Tài liệu A simulation model of the microbiological and chemical changes accompanying the initial stage of aerobic deterioration of silage

Crass and Forage Science (1990). Volume 45, 1 5 3 - 165 A simulation model of the microbiological and chemical changes accompanying the initial stage of aerobic deterioration of silage the phase between opening of the silo and feeding the silage to the animal. Silage exposed to air after a period of anaerobic storage shows large differences in susceptibility to aerobic spoilage. Aerobic deterioration is a microbial process carried out by aerobic microorganisms that cannot proliferate in the anaerobic environment of a sealed silo (Honig and Woolford, 1980). The growth of these organisms commonly results in a rise in pH and temperature, and the disappearance of fermentation acids. The losses in dry matter (DM), and hence nutritional value that accompany aerobic deterioration, can be up to 3O<7o (Honig, 1975; Woolford et at., 1978). Previous work showed that the growth of yeast often coincides with the heating of silages (Weise, 1963; Beck and Gross, 1964; Daniel et at., 1970; Ohyama and McDonald, 1975; Moon and Ely, 1979; Pahlow, 1982; Jonsson and Pahlow, 1984; Middelhoven and Franzen, 1986). However, there is evidence that aerobic deterioration can be initiated by bacteria (Woolford and Cook, 1978; Woolford et at.. 1978; Barry et ai. 1980; Crawshaw et ai, 1980). Spoelstra et at. (1988) found that acetic acid bacteria could initiate heating in whole-crop maize silage. The factors that determine which organisms will proliferate in a silage upon exposure to air are not yet fully understood. M. G. COURTIN* AND S. F. SPOELSTRA Insiiiitie for Livestock Feeding and Nutrition Research, Lelystad. The Netherlands Abstract A mathematical model is presented that predicts the time-course of aerobic deterioration in grass and whole-crop maize silages. The model predicts the stability of the silage taking into account the buffering capacity of the silage, the initial contents of organic acids and ethanot, pH, the initial temperature and the initial populations of the microorganisms. The specific processes simulated include the growth of yeast and acetic acid bacteria, the oxidation of fermentation products, the consumption of oxygen and the production of carbon dioxide, the rise in temperature, and the increase in pH. The deterioration of silage is seen to be initiated by acetic acid bacteria or by yeast, or by both groups together. The factors that determine which groups will prevail are the dry matter contents and the chemical composition of the silage. The output of the model is validated by comparison of the simulated data with data from published work on the deterioration of silage. Introduction The objective of this study was to develop a predictive simulation mode! of the basic processes that occur during aerobic deterioration considering the competition between yeasts and acetic acid bacteria, and the chemical composition of the silage. Emphasis was placed on modelling the initial stages of deterioration in an effort to predict the stability of silage in air. The model simulates the growth of yeasts and acetic acid bacteria, the oxidation of fermentation acids, the production and consumption of gases, the generation of heat through microbial Neal and Thornley (1983) provided a qualitative model of the fermentation of silage. This work was followed by quantitative models from Pitt et at. (1985), Leibensperger and Pitt (1987) and Meiering et at. (1988). As yet, no attempt has been made to extend the simulation to include Correspondence: Dr S. F. Spoelstra. Institute Tor Livestock Feeding and NutrJiion Research (iVVO), PO Box 160. 