally, rye breads had the highest (80-120%) bioaccessibility of folate, while oat breads had the lowest (31-102%). The high proportion of 5-methyltetrahydrofolate could result in low bioaccessibility because of its relatively low stability during digestion in bread matrices. An increase in 10-formylfolic acid content was observed for all the breads, but 10-formyldihydrofolate seemed to be more stable in rye breads than in oat and wheat breads. The results showed that folates undergo significant changes during digestion and that food matrices could be modified to affect these changes towards better folate bioaccessibility.

5-Methyltetrahydrofolate Is a Crucial Factor in Determining the Bioaccessibility of Folate in Bread.

This study investigated the bioaccessibility of folate in wheat bread baked with different ingredients and processing methods. Next, different matrices were spiked with 5-methyltetrahydrofolate, gallic acid (GA), or both to investigate the stability of 5-methyltetrahydrofolate during in vitro digestion. The folate bioaccessibility in bread varied from 44 to 96%. The inclusion of whole-grain or faba bean flour significantly improved both folate content and bioaccessibility. Baking with yeast increased the folate content by 145% in bread but decreased folate bioaccessibility compared to the bread without added yeast because of the instability of 5-methyltetrahydrofolate. Spiking experiments confirmed oxidation as a critical reason for 5-methyltetrahydrofolate loss during digestion. However, GA protected this vitamer from degradation. Additionally, 5-methyltetrahydrofolate was less stable in whole-grain wheat matrices than other matrices. This study demonstrated that the stability of 5-methyltetrahydrofolate is crucial for folate bioaccessibility in bread, and methods for stabilizing this vitamer should be further studied.

Bioaccessibility of folate in faba bean, oat, rye and wheat matrices.

Cereals and legumes are rich in folate. However, due to the instability of folate, processing and digestion can induce significant folate loss. In this paper, folate bioaccessibility of faba bean, oat, rye and wheat flours and pastes was studied using a static in vitro digestion model. Folate bioaccessibility depended on food matrices, varying from 42% to 67% in flours and from 40% to 123% in pastes. Digestion was associated with the interconversion of formyl folates, as well as the increase of oxidised vitamers and decrease of reduced vitamers. Especially in faba bean, 5-methyltetrahydrofolate showed surprisingly good stability both in digestion and heat treatment, resulting in high bioaccessibility. The physiological concentration of ascorbic acid did not stabilise folate in digestion; however, a higher level helped to maintain reduced vitamers. Heat treatment (10-min paste making) could improve folate bioaccessibility by liberating folate from the food matrices and by altering folate vitamer distribution.

Collaborative study: Quantification of total folate in food using an efficient single-enzyme extraction combined with LC-MS/MS.

Quantification of the specific folate vitamers to estimate total folate in foods is not standardized. A collaborative study, including eight European laboratories, was conducted in order to determine the repeatability and reproducibility of the method for folate quantification in foods using the plant-origin γ-glutamyl hydrolase as part of the extraction procedure. The seven food samples analyzed represent the food groups; fruits, vegetables, dairy products, legumes, offal, fish, and fortified infant formula. The homogenization step was included, and six folate vitamers were analyzed using LC-MS/MS. Total folate content, expressed as folic acid equivalent, was 17-490 µg/100 g in all samples. Horwitz ratio values were within the acceptable range (0.60-1.94), except for fish. The results for fortified infant formula, a certified reference material (NIST 1869), confirmed the trueness of the method. The collaborative study is part of a standardization project within the Nordic Committee on Food Analysis (NMKL).

Lactobacillus plantarum P2R3FA Isolated from Traditional Cereal-Based Fermented Food Increase Folate Status in Deficient Rats.

Folate deficiencies are widespread around the world. Promoting consumption of folate-rich foods could be a sustainable option to alleviate this problem. However, these foods are not always available. Cereals, being a staple food, could contribute to folate intake. They are fermented prior to consumption in many African countries, and fermentation can modify the folate content. In Ethiopia, injera is a widely consumed fermented flat bread. The main drivers of its fermentation are lactic acid bacteria (LAB). The aim of this work was to isolate and identify folate-producing LAB from injera fermented dough and to evaluate their ability to increase folate status after depletion in a rat model. Among the 162 strains isolated from 60 different fermentations, 19 were able to grow on a folate-free culture medium and produced 1 to 43 µg/L (24 h, 30 °C incubation). The four highest folate producers belonged to the Lactobacillus plantarum species. The most productive strain was able to enhance folate status after depletion in a rat model, despite the relatively low folate content of the feed supplemented with the strain. Folate-producing L. plantarum strain has potential use as a commercial starter in injera production.

