ABSTRACT
A dairy farmer's decision to cull or keep dairy cows is likely a complex decision based on animal health and farm management practices. The present paper investigated the relationship between cow longevity and animal health, and between longevity and farm investments, while controlling for farm-specific characteristics and animal management practices, by using Swedish dairy farm and production data for the period 2009 to 2018. We used the ordinary least square and unconditional quantile regression model to perform mean-based and heterogeneous-based analysis, respectively. Findings from the study indicate that, on average, animal health has a negative but insignificant effect on dairy herd longevity. This implies that culling is predominantly done for other reasons than poor health status. Investment in farm infrastructure has a positive and significant effect on dairy herd longevity. The investment in farm infrastructure creates room for new or superior recruitment heifers without the need to cull existing dairy cows. Production variables that prolong dairy cow longevity include higher milk yield and an extended calving interval. Findings from this study imply that the relatively short longevity of dairy cows in Sweden compared with some dairy producing countries is not a result of problems with health and welfare. Rather, dairy cow longevity in Sweden hinges on the farmers' investment decisions, farm-specific characteristics and animal management practices.
Key words
INTRODUCTION
The longevity of dairy cows has gained increased attention in recent years, largely due to the environmental (
Zehetmeier et al., 2014
; Bell et al., 2015
; Bergeå et al., 2016
; Grandl et al., 2019
) and economic consequences (De Vries, 2017
; Grandl et al., 2019
; Dallago et al., 2021
) associated with short longevity. The culling of dairy cows early on in the production cycle (i.e., before the cow is biologically old) is regarded as an unsustainable practice from both an economic (Vredenberg et al., 2021
; Gambonini et al., 2022
) and environmental (Axelsson, 2013
) point of view. The concept of longevity has also been linked to animal welfare (- Axelsson H.H.
Breeding for sustainable milk production: From nucleus herds to genomic data.
Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences,
2013
https://pub.epsilon.slu.se/10438/
Date accessed: September 2, 2021
Röcklinsberg et al., 2016
). Given the growing concern for and awareness of the treatment of animals in modern agriculture (Ingenbleek and Immink, 2011
; Lagerkvist and Hess, 2011
; Thorslund et al., 2017
), culling dairy cows at a young age raises animal welfare concerns. Short longevity and the practice of culling dairy cows at a young age signal that animals are not kept in such a way that they can function in production over an extended period of time (Röcklinsberg et al., 2016
).Culling of dairy cows is the act of removing cows from the herd. Culling decisions are usually influenced by intrinsic and extrinsic factors (
Bergeå et al., 2016
; Rostellato et al., 2021
). Intrinsic factors that influence cow longevity include health, milk yield, and reproductive status (De Vries, 2013
, - De Vries A.
Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.
https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/
Date: 2013
Date accessed: August 12, 2021
De Vries, 2017
; Zehetmeier et al., 2014
). Extrinsic factors influencing cow longevity include availability of replacement heifers, milk yield, land availability, and prices (Grandl et al., 2019
; Adriaens et al., 2020
; Rostellato et al., 2021
). The reasons for culling dairy cows have received considerable attention in the scientific literature. Grandl et al., 2019
indicated that a large number of cows are removed from the herd early in lactation mainly because of metabolic health reasons. De Vries, 2013
added that average cow longevity is based on extrinsic economic factors that are not related to the cow's health and performance.- De Vries A.
Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.
https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/
Date: 2013
Date accessed: August 12, 2021
Previous research has found that the amount of time a dairy cow stays in the herd is associated with the cow's welfare, farm economic outcome, and environmental impacts (
Brickell and Wathes, 2011
; Boulton et al., 2017
). Previous research has also attempted to examine to what extent cow longevity affects farm economic outcomes. In the United States, De Vries, 2013
found that when cows leave the herd early in lactation it costs the dairy farmer approximately €380 to €760 per cow, and this cost does not include milk losses. Culling cows early in the lactation (e.g., less than 60 DIM) leads to great economic losses (- De Vries A.
Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.
https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/
Date: 2013
Date accessed: August 12, 2021
Lunak, 2020
). In the UK and Denmark, studies have shown that a high proportion of income losses incurred by dairy farmers are linked to early culling of dairy cows (- Lunak M.
Cull rates: How is your farm doing? Penn State Extension.
https://extension.psu.edu/cull-rates-how-is-your-farm-doing
Date: 2020
Date accessed: November 21, 2022
Orpin and Esslemont, 2010
; Wright and Rusk, 2020
). In Germany, - Wright R.
- Rusk F.
What could an increase in dairy cow culling mean for the markets?.
Agriculture and Horticulture Development Board (AHDB),
2020
https://ahdb.org.uk/news/what-could-an-increase-in-dairy-cow-culling-mean-for-the-markets
Date accessed: November 20, 2022
Grandl et al., 2019
found that increasing dairy cow longevity improves the profitability of dairy farms.A growing body of literature has focused on intrinsic, genetic, and trait factors (
De Vries, 2013
; - De Vries A.
Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.
https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/
Date: 2013
Date accessed: August 12, 2021
Zehetmeier et al., 2014
; Haile-Mariam and Pryce, 2015
; Interbull, 2020
); extrinsic factors (Grandl et al., 2019
; Rostellato et al., 2021
) to explain longevity. However, the longevity of dairy cows in the herd is determined by culling decisions made by the dairy farmer, which are based on a complex combination of different factors (De Vries, 2013
, - De Vries A.
Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.
https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/
Date: 2013
Date accessed: August 12, 2021
De Vries, 2017
). Although several factors related to the animal inform the decision to cull a dairy cow, the decision is likely to include an economic component relating to how many and which cows should be kept in the herd to maximize profit. In terms of production economics, one reason for not culling the oldest cows could be a recent farm expansion, whereby a farmer may need to retain all animals in the herd to fill the places. This suggests that the amount a farmer invests in the farm can influence longevity. A recent study by Reimus et al., 2020
supports the economic reasons for culling decisions. The study indicated that farmer characteristics and farm management practices can have a substantial effect on health and longevity. Furthermore, the authors suggested that good management practices, including investments in good housing, can lead to better animal health, welfare, and longevity. Still, no study to date has examined the relationship between investments made by the farmer and cow longevity. As a result, there is currently a limited understanding of how cow longevity is related to strategic decisions by the farmer regarding enlargements of the operation.Furthermore, to the best of our knowledge, none of the previous studies mentioned above have examined causality between a farm's specific animal management practices and cow longevity. Instead, from a methodological point of view, most of the studies that have attempted to investigate culling decisions and methods to optimize longevity, as well as how these affect economic outcomes, have relied on computer-optimized models (
De Vries, 2004
; Cha et al., 2010
; Nielsen et al., 2011
). These computer-optimized models estimate profit per animal per year and assume that the profit is maximized for a fixed number of animals in the herd, without considering replacement heifers. In addition, these models do not account for time effects, given that the models are mostly set up for a specific time period or production season. Furthermore, the optimized models do not establish causality, and as a result, there is little information or no consensus on the causal link between farm-specific characteristics and animal management practices and cow longevity.Accordingly, the aim of our study was to investigate whether there is a causal relationship between animal health, cow longevity, and farm investments, while controlling for farm-specific characteristics and animal management practices by using Swedish dairy farm and production data from 2009 to 2018. The farm-specific and animal management factors include milk yield, breed, type of housing and milking system, and production system (conventional vs. organic). Specifically, we first investigate to what extent animal health affects cow longevity. Second, we emphasize in particular the extent to which farms made investments to increase the size of their operation and to what extent the rate of culling is determined by farmers' choices to increase the number of places or to keep their operation at a constant size at a particular point in time. Furthermore, we differentiate between average estimated effects and the heterogeneous effects estimated by ordinary least square (OLS) and unconditional quantile regressions (UQR), respectively.
The scientific novelty of our study lies in its investigation of the mean and heterogeneous effects of how farmers' investment decisions, animal health, management practices (e.g., choice of breed, housing, and milking systems) and production characteristics (e.g., milk yield, herd size) affect cow longevity. An understanding of these determinants can simultaneously function to enhance herd health, welfare, and farm economic outcome. In addition, this study uses on-farm recording data and employs a rigorous econometric framework.
This paper contributes to existing knowledge in the following ways: first, this study is the first to use long panel data on farm investments to examine the effect of investments in farm building on longevity of dairy cows. In this way, we highlight how farmers' strategies regarding dairy cow longevity can function to simultaneously enhance dairy herd welfare status. The study also contributes significant insights that are important for an informed discussion about ethical considerations linked to dairy cow longevity and to possible synergies or trade-offs between the wishes to keep animals for a long time for ethical reasons and consequences in terms of herd health and welfare. In addition, findings from this study contribute to the debate on wishes to keep animals in production for a long time to spread environmental burdens of raising heifers over a larger amount of total production, and how this relates to the herd health and welfare status. Finally, findings from the study contribute to dairy herd lifetime resilience, which is very relevant for the sustainability of the dairy industry.
