DEVELOPMENT OF A WHOLE FARM GROW-FINISH SWINE OPERATION
DECISION SUPPORT TOOL THROUGH MODELLED NUTRIENT LOSS
By
Brooke Latack

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Animal Science – Master of Science
2017

ABSTRACT
DEVELOPMENT OF A WHOLE FARM GROW-FINISH SWINE OPERATION
DECISION SUPPORT TOOL THROUGH MODELLED NUTRIENT LOSS
By
Brooke Latack
The need for animal production decision-making tools is essential to ensuring
sustainable animal protein production. Of the decision-making tools available, there is a
lack of consideration for year-after-year consequences of and the interaction between
animal production, manure storage, manure land application, and crop nutrient uptake.
A model was created to address this gap using a grow-finish swine system with respect
to nitrogen and phosphorus losses. Empirical formulas and published information of
nutrient retention or losses were used in the development of a management decision
driven model. Observed results from a previous study at Michigan State University were
compared to calculated results to assess the validity of model outputs. Calculated
outputs for total manure excretion, nitrogen excretion, and nitrogen loss from the system
were within the error found in the observed data. The model was tested further by
modifying the baseline scenario to test model sensitivity. Changes in manure storage
and dietary phosphorus concentrations had a greater impact on management longevity
than manure application method and field location. The results of the model assessment
indicate high variability in outputs, but this may reflect high variability in animal
production observation, indicating that model outputs still portray a realistic vision of a
grow-finish swine facility. This model allows the identification of critical control points
within the system to help direct future research and support producers and advisors in
creating a long-term, sustainable animal protein production system.

ACKNOWLEDGEMENTS

This thesis would never be possible without the support and assistance of many people.
First, I would like to thank my professor Dr. Wendy Powers for allowing me the
opportunity to pursue my Master’s degree, her guidance throughout my entire project,
and her unwavering belief in my ability to accomplish such a task.
I would also like to thank my committee members Dr. Steven Safferman, Dr. Chris
Saffron, Dr. Dale Rozeboom, and Dr. Brent Auvermann. Their invaluable depth of
knowledge and willingness to meet and offer assistance throughout my program
allowed me to grow substantially.
A great deal of gratitude is owed to the staff of the Animal Air Quality Research Facility.
The friendships and experiences garnered at the facility are largely due to the
willingness of the staff to guide and teach new skills as well as taking their own time to
improve my public speaking skills. Those experiences are some I will never forget.
Finally, I would like to thank my family for their patience, understanding, and faith in my
abilities throughout this entire process.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................vi
LIST OF FIGURES ......................................................................................................... vii
1 Introduction .................................................................................................................. 1
2 Objective ...................................................................................................................... 3
3 Literature Review ......................................................................................................... 4
3.1 Livestock production in the United States .............................................................. 4
3.2 Swine Production ................................................................................................... 4
3.3 Animal Housing ...................................................................................................... 5
3.4 Manure Storage ..................................................................................................... 6
3.5 Land Application of Manure ................................................................................... 6
3.6 Crop Nutrient Use .................................................................................................. 7
3.7 Nitrogen and Phosphorus Impacts on the Environment ......................................... 8
3.8 Computer modelling of nutrient flow of swine production ....................................... 9
3.9 Attributes for nutrient flow through animal production ............................................ 9
3.10 Choosing software for the model........................................................................ 11
4 Materials and Methods ............................................................................................... 13
4.1 Project Scope ....................................................................................................... 13
4.2 Functionality criteria of the model......................................................................... 14
4.3 Software platform identification ............................................................................ 17
4.4 Modeling process ................................................................................................. 18
4.4.1 Soil texture determination through zip codes ................................................. 18
4.4.2 Nutrient excretion and resulting manure composition and assumptions ........ 18
4.4.3 Nutrient loss during manure storage .............................................................. 19
4.4.4 Land application of manure ........................................................................... 19
4.4.5 Plant nutrient uptake ...................................................................................... 20
4.4.6 Data outputs .................................................................................................. 21
4.4.7 Validation of the model .................................................................................. 21
4.4.8 Monte Carlo analysis ..................................................................................... 26
5 Results ....................................................................................................................... 27
5.1 Baseline Scenarios .............................................................................................. 27
5.2 Manure Spreading Method Modification (Scenario 2) Outputs ............................. 29
5.3 Field Location and Critical P Threshold Modification (Scenario 3) Outputs.......... 31
5.4 Manure Storage Modification (Scenario 4) Outputs ............................................. 33
5.5 Dietary P Modification (Scenario 5) Outputs ........................................................ 35
5.6 Management practice influence on long-term sustainability ................................. 36
5.6.1 Manure spreading to meet crop N needs....................................................... 36
5.6.2 Manure spreading to meet crop P needs ....................................................... 38
iv

5.6.3 Monte Carlo results........................................................................................ 38
5.7 Software Platform Efficacy ................................................................................... 39
6 Conclusions ................................................................................................................ 41
APPENDICES ............................................................................................................... 43
APPENDIX A – Full table of all scenario outputs ....................................................... 44
APPENDIX B - Monte Carlo Yearly Output Statistics ................................................. 45
APPENDIX C - Program Screenshots........................................................................ 46
BIBLIOGRAPHY ........................................................................................................... 52

v

LIST OF TABLES

Table 1. Modeling software platform requirements and the applicability of LabVIEW, Matlab,
and Stella programming software on each requirement ............................................................17
Table 2. User inputs for baseline validation scenarios ..............................................................22
Table 3. Baseline scenario inputs for manure storage and land application ..............................22
Table 4. Baseline scenario inputs for crop growth.....................................................................23
Table 5. Modification scenarios to test sensitivity of the model .................................................25
Table 6. Monte Carlo assumptions for analysis.........................................................................26
Table 7. Model outputs for baseline scenarios ..........................................................................28
Table 8. Model outputs for scenario altering manure application method from surface broadcast
to injection (scenario 2) .............................................................................................................30
Table 9. Model outputs for scenario altering field location and P critical limits (scenario 3) .......32
Table 10. Model outputs for scenario altering manure storage method (scenario 4) .................34
Table 11. Model outputs for scenario increasing dietary P (scenario 5) ....................................35
Table 12. Complete table of all scenario outputs analyzed .......................................................44
Table 13. Output statistics for monte carlo analysis for all scenarios ........................................45

vi

LIST OF FIGURES

Figure 1. Model boundaries and scope ....................................................................................14
Figure 2. User inputs required by the model .............................................................................16
Figure 3. Model outputs and observed study data comparison for annual manure production ..28
Figure 4. Model outputs and observed study data comparison for annual N excretion and N loss
.................................................................................................................................................29
Figure 5. Monte Carlo analysis histograms for A: Baseline scenario, B: Scenario 2 (modified
manure application), C: Scenario 3 (modified field location), D: Scenario 4 (modified manure
storage), and E: Scenario 5: (modified dietary P) ......................................................................39
Figure 6. Front panel user interface for location, manure spreading, growth additive, and animal
performance data inputs ...........................................................................................................46
Figure 7. Front panel user interface for manure storage, manure application, and field
information. Image shows inputs for only 5 fields, but the model is programmed to allow up to 30
fields. ........................................................................................................................................47
Figure 8. Model outputs as seen on front panel of the program. ...............................................48
Figure 9. Block diagram of one of six available phase available in the program. The inputs
(seen on the left of the case structure) are dictated by the user on the front panel. The inputs
feed into a case structure to calculate N and P excretion. .........................................................49
Figure 10. Nutrient loss calculation due to manure storage and land application block diagram
.................................................................................................................................................50
Figure 11. Crop nutrient uptake and write to file portion of the block diagram. Remaining
nutrients flow to this section and crop nutrient use is calculated for each yearly iteration. .........51

vii

1 Introduction
The need for animal production decision-making tools is essential to ensuring
sustainable production. Of the decision making tools available, there is a lack of
consideration for year-after-year consequences of and the interaction between animal
production, manure storage, manure land application, and crop nutrient uptake. The
ability of a decision-making tool to address this gap can influence management
decisions and, therefore, long-term farm sustainability.
Two nutrients of specific interest are nitrogen (N) and phosphorus (P) because of
their negative environmental impact (i.e. leaching to groundwater and runoff leading to
pond or lake eutrophication). Both nutrients are important for animal growth. If fed in
insufficient quantities production decreases significantly. If fed in excess of nutrient
needs animals may meet production goals, but additional land is needed to apply these
excess nutrients in order to avoid negative environmental consequences. Manure
storage contributes to greater air emissions, including nitrous oxide and ammonia.
State-specific nutrient application thresholds influence producer decisions. Access to
improved decision tools will make it easier to make better decisions that are consistent
with state guidelines.
A software program that follows N and P flows through a swine production
system from excretion of nutrients through manure storage, manure land application,
and crop nutrient uptake allows researchers opportunities to identify areas for further
research to improve understanding of sustainable animal production. While many
programs currently available offer portions of the complete system, a comprehensive
program to analyze the entire system does not exist. There is a need for a tool that

1

analyzes a complete system through the four main stages of manure handling.
Advanced software to simulate a farm and to estimate the nutrients remaining in each
field following crop harvest and the number of years the current management practices
could continue before soil nutrient concentrations reach regulatory thresholds will
improve the system sustainability.