In situ production of active vitamin B12 in cereal matrices using Propionibacterium freudenreichii.

The in situ production of active vitamin B12 was investigated in aqueous cereal-based matrices with three strains of food-grade Propionibacterium freudenreichii. Matrices prepared from malted barley flour (33% w/v; BM), barley flour (6%; BF), and wheat aleurone (15%; AM) were fermented. The effect of cobalt and the lower ligand 5,6-dimethylbenzimidazole (DMBI) or its natural precursors (riboflavin and nicotinamide) on active B12 production was evaluated. Active B12 production was confirmed by UHPLC-UV-MS analysis. A B12 content of 12-37 µg·kg-1 was produced in BM; this content increased 10-fold with cobalt and reached 940-1,480 µg·kg-1 with both cobalt and DMBI. With riboflavin and nicotinamide, B12 production in cobalt-supplemented BM increased to 712 µg·kg-1. Approximately, 10 µg·kg-1 was achieved in BF and AM and was increased to 80 µg·kg-1 in BF and 260 µg·kg-1 in AM with cobalt and DMBI. The UHPLC and microbiological assay (MBA) results agreed when both cobalt and DMBI or riboflavin and nicotinamide were supplemented. However, MBA gave ca. 20%-40% higher results in BM and AM supplemented with cobalt, indicating the presence of human inactive analogues, such as pseudovitamin B12. This study demonstrates that cereal products can be naturally fortified with active B12 to a nutritionally relevant level by fermenting with P. freudenreichii.

Effects of yeasts and bacteria on the levels of folates in rye sourdoughs.

Fermentation of rye dough is often accompanied with an increase in folate content. In this study, three sourdough yeasts, Candida milleri CBS 8195, Saccharomyces cerevisiae TS 146, and Torulaspora delbrueckii TS 207; a control, baker's yeast S. cerevisiae ALKO 743; and four Lactobacillus spp., L. acidophilus TSB 262, L. brevis TSB 307, L. plantarum TSB 304, and L. sanfranciscensis TSB 299 originally isolated from rye sourdough were examined for their abilities to produce or consume folates. The microorganisms were grown in yeast extract-peptone-d-glucose medium as well as in small-scale fermentations that modelled the sourdough fermentation step used in rye baking. Total folate contents were determined using Lactobacillus rhamnosus (ATCC 7469) as the growth indicator organism. The microorganisms studied did not excrete folates into the media in significant amounts. Yeasts increased the folate contents of sterilised rye flour-water mixtures from 6.5 microg/100 g to between 15 and 23 microg/100 g after 19-h fermentation, whereas lactic acid bacteria decreased it to between 2.9 and 4.2 microg/100 g. Strains of Lactobacillus bulgaricus, L. casei, L. curvatus, L. fermentum, L. helveticus, Pediococcus spp., and Streptococcus thermophilus that were also tested gave folate contents after fermentation that varied between 2 and 10.4 microg/100 g. Although the four Lactobacillus spp. from sourdough consumed folates their effect on folate contents in co-cultivations was minimal. It was concluded that the increase of folate content during fermentation was mainly due to folate synthesis by yeasts. Fermentation of non-sterilised flour-water mixtures as such resulted in three-fold increases in the folate contents. Two folate producing bacteria were isolated from the non-sterilised flour and identified as Enterobacter cowanii and Pantoea agglomerans.

Effect of germination and thermal treatments on folates in rye.