MATERIALS AND METHODS
Data and Variable Description
The study used Swedish Dairy Cow Recording Scheme data from Växa Sverige and Farm Economic Survey (FES) from the Swedish Board of Agriculture for the years 2009 to 2018. The Swedish Dairy Cow Recording Scheme data contained detailed information on cow health variables, including the proportion of cows receiving veterinary treatment, mastitis, leg and hoof diseases, SCC, herd size, age at first calving (d), calving interval (mo), age at culling (d), number of cows culled, breed, milking system, and ECM were also included in this data set. The FES data contains variables such as investment in farm building, fixed and variable cost, system of production, milk and meat output, just to mention a few. The FES serves as a basis for Sweden's participation in the European Union Farm Accounting Data Network, which targets population farms containing at least 8 European size units (
Hansson et al., 2013
). Thus, FES data in Sweden target large farms. The 2 data sets were merged using unique identification numbers. Both data sets contained data from 2009 to 2018.De Vries, 2013
defined dairy cow longevity as the length of life of the dairy cow in the herd. Dairy cow longevity can be broken into the time before first calving and the time after first calving. In this study, dairy cow longevity was measured using the following 2 indicators: (1) productive lifespan of the cow, computed as the age at culling minus age at first calving, and (2) total lifespan (i.e., time from birth to culling). Farm management variables in the data set include breed of herd, type of housing, type of milking system, production system (conventional or organic), and investment costs (farm building). Figure 1 presents average milk yield per cow in the Nordic countries and shows that Sweden has high average milk yield per cow, relative to the other Nordic countries. There is an increasing trend in milk yield per cow in Sweden, Denmark, and Norway. From the year 2017, there has been a sharp increase in milk yield per cow in Sweden. Despite the high milk yield per cow, Sweden has short dairy herd longevity, and as such, there is the need to investigate what influences dairy farmers decision relating to how long herd stay in the dairy production system.- De Vries A.
Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.
https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/
Date: 2013
Date accessed: August 12, 2021
Figure 1Average raw milk yield per cow in hectogram across Nordic countries. Denmark, Norway, and Finland values are from . Sweden values are from .
FAOSTAT, 2022
Växa Sverige, 2020
Empirical Framework
We applied OLS regression and quantile regression models. These models were selected because they allowed us to analyze both the mean and heterogeneous effects of animal health and farm investments on dairy cow longevity, while controlling for production and farm management characteristics. We started from the following baseline OLS regression functions:
where the variable Healthi captures animal health proxied by the proportion of cows receiving veterinary treatment (
where LGVTi denotes longevity measured using the following 2 indicators: (1) age at culling and (2) productive lifespan. P_Health is the predicted health variable from the Equation [1]. This predicted health variable was used to avoid endogeneity bias, which would have occurred if the actual values were used (). Thus, the actual health variable could not be included directly in the Equation (2) because the proportion of cows receiving veterinary treatment correlates with some of the covariates in the equation (e.g., housing system, milking system). The rest of the variables were defined as above.
Healthi = α0 + Investiα1 + Prodiα2 + Mangmtiα3 + Yeariα4 + εi,
[1]
where the variable Healthi captures animal health proxied by the proportion of cows receiving veterinary treatment (
Fall et al., 2008
). Before this estimation, the variable, proportion of cows receiving veterinary treatment, was tested to see if the observed values were censored or not. The test revealed that the variable was uncensored, and OLS regression was therefore estimated instead of a censored regression (e.g., Tobit model), which established the relationship between variables when the dependent variable is censored. Investi captures positive change in investment in farm buildings. Prodi is a vector of production variables milk yield, herd size, calving interval, age at first calving, and SCC. Mangmti is a vector of farm management characteristics type of breed, type of housing, type of production system, and type of milking system. Yeari refers to a vector of year dummies used for capturing the time-invariant and time-variant heterogeneities; α0 is a constant; α1, α2, α3, and α4 are parameters to be estimated; and εi is an error term.
LGVTi = α0 + α1P_Healthi + α2Investi + α3Prodi + α4Mangmti + α5Yeari + εi,
[2]
where LGVTi denotes longevity measured using the following 2 indicators: (1) age at culling and (2) productive lifespan. P_Health is the predicted health variable from the Equation [1]. This predicted health variable was used to avoid endogeneity bias, which would have occurred if the actual values were used (). Thus, the actual health variable could not be included directly in the Equation (2) because the proportion of cows receiving veterinary treatment correlates with some of the covariates in the equation (e.g., housing system, milking system). The rest of the variables were defined as above.
In Equation [2], α1 captures the average (mean) effect of animal health on dairy cow longevity, which only illustrates a partial picture of the association between the 2 variables. Second, α2 captures the average (mean) effect of investment in housing on dairy cow longevity. From a policy perspective, policymakers are more likely to be interested in understanding the influence of a program intervention (e.g., investment support for reconstruction of animal housing or support for extended hoof health care) on the distributions of animal health and welfare. Therefore, a quantile regression model should be further estimated to identify the heterogeneous relationship between health and investment variables on dairy cow longevity, as well as control variables.
Previous studies have applied both the conditional quantile regression (CQR) model (
Mishra and Moss, 2013
; Agyire-Tettey et al., 2018
; Derbali et al., 2020
) and UQR model (Mishra et al., 2015
; Hernæs, 2020
; Pérez-Rodríguez and Ledesma-Rodríguez, 2021
). It should be noted here that differences exist between the CQR and UQR models. In particular, the CQR model offers a narrower interpretation of the effect of health and investments on longevity, because the estimated effect of a covariate is largely conditional on the selection of other variables (e.g., production, management) included in the model (Borah and Basu, 2013
; Pérez-Rodríguez and Ledesma-Rodríguez, 2021
). In contrast, the effect of health and investment variables in the UQR model estimates does not implicitly rely on the level of other conditional variables in the quantile models. The UQR model can estimate more reliable results, and thus, it is employed in the present study.The UQR model is based on the concept of the recentered influence function (RIF;
where F refers to the cumulative distribution function for dairy cow longevity, LGVT, and εδLGVT represents a distribution that puts mass at the value LGVT. For the τth quantile of LGVT, the influence function can be written as
where Qτ is the τth quantile of the unconditional distribution of LGVT, and the probability density function of LGVT evaluated is represented by fLGVT(Qτ). I(LGVT ≤ Qτ) indicates whether the observed value is less than or equal to Qτ. The RIF is obtained by adding the relevant statistic (i.e., the quantiles) to its IF:
In the present study, the UQR model estimates Equation [5] conditional on a vector of explanatory variables as follows (
where LGVTi is the dependent variable, referring to longevity; Qτ demotes the τth quantile of the dependent variable's cumulative distribution FLGVT; the remaining variables are defined as above; β0 is a constant; β1 and β2 are the parameters we are interested in, which is the unconditional marginal effect of health and investments; β3, β4, and β5 are corresponding parameters to be estimated; and μi is an error term. The UQR estimates from Equation [6] have a similar interpretation as the coefficients from an OLS regression and are estimates of unconditional quantile marginal effects.
Firpo et al., 2009
; Agyire-Tettey et al., 2018
). The influence function (IF) assesses the effect of an individual observation on a distributional statistic, v(F), such as the mean, median, or any quantiles, without having to recalculate that statistic. The IF can be defined as
[3]
where F refers to the cumulative distribution function for dairy cow longevity, LGVT, and εδLGVT represents a distribution that puts mass at the value LGVT. For the τth quantile of LGVT, the influence function can be written as
[4]
where Qτ is the τth quantile of the unconditional distribution of LGVT, and the probability density function of LGVT evaluated is represented by fLGVT(Qτ). I(LGVT ≤ Qτ) indicates whether the observed value is less than or equal to Qτ. The RIF is obtained by adding the relevant statistic (i.e., the quantiles) to its IF:
RIF (LGVT; Qτ) = IF(LGVT; Qτ).
[5]
In the present study, the UQR model estimates Equation [5] conditional on a vector of explanatory variables as follows (
Firpo et al., 2009
; Agyire-Tettey et al., 2018
):
RIF (LGVT; Qτ, FLGVT) = β0 + β1P_Healthi + β2Investi + β0Prodi + β4Mangmti + β5Yeari + μi,
[6]
where LGVTi is the dependent variable, referring to longevity; Qτ demotes the τth quantile of the dependent variable's cumulative distribution FLGVT; the remaining variables are defined as above; β0 is a constant; β1 and β2 are the parameters we are interested in, which is the unconditional marginal effect of health and investments; β3, β4, and β5 are corresponding parameters to be estimated; and μi is an error term. The UQR estimates from Equation [6] have a similar interpretation as the coefficients from an OLS regression and are estimates of unconditional quantile marginal effects.
