Content uploaded by Elias Maritan
Author content
All content in this area was uploaded by Elias Maritan on Jun 06, 2024
Content may be subject to copyright.
1
An Economic Analysis of Bolus-sensor Technology for Precision Dairy Cattle
Management
Elias Maritan a*, Gundula Hoffmann b, Friederike Schwierz c, S. Mark Rutter d, Andreas Meyer-Aurich c, James
Lowenberg-DeBoer a, Karl Behrendt a
a Food, Land & Agribusiness Management Department, Harper Adams University, Newport, United Kingdom
b Sensor and Modelling Department, Leibniz Institute of Agricultural Engineering and Bioeconomy, Potsdam, Germany
c Technology Assessment Department, Leibniz Institute of Agricultural Engineering and Bioeconomy, Potsdam, Germany
d Animal Health, Behaviour and Welfare Department, Harper Adams University, Newport, United Kingdom
* Corresponding author. Email: emaritan@harper-adams.ac.uk
Abstract
Bolus-sensors are a precision livestock farming tool measuring physiological, behavioural, and production
indicators of individual animals to improve herd productivity. Bolus-sensors collect data such as animal activity,
rumination, and body temperature. On dairy farms, these data may be used to assess general cattle health, detect
oestrus, and monitor presence of disease.
This study focuses on oestrus detection and clinical mastitis (CM) monitoring via bolus-sensors on two dairy
farms in northeastern Germany. The two farms use an automatic (AMS) and a conventional milking system (CMS),
respectively. A Monte Carlo partial budgeting analysis is performed to quantify the probability of positive net
economic outcomes after bolus-sensor adoption. Following previous economic research, the bolus-sensor is
assumed to affect milk production and herd growth rates. Annual net economic outcomes are calculated for the
two target functions and combined to assess the overall outcome when the bolus-sensor is utilised as a multi-
functional tool.
Analysis results show that economic outcomes differ across farm types and functions considered. For CM
monitoring, the probability of bolus-sensor adoption leading to increased profits is 58% on AMS farms and 99%
on CMS farms. Conversely, the probabilities for oestrus detection on AMS and CMS farms are 80% and 59%,
respectively. When the two functions are combined, the likelihood of increased profit is 75% on AMS farms and
93% on CMS farms. Annual economic benefits span from €26,470 to €110,301 on the AMS farm, and from
€27,420 to €197,763 on the CMS farm. The mean economic advantage for both target applications is €126,257
year-1 on the AMS farm and €214,670 year-1 on the CMS farm. However, these outcomes are highly variable and
more data on the effects of bolus-sensors on herd productivity are required. Additionally, further research is
needed to identify potential trade-offs among economic benefits, animal welfare, and other non-economic aspects.
Keywords: partial budgeting, Monte Carlo simulation, mastitis monitoring, oestrus detection, precision livestock
farming
1. Introduction
Precision livestock farming (PLF) is a management system relying on continuous, real-time, and automatic
collection of data related to individual animal behaviour, health, welfare, production, reproduction, and possibly
environmental impact (Mayo et al., 2019; Herlin et al., 2021). PLF technologies are adopted to optimise livestock
system efficiency (Mayo et al., 2019). Since the advent of milking robots in the 1990s, sensor technologies capable
of measuring physiological, behavioural, and production indicators of individual animals have gained increased
attention in PLF (Steeneveld et al., 2015; Herlin et al., 2021). These tools may replace or even enhance human
inspection, thus leading to substantial labour savings, while also enabling more intensive livestock management
by markedly increasing the amount of collected data (Herlin et al., 2021). Frequent, precise, and systematic data
collection may also improve farmers’ ability to abide with animal welfare standards, which are increasingly being
imposed via regulation globally (Herlin et al., 2021).
Sensors can be categorised as animal- and non-animal-based (Herlin et al., 2021). Animal-based sensors may
be fixed to ear tags, worn as collars or leg straps, or administered to the animal in the form of boluses (Herlin et
al., 2021). Non-animal-based sensors are usually developed for indoor systems and installed in the animals’
vicinity (e.g., heat stress detection cameras) (Herlin et al., 2021). On dairy cattle farms, sensor systems are
increasingly being used on a large scale to optimise farmers’ decision-making processes (Steeneveld et al., 2015).
These sensors may provide information on feed and water intake, oestrus, calving, and presence of disease (Herlin
et al., 2021). Certain diseases may even be detected before the manifestation of clinical signs, thus helping reduce
milk losses and antimicrobials use (Kim et al., 2019; Herlin et al., 2021). Because most research efforts have
concentrated on single-use sensors (Benaissa et al., 2020), this analysis focuses on two target applications in dairy
2
cattle systems to understand the economic performance of multi-functional sensor technology. The type of sensor
studied is an animal-based sensor-bolus placed in the cattle reticulorumen.