Effects of germination conditions and thermal processes on folate contents of rye were investigated. Total folate contents were determined microbiologically with Lactobacillus rhamnosus (ATCC 7469) as the growth indicator organism, and individual folates were determined by high-performance liquid chromatography after affinity chromatographic purification. Germination increased the folate content by 1.7-3.8-fold, depending on germination temperature, with a maximum content of 250 micro g/100 g dry matter. Hypocotylar roots with their notably high folate concentrations (600-1180 micro g/100 g dry matter) contributed 30-50% of the folate contents of germinated grains. Germination altered the proportions of folates, increasing the proportion of 5-methyltetrahydrofolate and decreasing the proportion of formylated folate compounds. Thermal treatments (extrusion, autoclaving and puffing, and IR and toasting) resulted in significant folate losses. However, folate levels in grains that were germinated and then were heat processed were higher than for native (nongerminated) grains. Opportunities to optimize rye processing to enhance folate levels in rye-based foods are discussed.

Stability of added and in situ-produced vitamin B12 in breadmaking.

Vitamin B12 exists naturally in foods of animal origin and is synthesised only by certain bacteria. New food sources are needed to ensure vitamin B12 intake in risk groups. This study aimed to investigate the stability of added cyanocobalamin (CNCbl, chemically modified form) and hydroxocobalamin (OHCbl, natural form) and in situ-synthesised vitamin B12 in breadmaking. Samples were analysed both with a microbiological (MBA) and a liquid chromatographic (UHPLC) method to test applicability of these two methods. Proofing did not affect CNCbl and OHCbl levels. By contrast, 21% and 31% of OHCbl was lost in oven-baking steps in straight- and sponge-dough processes, respectively, whereas CNCbl remained almost stable. In sourdough baking, 23% of CNCbl and 44% of OHCbl were lost. In situ-produced vitamin B12 was almost as stable as added CNCbl and more stable than OHCbl. The UHPLC method showed its superiority to the MBA in determining the active vitamin B12.

Ultra-high performance liquid chromatographic and mass spectrometric analysis of active vitamin B12 in cells of Propionibacterium and fermented cereal matrices.

A sensitive and selective method is needed to analyse in situ produced vitamin B12 in plant-based materials, potential new dietary sources of vitamin B12. A UHPLC/UV method was developed and validated for the determination of human active vitamin B12 in cell extracts of Propionibacterium freudenreichii subsp. shermanii and after immunoaffinity purification in extracts of cereal matrices fermented by P. freudenreichii. An Acquity HSS T3 C18 column resulted in a baseline separation, a calibration curve of excellent linearity and a low limit of detection (0.075 ng/5 µL injection). As confirmed by UHPLC-MS, the active vitamin B12 could be separated from pseudovitamin B12. The recovery of vitamin B12 from purified spiked cereal matrices was good (>90%; RSD<5%). A nutritionally relevant amount of active vitamin B12 was produced by P. freudenreichii in cereal malt matrices (up to 1.9 µg/100 g) in 24h at 28 °C.

In situ enrichment of folate by microorganisms in beta-glucan rich oat and barley matrices.

The objective was to study folate production of yeast strains, bacteria isolated from oat bran, and selected lactic acid bacteria as well as one propionibacterium in oat and barley based models. Simultaneously, we aimed at sustaining the stability of viscosity, representing the physicochemical state of beta-glucan. Total folate contents were determined microbiologically and vitamers for selected samples by UHPLC. Folate in yeast cells comprised mainly 5-methyltetrahydrofolate and tetrahydrofolate. Folate production by microbes in YPD medium was different to that in cereal fermentations where vitamers included 5-methyltetrahydrofolate, 5,10-methenyltetrahydrofolate and formylated derivatives. Microbes producing significant amounts of folate without affecting viscosity were Saccharomyces cerevisiae ALKO743 and Candida milleri ABM4949 among yeasts and Pseudomonas sp. ON8 and Janthinobacterium sp. RB4 among bacteria. Net folate production was up to 120 ng/g after 24 h fermentation and could increase during 2-week storage. Glucose addition increased the proportion of 5-methyltetrahydrofolate. Streptococcus thermophilus ABM5097, Lactobacillus reuteri, and Propionibacterium sp. ABM5378 produced folate but in lower concentrations. Both endogenous and added microbes contribute to folate enhancement. Selection of microbes with folate producing capability and limited hydrolytic activity will enable the development of products rich in folate and beta-glucan.