RESULTS
Descriptive Results
Table 1 presents the descriptive statistics of production, health, and investment variables. Milk yield in Swedish dairy cows has increased in recent decades because of improved genetics, feeding practices, and management (
Växa Sverige, 2020
). The average age at first calving in the Swedish data is about 28.5 mo. The results indicate that, on average, dairy cows are culled when they are 63 mo old.Table 1Summary statistics of production, health, and investment variables
| Item | Mean | Median | SD | Minimum | Maximum |
|---|---|---|---|---|---|
| Production variable | |||||
| Number of cows per herd | 88 | 63 | 95 | 10 | 1,335 |
| Age at first calving (d) | 856 | 835 | 94 | 689 | 1,644 |
| Calving interval (mo) | 13 | 13 | 1 | 11 | 23 |
| Milk yield (kg ECM) | 9,656 | 9,762 | 1,436 | 523 | 14,264 |
| Age at culling (d) | 1,884 | 1,853 | 267 | 1,114 | 3,769 |
| Productive lifespan (d) | 1,028 | 999 | 243 | 385 | 2,931 |
| Number of culled cows per year | 33 | 22 | 35 | 0 | 469 |
| Health variable | |||||
| Veterinary treatment (proportion of cows, %) | 24 | 20 | 18 | 1 | 100 |
| Mastitis treatment (proportion of cows, %) | 10 | 9 | 10 | 0 | 84 |
| Hoof and leg treatment (proportion of cows, %) | 2 | 0 | 4 | 0 | 46 |
| BMSCC (×1,000/mL milk) | 244 | 234 | 84 | 60 | 713 |
| Farm investment | |||||
| Farm building (SEK) | 2,692,089 | 1,224,094 | 4,014,146 | 1,144,164 | 4.45e+07 |
| Investment (1 if farmer made investment in farm building, 0 otherwise | 0.92 | — | — | 0 | 1 |
1 Source: Sveriges Riksbank (https://www.riksbank.se/en-gb/statistics/search-interest-exchange-rates/annual-average-exchange-rates/?y=2022&m=5&s=Comma&f=y).
2 Bulk milk SCC.
3 Average exchange rate 2022 (€1: SEK10.46).
In terms of productive life, we found that, on average, dairy cows are productive for 34 mo, with a median of 33 mo. Regarding the health variables, the results show that, on average, about 24% of the dairy cows in the herds receives veterinary treatment. The average bulk tank SCC observed in this study for the period from 2009 to 2018 was 244,000/mL milk, with a median of 234,000/mL.
To capture investment dynamics, in particular how a change in the level of investment affects cow longevity, we did not use the absolute values of the farm building in the empirical analysis. Changes in investment in farm infrastructure from the start year (t) to year (t + n) were computed and from this variable, a dummy variable was computed, taking on a value of 1 for farms with a positive change in investment in farm infrastructure and 0 for farms that did not make investments in farm infrastructure over the time period. The descriptive statistics show that the majority of the farms (92%) made investments in farm infrastructure. In terms of investment size, the results show that an average of SEK 2,692,089 (€257,616) was invested in farm infrastructure for the herd.
Table 2 presents the distributions of farm management characteristics. The majority of the dairy farms in the present study employed conventional production practices. Only about 16% of the dairy farms were certified by KRAV as organic producers. The percentage of organic producers does not differ significantly from the figure of 17% reported by the Swedish Board of Agriculture in 2018 (
Swedish Board of Agriculture, 2019
).Table 2Summary statistics (number and percentage of farms) of farm management variables
| Variable | No. of farms | Percentage |
|---|---|---|
| Production system | ||
| Conventional | 1,842 | 84 |
| Certified organic by KRAV | 347 | 16 |
| Milking system | ||
| Automatic milking system | 627 | 29 |
| Milking parlor | 436 | 20 |
| Rotary | 34 | 1 |
| Tiestall milking | 1,092 | 50 |
| Housing type | ||
| Freestall housing, noninsulated | 337 | 15 |
| Freestall housing, insulated | 745 | 34 |
| Tiestall | 1,107 | 51 |
| Breed | ||
| 1 = SR (SR ≥ 80%) | 468 | 21 |
| 2 = SH (SH ≥ 80%) | 548 | 25 |
| 3 = SR + SH ≥ 50% | 844 | 39 |
| 4 = Other breeds | 329 | 15 |
1 KRAV is the main Swedish organization that develops and maintains regulations for ecological sustainable agriculture. KRAV is the Swedish organic market's private label.
2 SR = Swedish Red; SH = Swedish Holstein. The numbers indicate the proportion of the breeds in the entire herd.
In terms of milking systems, the tiestall milking system is the most popular, used by 51% of the dairy farms, followed by automatic milking systems (AMS) and milking parlors. The rotary system is the least common system. Half of the dairy farms had tiestalls, and this is consistent with the use of a tiestall milking system. The remaining farms had freestall housing systems: 34% had insulated housing and 15% had noninsulated housing. In terms of breed, 39% of the farms had a herd with cross breeding between the Swedish Red and the Swedish Holstein, which together constituted more than 50% of the herd.
In Table 3, the correlations between health and longevity variables are presented. The proportion of cows receiving veterinary treatment was positive and significantly correlated with the proportion of cows treated for mastitis, as well as with the proportion of cows with hoof and leg disorders. Moreover, the proportion of treatments for hoof and leg disorders was positively and significantly correlated with the proportion of mastitis treatment. Culling age and productive lifespan were highly correlated. Milk yield is positively correlated with share of cows with veterinary, mastitis, and hoof and leg treatments, but negatively correlated with SCC, productive life span, and culling age.
Table 3Correlation of health and longevity variables
| Variable | Veterinary treatment | Mastitis treatment | Hoof and leg treatment | BMSCC | Productive lifespan | Culling age | Milk yield |
|---|---|---|---|---|---|---|---|
| Veterinary treatment | 1 | ||||||
| Mastitis treatment | 0.84 | 1 | |||||
| Hoof and leg treatment | 0.51 | 0.29 | 1 | ||||
| BMSCC | −0.13 | −0.08 | −0.08 | 1 | |||
| Productive lifespan | −0.05 | −0.05 | −0.03 | 0.07 | 1 | ||
| Culling age | −0.05 | −0.04 | −0.04 | 0.13 | 0.94 | 1 | |
| Milk yield | 0.16 | 0.11 | 0.11 | −0.19 | −0.23 | −0.16 | 1 |
1 BMSCC = bulk milk SCC.
** indicates statistical significance at 5% level.
In addition, bulk tank SCC was correlated with culling age, whereas the proportion of cows receiving veterinary treatments had no significant correlation with productive life or culling age. It is important to note that these results do not show cause-and-effect relationships. Hence, in the next section, we present OLS and UQR results to show how the different covariates affect health and longevity variables.
Average Effects of Farm-Specific Characteristics and Animal Management Practices on Animal Health and Productive Lifespan of Dairy Cows
Table 4 presents the empirical results of the mean-based analysis of the effects of farm-specific characteristics and animal management practices on dairy cow health using the OLS regression model. The results show that the proportion of cows receiving veterinary treatments increases by 0.002 percentage units for 1 kg of ECM increase in herd average milk yield. On the other hand, for one increase in bulk milk SCC (BMSCC; i.e., for an increase of 1,000 cells/mL), the proportion of cows receiving veterinary treatment decreased by 0.02 percentage units, and having organic production, compared with conventional, reduced the proportion by about 10 percentage units.