8200 AD Lelystad, The Netherlands, •Preseni address: PO Box 1214, Blind River. Oniario, Canada POR IBO, 153 154 M. G. Courtin and S. F. Spoelstra activity; and the change in pH for both maize and grass silages. Generai approach It was hypothesized that the growth of yeast and acetic acid bacteria are the principal processes occurring during the onset of aerobic deterioration and that the DM content and the chemical composition of the silage determine growth rates. The mathematical approach was an unsegregated model of microbial growth that treats the culture mass as the fundamental variable and ignores the presence of different strains and individual cells. The microorganisms compete for the available substrates, namely lactic acid, acetic acid and ethanol. The oxidation of the organic acids decreases the buffering capacity of the silage causing a rise in pH, production of carbon dioxide and the release of heat. These interactions were represented by a set of differential equations that were solved by numerical integration to predict the time-courses of the component variables of the system. The model was designed to operate on both grass and whole-crop maize silages. By way of simplification the actions of bacilli and moulds normally associated with the later stages of aerobic deterioration (Barry et ai., 1980; Lindgren el at.. 1985). were disregarded. It was assumed that the residual sugars are not utilized for the growth of organisms responsible for the onset of aerobic deterioration. U was also assumed that the concentrations of oxygen and carbon dioxide do not limit growth. In this way the model simulates the results that can be expected from an aerobic stability test, or on the loosened face of the silage clamp, where convection of air is freely occurring (Rees, 1982), rather than the conditions that will exist inside a silage clamp. However, combined with a description of the flow of gases through a silo, the model will predict the aerobic stability at any point in the clamp. Description of the model Growlh of yeast The growth of yeasts was formulated in the model in terms of the rate of change of the mass of cells per unit mass of silage; dCy/dt=Ugy-,i,y)XCy (1) where CY = mass of yeast (g yeast (g silage)"'; ^gY = specific growth rate (g new yeast (g totalyeast) " ' X h " ' ) ; and ;idY = specific death rate (g dead yeast (g total yeast)" ' x h " ' ) . Calculation of growth rate. The growth rate of yeast (n^y) in silage exposed to air was formulated in a similar way to that used by Meiering et at. (1988) for silage fermentation. Here the limiting substrate in the growth of yeast was considered to be lactic acid (compare Mankad and Bungay, 1988); I max I /*gY 'x '^ C, "-t (2) + Cl where /^l^y^'= maximum growth rate (h '); C| = concentration of lactic acid (g acid (g silage)"'); K5y| = Michaelis-Menten saturation constant (g lactic acid (g silage)"'). Maximum growth rate. The maximum growth rate was calculated by reducing a reference maximum growth rate with an Arrhenius temperature function, as in Meiering et al. (1988), and by factors representing the inhibition of growth due to the concentration of organic acids. Water activity (a^,) was considered not to influence the growth of yeast in silage. (3) where Ev = activation energy of yeast growth (kj (g °K)"'); R= universal gas constant (kJ (kmol " K ) " ' ) ; TOY = optimum growth temperature of yeast (°K); and /^y = inhibition factor of acetic acid on growth of yeast. Inhibition of yeasts by organic acids. An estimation of the degree of inhibition exerted by the concentration of lactic and acetic acid in silage was made from the growth of a yeast strain in liquid media. A strain of Gandida krusei, a yeast commonly occurring in silage exposed to air (Middelhoven and van Baalen, 1988; Lindgren et al.. 1985) was grown in media containing increasing amounts of lactic and acetic acids adjusted to pH 4-0, 4-5 and 5-0 A model of changes during aerobic deterioration of silage 155 (G. Pahlow, personal communication). The pH was removed as a variable by converting the total acid concentrations to concentrations of undissociaced acid at a particular pH, because the undissociated acid is the major source of inhibition (Eklund, 1983; Nodaetat., 1982). A non-linear regression of the resulting data yielded the following relationship; f = _ 1 J. e-''6WxCa^- for C|^,<0-0O819x((0-O5-C|^,)/0-05) (14) where C^^ = concentration of undissociated acetic acid (g acid (g silage)''); and C|u = concentration of undissociated lactic acid (g acid (g silage)" '). Growth of acetic acid bacteria Spoelstra et al. (1988) showed that acetic acid bacteria are capable of initiating aerobic deterioration in whole-crop maize silage. Such an active involvement has not yet been shown for grass silage although Acetobacter