This study involved analysis of monthly data collected from ten diverse reduced input dairy operations and a prototype farm. Practices utilized on these dairy operations included management intensive grazing, outdoor winter housing of the dairy herd, group rearing of baby calves, cross breeding, and value added sales. Data pertaining to pasture production, feed supplementation, reproductive performance, mortality rates and involuntary culling rates was collected and analyzed. A large amount of variability was observed across the operations. The impact of this variability on the long-term economic sustainability of a prototype dairy was estimated with stochastic simulation modeling. Results showed that the long-term impacts of high rates of involuntary culling and mortality outweighed the long-term impacts of feed supplementation.
Many dairy operations located in the upper Midwest use grazing systems to meet a portion of their annual forage needs. Buttel et al. (2003) found that 23% of all Wisconsin dairy farms surveyed utilized some type of management intensive rotational grazing system. However, the survey didn’t specify whether this grazing system was used for lactating cows, non-lactating cows and/or heifers. The United Stated Department of Agriculture (USDA, 2002) found that 15.1% of United States’ dairy operations, representing 9% of the lactating cows in the U.S., used rotational grazing systems. Thirty three percent of dairy operations, representing 16% of lactating cows used other grazing systems.
An important goal of any grazing dairy is to be profitable. Many studies have examined the impact of feed supplementation level on profitability. Soder and Rotz, (2001) conducted a farm simulation study testing the effects of various levels of concentrate supplementation under different management constraints. They found the highest net return to management was achieved at the highest concentrate supplementation level the tested, a maximum of 9 kg dry matter (DM) per cow per day in early lactation. Tozer et al. (2004) used partial budgeting to compare income over feed costs of high yielding Holsteins fed 2 different pasture allowances and either supplemented with 1 kg of concentrate per 4 kg of milk produced or not. Regardless of pasture allowance, higher income over feed costs was realized for concentrate supplementation.
While level of feed supplementation may impact the level of profitability in a grazing herd, other herd characteristics, such as the rates of replacement mortality (stillbirth and replacement heifer deaths), adult mortality, involuntary culling, and incidents of clinical mastitis, also influence profitability. Survey studies found that the rates of herd health characteristics were quite variable across herds (Gardner et al., 1990; Kelton et al., 1998; Rougoor et al., 1999; USDA, 2002; Ingvartsen et al., 2003).
The long term effects of different levels of herd characteristics on production and profitability may be estimated through simulation modeling. SIMHERD III, a dairy herd simulation model developed by researchers in Denmark, is an analytical tool which can be used to investigate dairy cattle production management, production systems and production diseases (Østergaard et al., 2004). SIMHERD III uses a set of decision variables, to define a management strategy, and an initial herd, to simulate the application of a management strategy for up to 10 years. Generally 10 or more iterations, replicate simulations, of the same time period, with the same management strategy, can be conducted. Output summarizing the mean values for economic and production results for a particular management strategy or scenario are generated. Comparisons can then be made between the outputs of different scenarios to test for significant differences (Østergaard et al., 2004).
- Identify herd characteristics that may impact the long term sustainability of reduced input dairy operations located within the upper Midwest.
Estimate long term production and economic effects of 5 herd characteristics through simulation modeling.
This study consisted of two parts, (1) farm data collection and analysis followed by (2) stochastic simulation modeling. Data were collected from 10 grazing dairy farms from June, 2003 to the beginning of the grazing season in 2004. The beginning of the grazing season was defined individually for each farm to be the date when the lactating cows were receiving a significant amount of their forage from pasture. Data from 9 of the farms was analyzed because we were unable to obtain data which was complete and accurate enough for analysis from one farm. During this time, data from the dairy herd at the West Central Research and Outreach Center (University of Minnesota, Morris, MN) were also collected and analyzed. Modeling was used to estimate the long term effects of different levels of management characteristics observed on MN grazing dairies to a base herd and management strategy based on the herd conditions and management strategies applied at the WCROC in 2003.
All lactating cows in the participating herds were grazed for two to seven months in 2003. Four, two and three herds were located in the north-central, central and southeastern areas of MN, respectively. There was a wide distribution in the size of the farms in terms of number of hectares operated and number of cows milked. The average number of cows milked per day during the data collection period ranged from a minimum of 41 to a maximum of 318 between the participating herds with a mean of 122. An average of 139 cows were milked per day at the WCROC. The number of total farm hectares ranged from a minimum of 59 to a maximum of 405 with a mean of 178 across farms. The breed make up of the animals in the participating herds ranged from 25 to 100% Holstein across herds with a mean of 73%. The length of the 2003 grazing season ranged from a minimum of 74 days for the farm which was located the furthest north to a maximum of 215 days for the farm which was located the furthest south. The mean across herds was 153 days. The length of the grazing season for the lactating herd at the WCROC was 195 days. The length of the grazing season on all farms may have been affected by the weather in 2003, which, as a whole was a warm and dry year in MN (Minnesota Agricultural Statistics, 2004). The total annual precipitation in the metro area of the state for 2003 was 14.3 cm below normal annual precipitation.
The amount of energy corrected milk (ECM = kilograms of milk * ((0.383*fat% +0.242*protein% + 0.7832)/3.14)), Østergaard et al., 2003) sold per cow pure day during the data collection period was 17 ± 3 kg (mean ± standard deviation) across herds. The WCROC herd sold a mean of 20 kg ECM per cow per day. The mean SCC of the milk sold from the herd was 387,000 ± 160,000.