Table 4Effects of farm-specific and animal management characteristics on dairy herd health ordinary least square estimates; dependent variable is proportion of herd with veterinary treatment
| Variable | Coefficient | R. SE | t-statistic | P-value |
|---|---|---|---|---|
| Constant | 26.51 | 9.92 | 2.67 | 0.01 |
| Production indicator | ||||
| Milk yield (kg ECM) | 0.00 | 0.00 | 7.53 | <0.001 |
| Calving interval (mo) | −0.05 | 0.36 | −0.13 | 0.90 |
| Age at first calving (d) | −0.00 | 0.00 | −0.16 | 0.87 |
| BMSCC (×1,000 cells/mL) | −0.03 | 0.01 | −3.75 | <0.001 |
| Herd size (no. of cows) | 0.01 | 0.01 | 1.55 | 0.12 |
| Production system (reference: conventional) | ||||
| Certified organic (KRAV) | −9.57 | 4.46 | −2.15 | 0.03 |
| Breed [reference: SR (SR ≥ 80%)] | ||||
| SH (SH ≥ 80%) | 0.17 | 1.20 | 0.14 | 0.89 |
| SR + SH ≥ 50% | −0.75 | 1.03 | −0.73 | 0.47 |
| Other_breed | −1.28 | 1.29 | −1.00 | 0.32 |
| Housing type (reference: freestall housing, noninsulated) | ||||
| Freestall housing, insulated | −4.81 | 1.42 | −3.39 | <0.001 |
| Tiestall | −7.35 | 4.96 | −1.48 | 0.14 |
| Milking system (reference: tiestall) | ||||
| Automatic milking system | 8.34 | 4.04 | −2.06 | 0.04 |
| Milking parlor | −9.29 | 3.96 | −2.34 | 0.02 |
| Rotary | 3.54 | 4.46 | −0.79 | 0.43 |
| Production system and milking system | ||||
| Conven_tiestall milking | −7.38 | 2.86 | −2.58 | 0.01 |
| Conven_parlor | −3.53 | 4.99 | −0.71 | 0.48 |
| Conven_AMS | −1.81 | 2.38 | −0.76 | 0.45 |
| Organic_tiestall milking | 9.67 | 4.25 | 2.28 | 0.03 |
| Organic_parlor | 10.73 | 5.19 | 2.07 | 0.04 |
| Year effects (reference: 2009) | ||||
| 2010 | −5.90 | 2.46 | −2.40 | 0.02 |
| 2011 | −7.32 | 2.46 | −2.97 | <0.001 |
| 2012 | −8.65 | 2.41 | −3.59 | <0.001 |
| 2013 | −4.01 | 1.39 | −2.88 | <0.001 |
| 2014 | −8.52 | 2.29 | −3.72 | <0.001 |
| 2015 | −11.28 | 2.27 | −4.98 | <0.001 |
| 2016 | −14.15 | 2.20 | −6.43 | <0.001 |
| 2017 | −12.96 | 2.22 | −5.85 | <0.001 |
| 2018 | −12.73 | 2.23 | −5.72 | <0.001 |
| Number of observations | 2,189 | |||
| Adjusted R2 | 0.56 | |||
| F-statistic | 9.40 | |||
| Mean variance inflation factor | 4.00 | |||
1 R. SE = robust standard error; BMSCC = bulk milk somatic cell count; SR = Swedish Red; SH = Swedish Holstein.
*** and ** show significance at 1% and 5% levels, respectively.
In terms of housing type, the results show that insulated freestall housing reduces the proportion of cows receiving veterinary treatment, compared with noninsulated freestall housing. This finding suggests that the use of AMS increases the proportion of veterinary treatment by 8, relative to a tiestall system. On the other hand, the use of milking parlors reduces the proportion of veterinary treatment by 9, relative to tiestall system.
The results show that herds raised using conventional systems with tiestall milking have a low proportion of veterinary treatment. On the other hand, herds kept on organic systems with tiestall and parlor milking have a high proportion of cows receiving veterinary treatment. Compared with 2009, veterinary cases among Swedish dairy herds have been declining from 2010 to 2018 and this may explain the declining proportion of veterinary treatments in the present study.
Table 5 presents the average effects of animal health, investments, farm-specific characteristics, and animal management variables on average productive lifespan. The results show that the average productive lifespan of dairy cows increases by about 39 d when a farm makes investments to increase the size of farm buildings.
Table 5Average effects of animal health, investment, farm-specific, and animal management variables on productive lifespan; dependent variable is productive lifespan of dairy herd
| Variable | Coefficient | R. SE | t-statistic | P-value |
|---|---|---|---|---|
| Health | −0.08 | 0.30 | −0.28 | 0.78 |
| Investment | 38.86 | 16.75 | 2.32 | 0.02 |
| Control variable | ||||
| Production indicator | ||||
| Milk yield | 0.02 | 0.00 | 4.73 | <0.001 |
| Calving interval | 37.19 | 5.75 | 6.47 | <0.001 |
| Age at first calving | −0.09 | 0.07 | −1.15 | 0.25 |
| BMSCC, ×1,000 cells/mL | 0.18 | 0.07 | 2.41 | 0.02 |
| Herd size | 0.06 | 0.06 | 0.98 | 0.33 |
| Production system (reference: conventional) | ||||
| Certified organic (KRAV) | −74.39 | 62.68 | −1.19 | 0.24 |
| Breed [reference: SR (SR ≥ 80%)] | ||||
| SH (SH ≥ 80%) | −21.79 | 16.05 | −1.36 | 0.18 |
| SR + SH ≥ 50% | −2.96 | 14.65 | −0.20 | 0.84 |
| Other_breed | 5.93 | 20.88 | 0.28 | 0.78 |
| Housing type (reference: freestall housing, noninsulated) | ||||
| Freestall housing, insulated | −7.57 | 15.77 | −0.48 | 0.63 |
| Tiestall | −44.73 | 57.68 | −0.78 | 0.44 |
| Milking system (reference: tiestall) | ||||
| AMS | −18.98 | 31.30 | −0.61 | 0.54 |
| Milking parlor | 10.13 | 30.99 | 0.33 | 0.74 |
| Rotary | 69.54 | 34.19 | 2.03 | 0.04 |
| Production system and milking system | ||||
| Conven_tiestall milking | −13.14 | 43.14 | −0.30 | 0.76 |
| Conven_AMS | −8.94 | 29.45 | −0.30 | 0.76 |
| Conven_parlor | 32.65 | 51.94 | −0.63 | 0.53 |
| Organic_tiestall milking | 35.59 | 64.39 | 0.55 | 0.58 |
| Organic_parlor | 82.66 | 58.71 | 1.41 | 0.16 |
| Year effects (reference: 2009) | ||||
| 2010 | 8.65 | 27.69 | 0.31 | 0.76 |
| 2011 | 12.41 | 27.07 | 0.46 | 0.65 |
| 2012 | 22.32 | 28.00 | 0.80 | 0.43 |
| 2013 | 16.83 | 26.70 | 0.63 | 0.53 |
| 2014 | 41.63 | 26.22 | 1.59 | 0.11 |
| 2015 | 72.28 | 26.56 | 2.72 | 0.01 |
| 2016 | 73.66 | 27.84 | 2.65 | 0.01 |
| 2017 | 67.42 | 26.86 | 2.51 | 0.01 |
| 2018 | 85.97 | 28.49 | 3.02 | <0.001 |
| Constant | 705.31 | 130.20 | 5.42 | <0.001 |
| Number of observations | 2,188 | |||
| Adjusted R2 | 0.62 | |||
| F-statistic | 6.13 | |||
| Mean variance inflation factor | 4 | |||
1 R. SE = robust standard error; BMSCC = bulk milk somatic cell count; SR = Swedish Red; SH = Swedish Holstein.
2 Health is the predicted proportion of the herd with veterinary treatment from the ordinary least square estimation of farm-specific and animal management characteristics on animal health (Equation 1).
*** and ** show significance at 1% and 5% levels, respectively.
Among the production variables, the results show that the average effect of a single unit increase in ECM milk yield equates to an increase in productive lifespan of 0.021 d. This implies that dairy cows with high milk yields remain in the herd longer. The results also show that a one-month increase in average calving interval on herd level leads to an increase in average productive lifespan on herd level of 37 d. The model showed that the average productive lifespan was higher in herds with higher BMSCC. Farmers do not cull the chronic infected cows, and hence, they will have a higher BMSCC. This finding is contrary to
Samoré et al., 2003
, who found high SCC to be associated with a high rate of culling. An increased productive lifespan was seen in herds with rotary systems compared with herds with tiestall milking systems. The results show that that the average productive lifespan of cows in Swedish dairy herds increased substantially from 2015 to 2018.The estimates of animal health, investments, farm-specific characteristics, and animal management variables on age at culling are provided in Appendix Appendix Table A1, Appendix Table A2. The results obtained do not differ from the results presented in Table 5, and in the interest of brevity, we will not discuss the results with age at culling as a dependent variable. The test statistics for variance inflation factor showed no severe multicollinearity problem in the covariates.
Heterogeneous Effects of Farm-Specific Characteristics and Animal Management Practices on Productive Lifespan of Dairy (UQR Estimates)
In this section, we present the UQR results. Before the discussion of the empirical results, we present the quantile statistics of productive lifespan, culling age, proportion receiving veterinary treatment, and investments (see Table 6). The unconditional regression results present some interesting findings. To begin with, the results in Table 7 show that the health variable (proportion receiving veterinary treatment), which was insignificant in the pooled sample (see Table 5), is statistically significant and negative in the 25th quantile. This result suggests that productive lifespan of dairy herds reduces as the proportion of veterinary treatments increases at the lower quantile (25th), where the average productive lifespan of a dairy cow is about 785 d, with a standard deviation of 102 d.