Effect of soybean processing on content and bioaccessibility of folate, vitamin B12 and isoflavones in tofu and tempe.

PURPOSE: To compare the content of bioaccessible folate, vitamin B12, and isoflavones in tofu and tempe, as influenced by soybean variety and food processing, particularly fermentation. PRINCIPAL RESULTS: Raw soybeans contained 2207-2671 µg/kg (dry matter) folate, cooked tempe 1493-4143, and cooked tofu 968-1273 µg/kg, the difference was attributed to the fermentation in tempe. Vitamin B12 was detected only in tempe (0.16-0.72 µg/kg). Isoflavone aglycones were formed during soaking of soybeans, with only minor differences between the contents in cooked tempe (average 1922-2968 µg/kg) or tofu (1667-2782 µg/kg) but strongly depending on bean variety. CONCLUSIONS: Folate and vitamin B12 contents were mainly influenced by microbial activity during fermentation, whereas isoflavone aglycone content was determined by bean variety. Tofu had lower folate and vitamin B12, but equal isoflavone contents as tempe. Bioaccessibility of folate (80-100%) and isoflavone aglycones (100%) were high.

Folate in oats and its milling fractions.

Total folate content in oat varieties from three harvesting years (2006-2008), and in oats milling fractions, was determined using microbiological assay. Furthermore, folate vitamer distribution in milling fractions were examined with the UPLC method, which was taken in use and validated. The total folate content of the cultivars varied moderately within each year. The average content in the 2008 samples was 685ng/gdm. The UPLC method proved fast and sensitive for determining seven folate monoglutamates in cereal samples. Folate content in fractions, which are normally discarded, such as flour from oat cutting and flaking, were 1.5- to 2.5-fold higher than in native grain. The main folate vitamers found in the oat fractions were 5-CH(3)-H(4)folate, 5-HCO-H(4)folate, and 5,10-CH(+)-H(4)folate. The UPLC results more closely matched the microbiological results compared to those that are usually achieved with HPLC methods. This study illustrates that oats and, especially, by-products of milling are good sources of folate.

Effects of environment and genotype on folate contents in wheat in the HEALTHGRAIN diversity screen.

This study examined the environmental and genetic variation in folate contents and compositions of wheat genotypes. The selected genotypes, 24 from winter wheat and 2 from spring wheat, were grown in Martonvasar, Hungary, for three consecutive years as well as at four locations (Hungary, France, United Kingdom, and Poland) in one year. Total folate contents were determined by microbiological assay, and folate vitamers were determined for selected genotypes by high-performance liquid chromatography. Statistically significant differences in folate content arose among both harvesting years and growing locations. Grains grown in Hungary had the highest average folate content and those from Poland the lowest. Altogether, a 2.8-fold difference in total folate content appeared, ranging from 323 ng/g of dm (Chinese Spring, grown in Hungary in 2005) to 889 ng/g of dm (Riband, grown in Hungary in 2007). In general, the total folate content varied more greatly among the four growing locations than among the three harvesting years. Environmental factors affected folate content more strongly than genetic factors. In addition, small grains with high bran yield and low thousand kernel weight had high folate contents. The dominant vitamer in wheat genotypes was 5-HCO-H4folate. Other formylated folates and 5,10-CH+-H4folate also existed in significant amounts. Variation in the proportions of 5-HCO-H4folate and 5-CH3-H4folate were mainly responsible for the variation in total folate content: samples with high total folate content had a high proportion of 5-CH3-H4folate and a low proportion of 5-HCO-H4folate. Genotypes with both low and high folate contents, as well as with narrow or broad range, were identified. Thus, the study produced important data for plant breeding to select lines with stable folate contents.

Isolation and characterization of folate-producing bacteria from oat bran and rye flakes.