Herd data collection and calculations
Herd data were collected from each farm during visits conducted at 2-week intervals during the grazing season and 1-month intervals during the non-grazing season. Data from herd records, recorded by the farm operators or by DHIA (Dairy Herd Improvement Association), were utilized. The number of cows culled, calves stillborn, replacement heifers which died, cows which died and cows which had at least one case of clinical mastitis in 2003 were determined for each herd through examination of available herd records or interviews with the farm operators.
The values for five herd characteristics were calculated for each herd. These include replacement mortality rate (RMORT), adult mortality rate (AMORT), involuntary culling rate (CULL), lactational incident risk of clinical mastitis (MAST) and average amount of net energy in the feed supplemented per cow per day during the grazing season (SUP). RMORT was calculated for each herd by summing the stillbirth rate for cows that freshened in 2003 with the death rate of live born heifer calves in 2003. Involuntary culling rate and AMORT for each herd were determined by dividing the total number of cows which were involuntarily culled or died, respectively, in 2003 by the total number of cow-years calculated for the herd. An involuntarily cull was defined as a cow which left the herd for purposes other than production or reproduction. A cow-year was defined as the average number of lactating and dry cows on inventory for each day over the period of a year. Data on the actual number of dry cows was difficult to obtain for seven of the herds. Therefore, the number of cow-years was calculated by dividing the average number of cows milked per day from May 1, 2003 to May 1, 2004 by 0.88, the mean proportion of cows that were milked each day at WCROC in 2003. The cow-years for two of the herds were calculated from DHIA data.
MAST was calculated for each herd by dividing the total number of cows that the farm operator reported had at least one case of clinical mastitis in 2003 by the total number of cows that calved in 2003. MAST refers to the risk of a cow developing at least one case of clinical mastitis in her current lactation.
Pasture samples from each farm were collected at 2 week intervals from the start of the study in June, 2003 until the end of the grazing season in 2003. End of the grazing season was defined separately for each herd as the date when the lactating cows were removed from the grazing paddocks and either put into a lot, a sacrifice paddock or a building; or November 15, 2003 when all pasture growth and/or stockpiled pasture forage was assumed to be exhausted. Pasture samples were dried and a 60º C oven for 48+ hr to determine DM content. These samples were then ground through a 1 mm screen in a Wiley mill (Swedesboro, NJ). Samples of feedstuffs supplemented to the lactating herd were collected monthly on each farm and included hay, hay silage, corn silage, and concentrate samples. Equal masses of monthly dried and ground samples were composited together by farm, feed type and season to create composite feed supplement samples. Each sample’s content of neutral detergent fiber, crude protein and ash were determined through lab analysis.
The kg DM of each feedstuff supplied per cow per day during the grazing and non-grazing seasons was determined through farmer interviews. The kg DM of each feedstuff fed during the grazing season was converted into Scandinavian Feed Units (SFU) and then summed to define the herd characteristic SUP for each herd. The amounts fed were converted in SFU to estimate the net energy of the ration fed and to aid modeling. The model required that the feeding regime be defined with SFU. SFU/kg DM for each feedstuff fed was determined with the following formula: [SFU/kg DM = -0.369 + 0.0989*DE(MJ/kg DM) – 0.347*CF(kg/kgDM)] (Møller et al., 2001).
The long term economic and production impacts of different levels of the five herd characteristics (RMORT, AMORT, CULL, MAST and SUP) were estimated by using a computer, herd simulation program, SIMHERD III. SIMHERD III can simulate the production changes in a dairy herd over time (Østergaard et al., 2000; Østergaard et al., 2003) through the application of a management strategy. Values of decision variables, parameters defining management strategies within the model, can be changed to different management strategies or scenarios (Østergaard et al., 2000; Østergaard et al., 2003). The model requires a base herd, to be acted upon by a set of decision variables which define a management strategy. The model is documented in detail by Østergaard et al. (2004).
The herd characteristics and management strategy applied to the WCROC herd in 2003 and 2004 were used to define the base herd and base herd management strategy for the simulations. This herd was selected because it has an extensive history of research data appropriate for use in this study. The WCROC herd is managed in an intensive grazing system during the grazing season and supplemented with TMR during the grazing and non-grazing seasons. The herd consisted of approximately 1/3 Holstein-Jersey crossbreds and 2/3 purebred Holsteins animals. Approximately 60% of the animals calve in the spring and 40% in the fall. The spring calving cows are housed outdoors during the winter on a bedded pack and the fall calving cows are wintered in a loose housing system. All cows are artificially inseminated within the breeding windows which corresponded to the spring and fall calving seasons. The calibration procedure is outlined extensively in Ballinger, 2005a.
The long term effects of the different levels of the herd characteristics RMORT, AMORT, CULL, MAST and SUP observed on farms were estimated by conducting a series of 17 different simulations. Mean, minimum and maximum values across the participating herds were calculated for each of the herd characteristics and used to define the mean, minimum and maximum levels for herd characteristics modeled. For a data set which consists of the values 5, 9, 12, 15, 22, and 27, the mean would be 10.5. The maximum would be 27. The minimum would be 5.
Simulations with the three different SUP levels were conducted by adjusting the SFU of feed supplemented to the minimum, mean and maximum amounts of SFU supplemented during the grazing season as averaged across the herds less a 10% wastage estimate. The effects of the different levels of each of these herd characteristics were estimated individually. The value for the decision variables for one herd characteristic were adjusted while leaving all the other values for the decision variables at the value calculated for the base herd management strategy.