spp. have also been isolated from grass silages (Spoelstra et ai, 1988). The activity of acetic acid bacteria was included in the model and is formulated in a similar fashion to the yeast; dC^/dt = (5) where C^ = mass of acetic bacteria (g bacteria (g silage) " '; MgA = specific growth rate of acetic acid bacteria (g new bacteria (g total bacteria)"'h"'); and ;idA = specific growth rate of bacteria (g dead bacteria (g total bacteria)"'h""'). Gatcutation of growth rate. Acetic acid bacteria are capable of oxidizing many organic compounds including lactic acid, acetic acid, butanediol ethanol and glucose (De Ley and Schell, 1958; 1959; Dupuy and Maugenet, 1963). The most important substrates with respect to the stability of silage are lactic and acetic acids, which contribute significantly to the buffering capacity. Work with acetic acid bacteria in silage (Spoelstra et al., 1988) and in other environments (Divies and Dupuy, 1969; Divies, 1972; Sh'imizu et al., 1977; Nanbae/a/., 1984; Jucker and Ettlinger, 1985) has shown that the ethanol concentration of the environment is important because the oxidation of acetic acid is repressed by ethanol. However, Lactic acid is oxidized under these conditions. When ethanol is consumed below the level that causes repression of the enzymes required for the oxidation of acetic acid, metabolization of acetic acid commences (Jucker and Ettlinger, 1985). The growth rate of the acetic acid bacteria (MgA) *3S modelled as the sum of the individual growth rates on the three most important substrates: acetic acid, ethanol and lactic acid (compare Kim et al., 1988); MgA. (6) where /ig^. i = growth rate of acetic acid bacteria on individual substrates ( h " ' ) with i = a for acetic acid, e for ethanol and 1 for lactic acid. The growth rates of acetic acid bacteria on ethanol and lactic acid were described by the following relationship with; MgA. (7) where ;i'g"!;"'i= maximum specific growth rate of acetic acid bacteria (g new bacteria (g total bacteria)"'h " ' ) ; K,,^| = saturation constant of growth rate (g substrate (g silage)"') with i = e for ethanol and 1 for lactic acid. The growth on acetic acid was formulated by the addition of an extra term to account for the repression of acetic acid oxidation by ethano!; (8) K. where /iij";^"a = maximum specific growth rate of acetic acid bacteria (g new bacteria (g total bacteria) • ' h " ' ) ; K,Aj = Michaelis-Menten saturation constant for acetic acid bacteria (g acetic acid (g silage}" '); K,= repression coefficient (g ethanol (g silage)"'). Maximum growth rate. The maximum growth rates of acetic acid bacteria were calculated as was done for the yeast with; I /*gA, (9) where ^^™"jo = relative maximum growth rate of acetic acid bacteria ( h ' ' ) with i = a for acetic 156 M. G. Courtin and S. F. Spoelstra acid, e for ethanol and 1 for lactic acid; EA = activation energy of acetic acid bacterial growth (kJ(kmol °K)"'); T^A = optimum growth temperature of acetic acid bacteria ("K); /a^ = inhibition factor of acetic acid on growth of acetic acid bacteria; Z,^ = inhibition factor of lactic acid on the growth of acetic bacteria; and /a*A = inhibition factor of reduced a^ on the growth of acetic acid bacteria. Inhibition of growth of acetic acid bacteria by organic acids. The effect of the level of organic acids on the growth of three different strains of Acetobacter spp. was studied in the process of formulating this model. Strains isolated from aerobically unstable maize silage were pregrown in media containing 3% ethanol and IVo yeast extract adjusted to pH 5. Suspensions of these cells were used to inoculate media containing 0-5% ethanol (preferential carbon source), 0 - 1 % yeast extract and concentrations of acetic acid ranging from 0 to 50 g 1"' and between 0 and 20 g 1" ' of lactic acid. Optical density readings were used to follow the growth of the suspension at 28 °C and hence the degree of inhibition caused by the two acids. The data were regressed to yield the following relationships. The inhibition factor, expressed in terms of undissociated acetic acid on the growth of Acetobacter sup., is (r- = 0-858): 27• I X C,^ -!M1-0 0-0 for 0-0035 (10) < C^, < 0•0406 C,,<0 •0035 C...