Tools for grass farmers to monitor grazing behavior and forage use in real time

This study was placed at Choiniere (Organic Dairy) Family Farm in Highgate VT. Choiniere farm has been certified organic since 2005 and it has pastures that have not been plowed for over 25 years. Before the experiment began, we learned that Choiniere Farm breeds year-round. Cowherd at Choiniere Family Farm is certified organic and do not receive any medicinal treatment other than nutrition consisting mostly of fresh forages and winter balage supplementation.  In May and October we collected balage samples (diet 1) to test for feed quality and fatty acids. Between June and September, we carried out weekly forage species diversity, pasture biology, botanical composition, yield and quality sampling of diets (2) diverse cool season forages and (3) pearl millet. We collected soil and forage samples where cows grazed, between May and November of 2014.

As a way to lower risks related to illness or pregnancies that could affect statistical viability of the experiment, we randomly selected eight-second and third lactation cows in relatively close milk production levels, with days in milk ranging from 50 to 90 days and milk fat between 2.8 and 3.6% which were offered three different diet treatments: (1) winter feed (balage and diverse cool season forages), (2) diverse cool season forages and (3) pearl millet summer annual pasture.

We used single-group repeated measures design, where eight cows strip grazed, diverse cool-season forages pastures (trt 1) and pearl millet (trt 2) and indoor feed (balage). Cows stayed for approximately two-alternated four-week periods on each diet respectively, in order to document changes. Samples (soils, forage, rumen and milk) were taken weekly. Grazing activity was measured for five different two-week periods during the course of the study.

Soil Assessment: Soil samples from each of the pastures where cows grazed were taken. Each sample was obtained by combining 15 sub-samples in a clean bucket. After mixing it, two sub-samples were taken from the mix, and placed in two sealed identified plastic bags, and stored inside a refrigerated cooler to be shipped overnight to two different soil labs, Earthfort in Oregon and Woods End in Maine. Earthfort analyzed relationships and dynamics of soil life groups such as, microbes, fungi, nematodes, protozoa, and bacteria. By using this test focused on biological measurement, we were able to compare soil life between highly tilled versus unplowed soils. Woods End performs regular chemical analyses, evaluating the quality of soil carbon content and respiration as a proxy for microbial activity, estimates N and P mineralization potential and nutrients likely to be plant relevant.

Forage Assessment: 1) Forage samples were taken in three stages. First, preceding forage sampling, we randomly threw a 0.1 m² quadrant 25 to 35 times within each plot or study site to rank the three most predominant forage species using the Dry Weight Rank Method. Second, with the quadrant on the ground, we monitored pasture and site ecology by using a pasture-soil biology monitoring technique. This technique helps quantifying soil-pasture ecosystem functions related to mineral and water cycle and community dynamics by randomly recording data from 25 to 100 points along a transect in the representative area of the paddock in study. These data points provide insights for management improvements by evaluating environmental indicators. These two previous methods were not invasive and did not damaged forages.  2) Botanical composition: Forage samples were collected along the study area, using an electric clipper at soil surface to trim within the quadrant to the estimated grazing height at 3m intervals. All samples were combined, and a representative subsample was separated into botanical components (grasses, legumes, forbs or dead material) to determine botanical composition. The fractions were bagged, identified and weighed fresh, then dried at 65°C for 48 h to determine the DM/A proportions forage yield (Kg/hectare) of the different botanical components. 3) Forage Quality: Grab samples were randomly collected from each plot before grazing, for analysis of forage quality factors (crude protein, acid detergent fiber, neutral detergent fiber, total digestible nutrients, fatty acid content, and net energy lactation). Samples were dried at 65°C for 48 h, milled and sent to UVM, NW Crop and Soil NIH Lab to be analyzed. 4) Pre- and post- grazing forage sward: A Filip’s folding plate meter (Jenquip, Fielding, New Zealand) was used to estimate pre- and post-grazing pasture mass by use of standard Jenquip equation for pastures with reasonable consistent rainfall. We used a standard equation for all pastures composed of a grass – legume mixture. For diverse pastures, we calibrated cuts to create individual standardized equations. Each of these calibration cuts were weighed fresh then oven dried at 65°C for 48 h and re-weighed to determine DM content. The calibration curve was calculated using the calibration cuts. Pre-and post-grazing plate meter measurements will then be used to calculate estimated herd daily dry matter intake. To determine the pre-grazing pasture mass of all pastures, the folding plate meter will collect plate meter data from a pasture within 24 hr. prior to grazing. Plate meter measurements will then be taken at every 2 paces on a diagonal of the paddock, and further measurements will be taken on the opposing diagonal. Plate meter data collection will be repeated after grazing for 24 h to determine the postgrazing pasture mass (kg DM per hectare).