Two scenarios designated as “best” and “worst” case scenarios were simulated by combining the levels for each of the five herd characteristics which resulted in the highest or lowest gross margin, respectively, when simulated individually. The base herd was used for all of the simulations. All simulations were conducted for a period of ten years and iterated (replicated) 40 times.
The output from the simulations includes a number of production results. The economic consequences of the scenarios were analyzed by applying a set of prices and costs that would likely be typical for Minnesota grazing dairies. Prices and costs were estimated from WCROC data and FINBIN data for year’s 2002 and 2003 (Center for Farm Financial Management, 2004). Prices and costs included prices for milk and animals sold and costs for pasture, feed, disease incident, adult death, insemination, and other costs. The other costs included the total costs for livestock supplies, DHIA record keeping, fuel and oil, repairs, custom hire, hauling and trucking and marketing.
The values for the production and economic results of the minimum, mean and maximum levels from simulations estimating the effect of herd characteristic individually were compared to each other. The effects of the “worst” and “best” case scenarios were compared to each other. Statistical significance between different scenarios was determined if the difference between the values for a result were greater than a least significant difference value calculated at the alpha level of 0.05.
The mean RMORT across herds was 19% and ranged from 4 to 45%. The RMORT for the WCROC herd was 20%. The herd with the highest level of RMORT had 20% stillbirths in 2003 and 25% of the live-born calves died. The mean AMORT across herds was 6% and ranged from 2 to 12%. The mean CULL across herds was 14% and ranged from 2 to 42%. One herd had the maximum values for AMORT, RMORT and CULL. One herd had the minimum values for both RMORT and CULL. The RMORT calculated in this analysis is the sum of the stillbirth rate and the mortality rate of live born heifer calves.
The stillbirth rate (percentage of calves born dead in 2003) ranged from 2 to 20% across the herds with a mean of 8.3% and a median of 6.5%. The median was near the maximum found for the range of stillbirths rates (1.4 – 6.3%) in 2264 herds in 4 papers reviewed by Stevensen and Call (1988). The mean stillbirth rate for the herds was greater than the mean of 5.0% calves born dead in Pennsylvania dairy herds surveyed by Heinrichs et al. (1987). Thirteen percent of calves born from WCROC 1st lactation cows were stillborn and 9% of calves from WCROC 2nd lactation and older cows were stillborn.
The number of replacement heifers which died, calculated as a percentage of the total number of live-born heifer calves in 2003, ranged from 2 to 25% for the herds with a mean of 11%. This mean was similar to a USDA (2002) finding. The mean mortality rate across surveyed U.S. dairy farms was 8.7% for dairy calves from birth to weaning (calculated as a percentage of dairy heifers born alive in 2002) and 1.9% from weaning to parturition (calculated as a percentage of the January 1, 2002 heifer inventory). The mean live-born heifer mortality rate across the herds is in agreement with results from a study of dairy heifer calves on Minnesota farms which found that the mortality rate of dairy calves in their first 16 weeks of life ranged from 0 to 33.6% and averaged 9.0% (Sivula et al., 1996).
Approximately 4.3% of the heifers on inventory at the WCROC from July 1, 2003 to June 30, 2003 died. This value was doubled to estimate the percentage of live-born heifers which would die before they calved.
The mean values found from the herds for AMORT and CULL a similar to the rates found by a USDA (2002) survey of dairy herds which found the rates of 5 and 13% for AMORT and CULL, respectively, when calculated as a percentage of the January 1, 2002 cow inventory. The mean AMORT for the participating herds was much higher than that found by Gardner et al. (1990) on California dairies where AMORT ranged from 0 to 5% per cow-year with an average of 2% across the dairies surveyed. The AMORT for the WCROC herd was 6%.
The MAST for herds in 2003 ranged from a minimum of 1% to a maximum of 57% with a mean of 12%. The MAST for the WCROC herd in 2003 was 52%. The distribution of MAST across herds was only calculated for seven of the nine herds because there were no data available on incidents of clinical mastitis for two herds. The mean rate and range in MAST for the herds were similar to those found by Kelton et al. (1998) who reviewed 62 citations between 1982 and 1996 and found that clinical mastitis ranged from a lactational incident rate of 1.7% to an annual rate of 54.6% with a median of 14.2%. The average mastitis rate found in the herds compared well with the average of 14.7% found by the USDA (2002). However, it was less than the average rates of 31 and 32% found on 43 Californian (Gardner et al., 1990) and 37 Dutch dairy operations (Rougoor et al., 1999), respectively. There was no correlation between the reported incidents of mastitis and the average somatic cell count for milk sold for the participating herds. This fact coupled with the incident rates of mastitis found in other studies indicates that the number of incidents may have been under reported for some if not all herds.
