Model zarządzania azotem w gospodarstwach specjalizującej się w produkcji zwierzęcej

http://www.degruyter.com/view/j/ssa (Read content)

Vol. 64 No 2/2013: 60–66

DOI: 10.2478/ssa-2013-0011

*e-mail: magdalena_szymanska@sggw.pl

INTRODUCTION

For many years, issues related to the use of nitro-gen in the broadly understood human activities have been simultaneously considered in the context of the negative impact of this element on the environment. Intensification of agriculture, guaranteeing increased production, leads at the same time to the dispersion of the components put into circulation. Estimates show that in farming areas of Europe the dispersion of nitrogen from soils is 10–50 times higher in com-parison with non-agricultural land. Studies also show that about 90% of the emitted ammonia comes from agricultural sources (Parris, 1998). The type of agri-cultural production on the farm determines the level of impact on the environment. More nutrients, in the form of animal feed and mineral fertilizers, flow into livestock producing farms in comparison with farms engaged in crop production only. In addition, on ani-mal farms there is often a surplus of manure produ-ced there in relation to the size of the arable land owned by them.

In view of these facts, particular attention should be paid to the proper management of nutrients on farms with livestock production. The lack of com-prehensive data on nutrient management in Polish agriculture and the many questions arising from the

research and experience abroad had prompted us to undertake a detailed study in this area.

The aim of this study was to develop a model of nitrogen management on farms specializing in live-stock production.

MATERIALS AND METHODS

The study was conducted on 20 farms located in 12 municipalities in the Mazowieckie province in central Poland in 2009–2012. A representative sam-ple of farms was selected in collaboration with the Mazowieckie Agricultural Advisory Centre in War-saw. The selected farms were divided, depending on the intensity and type of production, into organic farms and farms specializing in the production of dairy cattle, pigs or poultry (four types, each represented by five farms). The farmers who agreed to cooperate gathered, on a regular basis, information on crop and livestock production during each year of the study. The information was collected in the form of: 1. records of the products purchased and sold (records

of the weight of mineral fertilizers, nutritive feed and bulky feed (succulent and dry) purchased by the farm, and of the plant and animal products sold), 2. records related to the fields in use (information on the surface area of the fields, types of crops, doses of mineral and natural fertilizers applied, the amo-unts of crops and aftercrops harvested),

MAGDALENA SZYMAÑSKA*_{, EWA SZARA, MARIAN KORC, JAN £ABÊTOWICZ}

Warsaw University of Life Sciences – SGGW, Department of Soil Environment Sciences, Division of Agricultural Chemistry, ul. Nowoursynowska 159, 02-776 Warszawa

Model of nitrogen management on farms specializing

in livestock production

Abstract: The aim of this study was to determine the relationship between the efficiency of agricultural production and selected parameters of farms and data describing the flow of nutrients on the farms. An analysis model was developed for nitrogen manage-ment on farms specializing in livestock production. The study was conducted on 20 farms located in 12 municipalities of the Mazo-wieckie province in central Poland in 2009–2012. The model was developed using multiple linear regression analysis in accordance with the backward stepwise method. Based on the regression analysis, the farm parameters that did not determine the dependent variable were eliminated. In the end, there were 10 independent variables included in the model. The model indicates that the efficiency of nitrogen management, expressed in cereal units per 1 kg of nitrogen (CU·kg–1 _{N), is significantly affected by: crop} rotation, the demand for purchasing animal feed, the intensity of livestock production, the nitrogen content of farmyard manure, the nitrogen doses applied in mineral and natural fertilizers, and nitrogen outflow from farms with the sale of plant and animal products. The developed model explains 70% of the variation in the coefficient of efficiency expressed in cereal units per 1 kg N.

61 Model of nitrogen management on farms specializing in livestock production

3. animal feeding records (data on the type and qu-antity of feed given to the livestock on a monthly basis, both summer and winter feeding),

4. records of livestock trading (inventory data such as the age of the animals, purchase and sale dates, weight of the animals).