2

2 Objective
The objective of this work is to use reference equations and data to model the
flow of N and P through grow-finish swine operations including nutrient absorption
during growth, losses during manure storage and land application, and crop nutrient
uptake. Unlike other tools available, a goal of the model development is to integrate
feedback mechanisms at various points of the system, providing a dynamic
understanding of how components of the system directly or indirectly interact over a
multi-year time-frame (iterative process).
The following goals were critical in the development of the model:
● Identify impacts and consequences of management decisions with respect to
N and P;
● Reveal critical control points to increase the sustainability of animal protein
production;
● Aid in the development of a collaborative model to understand the broader
context of a sustainable of animal protein production system (social and
economic impacts).

3

3 Literature Review
3.1 Livestock production in the United States
The anticipated 33% to 71% population increase between now and 2050 calls into
question food security (Gerland et al., 2014). Because of the expected increase in
demand for livestock products it is important to find sustainable livestock and crop
management systems (Nardone et al., 2010; Wu et al., 2014). Balancing economic,
environmental and social factors is key to achieving efficient livestock production while
leaving the smallest environmental impact and adhering to social constructs (Hansmann
et al., 2012; Hoffmann, 2011; Honeyman, 1996). Excess nutrients, emitted through air,
land, and water, create ecosystem imbalances and poor living conditions (Jongbloed
and Lenis, 1998).

3.2 Swine Production
Modern swine production systems group animals by production phase often with
separate buildings for different phases (breeding/gestating, nursery, and grow-finish life
stages) and each phase is fed N levels appropriate for productive lean growth at that life
phase (Kliebenstein and Lawrence, 1995). An essential nutrient for both animals and
plants is N (Havlin et al., 2014; Wu et al., 2014). It is used animals to build amino acids
and subsequently construct proteins (Wu et al., 2014). Each production phase has
specific N and P needs to maximize productivity without excessive nutrient excretion in
urine and feces (Lange, 2013). Phase feeding has been shown to reduce N excretion by
4.4% (Sutton et al., 1999). The difference in nutrient recommendations between
production phases is based on the differing N retention values for the different phases.

4

Gilts retain less N, at a rate of protein deposition of 234 g/day, than barrows with a
retention of 277 g/day. Thus, if gilts are fed a level of N similar to barrows, the gilts
would excrete more N compared to the barrows (Lange, 2013).

3.3 Animal Housing
Many factors in swine housing will affect nutrient losses, such as manure collection
methods, time between manure collections, and housing management. Flooring in the
swine housing will affect how and how often of manure collection occurs
(Olesen and Sommer, 1993). Common styles of flooring include partially or fully slatted
flooring or solid flooring covered in shavings or straw (Arogo et al., 2003). Slatted floors
allow manure to collect for hours to months following excretion depending upon depth of
storage below the slatted floor (Arogo et al., 2003). Litter from solid floors may also be
collected at different time periods (Arogo et al., 2003). Philippe et al. (2007) found that
slatted floor collection systems emitted 6.2 g/day of ammonia while solid floor covered
with straw emitted 13.1 g/day when left for a full 120 day grow-finish phase. The
differences in flooring and in-house manure storage will lead to differences in the buildup of ammonia (Herber et al., 2000; Lim et al., 2002; Zhu et al., 2000). Drummond et al.
(1980) observed that ammonia levels of 50, 100, and 150 ppm reduced swine growth by
12, 30, and 29 percent, respectively, compared to a well-ventilated chamber with
minimal aerial ammonia present. The decrease in swine growth due to ammonia levels
is primarily in lean muscle accretion, resulting in an increased excretion of nutrients
consumed.

5

3.4 Manure Storage
Manure storage method and structure varies by farm and each type of storage reflects
different rates of N emission (Liu et al., 2014). Collection conditions (e.g. warm or cool
temperatures, air flow over the surface of the storage area, and time stored) impact the
N gas lost as nitrous oxide and ammonia (Havlin et al., 2014; Liu et al., 2014). One
study found that uncovered manure storage lost 14.7% of the initial N while covered
manure storage lost only 5.6% of initial N due to active air movement over the
uncovered manure (Hansen et al., 2006). A meta-analysis of available data revealed
that a lagoon storage system at a finishing farm emitted a mean of 3.70 ± 3.74
kg/year/pig while a slurry tank at a finishing farm emitted a mean of 1.85 ± 2.28
kg/year/pig (Liu et al., 2014). Warmer storage systems will emit more ammonia than a
cooler storage system (Liu et al., 2014). These differences in emissions are an integral
part of understanding the whole system and where N losses occur.

3.5 Land Application of Manure
Movement from manure storage into the manure spreader and onto the field will result
in some gaseous N loss. Differing manure application methods result in greater
variation in N emissions (Havlin et al., 2014). Ammonia loss following land application of
manure using surface broadcast, surface incorporation, and deep placement methods
was 68%, 17%, and 2% of total N applied, respectively (Huijsmans et al., 2003). The
same study found that increasing ambient temperature increased the N lost from 35.3%
at 10ºC to 56.2% at 20ºC when manure was surface broadcasted (Huijsmans et al.,

6

2003). Incorporation and deep placement methods resulted in similar increases in N lost
as ambient temperature increased (Huijsmans et al., 2003). Precipitation around the
time of manure spreading contributes to leaching and runoff because nutrient-rich water
moves faster than the rate at which plants take up N (Innes, 2000). Excess P in the soil
increases the chance of run off, ultimately increasing the chance of eutrophication in
nearby water sources (Havlin et al., 2014).

3.6 Crop Nutrient Use
Crops absorb nutrients throughout the growing season (Havlin et al., 2014). Each crop
has a specific timeframe during which nutrient absorption rate is greatest (Havlin et al.,
2014). Similarly, each crop has a different rate of nutrient uptake and nutrient
requirement so it is imperative to be able to make crop-specific nutrient
recommendations. At the time of greatest nutrient absorption, the crop will absorb more
N at a faster rate reducing environmental risks (Havlin et al., 2014). Soil type also plays
a role in the availability of nutrients for crop uptake. Fields with sandy or loam types of
soil had less N and P available for crop uptake than the fields composed mainly of clay,
though sandier soils had higher N availability in the winter compared to clay soil (Havlin,
et al., 2014). All of the variables affecting N uptake by crops are essential when
understanding the amount of N being used in a system and remaining amount of N is
left in the soil. Excess N in the soil is absorbed by a plant and can cause weakened
fiber within the plant due to increased carbohydrate movement (Havlin et al., 2014).
Excess N uptake can cause delayed maturity in plants (Havlin et al., 2014). If the soil is

7

in excess of N to the extent that the crop is no longer absorbing N, a stock will be left in
the soil (Havlin et al., 2014).

3.7 Nitrogen and Phosphorus Impacts on the Environment
Leaching and runoff of N are damaging to the environment (Daniel et al., 1994). If
leaching occurs vertically it can drain into the water supply, making it dangerous for
human consumption. If leaching or run off occurs into a river, lake, or pond the N will
cause eutrophication resulting in uncontrolled algae growth and depletion of the oxygen
in the water supply (Daniel et al., 1994).
Gaseous emissions of N are mainly in the form of nitrous oxide, a greenhouse gas, and
ammonia (Havlin et al., 2014). Both can be emitted during manure storage and manure
land application (Brunke et al., 1988; Havlin et al., 2014; Morken and Sakshaug, 1998).
Emissions can also create human health issues. Studies found that prolonged
exposure to even low levels of ammonia led those working within the barns to be
susceptible to regular upper respiratory irritation, inflammation, and infections (Crook et
al., 1991).
An important nutrient in crop growth is P. It is important to keep fields supplied with P,
but not all P is in a form that can be used by plants readily (Havlin et al., 2014). For this
reason crops may appear deficient when the soil has adequate amounts of P (Havlin et
al., 2014). Over applying manure to fields can lead to dangerously high levels of P in the
soil, which could move into nearby water and cause eutrophication (Jongbloed and
Lenis, 1998; van der Werf and Petit, 2002).

8

Soil P concentration is often used as the benchmark for constructing manure application
regulations because P concentrations often pose an environmental risk well before N or
potassium concentrations (Ribaudo et al., 2003). Many states employ maximum P
application rates, including allowance for multiple years’ worth of P application allowed
provided the field is not returned to for a period of time and the starting soil P
concentration does not exceed a prescribed threshold.

3.8 Computer modeling of nutrient flow of swine production
Understanding the flow of nutrients through a swine production system is an important
step in identifying ways to reduce environmental impact of meat production. Equations
from well-developed methods can be used to predict animal nutrient retention and
excretion (National Research Council, 2012), manure storage losses, manure land
application (Midwest Plan Service), and plant nutrient uptake (Havlin et al., 2014;
Lange, 2013). Optimization of management practices contribute to decreases in nutrient
leaching, air emissions, and soil fertility issues – all major environmental concerns for
livestock and crop production (Salois, 2015; Tilman et al., 2011).