The average amount of feed supplemented to the herds per cow/day was 9 ± 3 kg DM (mean as calculated across herds ± standard deviation) and 20 ± 3 kg DM during the grazing and non-grazing season, respectively. The average amount of feed supplemented to the WCROC herd was 12 kg DM and 21 kg DM during the grazing and non-grazing seasons, respectively. The average amount of net energy supplemented to the cows within the lactating herd during the grazing season ranged from a minimum of 4.0 SFU/cow/day to a maximum of 12.0 SFU/cow/day with a mean of 8.2 SFU/cow/day. The average amount of net energy supplemented to the cows in the WCROC lactating herd during the grazing season was 12.3 SFU. The types of feedstuffs supplemented varied quite widely between herds during the grazing and non-grazing seasons. Three of the herds located in north central Minnesota were supplemented with wet potato waste and two of these herds were also supplemented with beet tailings. These feedstuffs were available at relatively low prices within the area. Different types, amounts and qualities of forages were supplemented to the lactating herds as the grazing season progressed and forage availability within the pastures diminished. Forages supplemented included, corn silage, hay silage and/or dry hay, alfalfa, alfalfa-grass mixtures, cool season grasses, sorghum-sudan grass, barlage, pealage and oatlage. The herd at the WCROC was supplemented with a TMR mix which contained corn silage, alfalfa silage, primarily alfalfa hay, ground corn grain, soybean meal, distiller’s grains, urea and a mineral mix.
Five of the herds were fed at least a portion of their feedstuffs by a TMR mixer wagon. Fourteen percent of the grazing dairies interviewed by Loeffler et al. (1994) used a TMR to supply their herds with all of the feedstuffs supplemented. The correlation across farms between SUP levels and the amount of ECM sold per cow was very small (0.19) and is similar to the findings of Loeffler et al. (1996) who didn’t find a direct correlation between milk production and feed supplementation levels.
The nutrient composition of pasture for a herd was calculated as a mean of monthly averages. Each monthly average was an average value for all samples collected within that month. The NDF content of the pasture supplied to the herds was 10.5 ± 5.9%. The nutrient composition of the pasture supplied to the WCROC herd was 43% NDF and 20% CP. The CP content of the pasture supplied to the herds was 20.8 ± 3.8%. The mean values calculated across the herds for pasture DM, NDF and CP fall within the range for high quality pasture forage, 18 to 24% DM, 18-25% CP and 40-50% NDF, as described by Clark and Kanneganti (1998). However, the NDF and CP content of the pasture varied quite widely between farms. There are a large number of factors related to the management and the traits of the individual farms which can explain the variance. Some of the factors are differences in the weather, sampling procedure, grazing intensity and plant species within the pastures. Two herds which were continuously grazed had values for CP among the three lowest for this group of herds. The DM of the pasture forage available to the herds decreased as the grazing season progressed.
Production results. The production results, average values from the last 5 years of the 40 iterations for each simulation, are presented in detail in Table 9, Ballinger, 2005a. A few of the trends observed in the results are outlined below. When simulated independently, increasing the levels of the herd characteristics RMORT, AMORT and MAST from the minimum to the maximum levels resulted in a significant increase in numbers of replacement deaths, numbers of dead cows, and number of mastitis incidents, respectively. The mean amount of feed consumed per cow-year increased as the level of SUP was increased.
Increasing the levels of RMORT, AMORT and CULL had similar effects on the size of the heifer and cow herds. The application of the minimum levels of these characteristics caused the cow herd to grow to the herd size constraint of 150 cows within 5 years. The size of the herd was set at a maximum of 150 cows and a minimum of 75 cows in the base herd management strategy. These constraints were used for all scenarios tested. Herd size was maintained within these limits through the sales or purchase of replacement animals. The size of the herd was reduced severely enough when the maximum levels of RMORT and CULL were applied to require the purchase of replacement heifers to maintain the size of the herd above the lower limit. Four and 16 springing heifers were purchased in the last 5 years of the simulations, for the maximum levels of RMORT and CULL, respectively. The maximum CULL level caused the number of cows within the herd to decline to the lower herd size constraint more rapidly than the maximum RMORT level. The number of cows within the herd fell to the 75 cow herd size constraint by the 6th year of simulation for the maximum CULL level and by the 10th year of simulation for the maximum RMORT level. The size of the heifer herd (number of heifer-years) also decreased as the mean and maximum levels for these herd characteristics were applied.
The incidents of milk fever increased by 0.6 incidents per 100 cow-years as RMORT was increased from the minimum to the maximum, and decreased by 0.5 and 1.6 incidents per 100 cow-years when the AMORT and CULL levels were increased, respectively. These impacts on milk fever incidents can be attributed in part to the impacts of the different levels of these herd characteristics on the average age of the cows within the herd. The average ages of the cows within the herd will increase as the RMORT levels are increased and decrease as the AMORT and CULL levels are decreased. Older cows are at a greater risk of developing milk fever than younger cows. This is modeled within SIMHERD III through increasing the odds of developing milk fever as the age of a cow increases (Østergaard et al., 2003).
The SFU of feed consumed per cow-year increased by 29 SFU as the RMORT level was increased from 4 to 45%. As RMORT increases, the average age of the cows within the herd will increase because a smaller proportion of them are replacement animals. This causes an increase in the average size of the animals within the herd, increasing the average amount of feed consumed per cow-year. Increasing the levels of AMORT and CULL had the opposite effect. Feed intake was reduced by 0.4 and 1.6% with the maximum levels for AMORT and CULL, respectively, compared to the minimum levels. The average age of the cows in the milking herd decreases as AMORT and CULL levels increase because the chances of an individual cow surviving into older ages diminishes as the rates for these herd characteristics increases, therefore, decreasing the amount of feed consumed per cow-year.