The nitrogen management model presented here was developed on the basis of the data recorded by the farmers and the results of analyses of samples of soil, manure, crop plants and bulky feed, which were collected directly on the analyzed farms in 2009–2011. The collected soil samples were analyzed for: the amounts of available forms of phosphorus and potassium by Egner-Riehm method, magnesium by Schachtschabel’s method, and soil pH in KCl at

1 mol·dm–3_{, by potentiometry.}

The collected samples of manure, bulky feed pro-duced on the farm, purchased feed and the harvested crops were analyzed for: dry matter content by oven-drying and weighing, total nitrogen content by Kjel-dahl method.

On the basis of the data obtained from the farmers and the results of laboratory analyses nitrogen balan-ces were calculated as ‘field surface balance’ and ‘farm-gate balance’, according to the methodology adopted by OECD (Organisation for Economic Co-operation and Development); also calculated was the efficiency of agricultural production on the individu-al farms expressed in cereindividu-al units (CU) per 1 kg of

nitrogen applied on the farm (CU·kg–1 _N).

In order to determine the characteristics of the in-dividual parameters, the basic descriptive statistics were used: mean, standard deviation, median, mini-mum-maximum, and the correlation matrix. Frequ-ency distributions of a particular feature were shown in histograms.

The model was developed using multiple linear regression analysis carried out to the point at which the smallest significant number of variables was ob-tained (for p ³0.05) that explained to the greatest

extent the variable y expressed in CU·kg–1 _{N in}

ac-cordance with the backward stepwise method, using for that purpose Statistica 5.1. In order to explain how much of the total variability in the dependent varia-ble was explained by the nitrogen management mo-del described here, the corrected coefficient of

deter-mination (R2_{) for polynomial regression was used.}

RESULTS AND DISCUSSION

The population of farms included in the study va-ried in terms of the system of managing fertilizer com-ponents. The differences in the management resulted in the differences in the pH of the soils and the levels

of available nutrients in them. On the other hand, high soil fertility determines the extent of utilization of fertilizer components by crop plants (Janssen, 2006). For this reason, knowledge of the chemical proper-ties of soils is a very important element in the deve-lopment of a nutrient management system. The pro-perties of the soils concerned have been described in detail in previous articles (Szymañska et al., 2011a, 2011b). Most of the analyzed soils were highly aci-dic or aciaci-dic, which is characteristic of Polish agri-culture. The most soils of this type were found on farms specializing in dairy cattle production, and the least on organic farms. In terms of soil fertility with respect to the available forms of phosphorus, potas-sium and magnepotas-sium, the most soils with low and very low levels of these components were found on organic farms. By contrast, the dominant soils of po-ultry farms were very rich in these elements (Szy-mañska et al., 2011a, 2011b). The type of animal pro-duction determined the structure of the crops grown on the farm. Crop structure affects the balance of nutrients, mainly nitrogen (Barszczewski et al., 2011), and that is why it was subjected to a detailed analysis (Table 1). In general, the crop structure of the analy-zed farms was dominated by cereals (about 57%); the second largest were grasslands (about 23%), which were mainly found on cattle farms. The smallest sha-re in the crop structusha-re was that of root crops (about 6%) and legumes (an average of less than 5%). Legu-minous crops were found primarily on organic farms. The different types of farms differed in the

inten-sity of livestock production (INSan), which is a

me-asure of livestock density expressed as livestock units

per hectare of arable land (LU·ha–1_{). According to}

the principles of Good Agricultural Practice, stocking

density should not exceed 2 LU·ha–1_{. If that is the}

case, the farm is able to produce the right amount of animal feed and utilize the resulting manure within the farm. The average livestock density in the

analy-zed population was 1.45 LU·ha–1_{. It should be noted,}

however, that there was a poultry farm supporting

more than 19 LU·ha–1 _{(Table 1). The intensity of}

li-vestock production was positively correlated with the amount of nitrogen introduced into the soil with na-tural fertilizers (NatN) (r = 0.53*), and with the

amo-unt of nitrogen entering the farm in the purchased

animal feed (PAFN) (r = 0.7*). This indicates that on

those farms there was a shortage of their own feed resulting from the mismatch between the size of the livestock population and the croplands owned, or the crop structure did not match the feeding needs of the animals (Table 2), which is further confirmed by a high positive correlation coefficient (r = 0.69* )