>0•0406 sensitive to lowered a^, since they are commonly isolated from liquid environments such as beer, wine and dairy products (Abadie, 1982; Swings and De Ley. 1981; Aries et ai., 1982; Sponholz and Dittrich, 1985). L. brevis was the most sensitive of the three organisms used by Pitt et al. (1985) and had the following relationship (r= = 0-999); - 18^33 x l 9 - 5 9 x a ^ - 2 4 6 - 2 x (12) (a^-0^97)-, 0^94450^995 0-0 a.< Rates of change in substrate concentrations Consumption of ethanol. Acetic acid bacteria are capable of oxidizing ethanol to acetic acid according to; C2H3OH + O: — CHjCOOH + H^O + 492^6 kJmoP ' The consumption rate of ethanol Is modelled as a function of the mass of acetic bacteria; where Y^^ j = growth yield of acetic acid bacteria on ethanol (g bacteria (g ethanol)" '). Consumption of lactic acid. Both yeasts and acetic acid bacteria are capable of oxidizing lactic acid; CH3CHOHCOOH -I- 3 0 , — 3CO2 + 3H,0-f 1368-2 kJmoP ' The concentrations of lactic acid were left in terms of the total acid yielding the relationship (r2 = 0-745); r -42-8xC,+ M for 0-0023 1-0 0-0 (11) ) ( Inhibition of growth of acetic acid bacteria by decreased a^. In the absence of any experimental data on the dependence of the growth of acetic acid bacteria on water activity the relationship given by Pitt et at. (1985) derived from data of Lanigan (1963) for Lactobacitlus brevis was adopted. The choice was based on the assumption that the acetic acid bacteria are The consumption rate was formulated as; dC,/dt = - U,A/Y,A_ 1X C^ + ;i,y/Y,y., x Cy) (14) where = growth yield of acetic acid , bacteria from lactic acid (g bacteria (g lactic acid) * '); and YJY. I = growth yield of yeast from lactic acid (g yeast (g lactic acid)""'). Consumption and production of acetic acid. Acetic acid bacteria can produce acetic acid and also use it as a substrate. The equation for the production of acetic acid from ethanol is given above. The oxidation of acetic acid as; CHjCOCH + 2 0 : "* 2 0 0 ; + 2H,O + 875-l k J m o r ' A model of changes during aerobic deterioration of silage Yeast can utilize acetic acid for growth in silage (Woolford and Wilkie, 1984; Middeihoven and Franzen, 1986). Thus, the rate of change of acetic acid was expressed as; + ^« where Y^^ ^ = growth yield of acetic acid bacteria on acetic acid (g bacteria (g acetic a c i d ) ' ' ) ; and 7^^ = conversion coefficient of ethanol to acetic acid. Change in pH The pH of a silage changes during aerobic deterioration as the result of the consumption of lactic and acetic acid. As these acids disappear the buffering capacity of the silage is reduced (Greenhill, 1964c). Change in pH affects the degree of dissociation of the acids. The rate of change of pH is formulated as proposed by Pitt el at. (1985); - 1 dC,,/dt -(- dC,,/dt d(pH)/dt = (16) where C,j = concentration of dissociated lactic acid (g acid (g silage)''); Cj(| = concentration of dissociated acetic acid (g acid {g silage)"'; Chd = concentration of dissociated butyric acid (g acid (g silage)"'; C^j = concentration of dissociated ammonia (g ammonia (g silage)''; Wj = molecular mass of acetic acid = 6O'O5 g m o l " ' ; Wh = molecular weight of butyric acid = 88-10 g m o l " ' ; w, = molecular weight of lactic acid = 90-08 g m o l " ' ; w^ = molecular weight of ammonia = 17-03 g m o l ' ' ; fl = buffer inde.\ of silage (Equivalent of acid pH (g silage)' '). Dissociated and undissociated acids. The ratio of dissociated acid, Cj, to undissociated acid, C^, varies with the pH and was expressed as presented by Ektund (1983); 157 (18) C,/C = It follows that the change in the concentration of dissociated acids in silage was represented by the following equation; (19) .