The maximum level of MAST caused a 2.1% reduction in milk production per cow-year compared to the minimum and mean levels. This effect is likely due in part to the modeled effect of a mastitis infection to cause a decrease in milk production level to 0.95 (Østergaard et al., 2003) of a cow’s healthy producing ability for the rest of her lactation following infection.
The minimum, mean and maximum levels of SUP tested were 3.6, 7.4 and 10.8 SFU of feed supplemented per cow per day, which was calculated from the averaged energy level of the feed supplemented by each herd during the grazing season and adjusted to 90% to correct for wastage. Increasing SUP levels from the minimum to the mean level resulted in an annual increase in milk production per cow-year of 236 kg ECM. Increasing the SUP levels from the mean to the maximum level resulted in an annual increase of milk production per cow-year of 123 kg ECM. The corresponding increases in feed intake were 142 and 129 kg DM, respectively. This calculates to an annual milk response to additional kg DMI of 1.67 and 0.95 kg ECM/kg of DM and as expected follows the law of diminishing returns. There were no significant differences between the amounts of feed consumed by the dry cows or the lactating cows during the non-grazing season for the different SUP levels. Therefore all differences in annual intake are reflected by the relative differences in the amounts of feed supplement consumed during the grazing season and amounts of pasture consumed. The net energy density of the average ration consumed (pasture + feed supplement) per lactating cow during the grazing season ranged from 0.78 SFU/kg DM at the minimum SUP level to 0.84 and 0.89 SFU/kg DM at the mean and maximum SUP levels, respectively.
A year round calving strategy was applied in all simulations in this study and all cows were fed totally from supplements for 5.1 months of the year, irregardless of their stage of lactation. Some MN grazing dairies seasonally calve, either one or two times per year. The effects of a level of feed supplemented during the grazing season may vary for each cow depending on the time of the year when she calves. A cow typically requires a higher level of energy in the diet consumed per day in early lactation compared to later lactation because she has a higher potential for milk production in early lactation than later. Therefore, the milk production response of a cow to higher levels of feed supplementation during the grazing season may be greater for a cow calving in the spring compared to a cow calving in the fall.
Therefore, it may be prudent to provide a spring calving herd with a higher level of feed supplementation during the grazing season than a fall or year round calving herd, because the milk production response may be larger. However, large increases in feed supplementation may reduce pasture forage intake (Bargo et al., 2003).
The annual milk yield response per cow to increasing SUP from the minimum to the mean amount, and from the mean to the maximum amount was an increase of 0.36 kg ECM and 0.21 kg ECM per additional kg DM of the supplement consumed, respectively. These responses are lower than that found by Bargo et al. (2003) of an overall response of 1 kg milk/kg of concentrate supplemented. However, the supplements in the studies reviewed by Bargo et al. (2003) were concentrates while the supplement within our study was a TMR which had a lower energy density then concentrates because of the inclusion of forage. Soriano, et al. (2001) conducted a study to test the effects of supplementing cows on pasture by allowing them ad libitum access to TMR in either the afternoon or in the morning and had TMR intakes of 20.3 and 17.5 kg DM/cow per day. These levels are much greater than the supplementation levels examined in this study. The marginal daily milk yield response between the two levels examined by Soriano, et al. (2001) averaged 0.21 additional kg milk per additional kg of TMR and is similar to the milk yield responses found in our study when analyzing on an annual basis. However, the study conducted by Soriano et al. (2001) wasn’t specifically designed to test different levels of TMR supplementation and milk yield response was measured on a daily basis where as we examined it on an annual basis.
A “best” case scenario was designed by combining the lowest levels for the herd characteristics AM, RMORT, CULL and MAST with the mean SUP level. A “worst” case scenario was designed by combining the maximum levels for the herd characteristics AM, RMORT, CULL and MAST and the low SUP level. The differences between the results of the simulations with these scenarios were significant for nearly all production results variables measured. The differences between the results for number of cow-years, milk production per cow-year, replacement rate, and the disease incident rates of milk fever, dystocia and mastitis were greater than the results from any of the simulations where the different levels of the herd characteristics were tested individually. The growth or reduction in the size of the herd resulting from the application of these scenarios reached the herd size constraints within the first 5 years of simulation. The replacement rate resulting from application of the “best” case scenario was 18% higher than that resulting from maximum CULL level, reflecting the additional impact of the maximum AMORT level when combined with the maximum CULL level. The large difference in annual milk produced per cow between the “worst” and “best” case scenarios (482 kg ECM) is a the result of a combination of the effects of feed supplementation, resulting in an increase of 236 kg ECM when tested individually, and the impacts of the herd characteristics AMORT and CULL on the average age and size of the cows in the herd.
The economic consequences of the management strategies were studied by applying “typical” MN economic values for the herd inputs and outputs based upon WCROC data and FINBIN data for year’s 2002 and 2003 (Center for Farm Financial Management, 2004). The “gross margin” values used within this study were calculated from “total sales” less “variable costs”. Total sales included sales of milk, springing heifers, cull cows and bull calves. Variable costs included the purchases of feed for cows and heifers, purchase of springing heifers, costs for disease treatments, inseminations and other costs for cows and heifers. Labor and management costs were not included in the calculations. The gross margins of the scenarios simulated were limited by the herd size limits in a few cases where limits were reached within the ten years of the simulation period.