TABLE 1. Descriptive statistics of selected parameters of the analyzed farms Explanations: P

, K – amount of available forms of phosphorus, potassium; C

str

– share of cereals in crop structure (%); R

str

– share of root crops in crop structure (%); L

str

– share of legumes in crop structure (%);

Gstr

– share of grasslands in crop structure (%); INS

an

– intensity of animal production (LU·ha

–1); M

N

– nitrogen content in manure; Bfs

N

– field surface nitrogen balance (kg N·ha

–1); Min

N

– amount of nitrogen

applied in mineral fertilizers (kg N·ha

–1); Nat

N

– amount of nitrogen applied in natural fertilizers (kg N·ha

–1); P

A

FN

– amount of nitrogen in purchased animal feed (kg N·ha

–1); SAP

N

– amount of nitrogen in sold

animal products (kg N·ha

–1); SPP

N

– amount of nitrogen in sold plant products (kg N·ha

–1); Bfg

N

– farm-gate nitrogen balance (kg N·ha

–1). evit pir cse D scit sit ats scit sir etc ara hc lio s mra FC rts Rrts Lrts Grts S NI na ni M N ta N N gk· U C 1–NF AP N P AS N PP S N sf B N gf B N Hp l C K P gk·gm 1– K gk·gm 1– g M gk·gm 1– na e M3 7.48 2.1 85 5.3 72 1.3 64 2.7 54 8.55 8.49 8.7 25 4.14 6.7 38 8.8 43 6.03 5.3 014 0.3 23 6.0 18 3.8 10 5.7 01 nai de M0 6.40 0.6 76 9.6 61 0.3 50 0.2 60 1.30 0.04 .7 23 1.10 0.6 30 6.4 42 4.05 6.6 77 7.77 7.65 9.7 13 1.9 8 noit aiv ed dra dn at S2 7.05 1.9 36 3.0 41 5.6 46 8.2 26 5.64 9.0 14 1.0 28 3.26 2.5 32 2.7 22 5.08 2.6 82 1.3 51 1.5 14 8.6 30 9.2 01 mu mini M7 4.35 7.4 16 2.73 5.87 1000 93. 00 20. 411 1.05 9.1 12 .10 38. 17-3 .2 81-mu mix a M4 1.74 .8 911 1.4 922 6.6 630 012 25 56 .8 75 3.9 16 217 2.9 712 4.37 3.2 053 6.5 525 4.8 217 8.1 84 9.0 43

plant products (SPPN). Based on the data analysis,

it can also be stated that the greater the share of ce-reals in the crop structure (Cstr) was, the more

nitro-gen was imported with the purchased animal feed

(PAF_N) (r = 0.59*), which concerned mainly pig and

poultry farms.

It was found that large amounts of nitrogen ente-red the farms in the purchased feedstuffs – an avera-ge of 103.5 kg N per ha of arable land (Table 1). For comparison, the amount of nitrogen coming in with

mineral fertilizers was about 38 kg N·ha–1_{. Both the}

inflow of nitrogen with the purchased feed and with the mineral fertilizers was characterized by a high coefficient of variation in the analyzed population of farms and was about 83 and 94%, respectively. This spread of results was due to the fact that in the analy-zed group of farms there were farms with livestock production at different levels of intensity, from the lowest – organic farms, to very high – poultry farms. It should be noted that the amount of nitrogen ap-plied in mineral fertilizers was not large (up to 126

kg N·ha–1_{). On the one hand, that was due to the fact}

that on those farms natural fertilizers were applied, although their doses were not large either, averaging

about 49 kg N·ha–1_{; on the other hand, that was a}

result of high prices of mineral fertilizers coupled with low prices of agricultural produce. Among the analy-zed farms there were also cases where the doses of natural fertilizers were higher than the permissible

limit of 170 kg N·ha–1_{. At the same time, the outflow}

of nitrogen from the farms was relatively small. In plant products, on average for the 20 farms, it was

10.6 kg N·ha–1_{, and in the sold animal products}

aro-und 23 kg N·ha–1 _{(Table 1). This is evidence of a low}

efficiency of utilization of the applied nitrogen – only about 32%.

