Mapping the bacterial ecology on the phyllosphere of grass hay and the potential hazards of soaking fodder for horse gut health

Globally hay is the preferred forage for stabled horses. Variable nutritional and hygienic quality stimulates pre-feeding soaking to reduce dust and nutrients to reduce respiratory and metabolic disorders in horses. However, this practice has potential negative impacts on horse health. The objectives of this study were to map the bacterial profile of different hays and determine how soaking alters this with the aim of recommending best practice when feeding fodder to stabled horses. Two meadow and one Perennial Ryegrass hays were soaked for 0, 1.5, 9 or 16 hours. Post treatment, hays were analysed for water-soluble carbohydrate (WSC) and total aerobic bacteria (TVC), with differences determined using ANOVA and least significant difference. Bacteria were identified via genomic DNA extraction (V3 and V4 variable region of the 16S rRNA gene) and 16S library preparation according to the Illumina protocol. Differences in phyla and family operational taxonomic units within hay types were identified via paired t-tests on the DESeq2 normalised data and false discovery rates accounted for using Padj (P<0.05). Mean WSC losses g/kg DM (+/-SE) increased with soaking time being 30 (10.7), 72 (43.7), 80 (38.8) for 1.5, 9 and 16 hours soak respectively. No relationship existed between WSC leaching and bacteria content or profile. Grass type influenced bacterial profiles. Soaking altered the epiphytic bacterial profile across all hays and 9 hours soaking increased richness and Shannon diversity indices. Clustering of bacteria was seen between meadow hays which differed from perennial rye grass and this difference increased post soaking. The normal industry practice of soaking hay for 9 hours pre-feeding cannot be recommended as it increases total bacteria content with noted increases of some potential pathogens. The alterations in bacteria profile and hygienic quality may explain why changing fodder or pre-feeding treatments can frequently precipitate colic in horses.

aerobic bacteria (TVC), with differences determined using ANOVA and least significant difference. Bacteria bacterial profile across all hays and 9 hours soaking increased richness and Shannon diversity indices. equid gut microbiome. A persistent (at least 6 weeks) core bacterial community was recorded within all regions of the hindgut, but no clustering was seen between individuals (β diversity) according to diet. Diet seemed to placed into a sterile plastic bag and placed in a laminar-flow cabinet (Bassaire, Duncan Rd, Swanwick, foil tray and placed in a forced draught oven at 60°C and dried until a constant weight was obtained for dry 50 g of the dried, milled sample was retained and stored in sterile plastic tubes (VWR, UK) for subsequent WSC 123 analyses.

125
Bacterial culturing and enumeration 126 127 Immediately post-treatment, sub-sample 1 was roughly chopped into 2cm lengths with scissors, (previously 128 wiped with ethanol, and allowed to dry) and thoroughly mixed. A one gram sub-sample was then weighed into a 129 sterile plastic bag (Seward BA6040) to which 79 ml of sterile peptone saline solution (MRD) was added. The 130 bag was then placed into a Lab Blender 80 model (Steward Laboratory, Blackfriars Rd, London). The mixture 131 was then 'blended' for 2 minutes in order to wash bacteria from the hay into the solution as for 3M petrifilms 132 (3M Microbiology, 2013). One millilitre of the blended solution was placed into a sterile screw-cap tube (VWR, 133 UK) containing 9 ml MRD. Serial dilutions were prepared to 10 -6 . A 1 ml sample was then taken from 10 -2 , 10 -134 4 10 -6 dilutions and separately placed onto pre-labelled 3M Aerobic TVC 20 cm 2 petrifilm, (3M Microbiology,

135
Carl-Schurz-Straβe 1, Germany). Petrifilms are a sample ready culture medium, containing nutrients, a cold 136 water-soluble gelling agent and a tetrazolium indicator. Three petrifilms were prepared for each sample and 137 incubated for 3 days at 32°C.

138
Colony numbers were enumerated using an illuminated magnifier. All vital stained colonies were counted.
When colony numbers were particularly dense and small and >100 per film, three representative 1 cm squares 140 were counted. The average was determined, and scaled up 20-fold as an estimation of the count per film.