3.9 Attributes for nutrient flow through animal production
Multiple tools exist to consider management practices in place on the farm and follow
the flow of nutrients through the farm. Each program has specific attributes that make
the software program beneficial to users. However, currently available programs lack
essential features that result in a complete understanding of nutrient movement and
flows. The Purdue Manure Management (MMP) software program allows the user to

9

enter nutrient and field information and receive manure management plans for
spreading on fields, but does not allow for a long term assessment of the environmental
impacts of the management practices and does not account for any feedback within a
system (Karmakar et al., 2007). Other programs, like the Ammonia Emissions
Estimator, provide an understanding of emissions from animal production facilities
though no other losses or manure calculations are included (Koelsch et al., 2005).
Whole farm nutrient balance software developed by Cornell University (version 2.0,
2012) provides a whole farm nutrient balance, but lacks the ability to project years into
the future as to the sustainability of the management practices. Other software exists
that have similarities to the programs outlined, but all programs available lack the yearto-year environmental analysis necessary for understanding longer term implications of
management.
The capability of the software platform to allow the following calculations and input
variables to be used are important for mapping nutrient flow:


Animal management: feed composition, growth modifiers used, number of
animals produced, animal nutrient needs for stage of production



Manure storage: collection, storage, time manure is stored



Manure field application: application method, soil conditions, ambient
temperature, precipitation patterns, dominant soil texture



Nutrient removal by crops: soil nutrients content, soil nutrient availability, crop,
crop yield, crop nutrient needs



Detailed output depicting management impacts: Amount of manure produced
by animals, nutrient loss during manure storage, nutrient loss during field

10

application of manure, manure application quantities (i.e. a spreading plan),
soil nutrient concentrations after each crop harvest, and number of years until
soil nutrient levels approach a specified limit.

3.10 Choosing software for the model
The user interface of a model is an important attribute of model acceptance by the
intended audience. Accessibility and familiarity are additional important considerations
when choosing the model platform. Widely-used software, such as Microsoft Excel,
gives the general public easy access. Because Excel is user-friendly programs that
interface with Excel are advantageous. Because Excel does not incorporate some of the
more complicated mathematical structures, a more complicated and robust program is
required.
Some software programs offer a wide range of programming capabilities. MATLAB
offers programming options as far as computation, data analysis, programming, and
developing models. This system is ideal for many hoping to program more complex
models. LabVIEW (2012) offers similar programing capabilities. Both programs allow
the user to delve deeper into more complex modeling options, but both programs
require purchasing of the software. Neither program is widely used outside of the
engineering discipline. The general public is unlikely to purchase software for the sole
purpose of a single model.
For the purposes of following N and P from animal production to manure storage to
manure application and finally crop nutrient uptake, software with robust modeling
capabilities, such as LabVIEW (2012), have the ideal user interface and mathematical

11

capabilities. Because the model intent is as a tool to further future research, software
cost was not deemed a critical decision point. Using the abilities of LabVIEW to export
data calculated into an Excel file and ultimately being able to import the data into visual
representation software allows a robust model to be built that aggregates four primary
sources of N and P movement on a swine operation. With these concepts at the
forefront in the need for a comprehensive software program, creating a program to
integrate animal N and P excretion, losses during manure storage and land application,
and residual nutrients remaining in soil after crop nutrient uptake is prudent to creating a
better understanding of the sustainability animal protein production practices over
multiple years.

12

4 Materials and Methods
4.1 Project Scope
Project boundaries are illustrated in Figure 1. The scope of this project is the
farmstead boundary and includes swine feed metabolism, on-farm manure storage, land
application of manure, and crop nutrient uptake. Nutrients considered are N and P.
Nutrient retention by pigs, nutrients lost to overland flow into surface waters and
leachate into groundwater, N lost to the atmosphere as NH3 and N2O, and nutrients
removed from the soil by crops are within the scope of this project. All feedstock used in
the swine production portion of the system is assumed to be brought in from outside the
system. The fate and impact of nutrients in harvested swine, soil and manure emissions
of NH3 and N2O emitted, and nutrients in crops grown that leave the system after the
crop growing season are not within the scope of this project. Site specific location is
important for determining nutrient loss based on dominant soil texture. The geographic
scope of the project is the United States (including Hawaii and Alaska), where dominant
soil texture data have been incorporated into the program.

13

Figure 1. Model boundaries and scope

4.2 Functionality criteria of the model
The model estimates manure excreted, nutrient loss from storage and land
application, nutrient uptake by crops, and the resulting amount of manure that needs to
be applied to each field to fulfill crop N or P needs. Feedback loops adjust spreading
habits until P threshold limits (specified by the user) are met. User inputs required by
the model are shown in Figure 2. The variables needed reflect primary management
considerations in swine production farms and within crop management. All variables
included have an impact on nutrient loss throughout the system. Default values for

14

variables within the swine phase feeding, manure storage, and land application sections
of the model are pre-programmed with an override option by the user. Model outputs
export into a Microsoft Excel file for the user.

15

Figure 2. User inputs required by the model

16

4.3 Software platform identification
Table 1 shows the requirements for the software platform. LabVIEW (National
Instruments, Austin, TX), Matlab (MathWorks, Natick, MA), and Stella (isee systems,
Lebanon, NH) were compared. LabVIEW (2012) programming software was selected as
the best option. Because feedback is a critical component to the model created and
future modification of equations is necessary to keep the model up to date, Matlab was
removed from consideration as a candidate for the model. Stella contains the
mathematical and modeling capabilities but lacks the ease of use needed. LabVIEW
programming software is deficient in visual friendly outputs, but the limitation was
circumvented by exporting data to Microsoft Excel.

Table 1. Modeling software platform requirements and the applicability of LabVIEW,
Matlab, and Stella programming software on each requirement
Requirement

LabVIEW

Matlab

Simple user interface

x

x

Capable of complicated
mathematical structures

x

x

Feedback capabilities

x

Robust modeling capabilities

x

x
x

Visually friendly outputs
Ease of equation modification
and update

Stella

x

17

x

x

x

x

4.4 Modeling process
In order to accurately model the behavior of the grow-finish swine operation,
equations outlining mass balances of nutrients were necessary for each process. This
section documents the equations used to calculate nutrient loss and remaining nutrients
at each step in the process and the variable assumptions used in an example scenario.
When the user identifies either N or P of which to meet the needs of crops when
applying manure to fields, the model performs separate functions. When spreading to
meet N needs, the program runs until all fields exceed the critical P soil concentrations
specified by the user. When spreading to meet crop P needs, the model runs for 50
years and specifies how much additional N fertilizer is required to meet crop needs.
4.4.1 Soil texture determination through zip codes
The model identifies the dominant soil texture when the user inputs the first three
digits of the field zip code. By using a zip code data base (Fry, 1999) the general area of
each three digit code was determined. The Web Soil Survey (NRCS, 2016) outlines the
area of the zip code and identifies the dominant soil texture. A case structure was coded
to allow the program to search for the soil texture within the three digits of the area code
and output a coefficient calculated by Griffin et al. (2002) and Powell et al. (1994) for
nutrient loss during manure application to fields based on identified soil texture.
4.4.2 Nutrient excretion and resulting manure composition and assumptions
Excreted manure mass and nutrient composition were estimated using ASABE
Standard D384.2 (2005) equations and the following input variables: initial and final
body weight (kg), average dressing percentage (%), fat free lean percentage (%),
average daily feed intake (kg), crude protein in feed (%), P content in feed (%), number

18

of animals, and days on feed. An option was included to account for inclusion of growth
promoting additives. Up to six feeding phases period can be programmed into the
model. If fewer than six phases are fed, the program omits the unused phases from
calculations. User inputs drive the conditional structures that process N and P
consumption and retention. The equations derived from ASABE Standard D384.2
(2005) are used to calculate total manure excretion (kg) based on N intake (kg), N
retention (kg), P intake (kg), and P retention (kg) as subroutines of the program.
Outputs of total manure excretion, N excretion, and P excretion flow into the manure
storage phase of the program.
4.4.3 Nutrient loss during manure storage
The manure storage method drives nutrient loss. The program calculates the
number of days that manure is stored based on the user-defined times of manure
application. By calculating the days between manure spreading, nutrient loss during
storage is quantified. Based on storage method and time the manure is stored,
conditional structures apply literature values for N and P losses (Benham et al., 2009).
The nutrients remaining after storage are available for spreading onto fields for crop
nutrient uptake.
4.4.4 Land application of manure
Manure spreading practices will affect nutrient losses during the application
process. The model prompts the user to choose whether the program is spreading to
meet N or P needs. In the case of spreading to meet N needs of the crop, P is typically
over-applied, which leads to a shorter time period before the field P content exceeds the
critical P limits. When applying to meet the P needs of crops, N needs are typically not

19

met, requiring additional N fertilizer to make up the deficiency. An output is created by
the program indicating the number of years before the critical P limit is met or the
amount of additional N needed to satisfy the crop needs. Manure spreading methods
and seasonal timing of manure spreading influence nutrient losses. By implementing the
user defined management decisions, seasonal data using conditional structures
determine soil moisture and soil temperature and estimates the loss of nutrients
primarily due to N volatilization (MWPS, 1993). Using the nutrients available after land
application and the dominant soil texture determined by the program, losses due to
dominant soil texture are calculated (Griffin et al., 2002; Powell et al., 1994). Nutrients
not lost to air emissions, overland flow into surface water, or leachate into groundwater
remain available for crop growth.
4.4.5 Plant nutrient uptake
Nutrients removed from the soil during crop growth are largely dependent on the
crop being grown, soil nutrients, and the nutrients available from the manure plus any
additional fertilizer applied. The model was programmed to consider initial soil nutrient
level inputs relative to specific crop needs and acreage to perform conditional
processes that calculate crop nutrient removal using International Plant Nutrition
Institute equations. If a field is initially over the user-defined critical nutrient
concentration, the program does not allow additional spreading of manure onto that
field. Because the field does not receive manure, the growing crops can only use
nutrients accumulated in the soil from previous fertilization. The program advances
nutrient concentration data into the sub-program for each consecutive calendar year
until all fields exceed the maximum nutrient concentration. When spreading to meet