The first target application of the selected technology is the monitoring of clinical mastitis (CM). CM has
been the first focus of sensor development in the dairy sector (Hogeeven et al., 2010). It usually occurs during
lactation, thus severely impairing milk quantity and quality parameters on dairy farms (Kim et al., 2019). Early
CM detection may reduce negative economic and animal welfare impacts (Kim et al., 2019). Using temperature
data for CM detection is a reliable approach (e.g., Kim et al., 2019). This is especially true when data are gathered
by sensor-boluses which, unlike external sensors, have no temperature interference from the surrounding ambient
(Kim et al., 2019; Herlin et al., 2021). When compared to other diseases, CM leads to short-lived internal body
temperature increases (Kim et al., 2019). Temperature data are usually augmented by other data types (e.g., cattle
activity parameters) to improve CM detection accuracy. CM is conventionally monitored via visual inspection of
pre-milk (Hogeeven et al., 2010; Kim et al., 2019). This makes it impossible to detect mastitis at the subclinical
stage because clinical symptoms must first manifest (Kim et al., 2019). On dairy farms using automatic milking
systems (AMS), CM monitoring via visual observation is even more challenging (Hogeeven et al., 2010).
Alternative approaches to identifying CM cases include milk colour and homogeneity tests, measurement of milk
electrical conductivity (EC), and quantification of L-Lactate dehydrogenase (LDH) or somatic cell counts (SCC)
in milk (Hogeeven et al., 2010). Of these, EC, LDH and SCC are the most reliable (Hogeeven et al., 2010), but
LDH and SCC usually require third-party laboratory tests whose cost hinders farmers’ motivation to diligently
manage this disease (Boker et al., 2023). Thus, when visual observation of pre-milk is infeasible or impractical
(e.g., on AMS farms), EC readings are frequently used to monitor CM and identify milk samples requiring
laboratory testing (Steeneveld et al., 2015). Since the EC of milk can also be affected by factors other than CM
presence, temperature and other data collected by sensors may further improve the efficiency of EC monitoring
systems (Hogeeven et al., 2010).
The second target application of the studied technology is oestrus detection. The reproduction efficiency of
cattle significantly affects profitability on dairy farms (Benaissa et al., 2020). The most common approach to
oestrus detection is visual observation of animals manifesting sexually receptive behaviours that are associated
with reproductive readiness (Mayo et al., 2019; Lodkaew et al., 2023). Because this method is labour-intensive,
costly, and prone to errors (Mayo et al., 2019; Lodkaew et al., 2023), sensor technologies are increasingly being
used to monitor such behaviours automatically and continuously on dairy farms (Mayo et al., 2019; Benaissa et
al., 2020; Lodkaew et al., 2023). If oestrus is not promptly detected, farmers may fail to artificially inseminate
cattle at the optimal time before ovulation and must wait for another reproductive cycle (Lodkaew et al., 2023).
Breeding failure results in fewer calf births and consequently in lower herd growth rates (Lodkaew et al., 2023).
Previous research indicates that economic losses amount to US$ 360 per missed oestrus (Mayo et al., 2019).
Sensor systems have been shown to detect 80-85% of oestrus events, compared to approximately 55% when these
are identified via visual methods (Steeneveld et al., 2015). Improving oestrus detection by adopting sensor
technologies may reduce calving intervals, which is a known contributor to increased milk production on dairy
farms (Steeneveld et al., 2015).
The sensor-bolus selected for the present economic analysis is commercialised by smaXtec Animal Care
GmbH (Graz, Austria). The smaXtec® sensor-bolus continuously collects body temperature, rumination, water
intake, rumen pH, and activity data (smaXtec, 2024). It is currently commercialised in the US, New Zealand, and
several European countries (smaXtec, 2024). Upon sensor-bolus administration and activation, data are wirelessly
transmitted to the smaXtec Cloud TruDTM and processed by artificial intelligence (AI) (smaXtec, 2024).
Subsequently, the user receives processed data on a mobile phone application as well as recommendations for
action (smaXtec, 2024). It is here hypothesised that the multi-purpose smaXtec® sensor-bolus could improve the
economic performance of dairy farms. Economic effects of its use for CM monitoring and oestrus detection are
assumed to be variations in milk production and herd growth rates. A Monte Carlo (MC) simulation is performed
to quantify these effects on two dairy farms located in Germany. The two farms are an AMS farm and a
conventional milking system (CMS) farm. Results are generated to answer the three following research questions
(RQ): (i) what is the probability that smaXtec® provides a positive economic net outcome compared to
conventional CM monitoring approaches? (RQ1); (ii) what is the probability that smaXtec® provides a positive
economic net outcome compared to oestrus detection via visual observation? (RQ2); and (iii) what is the
probability that smaXtec® provides a positive economic net outcome when simultaneously considering CM
monitoring and oestrus detection applications? (RQ3). Interpretation of results is conducted from an
agroecological perspective by discussing some implications of sensor-bolus adoption that go beyond economic
considerations.
3
2. Materials and Methods
2.1. Description of the modelled farms
The two fictitious farms modelled in this study are located in the State of Brandeburg in northeastern Germany.