Grazing activity data loggers

Eight cows were enrolled from May to October 2014 in a study of the impact of pasture species diversity on milk composition and quality. A Hobo Pendant data logger was attached above the right hind ankle (metatarsus) to each cow for 5 different 2 week periods during the course of this study.  The loggers were attached using flexible bandage material and white medical tape. The loggers were removed after 2 weeks and cows were given at least one week before reattaching a data logger. Data was downloaded from the loggers using the Hobo USB Base Station data port connected to a personal computer and data was managed using Hoboware Pro software (Onset Computer Corporation). Data was exported to Microsoft Excel spreadsheets and lying behavior was summarized by day for each cow using SAS software and a set of previously published algorithms. Outputs from the activity data for each cow included daily lying and standing time, number of lying and standing bouts, and average duration of the daily lying and standing bouts. These data were then used to create averages for periods of time the cows were grazed on either a diverse pasture of a monoculture of millet. A minimum of 3 days data was compiled into the average for each feed source, and during 2 monitoring periods activity was recorded during a transition from one forage source to another forage source. Pre- and post-transition periods were defined, respectively, as the last 4 days grazing one type of pasture and the first 4 days grazing another pasture source.  

Feed samples: Feed composites were analyzed for dry matter (105°C for 8 h), for chemical composition, and for fatty acid content and profile. For the latter, a subset of the feed composites were dried at 55°C in a forced-air oven, ground to pass through the 1-mm screen of a Wiley mill. Fatty acid content and profile were analyzed using a modified version of the direct transesterification method. Fatty acids were transesterified directly using 2% H2SO4/methanol and a combination of acetone and toluene (2 h at 70°C). Samples were subsequently purified using a mixture of charcoal and silica gel. The fatty acid profiles (as fatty acid methyl esters, FAME) were analyzed by gas liquid chromatography (GLC) with flame ionization detection (FID; Shimadzu, Kyoto, Japan) using a highly polar 100 m fused‐silica capillary column. FAME were identified by comparison of retention times with known FAME standards. Total fatty acid content was determined using C13:0 as an internal standard.

Milk samples: One subset of weighted composite milk samples from each cow was be analyzed for milk components (fat, true protein, lactose, total solids, and milk urea N) by Lancaster DHIA Labs (Lancaster, PA) using mid-infrared technology. The second subset was analyzed for fatty acid content and profile using established methods. Briefly, cream is separated from milk composite samples by centrifugation. After extraction of total lipids from the cream using a mixture of n-hexane/isopropanol (3:2 v/v), FAME were prepared with 0.5 M NaOCH3/methanol (5 min at room temperature). Total fatty acid profile, covering circa 80 fatty acids in the range of C4:0 to C26:0, determined by GLC‐FID as described above. The short-chain FAME are corrected for mass discrepancy using correction factors. Fatty acid yield (g/d) is determined by multiplying the percentage of each fatty acid by milk fat yield and by 0.86, a value corresponding approximately to the percentage of fatty acids in triacylglycerides.

Rumen Health: Eight sentinel cows had rumen fluid extracted by the project team to facilitate measurement of critical rumen health outcomes related to three diet intakes. Samples included rumen pH, rumen microbial diversity, and consumed forage composition.

Rumen samples were collected weekly. Initially, we planned to fistulate six cows’ however the Northeast Organic Farmers Association (NOFA) advised against the method because of the stress (and other health complications) this method could impose to cows which in turn, could result in the loss of their organic status. Instead, we used a speculum --a 16 inch metal tube that is secured by an operator in the cow’s mouth while a second operator inserts a plastic flexible transparent hose through the speculum into cow’s stomach. The second operator pumps and extracts rumen fluid. To support rumen data, we planned to install rumination collars to measure rumination activity. However, because of schedule difficulties and high demand, the company responsible for installing these collars could only do it later in the season, which would inviabilize data collection in early season. As an alternative, we recorded events manually.

Bacterial and protozoal cells were isolated by differential centrifugation of rumen samples. Total lipids from rumen fluid as well as bacterial and protozoal cells wre extracted with chloroform/methanol (2:1 v/v). Fatty acids were transesterified with a combination of 0.5 M NaOCH3/methanol (10 min at 50°C) and 3% HCl/methanol (10 min at 80°C). Analysis of the resulting FAME is performed by GLC‐FID as described above.

Bacteriology: We collected weekly samples from quarter milk, right hind teat end skin swab, right inner hock skin swab, fecal material, and nasal swab for bacteriological analysis. All swab samples were transported to the laboratory on ice and frozen at -20 C for subsequent analysis. Bacteriological cultures have not been analyzed yet.

(See PowerPoint of final results:Biodiversity presentation results at end of report for original images/better quality.)