20

crop N needs, when all the fields reach critical the P concentration, the program
terminates, calculating the number of years the management practices occurs until the
endpoint is reached. When spreading to meet crop P needs, the model iterates 50 years
to estimate the annual manure spread and the annual amount of supplemental
commercial N fertilizer needed for maximum crop production.
4.4.6 Data outputs
Because user skills and platform experience will vary, clarity of output data is
essential. Coding was developed for automatic export of some data to Microsoft Excel.
These data include:


Gallons of manure excreted per year (gal/year)



N excreted (kg/year)



P excreted (kg/year)



Quantity of manure spread on each field annually (gal/acre)



Annual excess or deficiency in manure excreted (kg)



Number of years a management practice can be used before soil nutrient
thresholds met

4.4.7 Validation of the model
Data from a swine feeding study was used to assess model validity. Two
scenarios were considered, each representing one of two diets fed to 60 barrows
weighing approximately 91 kg at approximately 140 d of age (5 pigs per group; 6 groups
per diet) in a grow-finish swine study conducted at the Animal Air Quality Research
Facility at Michigan State University. The two diets contained different amounts of

21

growth additive; 0 g/ton and 9 g/ton. The animals were fed for 28 days then harvested.
Collected data was scaled to 4,000 animals in order to provide sufficient manure for
model-simulated field application. Example baseline validation parameters for the two
diets, based on study data, are depicted in Table 2.
Table 2. User inputs for baseline validation scenarios
Quantitative
Effect of
Growth
Additive

Diet N
Conte
nt (%)

Diet P
Content
(%)

Fat Free
Lean
Gain (%)

Start
Weight
(kg)

End
Weight
(kg)

Average
Dressing
Percentage
(%)

Daily
Average
Feed Intake
(kg/animal)

Days
on
Feed

4,000

0

16.2

0.5

56

91.7

133

74

3.1

35

4,000

0.2

16.2

0.5

56

92.8

140

74

3.1

35

Diet
Name

Number
of
Animals

A
B

The model requires user inputs for manure storage method, manure spreading method,
and timing of spreading in addition to the location of the fields. Table 3 outlines an
example baseline scenario of inputs.
Table 3. Baseline scenario inputs for manure storage and land application
Field Location

Nutrient to Base
Spreading On

Soil P Limits
(ppm)

Storage Method

Application
Method

Timing of
Application

Lansing, MI

N

150

Liquid/slurry
uncovered

Surface
broadcast

August/May

The baseline scenarios represented the study parameters as well as a realistic
extension of the observed study data to include manure management. Assumptions for
cropping inputs reflect common industry practices geographically proximate to the study
location (Table 3). Table 4 depicts the crop production inputs used for both baseline
scenarios.

22

Table 4. Baseline scenario inputs for crop growth
Field ID

Crop

Acres of
Field

Expected Yield
(bu/acre)

N Soil Test
(lb/acre)

P Soil Test
(ppm)

Frequency of
Manure
Spreading

1

Soybeans

50

45

50

25

Every year

2

Corn

50

160

50

25

Every year

3

Soybeans

75

45

75

35

Every year

4

Corn

75

160

75

35

Every year

5

Soybeans

100

45

30

20

Every year

To test the sensitivity and behavior of the model a series of scenarios using data in
Tables 1, 2, and 3 were developed as model inputs. Table 5 outlines the scenario
assumptions for each of the two study diets (baseline scenarios). Scenario 2 used the
same inputs as the baseline scenario, but altered the method of manure application
onto fields; changing the application from surface broadcast to injection, which have
different volatilization rates and will affect nutrients available for crop use. Scenario 3
altered the field location from Lansing, MI (baseline) to Greenville, MS (Scenario 3).
Regulations on field P limits vary by state, therefore the soil P limit was modified to
reflect regulations applicable to Greenville, MS. This scenario was intended to illustrate
the impact soil texture and P limits have on management longevity, so a location that
differed in both respects was selected. Scenario 4 modified the manure storage method
from uncovered liquid/slurry storage to bedded pack while keeping the other baseline
inputs the same. These two manure storage methods are very different with respect to
nutrient loss and would provide an understanding for system sensitivity to change in
manure storage method. The final modified scenario (5) used the same inputs as the
baseline scenario with a change in the P content of both diets. The modified scenario
increased the dietary P from 0.5% to 0.75%. These modified scenarios are important to

23

exhibit how the program responds to changing situations and its applicability to different
management practices.
After determining the validity of model outputs for situations where manure was
applied to meet N needs of the crops, the same scenarios were considered using
manure based on crop P needs. This provided a better understanding of program
sensitivity for both management decisions.

24

Table 5. Modification scenarios to test sensitivity of the model
Trial
name

Number
Diet N Diet P Fat free Starting
End
Growth
of
content content lean gain weight weight
additive
animals
(%)
(%)
(%)
(kg)
(kg)

Average
dressing
percentage
(%)

Daily
average
feed intake
(kg/day)

Days
on
feed

Field location

Nutrient to Soil P
base
limits
spreading on (ppm)

Manure
storage
method

Manure
application
method

Time(s) of
manure
application

Number
of fields

Crops

Acres
of
fields

Expected
crop
yield
(bu/acre)

N soil
P soil
test
test
(lbs/ac
(ppm)
re)

Frequency of
manure spreading
on field

1A1,2

4000

0

16.2

0.5

56

91.7

133

74

3.1

35

Lansing, MI

N

150

Liquid/Slurry
Uncovered

Surface
Broadcast

August
May

5

Soybeans
Corn
Soybeans
Corn
Soybeans

50
50
75
75
100

45
160
45
160
45

50
50
75
75
30

25
25
35
35
20

Every year
Every year
Every year
Every year
Every year

1B1,2

4000

0.2

16.2

0.5

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Liquid/Slurry
Uncovered

Surface
Broadcast

August
May

5

Soybeans
Corn
Soybeans
Corn
Soybeans

50
50
75
75
100

45
160
45
160
45

50
50
75
75
30

25
25
35
35
20

Every year
Every year
Every year
Every year
Every year

2A

3

4000

0

16.2

0.5

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Liquid/Slurry
Uncovered

Injection

August
May

5

2B 3

4000

0.2

16.2

0.5

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Liquid/Slurry
Uncovered

Injection

August
May

5

3A

4

4000

0

16.2

0.5

56

92.8

140

74

3.1

35

Greenville,
MS

N

100

Liquid/Slurry
Uncovered

Surface
Broadcast

August
May

5

3B 4

4000

0.2

16.2

0.5

56

92.8

140

74

3.1

35

Greenville,
MS

N

100

Liquid/Slurry
Uncovered

Surface
Broadcast

August
May

5

4A

5

4000

0

16.2

0.5

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Bedded
Pack

Surface
Broadcast

August
May

5

4B 5

4000

0.2

16.2

0.5

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Bedded
Pack

Surface
Broadcast

August
May

5

5A6

4000

0

16.2

0.75

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Liquid/Slurry
Uncovered

Surface
Broadcast

August
May

5

5B 6

4000

0.2

16.2

0.75

56

92.8

140

74

3.1

35

Lansing, MI

N

150

Liquid/Slurry
Uncovered

Surface
Broadcast

August
May

5

Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans
Soybeans
Corn
Soybeans
Corn
Soybeans

50
50
75
75
100
50
50
75
75
100
50
50
75
75
100
50
50
75
75
100
50
50
75
75
100
50
50
75
75
100
50
50
75
75
100
50
50
75
75
100

45
160
45
160
45
45
160
45
160
45
45
160
45
160
45
45
160
45
160
45
45
160
45
160
45
45
160
45
160
45
45
160
45
160
45
45
160
45
160
45

50
50
75
75
30
50
50
75
75
30
50
50
75
75
30
50
50
75
75
30
50
50
75
75
30
50
50
75
75
30
50
50
75
75
30
50
50
75
75
30

25
25
35
35
20
25
25
35
35
20
25
25
35
35
20
25
25
35
35
20
25
25
35
35
20
25
25
35
35
20
25
25
35
35
20
25
25
35
35
20

Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year
Every year

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting additive. Label B indicates a diet that does include growth at a level indicated further in the table.
2 The baseline scenario tested by the model. The scenario variables include a phosphorus field concentration limit of 150 ppm, uncovered liquid/slurry manure storage, surface broadcast manure application, feed phosphorus
concentrations of 0.5%, and a field location of Lansing, MI
3 All variables for the second scenario are the same as the baseline scenario except for manure application method. Manure application method is modified from surface broadcast to injection.
4 All variables for the third scenario are the same as the baseline scenario except for the location of the fields and field phosphorus limits. The location of the fields is modified from Lansing, MI to Greenville, MS. Due to differences in
state regulations the field phosphorus limit was modified from 150 ppm to 100 ppm.
5 All variables for the fourth scenario are the same as the baseline scenario except for manure storage method. The manure storage method is modified from liquid/slurry to bedded pack.
6 All variables for the fifth scenario are the same as the baseline scenario except for the phosphorus levels in the diet. Diet phosphorus levels were modified from 0.5% to 0.75% for both diets.