On the AMS farm, conventional CM monitoring and oestrus detection methods are assumed to be EC readings
and visual observation of cattle behaviour, respectively. The EC readers are installed on the AMS. AMSs may be
sold with an optional SCC sensor (e.g., GEA Group, 2024), but these sensors often rely on California Mastitis
Test protocols, which have been questioned for their low specificity rate (e.g., Lam et al., 2009; Kim et al., 2019).
Besides, a farm survey of 414 dairy farms conducted by Steeneveld et al. (2015) in the Netherlands found that EC
sensors were the most widely used sensor types on AMS farms. Thus, this analysis assumes that the AMS is not
equipped with SCC sensors and that the farm manager sends milk samples identified via EC readings to a
laboratory for diagnosing CM via more reliable tests such as the Porta SCC (Lam et al., 2009). The adoption rates
of AMSs are difficult to quantify because equipment retailers do not make their data publicly available (Eastwood
and Renwick, 2020). A recent estimate reported that 15% of German dairy farms used AMSs as of 2018 (Eastwood
and Renwick, 2020). Thus, a second CMS farm was included in this study because CMSs are still widely used in
Germany. On the modelled CMS farm, conventional CM monitoring is conducted via visual observation of pre-
milk (Lam et al., 2009; Boker et al., 2023), while conventional oestrus detection follows the same method of the
AMS farm. Labour times for conventional CM monitoring and oestrus detection are based on authors’ experience.
It is assumed that it takes 15 minutes to monitor the EC sensors at each milking event for the entire herd on the
AMS farm, and 15 seconds per cow at each milking event to visually assess pre-milk on the CMS farm. Average
milkings per day are assumed to be 2.7 and 2.2 on the AMS and CMS farms, respectively. For oestrus detection,
visual observation of cattle behaviour requires 50 minutes per day for the entire herd. The farm labour hourly rate
is assumed to be € 21.00 per hour (Achilles et al., 2020). The herd size is 227 milking cows, coinciding with the
mean herd size in Brandeburg (Tergast et al., 2023).
Both the AMS and CMS farms adopt the smaXtec® bolus-sensor system and move away from conventional
CM monitoring and oestrus detection practices. Milk production and herd growth rate variations pre- and post-
adoption of the bolus-sensor system are obtained from Steeneveld et al. (2015). Milk production is simulated via
10,000 MC iterations using mean, standard deviation, and correlation parameters provided in Steeneveld et al.
(2015). The assumed milk price is the weighted average price in Germany in 2023 (CLAL, 2024). Herd growth
rate increases are 2.50% and 2.20% on the AMS and CMS farms, respectively (Steeneveld et al., 2015). Calf
value is € 136.65 head-1 (Saxon State Ministry for Energy, Climate Protection, Environment and Agriculture,
2024). This value is for bull calves, which are assumed to be sold after four weeks following common practice in
the study area (Tergast et al., 2023). On the other hand, heifer calves are usually kept on farm as replacement
cattle (Tergast et al., 2023). Calf management costs are assumed at € 69.81 per bull calf from birth to sale (Redman,
2023). The latter include direct labour and variable costs such as milk substitute, concentrate, and bedding
(Redman, 2023).
2.2. Description of the adopted bolus-sensor system
The smaXtec® bolus-sensor has a size of 105 x 35 mm (Figure 1) (smaXtec, 2024). It is administered orally
using conventional bolus applicators (smaXtec, 2024). It has a data measurement frequency of 10 minutes and an
internal memory capacity of 6 days (smaXtec, 2024). The smaXtec® classic bolus collects the following data
types: (i) inner body temperature (accuracy of ± 0.01°C), (ii) water intake, (iii) rumination, and (iv) animal activity
(smaXtec, 2024). The adoption of the smaXtec® system involves three cost components. The upfront cost per
sensor-bolus in Germany is € 29.99 (smaXtec, 2024. Pers. Comm.). The sensor-boluses have an average useful
life of 5 years (smaXtec, 2024. Pers. Comm.). Subscription costs depend on the herd size, but these are on average
€ 2.99 per sensor per month (smaXtec, 2024. Pers. Comm.). Subscription costs cover data processing, internet
connection, customer support and sensor replacement charges in case of technical failure (smaXtec, 2024. Pers.
Comm.). Based on these figures, the per-cow annual cost of the smaXtec® technology is estimated at € 41.88.
Additionally, the smaXtec® system requires the installation of a base station costed at € 6,500, including
infrastructure, installation labour, and farmer training fees (smaXtec, 2024. Pers. Comm.). The base station is
where the data continuously collected by the sensor-boluses are processed. It is equipped with a common
subscriber identity module card, an antenna, a bolus applicator and climate sensors collecting ambient temperature,
air pressure, and humidity data (smaXtec, 2024. Pers. Comm.). The base station communicates with the smaXtec
Cloud TruDTM and provides the user with processed data, alerts, and recommendations for action via a mobile
application (smaXtec, 2024. Pers. Comm.). With an assumed useful life of 10 years, base station costs are
estimated at € 1007.50 year-1 herd-1. This figure includes annual insurance (assumed at € 65.00 year-1, i.e. 1% of
the base station cost) and opportunity cost of capital (€ 292.50 year-1 based on the most recent European Central
Bank fixed rate of 4.50%).