Table 6Descriptive statistics of the productive life, age at culling, veterinary treatment, and investment per quantile
| Variable | Observation | Mean | SD | Minimum | Maximum |
|---|---|---|---|---|---|
| 25th quantile | |||||
| Productive life (d) | 548 | 785.15 | 101.61 | 388.50 | 1,006.67 |
| Age at culling (d) | 548 | 1,595.08 | 93.80 | 1,114.00 | 1,713.22 |
| Veterinary treatment (%) | 548 | 25.98 | 18.64 | 0 | 93.07 |
| Investment | 548 | 0.89 | 0.31 | 0 | 1 |
| 50th quantile | |||||
| Productive life (d) | 547 | 942.23 | 74.89 | 631.40 | 1,110.29 |
| Age at culling (d) | 547 | 1,780.79 | 39.84 | 1,713.55 | 1,852.52 |
| Veterinary treatment (%) | 547 | 23.75 | 17.88 | 0 | 100 |
| Investment | 547 | 0.92 | 0.28 | 0 | 1 |
| 75th quantile | |||||
| Productive life (d) | 547 | 1,068.83 | 87.56 | 550.73 | 1,288.21 |
| Age at culling (d) | 547 | 1,927.31 | 44.69 | 1,852.97 | 2,012.26 |
| Veterinary treatment (%) | 547 | 23.51 | 18.18 | 0 | 100 |
| Investment | 547 | 0.91 | 0.29 | 0 | 1 |
| 95th quantile | |||||
| Productive life (d) | 547 | 1,316.64 | 249.30 | 384.50 | 2,930.67 |
| Age at culling (d) | 547 | 2,233.70 | 233.49 | 2,013.08 | 3,768.67 |
| Veterinary treatment (%) | 547 | 22.69 | 18.42 | 0 | 100 |
| Investment | 547 | 0.92 | 0.27 | 0 | 1 |
Table 7Heterogeneous effects of animal health, investment, and farm management variables on productive life of dairy cows; dependent variable is productive lifespan of dairy herd
| Variable | 25th quantile | 50th quantile | 75th quantile | 95th quantile |
|---|---|---|---|---|
| Health | −0.72 (0.33) | −0.14 (0.29) | −0.01 (0.39) | 0.03 (1.01) |
| Investment | 11.41 (20.52) | 9.91 (18.56) | 43.26 (20.09) | 88.19 (68.56) |
| Control variable | ||||
| Production indicator | ||||
| Milk yield | 0.01 (0.00) | 0.01 (0.00) | 0.02 (0.01) | 0.06 (0.02) |
| Calving interval | 30.37 (5.59) | 33.13 (5.06) | 46.96 (6.56) | 31.28 (17.67) |
| Age at first calving | −0.16 (0.07) | −0.04 (0.06) | 0.01 (0.08) | 0.01 (0.22) |
| BMSCC, ×1,000 cells/mL | 0.29 (0.07) | 0.25 (0.07) | 0.11 (0.09) | −0.12 (0.25) |
| Herd size | 0.28 (0.09) | 0.04 (0.08) | 0.07 (0.11) | −0.03 (0.31) |
| Production system (reference: conventional) | ||||
| Certified organic (KRAV) | −56.83 (78.55) | −75.28 (71.07) | −42.78 (92.26) | 169.69 (262.51) |
| Breed [reference: SR (SR ≥ 80%)] | ||||
| SH (SH ≥ 80%) | −15.59 (18.43) | −28.60 (15.68) | −47.18 (21.65) | 23.01 (61.59) |
| SR + SH ≥ 50% | 7.46 (15.88) | −5.85 (14.37) | −14.85 (18.65) | 58.67 (53.08) |
| Other_breed | 2.18 (20.79) | −17.25 (18.81) | 17.44 (24.42) | 48.73 (69.49) |
| Housing type (reference: freestall housing, noninsulated) | ||||
| Freestall housing, insulated | 17.04 (21.48) | −37.76 (19.44) | −3.51 (25.23) | 17.64 (71.79) |
| Tiestall | 3.00 (71.22) | −33.66 (64.43) | 8.27 (83.64) | 74.83 (237.99) |
| Milking system (reference: tie stall) | ||||
| AMS | 10.01 (61.76) | −37.92 (55.88) | 56.68 (72.54) | 101.06 (206.39) |
| Milking parlor | 44.35 (61.16) | 13.69 (55.33) | 80.53 (71.83) | 117.07 (204.37) |
| Rotary | 81.66 (67.73) | 27.42 (61.28) | 121.96 (79.55) | 343.80 (226.34) |
| Production system and milking system | ||||
| Conven_tiestall milking | −15.19 (52.62) | 0.73 (47.60) | 22.88 (61.80) | 41.88 (175.83) |
| Conven_AMS | 18.98 (40.02) | 18.00 (36.21) | −31.23 (47.00) | −9.39 (133.74) |
| Conven_parlor | −28.79 (68.63) | −35.26 (62.09) | 22.09 (80.60) | 73.66 (229.33) |
| Organic_tiestall milking | 81.05 (73.46) | 21.95 (68.67) | 10.15 (89.14) | −162.98 (253.64) |
| Organic_parlor | −0.87 (75.90) | 93.19 (66.46) | 47.26 (86.27) | −126.53 (245.48) |
| Year effect | ||||
| 2010 | 25.71 (32.73) | −16.35 (29.62) | −34.02 (38.44) | 67.33 (109.39) |
| 2011 | 10.28 (31.95) | −6.07 (28.91) | 2.51 (37.53) | 33.12 (106.78) |
| 2012 | 3.65 (31.01) | −2.78 (28.06) | 1.05 (36.42) | 98.89 (103.63) |
| 2013 | 18.22 (30.39) | 0.14 (27.49) | −12.47 (35.67) | 47.59 (101.57) |
| 2014 | 38.29 (30.06) | 21.26 (27.19) | 30.53 (35.30) | 64.64 (100.44) |
| 2015 | 48.87 (30.36) | 50.24 (25.46) | 52.46 (35.65) | 141.95 (101.45) |
| 2016 | 55.54 (28.35) | 49.42 (23.46) | 44.35 (35.65) | 127.90 (101.43) |
| 2017 | 34.78 (10.82) | 71.81 (27.88) | 40.111 (36.20) | 144.92 (102.99) |
| 2018 | 56.04 (28.89) | 77.01 (27.94) | 68.79 (33.27) | 109.99 (103.21) |
| Constant | 543.69 (143.33) | 89.87 (129.68) | 566.46 (168.34) | 1,070.29 (478.98) |
| Observation | 2,188 | 2,188 | 2,188 | 2,188 |
| Pseudo R2 | 0.24 | 0.21 | 0.26 | 0.25 |
1 BMSCC = bulk milk somatic cell count; SR = Swedish Red; SH = Swedish Holstein.
***, **, * show significance at 1%, 5%, and 10% levels, respectively.
Another interesting result relates to investments in farm infrastructure. The results show that investments in farm buildings prolongs the productive lifespan of the dairy herd. This effect is realized only in the 75th quantile, where the average productive lifespan is about 1,069 d, with a standard deviation of 86 d.
The estimated effects of milk yield monotonically increase from the lowest (25th) to the highest (95th) quantile. This result indicates that across the quantiles, an incremental increase in milk yield prolongs the productive lifespan of dairy cows (cows with high milk yield are kept in the herd). Similarly, the results show that calving interval is significantly positive across the 4 quantiles. The estimated effects continually increase from the 25th to 75th quantile. The estimated effect of age at first calving is significant and negative only in the lowest quantile (25th). This calving interval had no significant effect in the mean-based results presented in Table 5.
The BMSCC, which was significant in the mean-based analysis, is found to be significant and positive only in the lower (25th) and median (50th) quantiles, indicating that higher BMSCC at the lower and median quantiles prolongs the lifespan of the herd. In herds with prolong lifespan, famers keep cows with high SCC and hence, the higher BMSCC. The estimated effect of herd size on productive life is statistically significant and positive only in the lowest quantile, suggesting that higher herd size at the lower quantile prolongs the productive lifespan of dairy herd. This variable was not significant in the mean-based results presented in Table 5.
In terms of breed, we found that the effect of having at least 80% Swedish Holstein on the productive life of dairy cows is significant and negative in the median and 75th quantiles, compared with herds with at least 80% Swedish Red. This means that the productive life of dairy herds dominated by Swedish Holstein cows reduces in the median and 75th quantiles, where the average productive life is about 942 and 1,069 d, respectively. In addition, the heterogeneous results show that the estimated effect of insulated freestall housing, compared with noninsulated freestall housing, on the productive life of dairy cows was negative and significant only in the median quantile. In terms of year fixed effects, the results indicate that the productive life of dairy cows in the lowest quantile (25th) increased from 2016 to 2018. In the median quantile, the results indicate that productive life of dairy cows increased from 2015 to 2018.