The gross margins from the simulations examining the minimum levels of RMORT, AMORT, CULL, MAST, and FS were $211,000, $202,000, $209,000, $190,000 and $174,000, respectively. The gross margin for the mean and maximum levels of RMORT were 16 and 52% less than the minimum level, respectively. The gross margin for the mean and maximum levels of AMORT were 10 and 34% less than the minimum level, respectively. The gross margin for the mean and maximum levels of CULL were 23 and 58% less than the minimum level, respectively. The gross margin for the mean and maximum levels of MAST were 1.3 and 4.6% less than the minimum level, respectively. The gross margin for the mean and maximum levels of SUP were 3.7 and 2.7% less than the minimum level, respectively.
A large portion of the differences in gross margins resulting from the simulations with the minimum, mean and maximum levels for each herd characteristic is a reflection of the impact that different levels of these characteristics had on size of the herd; therefore, the volume of milk which the herd was capable of producing. Increasing the levels of RMORT, AMORT and CULL from the minimum to the maximum levels significantly reduced gross margin. Annual milk sold from the herd decreased from 1100, 1074, and 1114 metric tonnes for the minimum levels of RMORT, AMORT and CULL, respectively, to 980, 781, and 903 metric tonnes for the mean levels, respectively, to 633, 781, and 578 metric tonnes for the maximum levels, respectively. Simulations with the minimum and maximum levels of CULL resulted in the largest difference in gross margin compared to the difference in gross margin between the minimum and maximum levels for the other herd characteristics. This is likely due in part to the magnitude in range of CULL compared with the range for RMORT and AMORT. The maximum CULL was 20 times larger than the minimum CULL.
The gross margin per cow decreased as the levels of RMORT, AMORT and CULL were increased, indicating that a portion of the decrease in gross margin for these herd characteristics was due to other factors besides herd size. These other factors included the purchase or sales of springing heifers. The purchase of springing heifers was required at the minimum levels of RMORT and CULL to maintain herd size. A large number of springing heifers were sold at the minimum levels of RMORT, AMORT and CULL creating an important revenue source. When compared to the resulting gross margin from the application of the minimum levels, gross margin per cow decreased by 6, 4 and 6% for the mean and 16, 14 and 28% for the maximum levels of RMORT, AMORT and CULL, respectively.
The effect of increasing the MAST level on gross margin was not as dramatic as the effects of RMORT, AMORT, and CULL. It was significant for gross margin between the minimum and maximum levels with a reduction of $9000. The reduction in gross margin per cow was significant between the mean and the maximum, and minimum and maximum levels with an average reduction of $70. The relatively low impact of the different levels of MAST on gross margin can be explained in part because the size of the herd wasn’t affected by increases in the clinical mastitis rate. A USDA survey (2002) found that the average cost for treating a case of clinical mastitis ranged from $102 to $162 across U.S. dairy farms surveyed, with most of the cost due to the loss of milk production. Using these values, the impact of increasing the mastitis rate from the minimum to the maximum level was an additional cost of $7,650 to $12,000 per year. The difference observed in our simulations for gross margin falls within this range. SIMHERD III cannot model the impacts of different milk quality premiums for different levels of somatic cell counts within the milk sold. If this could be modeled perhaps a higher impact on gross margin due to differences in the incidents of mastitis could be realized.
The responses in gross margin to increasing the level of SUP were mixed. Numerically, gross margin increased when SUP was increased from the minimum to the mean, but, gross margin decreased numerically as the level of SUP was increased from mean to maximum. The difference in gross margin between SUP levels wasn’t significant between any of the levels tested.
The results of increasing the level of SUP on gross margin are in partial agreement with the findings of other studies. Tozer et al. (2004) found that feed supplementation of grazing, high-yielding Holstein cows at the level of 1 kg of concentrate per 4 kg of milk produced improved their income over feed costs when compared to no supplementation. However, Tozer et al. (2004) analyzed the effects of feed supplementation only over the grazing season, where as, we examined the effect over a period of 10 years.
Our study agreed in part with the findings of Soder and Rotz’s (2001) simulation study in that increasing the level of feed supplementation caused a numeric increase in gross margin from the minimum to the mean SUP levels. However, there was a numeric decline in gross margin as measured in our study when SUP level was increased from the mean to the maximum. Soder and Rotz (2001) found that profitability as measured by net return to management increased as the level of grain supplementation was increased from a maximum of 0 to 9 kg DM per day. There could be several reasons for this, not the least of which is the different measures of profitability. Net return to management takes into account fixed costs, machinery cost and labor costs which are not accounted for in the gross margin analysis within this study. Increasing the level of feed supplementation from 3 to 6 kg DM per day caused an annual increase in milk production of 900 kg in Soder and Rotz’s study (2001) compared to the response of 236 kg ECM observed within our study resulting from increasing SUP from 3.6 SFU (3.6 kg DM) to 7.4 SFU (7.5 kg DM). The supplements used within the studies could partially explain this response; TMR was used in this study while concentrate was used in Soder and Rotz’s (2001) study. Soder and Rotz’s (2001) simulations were conducted at the farm level, and took into account changes in the usage of farm area as the level of concentrate supplementation (purchased off farm) was increased. Pasture area was reduced and more area was available for forage production as the level of concentrate supplementation increased. Some of the decrease in pasture intake could have been substituted by increases in supplemented forage intake or visa versa in Soder and Rotz’s farm simulations. SIMHERD III only simulates at the herd level, in which all feed is given the same price and production from farmland area isn’t considered. Soder and Rotz (2001) found that the increase in net return to management decreased as grain supplementation levels were increased. This finding agrees with the law of diminishing returns and the findings from the SUP simulations. The increase in gross margin decreased as feed supplementation levels were increased.