142
Water soluble carbohydrate analyses 143 144 Immediately post-treatment, approximately 300 g of hay was weighed out into pre-weighed foil trays. These 145 were placed into a forced-draught oven and dried for a minimum of 48 hours at 65 o C until constant weight was 146 reached. Post-drying samples were milled through 0.75 mm mesh and re-bagged into 100g DM sub-sample 147 batches. Water soluble carbohydrate (WSC) analyses was then carried out on 3 replicates per sample using the 148 Phenol-sulphuric acid method (28) . The remaining 30 kg of stored dry hay from each hay type was sub-sampled 3 times, taking approximately 100 153 g for each sample. Each of the three replicate samples underwent the following procedure. A 0.5g sub-sample 154 was placed in a 50 ml glass tube. Seven and a half ml of tap water at 16 o C was added to each tube, covered with foil and placed in an incubator at 16 o C.After soaking, for 0, 1.5, 9 or 16 hours, samples were placed on to the manufacturer's protocol. 16S

213
There was no significant difference in the abundance of bacteria as measured by TVC (CFU/g) between the 214 three dry hays. The geometric mean (

219
The effect of soaking time on the dry matter, water soluble carbohydrate and microbial content in hay 220 Post-soaking, the forages absorbed between 50 and 62% additional moisture with no pattern emerging according 221 to soaking time.

222
All hays lost progressively more WSC up to 9 hours soaking.

226
When looking at individual hays and treatments ( 229 demonstrating that in this study no relationship existed in these three hays between WSC content and WSC Soaking for 9 hours produced a wide range of increases in % of CFU/g in MS and PRG of 1.24 (MS) to 19 PRG, but response to 9 and 16 hours soaking were less consistent with some increases at longer soaking times.

236
Therefore, as with WSC levels, the quantitative response of bacteria to soaking in different hays as determined 237 by CFU / g of hay was highly variable and showed no pattern according to soaking time.

275
The effect of soaking on the richness of bacterial families and the H index can be seen in Table 5

298
Of the remaining bacterial families that comprised between 31 to 51% of the bacteria present

299
Xanthomonadaceae, a family containing important animal and plant pathogens, increased in all hays at 9 hours.

354
The absorption of water of between 50 and 62% noted here were slightly lower than the 73% recorded by

369
There was also a highly variable response in bacterial growth (CFU/g) and in phyla and family profile across

411
Although highly variable between horses, Ericsson et al. [25] reported an abundance of α-Proteobacteria in the 412 upper gastrointestinal tract of 9 healthy horses. No information is available on what the horses were fed pre-413 euthanasia, but it is conceivable that like the forages in this study, Proteobacteria was present in significant 414 numbers. Epiphytic bacteria on forage may therefore have an impact on foregut bacterial profiles, but simultaneous profiling of feed and gut bacteria would have to be undertaken to determine the existence and these were less abundant than Bacteroidetes and Firmicutes which were the major phyla found in the gut.
Proteobacteria, Cyanobacteria, Actinobacteria and Bacteriodetes were present in the equid gut but at different 421 proportions to that found in the gut. The fact that the equid core gut community, particularly that in the upper 422 gastro-intestinal tract, which is composed of many small OTUs, lacks commonality between horses on similar 423 diets, suggests that horse response to diet is unique and this could explain the susceptibility of some animals to 424 digestive upset when fed similar diets to those that have no problems.

425
Clearly local environmental conditions contrive to favour the proliferation of some bacteria over others. For 426 example, some bacteria such as Enterobacteriaceae Rhodobacteraceae Bacillaceae Streptococcaceae, while 427 present in small numbers on dry leaves rapidly proliferated when wet, thereby altering the microbial profile of 428 the leaf. Therefore, the distribution of common opportunistic bacteria together with the more specific residents 429 to that particular phyllosphere under different environmental conditions can produce an ever-changing profile 430 [52,53]. In hays this could be further altered by a pre-feeding treatment such as soaking.

431
The relationship between feed and foregut bacterial profiles in particular requires further investigation. The