25

4.4.8 Monte Carlo analysis
Because the model represents a conglomeration of many variables impacting the
behavior of the much larger system, outputs from each scenario model run was
compared to literature values from studies addressing similar practices. These
comparisons provided a sense of suitability of the model to represent observed
situations. To understand the total margin of error the model creates through each run,
maxima and minima for selected variables were drawn from literature in order to run
Monte Carlo simulations. Using a random number generator to select values within the
maximum and minimum of the variable based on literature values (Table 6), the model
was run 1000 times. The data from the model runs were fit to a histogram to test the
normalcy of the distribution and variability of the model outputs.

Table 6. Monte Carlo assumptions for analysis
Variable

Default

Min

Max

Dietary N (%)

16.2

14

17

Dietary P( %)

0.5

4

7.5

Fat Free Lean Gain (%)

56

52

58

Start Weight (kg)

91.7

90

105

End Weight (kg)

133

120

135

Average Dressing Percentage

74

72

78

Daily Feed Intake

3.1

2.5

3.5

26

5 Results
5.1 Baseline Scenarios
Based on the calculations from the model, the diet containing the feed additive
resulted in less manure, reduced N excretion resulting in less N available for crop use.
Both diets produced the same P outputs (table 7). Manure excreted, N excreted, P
excreted, and N and P available for crop uptake is similar whether running the model to
spread manure to meet crop N needs or to meet crop P needs for the baseline scenario
as well as the subsequent scenarios analyzed. Figures 3 and 4 compare model outputs
to the observed data for both baseline diets. The model was able to generate outputs
similar to those observed in the study. Differences between the calculated values and
the observed values fall within the error of the observed data, though the calculated
values are very close to the high end allowed by the observed data plus standard
deviation (figure 3). This indicates that the data generated by the model is an
acceptable representation of what may be seen in a real world situation. Both calculated
outputs and observed data depict that diet B predicted less N excretion than that of diet
A (figure 4) due to the metabolic enhancing additive used in diet B increasing muscle
accretion, thus increasing N retention.

27

Table 7. Model outputs for baseline scenarios
Model outputs when
spreading to meet
nitrogen needs

Manure
Excreted
(MM
gal/year)

Scenario

Nitrogen
Excreted
(Mg/year)

Phosphorus
Excreted
(Mg/year)

Nitrogen
Available
for Crop
Uptake
(Mg/year)

Model outputs
when spreading
to meet
phosphorus
needs

Phosphorus
Available
for Crop
Uptake
(Mg/year)

Years
Before
Critical
P
Limits
Met

Total
Acres
Required
to Use All
Manure
Excreted

Additional
Nitrogen
Fertilizer
Required
(lb)

1,2

2.3

64.0

13.1

24.0

12.4

38

470

6900

1,2

1.8

53.3

13.1

20.0

12.4

20

450

10200

1A
1B

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting additive.
Label B indicates a diet that does include growth at a level indicated further in the table
2 The baseline scenario tested by the model. The scenario variables include phosphorus field concentration limit of 150 ppm, uncovered
liquid/slurry manure storage, surface broadcast manure application, feed phosphorus concentrations of 0.5%, and a field location of Lansing, MI

Figure 3. Model outputs and observed study data comparison for annual manure
production

Manure Excreted (KGal/year)

2500
Calculated
Model Output
Observed
Study Data

2000

1500

1000

500

0

Diet Without Growth
Promoting Additive

Diet With Growth Promoting
Additive

28

Figure 4. Model outputs and observed study data comparison for annual N excretion
and N loss
70

Calculated Model
Output

60

Observed Study
Data

Excreted (Mg)

50
40
30
20
10
0
Nitrogen Excreted Diet
Without Additive

Nitrogen Excreted Diet
With Additive

Nitrogen Loss From the
System Diet Without
Additive

Nitrogen Loss From the
System Diet With
Additive

5.2 Manure Spreading Method Modification (Scenario 2) Outputs
The modification of the manure spreading technique affected the N and P
available to crops. The calculated values are shown in table 8. By changing the method
from surface broadcast to injection, the nutrients available for crop uptake were
increased. Because surface broadcast allows for greater volatilization, more of the N
was lost. Given the manure spreading parameters of the baseline scenario, the model
was programmed to calculate a loss of 58% of the available N using surface broadcast
manure spreading. By modifying the manure spreading method to injection, the model
calculated that the N loss was decreased to 1% of available N.

29

Table 8. Model outputs for scenario altering manure application method from surface
broadcast to injection (scenario 2)
Model outputs
when spreading to
meet nitrogen
needs

Model outputs
when spreading
to meet
phosphorus
needs

Phosphor
us
Available
for Crop
Uptake
(Mg/year)

Years
Before
Critical
P
Limits
Met

Total
Acres
Require
d to Use
All
Manure
Excreted

Additional
Nitrogen
Fertilizer
Required
(lb)

Manure
Excreted
(MM
gal/year)

Nitrogen
Excreted
(Mg/year)

Phosphorus
Excreted
(Mg/year)

Nitrogen
Available
for Crop
Uptake
(Mg/year)

1,2

2.3

64.0

13.1

24.0

12.4

38

470

6900

1,2

1.8

53.3

13.1

20.0

12.4

20

450

10200

1,3

2.3

64.0

13.1

48.0

13.0

40

940

3400

1,3

1.8

53.3

13.1

34.0

13.0

36

765

5500

Scenario

1A
1B
2A
2B

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting
additive. Label B indicates a diet that does include growth at a level indicated further in the table
2 The baseline scenario tested by the model. The scenario variables include phosphorus field concentration limit of 150 ppm, uncovered
liquid/slurry manure storage, surface broadcast manure application, feed phosphorus concentrations of 0.5%, and a field location of Lansing,
MI
3 All variables for the second scenario are the same as the baseline scenario except for the manure application method. Manure application
method is modified from surface broadcast to injection

Huijsman (2003) conducted a study comparing various manure management
and application methods to understand N loss after manure application to fields. The
study used production methods similar to those used in the model calculations and
found that surface broadcast application methods had a mean loss of 56.2% of the total
N applied to the fields. Total N loss values observed in the study ranged from 44% to
70% of the total N applied. Injecting manure resulted in a mean loss of 0.87% of the
total N applied to the field, with the 95% confidence interval range of 0.58% to 1.15%
loss of total applied N. Bittman et al. (2005), Hansel et al. (2003), and Smith et al.
(2000) observed similar ranges in total N lost, all of which comfortably encompass the
total N losses due to the injection manure method calculated by the model. The ranges

30

reported in literature indicate that the calculated outputs and equations to estimate N
losses due to land application method appropriately represent what would be
reasonably be expected in field observations. Losses are attributed to the volatilization
of ammonia. The greater extent of volatilization resulting from the surface broadcast
technique reflects greater exposure to wind, rain, and sun, leading to increased mass
transfer and, ultimately, N loss. Manure injection decreased exposure to the elements
that increase mass transfer and N loss.
5.3 Field Location and Critical P Threshold Modification (Scenario 3) Outputs

The third scenario assessed the location of production as well as the critical
threshold of field P concentration. By modifying the location of the fields, the dominant
soil texture was changed from loam, found in Lansing, MI, to sandy loam, found in
Greenville, MS. The critical threshold of P decreased from 150 mg/kg in Michigan to 100
mg/kg in Mississippi. Due to the change in the dominant soil texture, the model
assumed N losses during manure spreading increased to approximately 68% from the
58% loss estimated in the baseline scenario.

31

Table 9. Model outputs for scenario altering field location and P critical limits (scenario
3)

Model outputs when
spreading to meet
nitrogen needs

Model
outputs when
spreading to
meet
phosphorus
needs

Manure
Excreted
(MM
gal/year)

Nitrogen
Excreted
(Mg/year)

Phosphorus
Excreted
(Mg/year)

Nitrogen
Available
for Crop
Uptake
(Mg/year)

Phosphoru
s Available
for Crop
Uptake
(Mg/year)

Years
Before
Critical
P Limits
Met

Total Acres
Required to
Use All
Manure
Excreted

Additional
Nitrogen
Fertilizer
Required
(lb)

1,2

2.3

64.0

13.1

24

12.4

38

470

6900

1,2

1.8

53.3

13.1

20.0

12.4

20

450

10200

1,4

2.3

64.0

13.1

20.4

11.1

42

400

8300

1,4

1.8

53.3

13.1

17.0

11.1

45

383

11000

Scenario

1A
1B
3A
3B

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting additive.
Label B indicates a diet that does include growth at a level indicated further in the table
2 The baseline scenario tested by the model. The scenario variables include phosphorus field concentration limit of 150 ppm, uncovered
liquid/slurry manure storage, surface broadcast manure application, feed phosphorus concentrations of 0.5%, and a field location of Lansing, MI
4 All variables for the third scenario are the same as the baseline scenario except for the the location of the fields and field phosphorus limits. The
location of the fields is modified from Lansing, MI to Greenville, MS. Due to differences in state regulations the field phosphorus limit was modified
from 150 ppm to 100 ppm