The genetic potential for milk production of the cows in the herd and effects on gross margin wasn’t considered within this study. Shalloo et al. (2004) used an Irish grass-based dairy simulation model to estimate the effect of 3 different levels of grain supplementation and 3 different genetic potentials for milk production under typical Irish conditions. Their findings suggest that the concentrate supplementation level that would return the highest total profit per farm would be a low level (336 kg DM/cow/year) for herds of cows with low genetic potential for milk production. Herds of cows with high genetic potential for milk production were predicted to return the highest total profit per farm at the highest grain supplementation level (1377 kg DM/cow/year) (Shalloo et al., 2004). This is important to consider because the genetic potential for milk production of cows in MN grazing herds may vary greatly.
The difference in the gross margin between the “worst” and “best” case scenarios was $179,000, or 47% greater than the largest difference in gross margin that resulted when the herd characteristics were evaluated individually. The gross margin of the “best” case scenario was 2.8 and 0.9 times greater than the “worst” case scenario when compared at the herd and cow levels, respectively. A large proportion of the difference in the gross margin was due to the difference in total sales. Total sales were 120% higher for the “worst” case scenario than the “best” case scenario while total costs were only 34% higher. It is clear from the results of the simulations that a combination of the levels of the herd characteristics that minimized or maximized gross margin when tested individually will result in even greater impacts on gross margin when combined. The differences in gross margin between “worst” and “best” case scenarios was limited by herd size and would have been larger if the herd size constraints were relaxed.
In interpreting these results, it is important to examine the relative differences in economic results between scenarios. Focusing on the actual gross margin values from the simulations can be misleading in a real life situation because costs vary between farms. However, relatively, the economic impacts of the ranges in the levels of RMORT, AMORT and CULL were much greater than the impacts of the ranges observed across the participating herds for the levels of SUP or MAST. This can be explained, in part, by the impact that each of these herd characteristics had on the amount of milk produced by the herd. The amount of milk produced by a herd has a strong impact on profitability.
Income from the sale of milk is the predominant income source for most dairy herds and accounted for at least 85% of the total sales within every simulation conducted in this study. Levels of RMORT, AMORT and CULL dramatically impacted the volume of milk sold from the herd through their impacts on herd size. Lactation incident risk of clinical mastitis and SUP impacted the amount of milk which could be produced per cow. The impacts of each of the herd characteristics on herd size outweighed their impacts on milk production per cow. It is important to note that the parameters of the simulations allowed for flexibility in the size of the herd allowing it to grow to 150 cows or shrink to 75 cows. This may have impacted the results. However, even if the flexibility in herd size was decreased, it is still very likely that the range in the levels of RMORT, AMORT and CULL would still have a large impact on gross margin through the sale or purchase of replacements.
While there is uncertainty regarding the validity of the outcomes from simulation this simulation program is a valuable tool to analyze relationships within a system and identify areas for further research. Results of different strategies modeled and their relative differences rather than direct comparisons are utilized because “relative differences between the simulations provide evaluations of systemic differences” (Soder and Rotz, 2001).
Bargo, F., Muller, L.D., Kolver, E.S., Delahoy, J.E., 2003. Invited Review: Production and digestion of supplemented dairy cows on pasture. J. Dairy Sci. 88:1-42.
Buttel, F.H., Jackson-Smith, D., Moon, S., 2000. A profile of Wisconsin’s dairy industry, 1999. No. 3. University of Wisconsin-Madison.
Clark, D.A., Kanneganti, V.R., 1998. Grazing management systems for dairy cattle. In: Cherney, J.H. and Cherney, D.J.R. (Eds.), Grass for Dairy Cattle. CABI Publishing, New York, New York, pp. 311-334.
Center for Farm Financial Management, FINBIN, Department of Applied Economics, University of Minnesota, 2004. http://www.agrisk.umn.edu/Finbin.asp (last accessed on October 29, 2004).
Gardner, I.A., Hird, D.W., Utterback, W.W., Danaye-Elmi, C., Heron, B.R., Christiansen, K.H. and Sischo, W.M., 1990. Mortality, morbidity, case-fatality, and culling rates for California dairy cattle as evaluated by the National Animal Health Monitoring System, 1986-1987. Prev. Vet. Med. 8: 157-170.
Heinrichs, A.J., Kiernan, N.E., Graves, R.E., Hutchinson, L.J., 1987. Survey of calf and heifer management practices in Pennsylvania dairy herds. J. Dairy Sci. 70, 896-904.
Ingvartsen, K.L., Dewhurst, R.J., Friggens, N.C., 2003. On the relationship between lactational performance and health: is it yield or metabolic imbalance that cause production diseases in dairy cattle? A position paper. Livest. Prod. Sci. 83, 277-308.
Kelton, D.F., Lissemore, K.D., Martin, R.E., 1998. Recommendations for recording and calculating the incident of selected clinical diseases in dairy cattle. J. Dairy Sci. 81, 2502-2509.
Loeffler, B., Murray, H., Johnson, D.G., Fuller, E.I., 1996. Knee Deep in Grass. Minnesota Extension Service, University of Minnesota.
Minnesota Agricultural Statistics, 2004. http://www.nass.usda.gov/mn (last accessed on November 24, 2004).