DISCUSSION
In this paper, we investigated how herd animal health and farm investment decision relate to dairy cow longevity on the herd level, while controlling for farm-specific characteristics and animal management practices. This study builds on previous literature that has investigated the relationship between longevity and intrinsic (i.e., genetics and traits) factors (
Zehetmeier et al., 2014
; Haile-Mariam and Pryce, 2015
; De Vries, 2017
) and the extrinsic reasons for culling dairy cows, such as availability of replacement heifers, land availability, and prices (Bergeå et al., 2016
; Grandl et al., 2019
; Gambonini et al., 2022
). This paper is one of the first to establish the extent to which a farms' decision to invest in farm infrastructure affects cow longevity. Thus, we examined the extent to which dairy cow longevity is affected by the farmers' decision to increase the number of places or keep the dairy operation at a constant size at a particular point in time. Similar to Ahlman et al., 2011
, De Vries, 2017
, and Rostellato et al., 2021
, cow longevity was considered using length of productive life and age at culling. These variables were continuous and allowed us to examine the average and heterogeneous effects of the variables of interest and the control variables.Regarding investment in farm infrastructure, the findings of the present study demonstrate that dairy cow longevity is significantly linked to changes in investments made by the farmer, namely, to increase farm building space and the number of places for dairy cows. The heterogeneous effect results show that the stage of productive life at which the farmer makes changes in farm infrastructure is crucial to cow longevity. This finding is in line with the Asset Replacement Theory (
Hartman and Tan, 2014
), which says that replacing assets is an important decision that every producer faces, and that the replacement decision is usually driven by increasing operating and maintenance costs associated with existing assets or improvements of existing assets in the market. In applying this basic economic theory to dairy cow longevity and the replacement of heifers, the best time to replace existing dairy cows would be determined by comparing the marginal revenue from the existing cows with the anticipated returns from replacement heifers (De Vries, 2017
; Vredenberg et al., 2021
; Gambonini et al., 2022
). Thus, the optimal time to replace existing cows will be the point where the annuity value of the cow falls below the maximum annuity value the farmer can attain from the replacement heifer. However, with an investment in farm infrastructure, there will be additional space for the replacement heifers, and the farmer will not have to cull old cows that are still productive. In the absence of investment in farm buildings, farmers would have to cull old cows to make room for new recruitment heifers, or they would have to bring in fewer recruitment animals. De Vries and Marcondes, 2020
, who argued that without space for new heifers, farmers would decide to cull existing cows to make space for new heifers, support this assertion.On average, animal health, proxied by the proportion of cows receiving veterinary treatment, negatively and insignificantly correlates with herd longevity. This finding is supported by the study by
Fall et al., 2008
, who also found no significant relationship between the length of productive life and the proportion of the herd receiving veterinary treatment in Sweden. However, heterogeneous findings from this study show that an increase in disease in the early stages of the herd's productive life (26 mo) significantly reduces longevity. For instance, primiparous cows with severe disease in early lactation are culled. This is in accordance with De Vries and Mercondes (2020) who opined that health problems are major drivers of culling at a young age.Among the control variables, increase in herd average milk yield significantly prolongs average herd longevity, and this positive relationship increases monotonically along all the 4 heterogeneous quantiles examined. This is supported by
Rostellato et al., 2021
who found that dairy farmers are more likely to cull herds with low output even if they had good health characteristics. Haile-Mariam and Pryce, 2015
found that increases in milk yield reduced culling in dairy cows in Australia. Longer calving interval prolongs herd longevity, and this is in line with the findings of Do et al., 2013
, who found calving interval and age at first calving to exert a positive effect on the longevity of Korean Holstein cows. Similarly, Remmik et al., 2020
found a longer calving interval to be positively associated with a higher productive life among dairy cows, also reducing the probability of early culling. The BMSCC was positively associated with cow longevity, on average, and this is contrary to the findings of Samoré et al., 2003
, who found high SCC to be associated with a high rate of culling. However, our heterogeneous findings showed that high BMSCC was positively associated with cow longevity in herds in the lower to medium quantiles of longevity. It is worth mentioning that it is not SCC per se that positively influences longevity, rather older cows have higher SCC. However, the disparity in the level of significance of SCC is in line with the Fuerst-Waltl et al., 2004
, who found inconsistent results for SCC across different parities.Increasing the age at first calving in a herd reduces the longevity of cows at an early stage of their productive life. This finding is consistent with a Swedish study by
Hultgren and Svensson, 2009
, who found that a Swedish dairy herd with an age at first calving of 27 to 28 mo to be 1.1 times more likely to have a shorter length of productive life relative to cows with an age at first calving younger than 25 mo. This could be attributed to an increased risk of complications at calving for older heifers as they become fatter. Sherwin et al., 2016
found a positive relationship between age at first calving and the length of productive life for cows in the United Kingdom.On average, the dominating breed of the herd had no significant effect on the length of productive life, and this is contrary to findings by
Garcia-Peniche et al., 2006
, who reported differences in survival probabilities between Holstein, Brown Swiss, and Jersey cows in the United States. However, Garcia-Peniche et al., 2006
support the heterogeneous findings relating to the negative relationship between Holstein herds and a herd average longevity around 31 to 36 mo. The authors found Holstein herds to have a lower longevity compared with Brown Swiss or Jersey breeds.In terms of animal health, this study shows that high milk yield is positively associated with the proportion of cows receiving veterinary treatment, whereas high SCC is negatively associated with the proportion of cows receiving veterinary treatment. Existing studies by
Cinar et al., 2015
found that the somatic cell count is negatively correlated with milk yield. Yalçın et al., 2000
and Król et al., 2010
indicated that a high SCC is positively associated with milk losses and milk quality. Thus, existing studies show that SCC affects milk output. Contrary to studies by Sutherland et al., 2013
, Weller and Bowling, 2000
, and Rutherford et al., 2009
, the present study showed that in herds with organic production, the proportion of cows with veterinary treatment was lower compared with herds with conventional system. The proportion of cows with veterinary treatments was lower in herds with an insulated freestall housing compared with herds with a noninsulated freestall housing. We do not know why that differs and will not make any speculation on that. More research is needed to clarify that association.Moreover, in herds with milking parlors compared with tiestalls, the proportion of the herd with veterinary treatment was reduced. Earlier findings by
van den Borne et al., 2021
and Frössling et al., 2017
indicated that AMS leads to poorer teat and udder health relative to conventional milking systems, such as the milking parlor. The positive correlation between the use of AMS and the proportion of the herd receiving veterinary treatment is not supported by earlier studies by Neijenhuis et al., 2004
and - Neijenhuis F.
- Bos K.
- Sampimon O.C.
- Poelarends J.
- Hillerton J.E.
- Fossing C.
- Dearing J.
Changes in teat condition in Dutch herds converting from conventional to automated milking.
in: Meijering A. Hogeveen H. de Koning C.J.A.M. Automatic Milking—A Better Understanding. Wageningen Academic Publishers,
2004: 141-147
Zecconi et al., 2004
found no significant change in udder health in cows with AMS. It is important to note that this study considered the proportion of the herd treated for all diseases, including udder diseases.- Zecconi A.
- Piccinini R.
- Casirani G.
- Binda E.
- Migliorati L.
Introduction of AMS in Italian dairy herds: Effects on teat tissues, intramammary infection risk, and spread of contagious pathogens.
in: Meijering A. Hogeveen H. de Koning C.J.A.M. Automatic Milking—A Better Understanding. Wageningen Academic Publishers,
2004: 161-167
From an empirical point of view, the UQL gave us the opportunity to better comprehend and assess the sole effects of animal health and farm investment on longevity, regardless of the effects of other control variables (
Mishra et al., 2015
). Indeed, the findings from this study suggest that the effect of animal health and farm investments on longevity varies significantly across quantiles.The implication of these findings is that the short longevity of dairy herd in Sweden relative to other dairy producing countries is not a result of problems with health and welfare. Rather, dairy cow longevity in Sweden hinges on the farmers' investment, farm-specific characteristics, and animal management decisions. Thus, productive lifespan of dairy herd is largely based on the dairy farmer's decision-making.
With the societal concern of farm animal welfare and environmental sustainability, it is probable that the society will require higher longevity of dairy herd in accordance with life expectancy. Our findings point out that higher longevity can be attained through investment in farm building to accommodate recruiting heifers as well as herd management. Extending the productive life of dairy herd could enhance environmental sustainability as well as building a resilient dairy sector using farm management and decision support.
CONCLUSIONS
Based on our findings, the following conclusions are drawn: first, investment in farm infrastructure prolongs dairy cow longevity in Sweden. The investment in farm buildings creates room for new or superior recruitment heifers, without the need to cull already existing dairy cows. This means that the culling decision is indeed a strategic decision related to investment choices made at the farms. Second, in Swedish dairy herds, there is no association between proportion of unhealthy cows in the herd reflected as the proportion receiving veterinary treatment and the average length of productive life of the cows. Animal health, on average, does not have a significant effect on the longevity of a dairy herd in Sweden, and culling is predominantly done for other reasons than poor health. In addition, production variables that were associated with higher dairy herd longevity include higher milk yield and long calving interval. Prolonged age at first calving reduces the longevity of the herd. The findings further show that the productive lifespan of dairy herds in Sweden started increasing significantly from 2015 to the present day.
ACKNOWLEDGMENTS
This study was funded by the Swedish Government Research Council for Sustainable Development (FORMAS; Stockholm, Sweden), grant no: 2020-02507, which is gratefully acknowledged. The authors have not stated any conflicts of interest.