Detailed analysis of the ratio of nitrogen inflows to outflows was made using the balance sheet me-thod involving calculations of field surface balance

(BfsN) and farm-gate balance (BfgN) (Table 1, Fig.

1and 2). Balance sheet balances are indicators of the quality of the nutrient management system, as well as being important agri-environmental indicators (Watson et al., 2003). The two balances, on average for the entire population of farms, were positive, about 18.4 kg N·ha–1 _{and 107.5 kg N·ha}–1_{, respectively. The}

average field surface N balance was smaller than the average balance for Poland, which is about +48.3 kg

N·ha–1_{. At the same time, it was about 50% lower}

than the balance in the Mazowieckie province, whe-re the analyzed farms wewhe-re located (Kopiñski, 2007). The distribution of field surface N balance (Fig. 1) shows that the largest number of fields falls within

ni-63 Model of nitrogen management on farms specializing in livestock production

trogen management, and 60–80 kg N·ha–1_{. The large}

number of data with a large N balance surplus (>60 kg N·ha–1_{) is an undesirable occurrence, especially as the}

predominant soils of the analyzed farms were light soils susceptible to the leaching of nutrients. This is parti-cularly important in the case of the nitrogen responsi-ble for the eutrophication of surface waters, and besi-des, represents a financial loss for the farmer.

The surplus in the case of the farm-gate N lance was much larger. The distribution of the ba-lances from that balance sheet (Fig. 2) shows that the largest number of objects fell within three

ran-ges: 50–100, 150–200 and 0–50 kg N·ha–1_{. A}

relati-vely large number of objects was also in the range of

250–300 kg N·ha–1_{; those were farms that}

speciali-zed in poultry production. The obtained amounts of the balance surpluses indicate significant retention of nitrogen within the farm and may also indicate si-gnificant losses of this element through the leaching of nitrates and emissions of ammonia.

The above characteristics of the parameters of the analyzed farms were used to develop a model of ni-trogen management on farms that raise livestock at different levels of production intensity. Because of FIGURE 1. The distribution of field surface N balance (BfsN in kgN·ha–1)

FIGURE 2. The distribution of farm-gate N balance (Bfg_N in kgN·ha–1₎

)LHOGVXUIDFH1EDODQFH                   @ @ @  @ @ @ @ @ ! )DUPJDWH1EDODQFH                    @  @@@@@@@@@@! Number of fields Number of observations

MAGDALENA

SZYMAÑSKA, EW

A

SZARA, MARIAN KORC, JAN £ABÊT

OWICZ H p KCl P g k · g m –1 K_mg·kg–1 Csrt Rsrt Lsrt Gsrt INSan MN BsfN MinN NatN PAFN SAPN SPPN BfgN H p _KCl 1 g k · g m P –1 _0.18* g k · g m K –1 _{0.27* 0.49*} Csrt -0.24* 0.31* 0.13 Rsrt 0.15* -0.06 0.19* -0.48* Lsrt 0.22* -0.18* -0.01 -0.10 -0.14 Gsrt -0.01 -0.20* -0.30* -0.70* 0.07 -0.36* S N I _an -0.01 0.16* 0.14 0.23* -0.16* -0.06 -0.24* N M -0.12 0.03 -0.02 0.11 -0.11 -0.08 -0.02 0.15* s s B N -0.33* 0.04 -0.13 0.40* -0.36* -0.13 -0.15* 0.19* 0.22* n i M _N -0.11 0.24* 0.15* 0.45* -0.17* -0.09 -0.22* 0.04 0.08 0.52* t a N _N -0.12 0.03 -0.14 0.30* -0.42* 0.03 -0.10 0.53* 0.33* 0.39* 0.10 F A P _N 0.06 0.37* 0.24* 0.59* -0.33* -0.10 -0.46* 0.70* 0.15* 0.29* 0.16* 0.53* P A S N -0.05 0.07 -0.03 0.37* -0.19* -0.12 -0.25* 0.15* -0.07 0.03 0.01 -0.04 0.14 P P S N 0.01 0.24* 0.16* 0.21* 0.04 -0.07 -0.22* 0.69* 0.02 -0.13 -0.15* 0.30* 0.52* 0.26* g f B N 0.04 0.32* 0.24* 0.42 -0.24* -0.04 -0.30* 0.42* 0.19* 0.42* 0.49* 0.45* 0.74* -0.44* 0.1 1