4
Figure 1. The classic smaXtec® bolus-sensor (smaXtec, 2024)
Following personal communication with smaXtec Animal Care GmbH, it is assumed that it requires
approximately 15 minutes per day to supervise the mobile application alerts received by the system for the entire
herd. For the herd size considered in this study, this is equivalent to 0.40 hours per cow per year for either CM
monitoring or oestrus detection. The smaXtec® system exploits AI algorithms to provide farmers with CM case
probabilities for individual animals based on body temperature, rumination, and activity data (smaXtec, 2024.
Pers. Comm.). CM can be confidently detected after a second alert but, depending on the specific CM pathogen,
certain CM cases may be detected at the first alert (e.g., E. coli CM) (smaXtec, 2024. Pers. Comm.). This may
enable substantial savings in antimicrobial use and laboratory testing even on AMS farms where the EC sensors
cannot detect EC changes in milk up to 48 hours after the first smaXtec® system alert (smaXtec, 2024. Pers.
Comm.). For oestrus detection, the smaXtec® AI algorithms rely on rumination activity and animal movement
patterns (smaXtec, 2024. Pers. Comm.). System alerts are triggered when an animal behaves in a substantially
different manner compared to the rest of the group (smaXtec, 2024. Pers. Comm.).
Although sensitivity and specificity percentages of the smaXtec® sensor are not available, this preliminary
analysis assumes that this system has a reasonably accurate performance for the two tested functions. Ingestible
sensors are an effective tool for CM monitoring, especially when body temperature data are correlated with
additional biometric data (Hogeveen et al., 2010; Kim et al., 2019). According to Hogeveen et al. (2010), the
minimum specificity and sensitivity of a CM detection system should be 80% and 99%, respectively. It is here
assumed that the smaXtec® system meets these requirements. Likewise, sensor-supported activity monitoring is
regarded as the most successful tool for oestrus detection (Benaissa et al., 2020). Thus, for oestrus detection, the
smaXtec® sensor is assumed to at least match the previously reported 80-85% oestrus detection rate (Steeneveld
et al., 2015). These assumptions will be revised in future research as more data become available.
2.3. Description of the partial budgeting model
The partial budgeting model used in the present economic analysis relies on the approach by Davies et al.
(2022). Davies et al. (2022) conducted a MC simulation with 10,000 iterations to explore the net economic
outcomes of transthoracic ultrasonography testing in subclinical cases of pulmonary adenocarcinoma in live sheep.
This methodology is applied to the use of the smaXtec® system for CM monitoring and oestrus detection in dairy
cattle over one year to quantify the probability of increased profit after the adoption of smaXtec®. The economic
factors considered include a higher milk production and an increased herd growth rate (Steeneveld et al., 2015).
The latter is a result of potentially shorter calving intervals. The net economic outcomes for CM monitoring
(Scenario 1) and oestrus detection (Scenario 2) are separately calculated and subsequently combined to estimate
an overall net outcome when smaXtec® is used as a multi-purpose technology (Scenario 3) (Table 1). The relative
influence of individual model parameters is explored via sensitivity analyses. The model was developed in
Microsoft® Excel® (Microsoft Corporation, 2024).
Table 1. Scenarios tested in the Monte Carlo partial budgeting model
Net economic outcome
AMS farm
CMS farm
Clinical mastitis monitoring
Scenario 1a
Scenario 1b
Oestrus detection
Scenario 2a
Scenario 2b
Overall net economic outcome
Scenario 3a
Scenario 3b
Net economic outcomes in Scenarios 1 and 2 are calculated via Eq.1. The individual components of Eq.1 are
estimated using Eq.2, Eq.3, and Eq.4 for Scenario 1 and Eq.5, Eq.6, and Eq.7 for Scenario 2. Overall net
outcomes (i.e., Scenario 3) are calculated via Eq.8.
5
(1)
(2)
where aCM is the effect on revenue of using smaXtec® for CM monitoring (€ year-1); Δα is a stochastic variable
accounting for the effect on milk production after adoption of the smaXtec® system (l cow-1); HS is the herd size
of the modelled farm; and p is the average weighted milk price in Germany in 2023 (€ kg-1).
(3)
where bCM is the cost increase incurred after adopting smaXtec® for CM monitoring (€ year-1); β1CM is the labour
time requirement to monitor the CM alerts on the smaXtec® application (h year-1 cow-1); lab is the hourly labour
rate in Germany (€ h-1); γ1 is the yearly adoption cost per cow of the smaXtec® system excluding infrastructure
costs (€ year-1 cow-1); γ2 is the yearly adoption cost per herd for the infrastructure costs of the smaXtec® system
(€ year-1 herd-1); and HS is described in Eq. 2.