-)-1 where Cni = concentration of dissociated acids (g acid (g silage)"' with i = a Tor acetic acid, b for butyric acid, I for lactic acid and n for ammonia; pK, = log of the dissociation constant with pK, = 4-76, pK(, = 4-86. pKi = 3-86 and pK^ = 9-00. By the same argument the concentration of undissociated acid {C,J was; 1 xCx (20) Buffer index of silage. The change in the buffering capacity of silage can be arithmetically represented in terms of a buffer index. The buffer index varies with the concentration of the buffering material and buffer indices are additive for all components at a given pH. Greenhill (1964c) observed that the buffer index of a silage could be approximated by the sum of the buffer indices of the original herbage and the acids produced during fermentation, weighted in proportion of the mass fractions of each component. The effect of ammonia was added to the relationship to provide buffering as the pH of the silage increased. Thus; ( 1 - C — C - C - C ) X Oh (21) where 0| = buffer index of individual components (Equivalents of acid, pH g ' ') with i = a for acetic acid, b for butyric acid, 1 for lactic acid and n for ammonia; and Of, = buffer index of the original herbage (Equivalents of acid pH (g herbage)"'). The buffer index of the individual components can be calculated from the derivative of the titration function with respect to pH; (17) and hence the ratio between undissociated and total acid (C.) is; B:=~ X (22) (1 + 158 M. G. Courtin and S. F. Spoelstra where i = a, b, 1 and n for acetic acid, butyric acid, lactic acid and ammonia. Relationships for (3^ are given by Pitt et al. (1985) for various crops. Those for ryegrass and corn were used after conversion to a fresh matter basis. Initial ammonia content of silage. The ammonia content of silage can be approximated from the following relationship as given by Leibensperger and Pitt (1987) with r- = 0-427; C,, = ( 4 0 4 x I O - ^ - 5 - 6 4 x 1 0 - ^ x d ) x d for 0 - 1 2 < d < 0 - 5 4 (23) where C^ = concentration of ammonia (g NH^ (g silage)" ') and d = dry matter content (g DM (g silage)''). This relationship was used for both grass and maize in the model. where Cco, = production of carbon dioxide (g CO; {g silage)''); W(^o., = molecular weight of carbon dioxide = 44-01 g m o l " ' ; Yce = molar ratio of CO, produced per mol of ethanol oxidized = 1 -0 mol CO; {mol ethanol)''; Yca = molar ratio of C02 produced per mole of acetic acid oxidized = 2-0 mol CO; (mol acid)" '; and Yt:i = molar ratio of CO; produced per mol of lactic acid oxidized = 3-0 mol CO2 (mol acid)" '. Production of energy Aerobic deterioration of silage is accompanied by heating of the surface exposed to the air. An estimation of the amount of heat released by the action of the microorganisms is possible using the heat of combustion of the individual substrates as follows; dE/dt = (dC,/dt X cf, , / w J + ( d C / d t x cf^, c/wj + (dC/dt X cf,. /W|) (26) Consumption of oxygen Oxygen is utilized in the oxidation of substrate by yeast and acetic bacteria In the reactions presented earlier. The total consumption of o.\ygen was calculated by converting the change in substrate concentrations to oxygen consumption; dCo,/di = ( d C / d t X wo,/w^ x Yo^) + {dQ/dt x w,,y w, X Y,,J + (dC/dt X wo^/w, X Yo|) (24) where C(,., = consumption of oxygen (g 0-. (g silage)"'); w^, = molecular weight of ethano! = 46-07 g(mol)"'; w,,, = molecular weight of oxygen = 32 00 g(mol)~ '; Yyj = molar ratio of oxygen to ethanol oxidized (mol oxygen (mol ethanol)''); Y[x. = mo!ar ratio of oxygen consumed to acetic acid oxidized (mol oxygen (mol acetic a c i d ) ' ' ) ; YQ = molar ratio of oxygen consumed to lactic acid oxidized (mol oxygen (mol lactic acid)"'). where E = heat production (kJ (g silage)"'), cf, = heat released during reaction (kj(mol)"') with i = a for acetic acid, e for ethanol and 1 for lactic acid. With this estimation of the energy released a formulation for the corresponding temperature rise was proposed; dT/dt = dE/dt (27) where c^, = specific heat of herbage (kJ(g °K)"'). The specific heat of herbage is a weighted average of the specific heat of water (CHIQ) and of herbage DM (c^); Ch = dxCd + ( l - d ) x c H 2 o (28) where CH,O = '*-19x 10"^ kJ(g ° K ) " ' ; Ca = l-89xlO-^kJ(g ° K ) " ' (McDonald, 1981) and d = D M content {g (g silage)''). Dry matter content and water activity Production of carbon dioxide Carbon dioxide is produced during the complete oxidation of ethanol, acetic acid and lactic acid as shown in earlier equation. As with the consumption of oxygen, the production of carbon dioxide was calculated from the consumption of substrates as follows: d Q ,,ydt = (dC,/dt X (25) The DM content of herbage is known to affect the course of silage fermentation by influencing the rate of fermentation and the numbers of bacteria found on