Abbreviations: see Table 1.

s el b ai r a V Intercept R_srt L_srt G_srt INS_an M_N Min_N Nat_N PAF_N SAP_N SPP_N r o rr e d r a d n a t S 0.0671 0.0024 0.0015 0.0009 0.0091 0.0117 0.0004 0.0006 0.0003 0.0003 0.0012

TABLE 3. Standard error of estimate for the direc-tional coefficient (slope) of each independent va-riable in the model

65 Model of nitrogen management on farms specializing in livestock production

the unfavourable relationship between the prices of the means of production and the prices of crops, as an indicator of the efficiency of nitrogen management it was decided to use the efficiency of production per kg of the component applied in the fertilizers and animal feed, i.e. the amount of plant and animal pro-ducts produced, expressed in cereal units per kg of

the component used (CU·kg–1 _{N). Similar indicators}

had been adopted in the development of nutrient ma-nagement efficiency on dairy farms (Schröder et al., 2005). This indicator was adopted as the dependent variable y in our nitrogen management model. The

average value of this indicator was 0.63 CU·kg–1 _N

(Table 1). The coefficient of variation of this trait was high, reaching 82%, which was due to the differen-ces in the intensity of production on the different ty-pes of farms.

Based on the multiple regression analysis perfor-med stepwise, those parameters that did not determi-ne the dependent variable were eliminated. In the end, there were 10 independent variables in the model (Equation 1).

Equation 1. Model of nitrogen management on

farms specializing in livestock production, with

R

_{= 0.7}

y = 1.0142– 0.0186 Rstr–0.0168 Lstr–0.0076 Gstr

+0.2251 INS_an–0.0407 M_N–0.0011 Min_N

+0.0053 NatN–0.0036 PAFN–0.0013 SAPN

–0.0081 SPPN

where:

Dependent variable:

y – efficiency of agricultural production in cereal units

per 1 kg of nitrogen applied on the farm (CU·kg–1 _N);

Independent variables:

Rstr – share of root crops in crop structure (%),

Lstr – share of legumes in crop structure (%),

Gstr – share of grasslands in crop structure (%),

INSan– intensity of animal production (LU·ha–1),

MN – nitrogen content in manure,

MinN – amount of nitrogen applied in mineral

ferti-lizers (kg N·ha–1_),

NatN – amount of nitrogen applied in natural

fertili-zers (kg N·ha–1_),

PAFN – amount of nitrogen in purchased animal feed

(kg N·ha–1_),

SAPN– amount of nitrogen in sold animal products

(kg N·ha–1_),

SPPN – amount of nitrogen in sold plant products

(kg N·ha–1_).