(4)
where cCM are the costs saved thanks to the adoption of the smaXtec® system for CM monitoring (€ year-1); β0CM
is the labour time requirement to monitor CM via visual observation of pre-milk (h milking-1); mpd are average
milkings per day; and HS and lab are described in Eq.2 and Eq.3, respectively.
(5)
where aOD is the effect on revenue of using smaXtec® for oestrus detection (€ year-1); ΔHS is the herd growth rate
increase after adoption of smaXtec® (%); cv is bull calf value in Germany as of April 2024 (€ head-1); and Δα,
HS and p are described in Eq.2.
(6)
where bOD is the cost increase incurred after adopting smaXtec® for oestrus detection (€ year-1); β1OD is the labour
time requirement to monitor the oestrus detection alerts on the smaXtec® application (h year-1 cow-1); cmc is the
bull calf management cost until its sale (€ head-1); lab, γ1, HS, and γ2 are described in Eq.3; and ΔHS is described
in Eq.5.
(7)
where cOD are the costs saved thanks to the adoption of the smaXtec® system for oestrus detection (€ year-1); β0CM
is the labour time requirement to detect oestrus via visual observation of cattle behaviour (h year-1 cow-1); and HS
and lab are described in Eq.2 and Eq.3, respectively.
(8)
where aCM, bCM, cCM, aOD, bOD and cOD are the outputs of the previous six equations; HS is described in Eq.2; and
γ1, γ2 are described in Eq.3. The latter are subtracted from the overall net outcome to avoid double counting of the
smaXtec® system adoption costs.
3. Results
The probability that smaXtec® provides a positive net economic outcome compared to conventional CM
monitoring approaches was 58% on the AMS farm and 99% on the CMS farm (RQ1). Annual mean net economic
outcomes for Scenarios 1a and 1b are shown in Table 2.
Table 2. Mean net economic outcomes of clinical mastitis (CM) monitoring via smaXtec®. Standard deviation
values are provided in parentheses.
AMS farm (Scenario 1a)
CMS farm (Scenario 1b)
Effect on revenue (aCM)
€ 33,671 (± € 138,444)
€ 194,234 (± € 86,930)
Increased costs (bCM)
€ 12,421 (± € 0)
€ 12,421 (± € 0)
Saved costs (cCM)
€ 5,220 (± € 0)
€ 15,950 (± € 0)
Mean net economic outcome
€ 26,470
€ 197,763
6
The probability that smaXtec® provides a positive net economic outcome compared to oestrus detection via
visual observation was 80% on the AMS farm and 59% on the CMS farm (RQ2). Annual mean net economic
outcomes for Scenarios 2a and 2b are shown in Table 3.
Table 3. Mean net economic outcomes of oestrus detection via smaXtec®. Standard deviation values are
provided in parentheses.
AMS farm (Scenario 2a)
CMS farm (Scenario 2b)
Effect on revenue (aOD)
€ 116,990 (± € 132,843)
€ 34,044 (± € 112,344)
Increased costs (bOD)
€ 13,077 (± € 0)
€ 13,012 (± € 0)
Saved costs (cOD)
€ 6,388 (± € 0)
€ 6,388 (± € 0)
Mean net economic outcome
€ 110,301
€ 27,420
The overall economic outcome of smaXtec® adoption was calculated by combining the mean net economic
outcomes shown in Table 2 and Table 3. The probability of overall increased profits was 75% on the AMS farm
and 93% on the CMS farm (RQ3). Cumulative distribution functions and descriptive statistics of overall net
economic outcomes for the two farms are provided in Figure 2. The cumulative distribution function of the CMS
farm was found to stochastically dominate its AMS counterpart in the first-degree sense. The expected shortfall
at 10% level on the AMS farm was substantially larger than that on the CMS farm indicating that adopting the
smaXtec® system was a greater investment risk for the former farm type.
Figure 2. Cumulative distribution functions and descriptive statistics of overall net economic outcomes for
the AMS and CMS farms when smaXtec® is used as a multi-purpose technology (Scenarios 3a and 3b)
Finally, sensitivity analyses were conducted to quantify the influence of the input data on the estimated net
economic outcome probabilities. Parameter sensitivities were tested between 50% and 150% of the correspondent
baseline value with a 25% step. Probability variations of up to 1% were considered negligible and were therefore
not reported in this paper. Positive net outcome probabilities were found to be slightly sensitive (-2%) to a 50%
milk price reduction on the AMS farm for CM monitoring (Scenario 1a) and on the CMS farm for oestrus detection
(Scenario 2b). In the latter scenario, the probability of a positive economic outcome was also sensitive to a 50%
smaller herd size (+2%), and a 50% reduction (+2%) or increase (-2%) in the per-cow costs of the smaXtec®
system (i.e., the γ1 parameter).