APPENDIX
Appendix Table A1Effects of animal health, investment, farm-specific, and animal management on age at culling; dependent variable is total lifespan
| Variable | Coefficient | R. SE | t-statistic | P-value |
|---|---|---|---|---|
| Health | −0.07 | 0.30 | −0.25 | 0.78 |
| Investment | 38.63 | 16.05 | 2.41 | 0.02 |
| Production indicator | ||||
| Milk yield (kg ECM) | 0.02 | 0.00 | 4.73 | 0.00 |
| Calving interval (mo) | 37.29 | 5.75 | 6.49 | 0.00 |
| Age at first calving (mo) | 0.82 | 0.07 | 11.01 | 0.00 |
| BMSCC, ×1,000 cells/mL | 0.28 | 0.07 | 3.72 | 0.00 |
| Herd size (no. of cows) | 0.06 | 0.06 | 0.98 | 0.33 |
| Production system (reference: conventional) | ||||
| Certified organic (KRAV) | −74.39 | 62.68 | −1.19 | 0.24 |
| Breed [reference: SR (SR ≥ 80%)] | ||||
| SH (SH ≥ 80%) | −21.79 | 16.05 | −1.36 | 0.18 |
| SR + SH ≥ 50% | −2.96 | 14.65 | −0.20 | 0.84 |
| Other_breed | 5.93 | 20.88 | 0.28 | 0.78 |
| Housing type (reference: freestall housing, noninsulated) | ||||
| Freestall housing, insulated | −7.57 | 15.77 | −0.48 | 0.63 |
| Tiestall | −44.73 | 57.68 | −0.78 | 0.44 |
| Milking system (reference: tiestall) | ||||
| AMS | −18.98 | 31.30 | −0.61 | 0.54 |
| Milking parlor | 10.130 | 30.99 | 0.33 | 0.74 |
| Rotary | 69.54 | 26.19 | 2.65 | 0.06 |
| Production system and milking system | ||||
| Conven_tiestall milking | −13.14 | 43.14 | −0.30 | 0.76 |
| Conven_AMS | −8.94 | 29.45 | −0.30 | 0.76 |
| Conven_parlor | −32.65 | 51.94 | −0.63 | 0.53 |
| Organic_tiestall milking | 35.59 | 64.40 | 0.55 | 0.58 |
| Organic_parlor | 82.66 | 58.71 | 1.41 | 0.16 |
| Year effects (reference: 2009) | ||||
| 2010 | 8.65 | 27.69 | 0.31 | 0.76 |
| 2011 | 12.41 | 27.07 | 0.46 | 0.65 |
| 2012 | 22.32 | 28.00 | 0.80 | 0.43 |
| 2013 | 16.83 | 26.70 | 0.63 | 0.53 |
| 2014 | 41.63 | 26.22 | 1.59 | 0.11 |
| 2015 | 72.384 | 26.56 | 2.79 | 0.01 |
| 2016 | 73.76 | 27.84 | 2.65 | 0.01 |
| 2017 | 67.52 | 26.86 | 2.51 | 0.01 |
| 2018 | 85.87 | 28.50 | 3.01 | 0.00 |
| Constant | 701.31 | 130.20 | 5.39 | 0.00 |
| Observation | 2,188 | |||
| Adjusted R2 | 0.24 | |||
| F-statistic | 18.56 | |||
1 R. SE = robust standard error; BMSCC = bulk milk somatic cell count; SR = Swedish Red; SH = Swedish Holstein.
*** and ** show significance at 1% and 5% levels respectively; dependent variable is age at culling.
Appendix Table A2Heterogeneous effects of animal health, investment and farm management variables on age at culling; dependent variable is total lifespan
| Variable | 25th quantile | 50th quantile | 75th quantile | 95th quantile |
|---|---|---|---|---|
| Health | −0.74 (0.32) | −0.15 (0.30) | −0.01 (0.39) | 0.056 (1.10) |
| Investment | 11.41 (20.52) | 9.91 (18.56) | 44.51 (22.10) | 88.19 (68.56) |
| Control variable | ||||
| Production indicator | ||||
| Milk yield (kg ECM) | 0.02 (0.00) | 0.01 (0.00) | 0.03 (0.01) | 0.06 (0.02) |
| Calving interval (mo) | 30.38 (5.57) | 33.14 (5.04) | 46.98 (6.54) | 31.28 (17.67) |
| Age at first calving (mo) | 0.84 (0.067) | 0.96 (0.06) | 1.01 (0.08) | 1.014 (0.22) |
| BMSCC (×1,000 cells/mL) | 0.30 (0.07) | 0.26 (0.07) | 0.11 (0.09) | −0.12 (0.25) |
| Herd size (no. of cows) | 0.29 (0.08) | 0.04 (0.08) | 0.07 (0.11) | −0.03 (0.31) |
| Production system (reference: conventional) | ||||
| Certified organic (KRAV) | −56.83 (78.55) | −75.28 (71.07) | −42.78 (92.26) | 169.69 (262.51) |
| Breed [reference: SR (SR ≥ 80%)] | ||||
| SH (SH ≥ 80%) | −15.59 (18.43) | −28.60 (14.68) | −47.18 (21.65) | 23.01 (61.59) |
| SR+SH ≥ 50% | 7.46 (15.88) | −5.85 (14.37) | −14.85 (18.65) | 58.67 (53.08) |
| Other_breed | 2.18 (20.80) | −17.25 (18.81) | 17.44 (24.42) | 48.73 (69.49) |
| Housing type (reference: freestall housing, noninsulated) | ||||
| Loose housing, insulated | 17.04 (21.48) | −37.76 (17.44) | −3.507 (25.230) | 17.64 (71.79) |
| Tiestall | 3.00 (71.22) | −33.66 (64.43) | 8.272 (83.643) | 74.83 (38.99) |
| Milking system (reference: tiestall) | ||||
| AMS | 10.01 (61.76) | −37.92 (55.88) | 56.68 (72.54) | 101.06 (206.39) |
| Milking parlor | 44.35 (61.16) | 13.69 (55.33) | 80.53 (71.83) | 117.07 (204.37) |
| Rotary | 81.66 (67.73) | 27.42 (61.28) | 121.955 (79.55) | 343.80 (226.34) |
| Production system and milking system | ||||
| Conven_tiestall milking | −15.19 (52.62) | 10.73 (47.60) | 22.88 (61.80) | 41.88 (75.83) |
| Conven_AMS | 18.98 (40.02) | 18.00 (36.21) | −31.23 (47.00) | −9.39 (13.73) |
| Conven_parlor | −28.80 (68.63) | −35.26 (62.09) | 22.09 (80.60) | 73.66 (39.33) |
| Organic_tiestall milking | −0.87 (5.90) | 21.95 (68.67) | 10.15 (89.14) | −162.98 (253.64) |
| Organic_parlor | 81.05 (73.46) | 93.19 (66.46) | 47.26 (86.27) | −126.53 (245.48) |
| Year effects (reference: 2009) | ||||
| 2010 | 25.71 (32.73) | −16.35 (29.62) | −34.02 (38.44) | 67.32 (49.39) |
| 2011 | 10.28 (31.95) | −6.07 (8.91) | 2.51 (3.53) | 33.12 (26.78) |
| 2012 | 3.65 (3.01) | 2.78 (2.06) | 1.051 (3.42) | 98.89 (103.63) |
| 2013 | 18.22 (30.39) | 0.14 (2.49) | −12.47 (35.70) | 47.59 (31.57) |
| 2014 | 38.30 (30.06) | 21.26 (27.19) | 30.53 (35.30) | 64.64 (40.44) |
| 2015 | 48.87 (28.36) | 50.24 (23.46) | 52.46 (35.65) | 141.95 (101.45) |
| 2016 | 55.54 (28.35) | 49.42 (24.46) | 44.35 (35.65) | 127.91 (101.43) |
| 2017 | 34.78 (10.8) | 71.81 (27.88) | 40.11 (36.20) | 144.92 (102.99) |
| 2018 | 56.037 (28.89) | 77.01 (27.94) | 68.798 (31.27) | 109.99 (103.21) |
| Constant | 543.79 (143.36) | 689.88 (129.68) | 565.32 (168.35) | 1,071.29 (478.97) |