Loam texture, being a finer soil, allows smaller amounts of nutrients deeper in the
soil, leaving more N and P available for plant use. Due to the addition of sand in the
sandy loam soil found in Greenville, MS, N and P readily move through the larger pores
in the soil, allowing for less of the applied nutrients to be available to the crops. Because
of the increase in nutrient movement in the soil for the modified scenario, the field
accumulated P at a slower rate due to the nutrient losses, which led to critical limits
being met later. The program took all of these factors into account when calculating the
availability of nutrients for uptake by crops after application. Griffin et al. (2002) applied
swine manure to both sandy loam soil and loam soils. The study found a difference of
volatilized N between the soil types, with loam soil associated with 77% of total N

32

applied volatilized while sandy loam corresponded to, on average, 99% of total N
applied volatilized. Similarly, Huijsmans et al. (2003) found that a range between 37%
and 100% of total applied N to sandy loam soils was volatilized. The values found in
these studies are greater than those calculated by the model, but the model calculations
fell within the error of both study values.
5.4 Manure Storage Modification (Scenario 4) Outputs
The fourth scenario compares N losses from an uncovered liquid/slurry system
versus a bedded pack. Cortus et al. (2009) found that by storing manure as a
liquid/slurry, N losses ranged from 17% to 137%. Model estimates fall within this range,
a range that was observed by Portejoie et al. (2003). The model estimated N losses
from an uncovered liquid/slurry system at 25% while it estimated that a bedded pack
lost 50% of available N. Thus, the developed model produced outputs within reported
values.

33

Table 10. Model outputs for scenario altering manure storage method (scenario 4)

Model outputs when
spreading to meet
nitrogen needs

Nitrogen
Excreted
(Mg/year)

Phosphor
us
Excreted
(Mg/year)

Nitrogen
Available
for Crop
Uptake
(Mg/year)

Phosphorus
Available
for Crop
Uptake
(Mg/year)

Years
Before
Critical
P Limits
Met

Total
Acres
Required
to Use All
Manure
Excreted

Additional
Nitrogen
Fertilizer
Required
(lb)

2.3

64.0

13.1

24.0

12.4

38

470

6900

1.8

53.3

13.1

20.0

12.4

20

450

10200

2.3

64.0

13.1

16.0

9.8

18

313

8200

1.8

53.3

13.1

13.3

9.8

3

299

14000

Manure
Excreted
(MM
gal/year)

1,2
1,2
1,5
1,5

Scenario

1A
1B
4A
4B

Model
outputs
when
spreading
to meet
phosphorus
needs

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting additive.
Label B indicates a diet that does include growth at a level indicated further in the table
2 The baseline scenario tested by the model. The scenario variables include phosphorus field concentration limit of 150 ppm, uncovered
liquid/slurry manure storage, surface broadcast manure application, feed phosphorus concentrations of 0.5%, and a field location of Lansing, MI
5 All variables for the fourth scenario are the same as the baseline scenario except for the manure storage method. The manure storage method
was changed from liquid/slurry to bedded pack

34

5.5 Dietary P Modification (Scenario 5) Outputs
The final scenario assessed the impact of increasing dietary P content from the
baseline 0.5% P to 0.75% P. This increase in dietary P content increased the P
excreted by the animals and subsequently the P available in soil for crop uptake. Table
11 shows the model outputs for the Scenario 5.
Table 11. Model outputs for scenario increasing dietary P (scenario 5)

Model outputs when
spreading to meet
nitrogen needs

Phosphorus
Excreted
(Mg/year)

Nitrogen
Available
for Crop
Uptake
(Mg/year)

Phosphorus
Available
for Crop
Uptake
(Mg/year)

Years
Before
Critical
P Limits
Met

Total
Acres
Required
to Use All
Manure
Excreted

Additional
Nitrogen
Fertilizer
Required
(lb)

64.0

13.1

24.0

12.4

38

470

6900

1.8

53.3

13.1

20.0

12.4

20

450

10200

2.3

64.0

24.4

48.0

18.3

16

940

8500

1.8

53.3

24.4

40.0

18.3

3

900

14000

Manure
Excreted
(MM
gal/year)

Nitrogen
Excreted
(Mg/year)

1,2

2.3

1,2
1,6
1,6

Scenario

1A
1B
5A
5B

Model
outputs when
spreading to
meet
phosphorus
needs

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting additive.
Label B indicates a diet that does include growth at a level indicated further in the table
2 The baseline scenario tested by the model. The scenario variables include phosphorus field concentration limit of 150 ppm, uncovered
liquid/slurry manure storage, surface broadcast manure application, feed phosphorus concentrations of 0.5%, and a field location of Lansing, MI
6 All variables for the fifth scenario are the same as the baseline scenario except for the phosphorus levels in the diet. Diet phosphorus levels
were modified from 0.5% to 0.75% for both diets

Because the crops produced in both scenarios were the same, nutrient needs
were the same. As a result, the modified scenario accumulated P in the soil at a faster
rate than the baseline scenario, resulting in the cessation of the management practice in
fewer years. The calculations yielded an 83% increase in P excretion from swine when
increasing the dietary P to 0.75%. Rodehutscord et al. (1999) found that when
increasing the dietary P that there was up to a 81% increase in P excretion when fed

35

diets containing 0.70% dietary P. Vipperman et al. (1974) observed an increase in P
output of 60% to 110% as a result of increasing dietary P by 50%.
5.6 Management practice influence on long-term sustainability
5.6.1 Manure spreading to meet crop N needs
Model outputs for the scenarios depict that the number of years a practice can be
used before critical P limits are met varies widely by management and site variables.
Due to the fact that all of the scenarios spread manure to meet N needs of crops, P was
always over-applied to the fields, allowing each scenario to hit the critical P thresholds
at different time frames. When compared to the baseline scenario for the diet not
containing growth additive, the scenario modifying manure application from surface
broadcast to injection (scenario 2A) and the scenario modifying field location and critical
P limits (scenario 3A) both have less than a 10% change in longevity of practice from
the baseline scenario. Longevity for the scenario modifying manure storage from
uncovered liquid/slurry to bedded pack (scenario 4A) and the scenario increasing
dietary P (scenario 5A) decreased longevity 53% and 57%, respectively, when
compared to the baseline scenario for the diet with no feed additive. This indicates that
management changes earlier in the system (manure storage and dietary changes) have
a greater effect on nutrient loss and management efficacy than those further postexcretion in the system.
The model calculated that management decisions could continue for 80% and
125% more time for the scenario modifying manure application from surface broadcast
to injection (scenario 2B) and the scenario modifying field location and critical P limits

36

(scenario 3B), respectively, when compared to the baseline scenarios of diet B.
Longevity of practices decreased 85% in both the scenario modifying manure storage
from uncovered liquid/slurry to bedded pack (scenario 4B) and the scenario increasing
dietary P (scenario 5B) compared to the baseline of diet B. These factors indicate that
the addition of metabolic enhancing feed additives create more sensitivity within the
system, allowing for more options and modifications having a greater effect when
practiced.
The results indicate that the growth additive regularly decreases the amount of time the
management style could be used, which is most evident in the scenario modifying
manure storage from uncovered liquid/slurry to bedded pack (scenario 4) and the
scenario increasing dietary P (scenario 5). The baseline scenario saw a similar
discrepancy between the two diets. When the growth additive was fed, less N was
excreted, so the concentration of N relative to the mass of manure excreted decreased
while the concentration of P relative to the mass of manure excreted increased.
Because manure was spread to meet the N needs of the crops produced, more mass of
manure was growth additive was fed, which meant that a greater amount of P was
applied yearly, leading to faster P accumulation in the soil compared to the scenario
without the addition of growth additive. The exception to this effect is the case of the
scenario changing farm location and critical P limits (scenario 3), where the diet with
growth additive can be practiced for a slightly longer period (45 compared 42 years); as
much due to state policy as any biological determinant.

37

5.6.2 Manure spreading to meet crop P needs
Baseline scenario calculations when spreading manure to meet P needs
revealed that the scenario supplying a growth additive to the animals ultimately required
a greater amount of additional N fertilizer after spreading due to lower N concentration
in the manure. The fields producing soybeans, for all scenarios assessed, did not
require additional N fertilizer due to N fixation by soybeans. The scenario modifying
manure spreading from surface broadcast to injection required less additional N fertilizer
due to the higher concentration of N in the manure, leading to a greater amount of N
applied when meeting phosphorus needs, when compared to the baseline. Changing
farm location and dominant soil texture, manure storage, and dietary phosphorus
content all increased the requirement for additional N fertilizer relative to the baseline
scenarios. This was due to the concentration of N in the manure as a percentage of
total manure content decreasing while P concentrations remained similar. This change
in the N to P ratio resulted in less N being applied when spreading manure to meet crop
P needs, thus requiring more supplemental N to be applied to the field.
All of the information gleaned from the comparison of the scenarios show that
there are distinct sensitivities found with each of the scenarios which ultimately affect
how the model behaves. The differences in spreading techniques, whether spreading to
meet N or P needs, and the supplemental fertilizer needed will be important when
deciding management options.
5.6.3 Monte Carlo results
Monte Carlo results for each scenario spreading manure based on N crop needs
are depicted in figure 5. From these figures, all scenario analyses show a non-normal