4. Discussion
The mean net economic outcomes shown in Table 2 and Table 3 are always positive on both the AMS and
the CMS farms. However, their magnitude depends on the milking system type considered. The AMS farm had a
higher probability of obtaining a positive economic outcome when smaXtec® was used for detecting oestrus,
whereas the CMS farm was more likely to achieve economic benefits when this technology was used for CM
monitoring. Indeed, it is known that farms equipped with different milking systems tend to adopt sensors for
distinct target applications (Steeneveld et al., 2015). When the AMS farm used smaXtec® for CM monitoring
besides oestrus detection, the probability of achieving a positive economic outcome decreased from 80% to 75%
(-5%). Similarly, on the CMS farm, utilising smaXtec® as a multi-functional sensor resulted in a 6% reduction of
7
positive economic outcome probability from 99% to 93%. Thus, both farms were more likely to benefit from the
smaXtec® system when this was used as a single-purpose technology targeting the most economically beneficial
function i.e., oestrus detection on the AMS farm and CM monitoring on the CMS farm.
The high standard deviation values in Scenarios 1 and 2 indicate that effects on revenue, and consequently
economic outcomes, were extremely variable. This was particularly the case on the AMS farm, where the expected
shortfall at 10% level in overall net economic outcomes (Scenario 3a) spanned from € -602,418 to € -114,842.
The high variability is also reflected in the remarkably wide range between minimum and maximum overall profits
for both farm types (Figure 2), though the first-degree stochastic dominance of the CMS curve indicated that
smaXtec® adoption was a less risky investment in Scenario 3b. This high variability is due to the uncertain effects
of sensor adoption on milk production. In the scientific literature, data on these effects are rarely available or do
not possess sufficient detail to correlate productivity parameters with specific sensor types. For example, in the
study by Steeneveld et al. (2015), the effects on milk production after adopting a CM monitoring sensor were
aggregated for colour, EC, SCC, LDH, or other sensors despite distinct approaches achieving substantially
different levels of accuracy. More field studies are required to fill this data gap.
To further explore the uncertain effects of sensor adoption on farm economic outcomes, sensitivity tests were
performed on all deterministic parameters. Depending on the milking system type and target sensor function,
positive net outcome probabilities were found to be only slightly sensitive to milk price, herd size and cost of the
smaXtec® bolus-sensors and monthly subscription costs. On the AMS farm, a 50% milk price reduction resulted
in a 2% decrease of the positive net outcome probability when monitoring CM (Scenario 1a). A comparable
decrease was found on the CMS farm when smaXtec® was used to detect oestrus (Scenario 2b). However, such
a milk price has not occurred since June 2016 (CLAL, 2024). The lowest annual average milk price after 2016
was encountered in 2020, corresponding to a 28% reduction in 2023 prices (CLAL, 2024). A 25% milk price
reduction reduced the probability of increased profits by 1%. Herd size and per-cow smaXtec® costs affected net
economic outcome probabilities in Scenario 2b. A 50% smaller herd size led to an increased probability of positive
economic outcomes by 2%. Considering the lack of influence of parameters such as herd growth rate, calf
maintenance costs, personnel costs and labour requirements, this was likely due to savings in the per-cow cost of
the smaXtec® system. Indeed, γ1 was ten times the base station cost per cow (i.e., γ2/HS) with a herd size of 223
and still about five times the base station cost per cow when herd size was 50% smaller. A 50% reduction and
increase in personnel costs and labour requirements led to a 2% increase and a 2% decrease in positive net outcome
probabilities, respectively. The relatively low influence of the tested parameters seems to corroborate that the milk
production effect of sensor adoption is the most important variable.
The cost increases across scenarios mainly resulted from the investment required to adopt the smaXtec®
technology. These were always greater than the saved costs except for Scenario 1b. Nevertheless, the initial
investment in the smaXtec® system appeared to be economically viable in all scenarios because of positive effects
on revenue resulting from higher milk production and/or herd growth rates. The largest cost savings (€ 15,950)
were achieved in Scenario 1b. Substituting visual observation of pre-milk for smaXtec® CM monitoring led to a
large reduction in labour requirements, which was an important contributor to the high probability of achieving
economic benefits after its adoption on the CMS farm. Labour savings are often regarded as a major driver for
adoption of sensors and automatic milking systems on conventional dairy farms (Steeneveld et al., 2015;
Eastwood and Redwick, 2020). A strong motivation for adopting sensor systems for CM monitoring on AMS
farms might be antimicrobial use and laboratory test savings when mastitis is treated at the sub-clinical stage.
However, these savings could not be quantified in this analysis in the absence of accurate data.
A lower reliance on antimicrobial use and lower laboratory testing requirements, but also an increased work
flexibility and a reduction in repetitive physical work are some of the potential non-economic and more
agroecologically based benefits of smaXtec® adoption. Assigning a monetary value to improved working
conditions is difficult. However, besides labour cost savings, reduction in drudgery and an improved work-life
balance are among the major drivers for farmers’ adoption of AMSs (Eastwood and Redwich, 2020) and animal
sensors (Steeneveld et al., 2015) across the world. The smaXtec® technology seems to be technically capable of
enabling these benefits while maintaining economic feasibility. Besides, continuous, unbiased and reliable data
collection from individual animals through sensors may facilitate abiding with animal welfare standards (Stygar
et al., 2021). Compliance with such standards is gaining a growing attention from European consumers, who are
willing to pay a 31% premium for milk produced on dairy farms proactively managing animal health and welfare
(Stygar et al., 2021). Due to this increased attention by consumers, animal welfare standards are increasingly
being mandated by regulation.