| Observation | 2,188 | 2,188 | 2,188 | 2,188 |
| Pseudo R2 | 0.27 | 0.38 | 0.16 | 0.22 |
1 BMSCC = bulk milk somatic cell count; SR = Swedish Red; SH = Swedish Holstein.
***, **, * show significance at 1%, 5%, and 10% levels, respectively. Values in parentheses are standard errors.
REFERENCES
- Productive life span and resilience rank can be predicted from on-farm first-parity sensor time series but not using a common equation across farms.J. Dairy Sci. 2020; 103 (32475663): 7155-7171
- An unconditional quantile regression based decomposition of spatial welfare inequalities in Ghana.J. Dev. Stud. 2018; 54: 537-556
- Culling reasons in organic and conventional dairy herds and genotype by environment interaction for longevity.J. Dairy Sci. 2011; 94 (21338822): 1568-1575
- Breeding for sustainable milk production: From nucleus herds to genomic data.(PhD Diss.) Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, 2013https://pub.epsilon.slu.se/10438/Date accessed: September 2, 2021
- Econometrics.Springer, 2011
- Effects of changing cow production and fitness traits on profit and greenhouse gas emissions of UK dairy systems.J. Agric. Sci. 2015; 153: 138-151
- Farmer awareness of cow longevity and implications for decision-making at farm level.Acta Agric. Scand. A Anim. Sci. 2016; 66: 25-34
- Highlighting differences between conditional and unconditional quantile regression approaches through an application to assess medication adherence.Health Econ. 2013; 22 (23616446): 1052-1070
- An empirical analysis of the cost of rearing dairy heifers from birth to first calving and the time taken to repay these costs.Animal. 2017; 11 (28173887): 1372-1380
- A descriptive study of the survival of Holstein-Friesian heifers through to third calving on English dairy farms.J. Dairy Sci. 2011; 94 (21426972): 1831-1838
- The cost of different types of lameness in dairy cows calculated by dynamic programming.Prev. Vet. Med. 2010; 97 (20801533): 1-8
- Effect of somatic cell count on milk yield and composition of first and second lactation dairy cows.Ital. J. Anim. Sci. 2015; 143646
- Keeping dairy cows for longer: A critical literature review on dairy cow longevity in high milk-producing countries.Animals (Basel). 2021; 11 (33805738): 808
- Economic value of delayed replacement when cow performance is seasonal.J. Dairy Sci. 2004; 87 (15375056): 2947-2958
- Cow longevity economics: The cost benefit of keeping the cow in the herd. Proceedings from the Cow Longevity Conference 2013 that took place at Hamra farm, Sweden in August 2013.https://www.thecattlesite.com/articles/3950/cow-longevity-economics-cost-benefits-of-keeping-a-cow-in-the-herd/Date: 2013Date accessed: August 12, 2021
- Economic trade-offs between genetic improvement and longevity in dairy cattle.J. Dairy Sci. 2017; 100 (28215896): 4184-4192
- Review: Overview of factors affecting productive lifespan of dairy cows.Animal. 2020; 14: s155-s164
- OPEC news and predictability of energy futures returns and volatility: Evidence from a conditional quantile regression.J. Econ. Finance Adm. Sci. 2020; 25: 239-259
- The effect of age at first calving and calving interval on productive life and lifetime profit in Korean Holsteins.Asian-Australas. J. Anim. Sci. 2013; 26 (25049735): 1511-1517
- Reproductive performance, general health, and longevity of dairy cows at a Swedish research farm with both organic and conventional production.Livest. Sci. 2008; 118: 11-19
- Crops and livestock products.
- Unconditional quantile regressions.Econometrica. 2009; 77: 953-973
- Incidence and duration of increased somatic cell count in Swedish dairy cows and associations with milking system type.J. Dairy Sci. 2017; 100 (28647335): 7368-7378
- Effect of maternal age on milk production traits, fertility, and longevity in cattle.J. Dairy Sci. 2004; 87 (15328244): 2293-2298
- Estimation and analysis of cow-level cumulative lifetime break-even on financial resiliency.J. Dairy Sci. 2022; 105 (35282908): 4653-4668
- Effects of breed and region on longevity traits through five years of age in Brown Swiss, Holstein, and Jersey cows in the United States.J. Dairy Sci. 2006; 89 (16899703): 3672-3680
- Impact of longevity on greenhouse gas emissions and profitability of individual dairy cows analysed with different system boundaries.Animal. 2019; 13 (29807552): 198-208
- Variances and correlations of milk production, fertility, longevity, and type traits over time in Australian Holstein cattle.J. Dairy Sci. 2015; 98 (26254520): 7364-7379
- Farmers' motives for diversifying their farm business – The influence of family.J. Rural Stud. 2013; 32: 240-250
- Equipment replacement analysis: A literature review and directions for future research.Eng. Econ. 2014; 59: 136-153
- Distributional effects of welfare reform for young adults: An unconditional quantile regression approach.Labour Econ. 2020; 65101818
- Heifer rearing conditions affect length of productive life in Swedish dairy cows.Prev. Vet. Med. 2009; 89 (19297046): 255-264
- Consumer decision-making for animal-friendly products: Synthesis and implications.Anim. Welf. 2011; 20: 11-19
- National genetic evaluation info – France – Combined longevity.
- Effect of somatic cell counts in milk on its quality depending on cow breed and season.Annales UMCS Zootechnica. 2010; 28: 9-17
- A meta-analysis of consumer willingness to pay for farm animal welfare.Eur. Rev. Agric. Econ. 2011; 38: 55-78
- Cull rates: How is your farm doing? Penn State Extension.https://extension.psu.edu/cull-rates-how-is-your-farm-doingDate: 2020Date accessed: November 21, 2022
- Modeling the effect of off-farm income on farmland values: A quantile regression approach.Econ. Model. 2013; 32: 361-368
- Impact of off-farm income on food expenditures in rural Bangladesh: An unconditional quantile regression approach.Agric. Econ. 2015; 46: 139-148
- Changes in teat condition in Dutch herds converting from conventional to automated milking.in: Meijering A. Hogeveen H. de Koning C.J.A.M. Automatic Milking—A Better Understanding. Wageningen Academic Publishers, 2004: 141-147
- Identifying risk factors for poor hind limb cleanliness in Danish loose-housed dairy cows.Animal. 2011; 5 (22440353): 1613-1619
- Culling and Wastage in dairy herds: an update on incidence and economic impact in dairy herds in the UK.Cattle Pract. 2010; 18: 163-172
- Unconditional quantile regression and tourism expenditure: The case of the Canary Islands.Tour. Econ. 2021; 27: 626-648
- Herd-level risk factors for cow and calf on-farm mortality in Estonian dairy herds.Acta Vet. Scand. 2020; 62 (32164740): 15
- Impact of calving interval on milk yield and longevity of primiparous Estonian Holstein cows.Czech J. Anim. Sci. 2020; 65: 365-372
- Understanding Swedish dairy farmers' view on breeding goals –Ethical aspects of longevity.in: Olsson I.A.S. Araújo S.M. Vieira M.F. Food Futures: Ethics, Science and Culture. Wageningen Academic Publishers, 2016: 61-66
- Influence of production, reproduction, morphology, and health traits on true and functional longevity in French Holstein cows.J. Dairy Sci. 2021; 104 (34593220): 12664-12678
- Lameness prevalence and risk factors in organic and non-organic dairy herds in the United Kingdom.Vet. J. 2009; 180 (18462961): 95-105
- Relationship between somatic cell count and functional longevity assessed using survival analysis in italian Holstein-Friesian cows.Livest. Prod. Sci. 2003; 80: 211-220
- The association between age at first calving and survival of first lactation heifers within dairy herds.Animal. 2016; 10 (27132864): 1877-1882
- Animal health and welfare issues facing organic production systems.Animals. 2013; 3: 1021-1035
- Harvest for organic and conventional cultivation in 2021.
- Perceived importance and responsibility for market-driven pig welfare: Literature review.Meat Sci. 2017; 125 (27886640): 37-45
- Trends in somatic cell count deteriorations in Dutch dairy herds transitioning to an automatic milking system.J. Dairy Sci. 2021; 104 (33612244): 6039-6050
- Husdjursstatistics. Cattle statistics.
- An empirical analysis on the longevity of dairy cows in relation to economic herd performance.Front. Vet. Sci. 2021; 8 (33912606)646672
- Health status of dairy herds in organic farming.Vet. Rec. 2000; 146 (10674698): 80-81
- What could an increase in dairy cow culling mean for the markets?.Agriculture and Horticulture Development Board (AHDB), 2020https://ahdb.org.uk/news/what-could-an-increase-in-dairy-cow-culling-mean-for-the-marketsDate accessed: November 20, 2022
- Estimation of milk yield losses from subclinical mastitis in dairy cows.Turk. J. Vet. Anim. Sci. 2000; 24: 599-604
- Introduction of AMS in Italian dairy herds: Effects on teat tissues, intramammary infection risk, and spread of contagious pathogens.in: Meijering A. Hogeveen H. de Koning C.J.A.M. Automatic Milking—A Better Understanding. Wageningen Academic Publishers, 2004: 161-167
- A dominance analysis of greenhouse gas emissions, beef output and land use of German dairy farms.Agric. Syst. 2014; 129: 55-67
Article info
Publication history
Published online: April 05, 2023
Accepted:
December 18,
2022
Received:
September 22,
2022
Identification
Copyright
© 2023 The Author(s).
User license
Creative Commons Attribution (CC BY 4.0) | How you can reuse 
Elsevier's open access license policy
Creative Commons Attribution (CC BY 4.0)
Permitted
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
Elsevier's open access license policy
00162-5&id=gr1.jpg){kind=link}