38

distribution of model outcomes. All analyses depict a skew in the model outputs. This
indicates that the model generates outputs with high variability. This is due to the high
variability in observed data used to develop the model.
Figure 5: Monte Carlo analysis histograms for A: Baseline scenario, B: Scenario 2
(modified manure application), C: Scenario 3 (modified field location), D: Scenario 4
(modified manure storage), and E: Scenario 5: (modified dietary P)

5.7 Software Platform Efficacy
The LabVIEW programming software presented challenges and benefits
throughout programming that may persist beyond the completion of the model. By using
LabVIEW software, multiple methods to achieve the same goal were found, allowing for
more flexibility during the programming phase. LabVIEW also allowed for sub-programs

39

to be created facilitating a more organized design, while making it easier for users to
navigate the wire diagram of the program. Though LabVIEW programming was
beneficial during the programming phase, the complexity of the method needed to
import and use large data sets as a way to create new equations or update out-of-date
equations within the LabVIEW program may be a deterrent for those who may update
the program down the road. If the user does not desire to invest the time to learn the
LabVIEW software enough to use the complex methods, creating an equation in
Microsoft Excel and then importing that equation in the LabVIEW program would be
required. Additionally, due to the complexity of the model created, many case structures
were used to drive the program based on user inputs. Because of this, there may be
difficulties encountered by users in following the logic of the wire diagram. In order to
counter this issue, it was necessary to ensure that all structures were labeled
adequately in order to make very clear what the role of each structure was. Overall, the
difficulties encountered in the process of programming and that may be encountered by
future users are largely due to the fact that the LabVIEW software was not developed as
a platform to build decision making tools. These difficulties were minimal compared to
the power the LabVIEW software afforded in making many calculations simultaneously
and iteratively. While users may encounter difficulties in modifying the program, the user
interface was programmed to allow ease of use while using it as a decision making tool,
satisfying the need for a program with an accessible, easy to understand user interface.

40

6 Conclusions
The developed model demonstrated an ability to accurately portray a realistic
vision of a long term whole farm nutrient balance for a swine grow-finish facility under
different management and site characteristic scenarios. When tested against literature
values, the outputs accurately depicted what was observed. Monte Carlo analyses
revealed high variability in the output of the model that mimicked observed variation.
The model allows the user to observe the behavior of the system at a level that
facilitates understanding of what variables impact nutrient loss and long-term
sustainability of management practices. Through model runs depicting different
management options, the model demonstrated its sensitivity to site management
conditions. This sensitivity allows the user to understand the downstream effects of
management. Based on the scenarios considered, the model has a greater sensitivity to
the on-farm management practices modified earlier in the system (i.e swine
feed/metabolism and manure storage) compared to off-farm practices and influences
(i.e. crop uptake and losses due to manure spreading). Given the ability of the model to
demonstrate the critical points of control within the system, this program has the ability
to lend an understanding not only to those aiming to expand the research of variables
within the system, but also to producers and advisors who aim to enhance adoptions of
practices and behaviors to create a long-term sustainable food system.
While the model created brought together many variables in order to give users
the ability to visualize the system as a whole and understand the feedback influences,
there is still much that could be developed to enhance the tool. Upstream decisions had
large influence on downstream losses and longevity; the model would benefit from to

41

enhanced feedback that ties to animal production decisions (i.e. decisions of herd size
and dietary modification based on performance).
Further research of the variables not validated by observed data from a study,
namely those associated with manure storage, manure application, and crop uptake,
would allow the model to increase the accuracy of the outputs and increase the ability to
accurately represent the system. Including an economic factor and environmental
factors beyond the current scope that will affect the system would provide a broader
understanding of the system.
The overarching goal of this model is to be able to use it in a much larger model
in order to understand the suitability of animal protein production through the systems
behaviors. Given the capabilities the model has demonstrated, there still remains
potential to expand the model to include more variables and feedback as well as include
more species involved in protein production in order to provide greater depth when used
in the more expansive model, though the software platform used may cause difficulties
when moving forward with the model.

42

APPENDICES

43

APPENDIX A – Full table of all scenario outputs

Table 12. Complete table of all scenario outputs analyzed

Scenario

Manure
Excreted
(MM
gal/year)

Nitrogen
Excreted
(Mg/year)

Phosphorus
Excreted
(Mg/year)

Nitrogen
Available
for Crop
Uptake
(Mg/year)

Phosphorus
Available
for Crop
Uptake
(Mg/year)

Years
Before
Critical
P Limits
Met

Total
Acres
Required
to Use All
Manure
Excreted

1A1,2

2.3

64

13.1

24

12.4

38

470

1B1,2

1.82

53.3

13.1

20

12.4

20

450

2.3

64

13.1

48

13

40

940

1.82

53.3

13.1

34

13

36

765

2.3

64

13.1

20.4

11.1

42

400

3B1,4

1.82

53.3

13.1

17

11.1

45

383

1,5

2.3

64

13.1

16

9.8

18

313

1,5

1.82

53.3

13.1

13.3

9.8

3

299

1,6

2.31

64

24.4

48

18.3

16

940

1,6

1.83

53.3

24.4

40

18.3

3

900

2A

1,3

2B1,3
3A

1,4

4A
4B
5A
5B

1 Label of A and B indicate the diet differences for each scenario. Label A indicates the diet that does not include a growth promoting additive. Label B indicates a diet that
does include growth at a level indicated further in the table.
2 The baseline scenario tested by the model. The scenario variables include a phosphorus field concentration limit of 150 ppm, uncovered liquid/slurry manure storage,
surface broadcast manure application, feed phosphorus concentrations of 0.5%, and a field location of Lansing, MI
3 All variables for the second scenario are the same as the baseline scenario except for manure application method. Manure application method is modified from surface
broadcast to injection.
4 All variables for the third scenario are the same as the baseline scenario except for the location of the fields and field phosphorus limits. The location of the fields is modified
from Lansing, MI to Greenville, MS. Due to differences in state regulations the field phosphorus limit was modified from 150 ppm to 100 ppm.
5 All variables for the fourth scenario are the same as the baseline scenario except for manure storage method. The manure storage method is modified from liquid/slurry to
bedded pack.
6 All variables for the fifth scenario are the same as the baseline scenario except for the phosphorus levels in the diet. Diet phosphorus levels were modified from 0.5% to
0.75% for both diets.

44

APPENDIX B - Monte Carlo Yearly Output Statistics

Table 13. Output statistics for monte carlo analysis for all scenarios
Scenario

1

2

3

4

5

Mean

36

28.3

19.6

8.17

20

Standard Error

1.26

1.13

0.86

0.25

0.58

Median

17

14

10

4

14

Mode

12

4

4

4

3

Standard Deviation

39.7

35.9

27.2

7.96

18.4

Kurtosis

10.6

4.93

2.08

10.7

5.29

Skewness

3.12

2.27

1.84

3.05

2.26

Range

378

378

177

94

171

Minimum

3

3

3

3

3

Maximum

381

381

180

97

174

Count

1000

1000

1000

1000

1000

45

APPENDIX C - Program Screenshots

Figure 6. Front panel user interface for location, manure spreading, growth additive,
and animal performance data inputs

46

Figure 7. Front panel user interface for manure storage, manure application, and field
information. Image shows inputs for only 5 fields, but the model is programmed to allow
up to 30 fields.

47

Figure 8. Model outputs as seen on front panel of the program.

48

Figure 9. Block diagram of one of six available phase available in the program. The inputs (seen on the left
of the case structure) are dictated by the user on the front panel. The inputs feed into a case structure to
calculate N and P excretion.

49

Figure 10. Nutrient loss calculation due to manure storage and land application block diagram

Field dominant soil
texture calculations

Storage nutrient
loss calculation
case structure

Field application nutrient loss
calculation case structure

50

Figure 11. Crop nutrient uptake and write to file portion of the block diagram.
Remaining nutrients flow to this section and crop nutrient use is calculated for each
yearly iteration.

51

BIBLIOGRAPHY

52

BIBLIOGRAPHY

ASAE, 2005, ASAE Standard D384.2: American Society of Agricultural Engineers.
Arogo, J., P. W. Westerman, and A. J. Heber, 2003, A review of ammonia emissions
from confined swine feeding operations: Transactions of the Asae, v. 46, p. 805817.
Benham, B. L., R. B. Ferguson, C. G. Henry, M. A. Reynolds, C. A. Shapiro, J. P. Stack,
and C. S. Wortman, 2009, Managing livestock manure to protect environmental
quality: UNL EC.
Bittman, S., L. J. P . van Vliet, C. G. Kowalenko, S. McGinn, D. E. Hunt, and F.
Bounaix, 2005, Surface-banding liquid manure over aeration slots: Agronomy
Journal, v. 97, p. 1304-1313.
Brunke, R., P. Alvo, P. Schuepp, and R. Gordon, 1987, Effect of meteorological
parameters on ammonia loss from manure in the field: Journal of Environmental
Quality, v. 17, p.431-436.
Cornell (2003). Nutrient Management Software version 2.0 [Computer Software]. Ithaca,
NY.
Cortus, E. L., S. P. Lemay, E. M. Barber, and G. A. Hill, 2009, Modelling ammonia
emission from swine slurry based on chemical and physical properties of the
slurry: Canadian Biosystems Engineering, v. 51, p. 6.9-6.22.
Crook, B., J. F. Robertson, S. A. T. Glass, E. M. Botheroyd, J. Lacey, and M. D.
Topping, 1991, Airborne dust, ammonia, microorganisms, and antigens in pig
confinement ouses and the respiratory health of exposed farm workers:
American Industrial Hygiene Association Journal, v. 52, p. 271-279.
Daniel, T. C., A. N. Sharpley, D. R. Edwards, R. Wedepohl, and J. L. Lemunyon, 1994,
Minimizing surface-water eutrophication from agriculture by phosphorus
management: Journal of Soil and Water Conservation, v. 49, p. 30-38.
DeCamp, S. A., S. L. Hankins, A. L. Carroll, D. J. Ivers, B. T. Richert, A. L. Sutton, and
D. B. Anderson, 2001, Effects of ractopamine and level of dietary crude protein
on nitrogen and phosphorus excretion from finishing pigs: Purdue University
Swine Research Report.
Drummond, J. G., S. E. Curtis, J. Simon, and H. W. Norton, 1980, Effects of aerial
ammonia on growth and health of young pigs: Journal of Animal Science, v. 50,
p. 1085-1091.