The aim of this research was to identify endogenous bacteria in commercial oat bran and rye flake products in order to study their folate production capability while maintaining the soluble dietary fibre components in physiologically active, unhydrolyzed form. Fourty-two bacteria were isolated from three different oat bran products and 26 bacteria from one rye flake consumer product. The bacteria were tentatively identified by sequence analysis of the 16S rRNA genes. The identification results revealed up to 18 distinct bacterial species belonging to 13 genera in oat bran, and 11 species belonging to 10 genera in rye flakes. The most common bacterial genus in oat bran was Pantoea, followed by Acinetobacter, Bacillus, and Staphylococcus. Pantoea species dominated also in rye flakes. The extracellular enzymatic activities of the isolates were studied by substrate hydrolysis plate assays. Nearly 80% of the isolates hydrolyzed carboxymethylcellulose, whereas starch-degrading activities were surprisingly rare (10%). Beta-glucan was hydrolyzed by 19% of the isolates. Protease, lipase or xylanase activity was expressed by 24%, 29%, and 16%, respectively, of the isolates. Representatives of the genera Bacillus, Curtobacterium, Pedobacter, and Sanguibacter showed the highest diversity of enzymatic activities, whereas members of Janthinobacterium and Staphylococcus possessed no hydrolytic activities for the substrates studied. Production capability for total folates was analyzed from aerobic cell cultures at the stationary growth phase. The amount of folates was determined separately for the cell mass and the supernatant by microbiological assay. For comparison, folate production was also examined in a number of common lactic acid bacteria. The best producers in oat bran belonged to the genera Bacillus, Janthinobacterium, Pantoea, and Pseudomonas, and those in rye flakes to Chryseobacterium, Erwinia, Plantibacter, and Pseudomonas. Supernatant folate contents were high for Bacillus, Erwinia, Janthinobacterium, Pseudomonas, and Sanguibacter. Compared to the endogenous bacteria, lactic acid bacteria were poor folate producers. The results of this work provide the first insight into the potential role of endogenous microflora in modulating the nutrient levels of oat and rye based cereal products, and pave way to future innovations of nutritionally improved cereal foods.

Production of folate by bacteria isolated from oat bran.

Twenty bacteria isolated from three commercial oat bran products were tested for their folate production capability. The bacteria as well as some reference organisms were grown until early stationary phase on a rich medium (YPD), and the amount of total folate in the separated cell mass and the culture medium (supernatant) was determined by microbiological assay. Folate vitamer distribution was determined for eight bacteria including one isolated from rye flakes. For seven bacteria the effect of temperature and pH on folate production was studied in more detail. Relatively large amount of folate was both produced in the cell biomass (up to 20.8microg/g) and released to the culture medium (up to 0.38microg/g) by studied bacteria. The best producers were characterized as Bacillus subtilis ON4, Chryseobacterium sp. NR7, Janthinobacterium sp. RB4, Pantoea agglomerans ON2, and Pseudomonas sp ON8. The level of folate released in culture medium was the highest for B. subtilis ON5, Chryseobacterium sp. NR7, Curtobacterium sp. ON7, Enterococcus durans ON9, Janthinobacterium sp. RB4, Paenibacillus sp. ON10, Propionibacterium sp. RB9, and Staphylococcus kloosii RB7. Marked differences in the distribution of folate vitamers among the bacterial strains were revealed by the HPLC analysis. The main vitamers were tetrahydrofolate, 5,10-methenyltetrahydrofolate, 5-methyltetrahydrofolate, and 5-formyltetrahydrofolate. Increase in the folate content during bacterial growth was accompanied by proportional increase in the 5-methyltetrahydrofolate content and decrease of 5-formyltetrahydrofolate. 10-Formylfolic acid dominated in the culture media of four bacteria, and Janthinobacterium sp. RB4 was also found to excrete 5-methyltetrahydrofolate. Intracellular folate content was higher when the bacteria were grown at 28 degrees C than at 18 degrees C or 37 degrees C and also higher at pH 7 than at pH 5.5.

The HEALTHGRAIN wheat diversity screen: effects of genotype and environment on phytochemicals and dietary fiber components.

Analysis of the contents of bioactive components (tocols, sterols, alkylresorcinols, folates, phenolic acids, and fiber components) in 26 wheat cultivars grown in six site x year combinations showed that the extent of variation due to variety and environment differed significantly between components. The total contents of tocols, sterols, and arabinoxylan fiber were highly heritable and hence an appropriate target for plant breeding. However, significant correlations between the contents of bioactive components and environmental factors (precipitation and temperature) during grain development also occurred, with even highly heritable components differing in amount between grain samples grown in different years on different sites.