the crop. The DM content also influences the stability of a silage upon exposure to air (Ohyama et al., 1981). Silage microorganisms live in the aqueous fraction of the silage. The water activity of the aqueous fraction affects the rate of bacterial development and depends largely on the moisture A model of changes during aerobic deterioration of silage 159 Table 1. Values oT parameters used in model Constant Value Uniis Reference fw 1-2 g acetic acid (g ethanol)'' (1) EA 65900 Jtnor' (2) Ey 67700 Jmol'' <3) MJA"™ 0-15 h-' M 'r« 0-22 h-' 0-08 h" 0-55 h-' 0-00001 g acetic acid (g silage)"' 0-0001 g ethanol (g silage)"' K,A.I 0-001 g lactic acid (g silage) ~' K,Y,I 0-005 g laciic acid (g silage)"' lA, t 0-000345 g ethanol (g silage)'' down to soluble compounds during fermentation the a^ decreases (Greenhill, 1964c). The compounds formed are predominantly lactic acid, acetic acid and ethanol: all products of the fermentation of simple sugars and proteins. The reduction in a^^ resulting from this conversion was estimated by calculating the freezing point depression induced by these compounds coming into solution. Fontan and Chirife (I98I) show that [he a^ of a solution can be approximated from the following equation, which is derived from the relationship between the freezing point of an aqueous solution and its a^. - I n a« = 10 (30) (4) V. 0-04 g bacteria (g acetic acid)"' Y.A.C 0-06 g bacteria (g ethanol)"' V I 0-06 g bacteria (g lactic acid)"' V. on g yeasts (g acetic acid)" ' V.i on g yeasts (g lactic acid)"' To A 301 "K (3) T.y 303 °K (6) where (^F = the freezing point depression (°C). Hence the change in a^ resulting from the products of silage fermentation was approximated as; -(9-694x10 (31) (I) Nomura et al.. 1988; (2) Wilson, 1986; <3) CornishBowden. 1976; (4) Divies, 1972; (5) Swing and De Ley, 1981; (6) McDonald, 1981. where i = a for acetic acid, e for ethanol and 1 for lactic acid. According to Ross (1975) the final a^^, becomes; conient of the forage. The initial a^ in silage depends on the crop species and the DM content of the forage. Greenhill (1964b) obtained the following expression for initial a^ (a^.^,); X a^i x = l-bxd/(!-d) (29) where b is a constant dependent on the crop. 0-03-0 05 for lucerne & white clover. 0-02-0-04 for ryegrass. In lhe model a value of b = 0-03 was assumed for grasses and b = 0-02 was used for wholecrop maize. The lower value for maize was taken to include the effect of the heterogeneity of maize. Whole-crop maize is mixture of low DM stover and high DM ears. Deinum and Knoppers (1979) measured the variation in the DM of these two constituents of maize and found a range of 17-I-21-4«/o DM for the stover and 37-3-53•7'Vo for the ears over a growing season. Microorganisms proliferate in the aqueous phase; hence, the availability of water would effectively be higher than indicated by the average DM content. As the biopolymers in the crop are broken (32) The depression of the freezing point resulting from the addition of lactic acid, acetic acid, and ethanol to an aqueous solution varies with the concentration of the solute. Data from Weasi (1987) were regressed to yield relationships between the freezing point depression and solute concentration between 0 and 50 g I" '; «Fa = 0• 3142 X (100 X CJ -1- 0• OOi 14, r^ = 0• 999 (33) 0Fi = O-1901 X (100xC,)-I-0-00087, r^ = 0-999 (34) (35) Numerical solution of model The above equations were programmed in FORTRAN on a VAX computer system. The differential equations were numerically solved using Euler's method. The values of the parameters used in running the simulations are presented In Table 1. The values were literature data. The values for lhe parameters describing growth were obtained in a similar manner 160 M. G. Courtin and S. F. Spoelstra 9 - I •o Z 6 5 s o o 3 - 24 48 72 96 120 144 168 192 216 Aerobic exposure (h) Figure 1. Experimental data for microbial counts and pH from Spoelstra el al., (1988 experiment 2) and the time courses predicted by the model, ( i ) acetic acid bacteria; ( 0 ) yeast; (D) pH; (—) simulated. 