In Germany, the Animal Welfare Act of 1972 and subsequent amendments requires livestock managers to
regularly supervise animal health (Bundesministerium der Justiz, 1972). When animal health is monitored with
the aid of technology, the Act imposes the implementation of precautionary measures in case of technical
8
malfunctions (Bundesministerium der Justiz, 1972). The smaXtec® system has a failure rate of 3% (smaXtec,
2024. Pers. Comm.). When data transmission from a sensor-bolus stops, the system sends an error message to the
user via the dedicated mobile application. After a second error message, the user receives an automated email.
The farmer has 6 days to replace a faulty sensor without losing animal data, which keep being collected even
when they are not wirelessly transmitted to the base station. Other issues may occur if cattle walk too far from the
base station, if the base station goes offline, or if the bolus-sensor battery runs out (smaXtec, 2024. Pers. Comm.).
In all these cases, alerts will be triggered so that the farm manager may promptly act to solve the issue and
implement precautionary measures if needed. However, more studies are required to validate sensor-based welfare
assessment, especially on commercial farms (Herlin et al., 2021; Stygar et al., 2021). Animal welfare is a
multidimensional concept which is unlikely to be captured by a single technology and potential trade-offs should
be taken into consideration (van Erp-van der Kooij and Rutter, 2020). For example, replacing human contact with
sensors for monitoring CM and oestrus may lead to increased stress when animals are manually handled for other
tasks (Herlin et al., 2021). Stygar et al. (2021) investigated into the commercial validation rate of available PLF
sensor technologies for animal welfare assessment. They found that only 14% of the 129 identified technologies
had been validated, with sensor-boluses representing the least validated category (7%).
5. Conclusions
The present analysis focused on the economic performance of the smaXtec® bolus-sensor system
implemented for CM monitoring and oestrus detection on fictitious AMS and CMS dairy farms in Germany.
Probabilities of positive net economic outcomes were estimated using a partial budgeting approach and a Monte
Carlo simulation of the effect of bolus-sensor adoption on milk production and herd growth rates. The probability
of achieving increased profits on the AMS farm for CM monitoring and oestrus detection were 58% and 80%,
respectively. On the CMS farm, the probabilities were 99% for CM monitoring and 59% for oestrus detection.
When smaXtec® was used as a multi-purpose tool targeting both functions, positive net outcome probabilities
were 75% on the AMS farm and 93% on the CMS farm.
Depending on the target application, the mean annual net economic outcomes of smaXtec® adoption ranged
from € 26,470 to € 110,301 on the AMS farm, and from € 27,420 to € 197,763 on the CMS farm. When the two
target applications were simultaneously considered, the mean overall net economic outcome was € 126,257 year-
1 on the AMS farm and € 214,670 year-1 on the CMS farm. The higher profits were mostly due to the increased
milk production, higher herd growth rates and labour savings achieved after adopting the studied technology.
However, the economic performance of the smaXtec® system was found to be highly variable and consequently
these results should be interpreted with care.
Agroecologically based benefits of bolus-sensor adoption on dairy farms may include an increased work
flexibility, a reduction in repetitive physical work and an improved compliance with animal welfare standards and
regulation. However, these aspects are not well documented in the scientific literature, especially as far as animal
welfare is concerned. Further research is required to identify potential economic, social, and environmental trade-
offs of sensor-bolus use on dairy farms in Germany and globally.
Acknowledgments
This study was co-funded by UK Research and Innovation (UKRI Reference No.10037994) and by the EU’s
Horizon Europe research and innovation programme (Grant Agreement No.101060759) as part of the
“Digitalisation for Agroecology” project (D4AgEcol | https://d4agecol.eu/). The authors wish to thank Dr. ir.
Wilma Steeneveld at Utrecht University and the UK SmaXtec Animal Care team for their kind support in
providing some of the data used in the present analysis.
References
Achilles, W., Eckel, H., Eurich-Menden, B., Frisch, J., Fritzsche, S., Funk, M., Gaio, C., Grebe, S., Grimm,
E., Grube, J., Hartmann, W., Horlacher, D., Kloepfer, F., Kron, K., Meyer, B., Sand, I., Schroers, J.O., Schultheiss,
U. and Wulf, S. 2020. Betriebsplanung Landwirtschaft 2020/21. Germany: Kuratorium für Technik und Bauwesen
in der Landwirtschaft
Benaissa, S., Tuyttens, F.A.M., Plets, D., Trogh, J., Martens, L., Vandaele, L., Joseph, W. and Sonck, B. 2020.
Calving and estrus detection in dairy cattle using a combination of indoor localization and accelerometer sensors.