53

Fry, B., 2004, zipdecode: www.benfry.com/zipdecode/.
Gerland, P., A. E. Raftery, H. Sevcikova, N. Li, D. Gu, T. Spoorenberg, L. Alkema, B. K.
Fosdick, J. Chunn. N. Lalic, G. Bay, T. Buettner, G. K. Heilig, and J. Wilmoth,
2014,World population stabilization unlikely in this century: Science, v. 346, p.
234-237.
Griffin, T., C. Honeycutt, Z. He, 2002, Effects of temperature, soil water status, and soil
type on swine slurry nitrogen transformations: Biology and Fertility of Soils, v. 36,
p. 442-446.
Hansen, M. N., K. Henriksen, and S. G. Sommer, 2006, Observations of production and
emission of greenhouse gases and ammonia during storage of solids separated
from pig slurry: Effects of covering: Atmospheric Environment, v. 40, p. 41724181.
Hansmann, R., H. A. Mieg, and P. Frischknecht, 2012, Principal sustainability
components: empirical analysis of synergies between the three pillars of
sustainability: International Journal of Sustainable Development and World
Ecology, v. 19, p. 451-459.
Harper, A. F, E. T. Kornegay, and T. C. Schell, 1997, Phytase supplementation of lowphosphorus growing-finishing pig diets improves performance, phosphorus
digestibility, and bone mineralization and reduces phosphorus excretion: Journal
of Animal Science, v. 75, p. 3174-3186.
Havlin, J. L., J. D. Beaton, S. L. Tisdale, and W. L. Nelson, 2014, Soil fertility and
fertilizers: an introduction to nutrient management: Soil fertility and fertilizers: an
introduction to nutrient management., p. x + 499 pp.
Herber, A. J., J. Q. Ni, T. T. Lim, C. A. Diehl, A. L. Sutton, R. K. Duggirala, B. L.
Haymore, D. T. Kelly, and V. I. Adamchuk, 2000, Effect of a manure on ammonia
emission from swine finishing buildings: ASAE, v. 43, p. 1895-1902.
Hoffmann, I., 2011, Livestock biodiversity and sustainability: Livestock Science, v. 139,
p. 69-79.
Honeyman, M. S., 1996, Sustainability issues of US swine production: Journal of Animal
Science, v. 74, p. 1410-1417.
Huijsmans, J. F. M., J. M. G. Hol, and G. D. Vermeulen, 2003, Effect of application
method, manure characteristics, weather and field conditions on ammonia
volatilization from manure applied to arable land: Atmospheric Environment, v.
37, p. 3669-3680.

54

Innes, R., 2000, The economics of livestock waste and its regulation: American Journal
of Agricultural Economics, v. 82, p. 97-117.
Jongbloed, A. W., and N. P. Lenis, 1998, Environmental concerns about animal
manure: Journal of Animal Science, v. 76, p. 2641-2648.
Karmakar, S., C. Lague, J. Agnew, and H. Landry, 2007, Integrated decision support
system (DSS) for manure management: A review and perspective: Computers
and Electronics in Agriculture, v. 57, p. 190-201.
Kliebenstein, J. B., and J. D. Lawrence, 1995, Contracting and vertical coordination in
the United States pork industry: American Journal of Agricultural Economics, v.
77, p. 1213-1218.
Koelsch, R., and R. Stowell, 2005, Ammonia emissions estimator [worksheet].
University of Nebraska.
Lange, C. F. M. d., 2013, New NRC (2012) nutrient requirements of swine: Banff Pork
Seminar Proceedings, Banff, Alberta, Canada, 15-17 January 2013, p. 17-28.
Liu, Z., W. Powers, J. Murphy, and R. Maghirang, 2014, Ammonia and hydrogen sulfide
emissions from swine production facilities in North America: A meta-analysis:
Journal of Animal Science, v. 92, p. 1656-1665.
Morken, J. and S. Sakshaug, Direct ground injection of livestock waste slurry to avoid
ammonia emission: Nutrient Cycling in Agroecosystems, v. 51, p. 59-63.
MWPS, 1993, Livestock waste facilities handbook.
Nardone, A., B. Ronchi, N. Lacetera, M. S. Ranieri, and U. Bernabucci, 2010, Effects of
climate changes on animal production and sustainability of livestock systems:
Livestock Science, v. 130, p. 57-69.
National Instruments (2012). LabVIEW [Computer Software], Austin, TX.
NRC, 2012, Nutrient Requirements of Swine: Eleventh Revised Edition: The National
Academies Press, Washington, DC.
Olesen, J. E. and Sommer, S. G., 1993,Modelling effects of wind speed and surface
cover on ammonia volatilization from stored pig slurry: Atmospheric Environment,
v. 27, p. 2567-2574.
Philippe, F. X., M. Laitat, B. Canart, M. Vandenheede, and B. Nicks, 2007, Gaseous
emissions during the fattening of pigs kept either on fully slatted floors or on
straw flow: Animal, v. 1, p. 1515-1523.

55

Portejoie, S., J. Martinez, F. Guiziou, and C. M. Coste, 2003, Effect of covering pig
slurry stores on the ammonia emissions process: Biosource Technology, v. 87, p.
199-207.
Powell Gaines, T. and S. T. Gaines, 1994, Soil texture effect on nitrate leaching in soil
percolates: Common. Soil Sci. Plant Anal., v. 25, p. 2561-2570.
Ribaudo, M. O., N. R. Gollehon, and J. Agapoff, 2003, Land application of manure by
animal feeding operations: Is more land needed?: Journal of Soil and Water
Conservation, v. 58, p. 30-38.
Richard, T.L., J. Harmon, M. Honeyman, J. Creswell. 1997. Hoop Structure Bedding
Use, Labor, Bedding Pack Temperature, Manure Nutrient Content, and Nitrogen
Leaching Potential. ASL- R1499. In: 1997 ISU Swine Research Report. AS-638.
Dept. of Animal Science, Iowa State University, Ames, IA. pp 71-75
Rodehutscord, M., M. Faust, E. Pfeffer, 1999, The course of phosphorus excretion in
growing pigs fed continuously increasing phosphorus concentrations after a
phosphorus depletion: Archiv fur Tierernaehrung, v. 52.
Salois, M. J., 2015, Mitigating climate change and the role of innovation and
productivity: an American perspective: Eurochoices, v. 14, p. 41-47.
Soil Survey Staff, Natural Resources Conservation Service, United States Department
of Agriculture. Web Soil Survey.
Sutton, A. L., K. B. Kephart, M. W. A. Verstegen, T. T. Canh, and P. J. Hobbs, 1999,
Potential for reduction of odorous compounds in swine manure through diet
modification: Journal of Animal Science, v. 77, p. 430-439.
Tilman, D., C. Balzer, J. Hill, and B. L. Befort, 2011, Global food demand and the
sustainable intensification of agriculture: Proceedings of the National Academy of
Sciences of the United States of America, v. 108, p. 20260-20264.
Van Cauwenbergh, N., K. Biala, C. Bielders, V. Brouckaert, L. Franchois, V. G. Cidad,
M. Hermy, E. Mathijs, B. Muys, J. Reijnders, X. Sauvenier, J. Valckx, M.
Vanclooster, B. Van der Veken, E. Wauters, and A. Peeters, 2007, SAFE - A
hierarchical framework for assessing the sustainability of agricultural systems:
Agriculture Ecosystems & Environment, v. 120, p. 229-242.
van der Werf, H. M. G., and J. Petit, 2002, Evaluation of the environmental impact of
agriculture at the farm level: a comparison and analysis of 12 indicator-based
methods: Agriculture Ecosystems & Environment, v. 93, p. 131-145.

56

Vipperman, P. E., E. R. Peo, and P. J. Cunningham, 1974, Effect of dietary calcium and
phosphorus level upon calcium, phosphorus and nitrogen balance in swine:
Journal of Animal Science, v. 38, p. 758-765.
Wu, G. Y., F. W. Bazer, H. R. Cross, and Anonymous, 2014, Land-based production of
animal protein: impacts, efficiency, and sustainability: Frontiers in Agricultural
Sustainability: Studying the Protein Supply Chain to Improve Dietary Quality, v.
1328, p. 18-28.

57