The above model indicates that among the diffe-rent parameters of the farm a decisive role in deter-mining the efficiency of nitrogen management in terms of cereal units obtained from 1 kg of applied nitrogen is played by: crop rotation and the crop struc-ture associated with it; the amount of nitrogen bro-ught to the farm with the purchased animal feed; the intensity of livestock production and digestibility of the feed indirectly associated with it (nitrogen con-tent in manure); nitrogen dose levels in natural and mineral fertilizers; and also nitrogen outflow from the farm in the plant and animal products sold. The presence in the model of the crop structure on the farm results from the different nutritional require-ments of the different groups of crop plants, which is reflected in different rates of nitrogen removal from the soil. The above model explains 70% of the variation

in the efficiency coefficient expressed in CU·kg–1 _N

(R2 _{= 0.7*). Another measure of how well the model}

fits the empirical data is the standard error of estima-te Se, which is 0.29. This means that the values of the

variable y in CU·kg–1 _{N, calculated on the basis of}

the above equation, differ from the empirical values by about ± 0.29. Standard errors of estimation of the partial regression coefficients of each independent variable in the model are shown in Table 3. The lar-gest error of estimate (about 35%) is for the nitrogen introduced with mineral fertilizers, while the smal-lest (4%) is for the intensity of livestock production.

CONCLUSIONS

1. The intensity of livestock production determines the efficiency of nitrogen management on the farm. 2. Among the four types of farms specializing in or-ganic production and the production of dairy cat-tle, pigs and poultry, the largest balance surpluses, and thus the greatest potential losses of nitrogen, occurred on the farms that specialized in poultry production.

3. Farms with intensive animal production are cha-racterized by a very large inflow of nitrogen in the form of purchased animal feed relative to the amo-unt of nitrogen contained in the animal products sold. The difference is the amount of nitrogen re-tained in the livestock and the loss of nitrogen thro-ugh leaching and gaseous emissions.

4. On the basis of the developed model it was shown that a decisive role in the determination of the ef-ficiency of nitrogen management expressed in CU·kg–1 _{N is played by: crop rotation; the}

intensi-ty of animal production; the nitrogen content in manure; the amount of nitrogen in the purchased feed; the size of nitrogen doses in natural and

mi-neral fertilizers; and also nitrogen outflow from the farm in the plant and animal products sold. 5. The model explains 70% of the variation in the

effi-ciency coefficient expressed in CUs per 1 kg N.

ACKNOWLEDGMENTS

The study was granted by the National Science Centre of Project No. N N310 089136.

LITERATURE

Barszczewski J., Jankowska-Huflejt H., Prokopowicz J., 2006. Bilanse azotu, fosforu i potasu w gospodarstwach ekologicz-nych o du¿ym udziale ³¹k i pastwisk. Woda Œrodowisko Ob-szary Wiejskie 6, 16: 35–46.

Janssen B.H., Willigen P., 2006. Ideal and saturated soil fertility as bench marks in nutrient management. 1. Outline of the fra-mework. Agriculture, Ecosystems and Environment 116: 132– 146.

Kopiñski J., 2007. Bilans azotu brutto dla Polski i województw w latach 2002-2005. Studia i Raporty IUNG-PIB. 5: 117– 131.

Parris K., 1998: Agricultural nutrient balances as agri-environ-mental indicators: an OECD perspective. Environagri-environ-mental Pol-lution 102, S1: 219–225.

Schröder J.J., Vertes F., Chadwick D.R., Sillebak Kristensen I., 2005. How to compare the nutrient use efficiency of dairy farms? Soil Use and Management 21: 196–204.

Szymañska M., £abêtowicz J., Czopowicz A., 2011a. Zawartoœæ dostêpnych form fosforu w glebie w zale¿noœci od kierunku produkcji gospodarstwa. Zesz. Probl. Post Nauk. Roln. 565: 323–330.

Szymañska M., £abêtowicz J., Czopowicz A., 2011b. Zawartoœæ dostêpnych form potasu w glebie w zale¿noœci od kierunku produkcji gospodarstwa. Zesz. Probl Post Nauk. Roln. 565: 331–338.

Watson C.A, Atkins T., Bento S., Edwards A.C., Edwards S.A., 2003. Appropriateness of nutrient budgets for environmental risk assessment: a case study of outdoor pig production. Eu-rop. J. Agronomy 20: 117–126.

Received: July 9, 2013 Accepted: August 29, 2013