Møller, J., Thogersen, R., Kjeldsen, A.M., Weisbjerg, M.R., Søegaard, K., Hvelplund, T., Børsting, C.F., 2001. Feedstuff Table. Composition and feeding value of feedstuffs for cattle. Report no. 91, 2000. English version.
Østergaard, S., Sorensen, J.T., Kristensen, A.R., 2000. A stochastic model simulating the feeding-health-production complex in a dairy herd. J. Dairy Sci. 83, 721-733.
Østergaard, S., Sorensen, J.T., Houe, H. 2003. A stochastic model simulating milk fever in a dairy herd. Prev. Vet. Med. 58, 125-143.
Østergaard, S., Sorensen, J.T., Enevoldsen, C., 2004. SIMHERD III: Users Manual. DIAS Internal Report no. 209. Danish Institute of Agricultural Sciences.
Rougoor, C.W., Hanekamp, W.J.A., Dijkhuizen, A.A., Nielen, M., Wilmink, J.B.M., 1999. Relationships between dairy cow mastitis, fertility management and farm performance. Prev. Vet. Med. 39, 247-264.
Shalloo, L., Kennedy, J., Wallace, M., Rath, M., Dillon, P., 2004. The economic impact of cow genetic potential for milk production and concentrate supplementation level on the profitability of pasture based systems under different EU milk quota scenarios. J. Agric. Sci. 142, 357-369.
Sivula, N.J., Ames, T.R., Marsh, W.E., 1996. Management practices and risk factors for morbidity and mortality in Minnesota dairy heifers calves. Prev. Vet. Med. 27, 173-182.
Soder, K.J., Rotz, C.A., 2001. Economic and environmental impact of four levels of concentrate supplementation in grazing dairy herds. J. Dairy Sci. 84, 2560-2572.
Soriano, F. D., Polan, C. E., Miller, C. N., 2001. Supplementing pasture to lactating Holsteins fed a total mixed ration diet. J. Dairy Sci. 84, 2460-2468.
Stevensen, J.S., Call, E.P., 1988. Reproductive disorders in the periparturient dairy cow. J. Dairy Sci. 71, 2572-2583.
Tozer, P.R., Bargo, F., Muller, L.D., 2004. The effect of pasture allowance and supplementation on feed efficiency and profitability of dairy systems. J. Dairy Sci. 87, 2902-2911.
USDA, 2002. Part 1: Reference of Dairy Health and Management in the United States, 2002.
Educational & Outreach Activities
Ballinger, E., 2003. Evaluation of the Components of Reduced Input Dairy Farms. Oral presentation. University of Minnesota. April 14, 2003.
Ballinger, E., 2004. Analysis of the Components of Reduced Input Dairy Farms. Oral presentation. University of Minnesota. March 29, 2004.
Ballinger, E., 2005a. Quantification of herd and management characteristics of Minnesota grazing dairies and evaluation of the technical and economic effects of important traits through stochastic simulation modeling. Masters thesis. University of Minnesota.
Ballinger, E., 2005b. Quantification of herd and management characteristics of Minnesota grazing dairies and evaluation through stochastic simulation modeling. Masters thesis presentation. University of Minnesota. February 25, 2005.
Johnson, D.G., M.V. Rudstrom, R. Imdieke, E. Ballinger, and G.J. Cuomo. 2005. Farmer-driven research for developing models of successful low input dairy farmers of small to medium size in the American Midwest. Proceedings of XX International Grasslands Conference: Offered Papers, page 820.
Johnson, D., E. Ballinger, J. Linn, M. Rudstrom. L. Hansen, A. Seykora, and B. Heins. Evaluation of low input dairy farming systems by on-farm research, prototyping, and simulation modeling, with emphasis on grazing management, seasonal calving, crossbreeding, and outwintering. 2005. Proceedings Midwest Section American Dairy Science Association, page 4.
The grazing dairies where this data was collected were quite variable in terms of size, milk production level, herd characteristics and management practices. The range in rates of replacement mortality, adult mortality, involuntary culling varied greatly between the herds, yet agreed with ranges found by other studies. Simulations with a base herd and the range of values for the herd characteristics replacement mortality, adult mortality and involuntary culling rates had much larger impacts on gross margin than the different levels of lactational incident rate of clinical mastitis or feed supplementation during the grazing season. The areas in which improvement would have the largest economic impact on grazing farms would likely be in the reduction of calf mortality, adult mortality and involuntary culling rates leading to greater sustainability on these farms.
The long-term economic impacts of the 5 characteristics investigated in this study are described at length in the Results and Discussion/Milestone section. Operations with high levels of involuntary culling, replacement mortality and adult mortality would likely benefit more from addressing these issues than in adjusting feed supplementation levels. This is because involuntary culling, replacement mortality and adult mortality impact the size of the herd will feed supplementation impacts the amount that can be produced per cow.
No data available
Areas needing additional study
Areas that require further investigation include the testing of the results of different levels of concentrate supplementation under different grazing forage quality conditions. What affects would the quality of the forage supplemented have? What interactions are there between the amount of forage supplemented, the quality of the forage, the amount of concentrate supplemented and the quality of concentrate and the quality of pasture and how they all interact with one another? The cost of applying control methods for reducing the rates of involuntary culling, and mortality need to be investigated within different systems to find what levels for these herd characteristics and financially sustainable. These are but a few of the areas where systems modeling techniques and other methods of analysis can be applied to find management techniques which can help improve the profitability and sustainability of grazing dairy farms located in the upper Midwest.