Computers and Electronics in Agriculture, 168, 105153
Boker, A.R., Bartel, A., Duc, P.D., Hentzsch, A., Reichmann, F., Merle, R., Arndt, H., Dachrodt, L., Woudstra,
S. and Hoedemaker, M. 2023. Status of udder health performance indicators and implementation of on farm
monitoring on German dairy cow farms: results from a large scale cross-sectional study. Frontiers in Veterinary
Science, 10:1193301
9
Bundesministerium der Justiz. 1972. Animal Welfare Act. [Online]. Germany Federal Government. Available
from: https://www.gesetze-im-internet.de/tierschg/BJNR012770972.html [Accessed 13 January 2024]
CLAL srl. 2024. Farm-gate milk prices, Germany. [Online]. CLAL srl. Available from:
https://www.clal.it/en/index.php?section=latte_germania [Accessed on 30 April 2024]
Davies, P., Strugnell, B., Thomas, L., Lovatt, F. and Willison, I. 2022. To scan or not to scan? The economics
of transthoracic ultrasonography for ‘whole-flock’ ovine pulmonary adenocarcinoma screening in UK sheep
flocks. Veterinary Record, e1980, pp.1-8
Eastwood, C.R. and Renwick, A. 2020. Innovation Uncertainty Impacts the Adoption of Smarter Farming
Approaches. Frontiers in Sustainable Food Systems, 4(24) https://doi.org/10.3389/fsufs.2020.00024
GEA Group. 2024. GEA DairyRobot R9500 Robotic Milking System. [Online]. GEA Group Aktiengesellschaft.
Available from: https://www.gea.com/en/products/milking-farming-barn/dairyrobot-automated-
milking/dairyrobot-r9500-robotic-milking-system/ [Accessed on 29 April 2024]
Herlin, A., Brunberg, E., Hultgren, J., Hogberg, N., Rydberg, A. and Skarin, A. 2021. Animal welfare
implications of digital tools for monitoring and management of cattle and sheep on pasture. Animals, 11(829)
Hogeveen, H., Kamphuis, C., Steeneveld, W. and Mollenhorst, H. 2010. Sensors and Clinical Mastitis – The
Quest for the Perfect Alert. Sensors, 10, pp.7991-8009
Kim, H., Min, Y. and Choi, B. 2019. Real-time temperature monitoring for the early detection of mastitis in
dairy cattle: Methods and case researches. Computers and Electronics in Agriculture, 162, pp.119-125
Lam, T.J.G.M., Olde Riekerink, R.G.M., Sampimon, O.C. and Smith, H. 2009. Mastitis diagnostics and
performance monitoring: a practical approach. Irish Veterinary Journal, 62, pp.34-39
Lodkaew, T., Pasupa, K. and Loo, C.K. 2023. CowXNet: An automated cow estrus detection system. Expert
Systems with Applications, 211, 118550
Mayo, L.M., Silvia, W.J., Ray, D.L., Jones, B.W., Stone, A.E., Tsai, I.C., Clark, J.D., Bewley, J.M. and
Heersche Jr., G. 2019. Automated estrous detection using multiple commercial precision dairy monitoring
technologies in synchronized dairy cows. Journal of Dairy Science, 102(3), pp.2645-2656
Microsoft Corporation. 2024. Microsoft® Excel®: Version 2403. Redmond, WA, USA: Microsoft
Corporation
Redman, G. ed. 2023. John Nix Pocketbook for Farm Management for 2024. 54th Edition. UK: The
Pocketbook
Saxon State Ministry for Energy, Climate Protection, Environment and Agriculture. 2024. Livestock prices for
calves in Central Germany and East Germany. [Online]. Government of Germany. Available from:
https://www.landwirtschaft.sachsen.de/nutztierpreise-kaelber-mitteldeutschland-und-ostdeutschland-54528.asp
[Accessed 16 January 2024]
SmaXtec. 2024. THE SMAXTEC HEALTH SYSTEM. [Online]. SmaXtec Animal Care GmbH. Available from:
https://smaxtec.com/ [Accessed 15 January 2024]
Steeneveld, W., Vernooij, J.C.M. and Hogeveen, H. Effect of sensor systems for cow management on milk
production, somatic cell count, and reproduction. Journal of Dairy Science, 98, pp.3896-3905
Stygar, A.H., Gomez, Y., Berteselli, G.V., Dalla Costa, E., Canali, E., Niemi, J.K., Llonch, P. and Pastell, M.
2021. A systematic review on commercially available and validated sensor technologies for welfare assessment
of dairy cattle. Frontiers in Veterinary Science, 8:634338
Tergast, H., Hansen, H. and Weber, E.C. 2023. Steckbriefe zur Tierhaltung in Deutschland: Milchkühe.
[Online]. Thünen-Institut für Betriebswirtschaft. Available from: https://www.thuenen.de/media/ti-
themenfelder/Nutztierhaltung_und_Aquakultur/Haltungsverfahren_in_Deutschland/Milchviehhaltung/Steckbrie
f_Milchku__he_2023.pdf [Accessed 16 January 2024]
van Erp-van der Kooij, E. and Rutter, S.M. 2020. Using precision farming to improve animal welfare. CAB
Reviews, 15, 051