0.02 O.Ol 24 48 72 96 120 144 168 192 216 Aerobic exposure (h Figure 2, Experimental daCa for organic acids and ethanol from Spoelstra et al., (1988, experimem 2) and the time courses predicted by ihe model. {•) lactic acid; (•) aceiJc acid; (A) ethanol; (—) simulated. A model of changes during aerobic deterioration of silage 161 Table 2. Sources of validation studies and comparison of experimental and simulated results Composition of silage before exposure to air Source Middeihoven and van Baalen (1988) Pahlow (1979) non-urea treated Spoelstra el al., (1985) Experiment 5, anaerobic Spoelstra et al.. (1988) Experiment 1, uninocj Experiment 2. inoculated Crawshaw et al.. (1980) Control Ohyama et al.. (1981) Experiment 2, DL Experiment 4. DL Pahiow (1982) K, anaerobic B. anaerobic Woolford et al.. (1979) Control, direct cut' Control, wilted Crop DM '( g k g - ' ) 1 pH Stabiiityt Yeast A A B HL HAC (: g k g - ' ) ET (LU g " ) Observed 1Predicted (h) maize 309 3'05 10'5 3'0 0-5 4-0 2-0' 65-90 105 maize 300 3-67 21'2 8'7 2 6 6-0 2'0* 60-72 140 grass 300 3-70 22-7 5 9 16 3-0 3-0' 24-48 75 maize maize 309 4 00 3 85 16-0 12'4 9-2 9-6 5-5 6-3 <2 3 5 <1 255 3-4 144- 168 129-142 149 148 grass 154 4-45 83 4-8 10 4-6 < I' n.d. 85 grass grass 518 339 4-62 4-49 16 5 24-9 3 9 4'2 0'5 10 6-1 1 1 < 1' < 1• 24-48Tt ! 2 0 - 144 45 128 grass grass 410 410 4-50 4-00 12 3 26-7 51 3-5 3-7 4'5 1-8 < 1 < 1' < 1" 72-168 >I68 220 grass grass 175 495 4'30 5-50 9'6 15 8 4-0 4-2 2-4 0-5 28 < 1' 4-0 <1- 216 168 142 94 94 HL, lactic acid; HAC. acetic acid; ET, ethanol; AAB, acetic acid bacteria; t, duration of aerobic exposure preceeding 0-5 unit rise in pH. Where a range is given for the observed stability data was insufficient to pinpoint the begin of the pH rise. t. used in model development; * , contained O-S^a DM butyric acid; £, estimates, not actual counts; ft, estimated based on temperature; n.d.—not determined. to Meiering et al. (1988) and adjusted by simulating experimental data. Model validation and results Validation experiments The validity of the model was investigated by comparing the output with the results of published and unpublished work using the experimental data as inputs to the model. Experiments were chosen in which the concentrations of lactic acid, acetic acid, ethanol, pH, temperature and the number of yeast are reported at intervals throughout the period of aerobic exposure. Where some ofthe data were not measured an estimate was made. The selected experiments included both grass and maize silages at a wide range of DM contents. The sources of these experiments are listed in Table 2. Model results For each set of experimental data a simulation was run that predicted the time courses of yeast and acetic acid bacteria growth, and the changes in concentration of acetic acid, lactic acid and ethanol. The change in pH caused by the changes in acid concentration was also predicted. A sample comparison between experimental data and the output of the simulation model is given in Figures I and 2. Figure I shows the growth of yeasts and acetic acid bacteria as taken from Spoelstra et al. (1988) Experiment 2. In this particular case, growth of both yeasts and acetic acid bacteria was observed. The growth predicted by the model corresponds closely to the observed counts, although the model did not predict the apparent lag in the growth of the acetic acid bacteria. Also presented in Figure 1 is the change in pH of the silage during exposure to air. The model predicts the pH rise to occur slightly after that observed in the experiment. The changes in chemical composition of the above silage upon exposure to air are shown in Figure 2. The levels of ethanol and lactic acid were observed to drop as deterioration progressed. The level of acetic acid was observed to increase above its initial level before quickly being reduced to zero. These trends were also reflected by the simulation based on the initial concentration measured in the experimental silage. 162 M. G. Couriin and S. F. Spoelstra Table 3, Comparison of simulated and experimental results for oxygen consumption and carbon dioxide production Crop DM I kg-') Pahlow (1982)* 410 Woolford et al.. (I979)tT 175 495 Crawshaw et al.. {1980)tt 154 Oj consumed (gkg D M - ' ) Exp.t Pred.t COj produced (gkg D M - ' ) Exp.t Pred.t 70 83 _l 105 — 130 68 105 143 166 93 161 174 170 t Experimental values; J Predicted values; ^ not determined; * 7 d after start of exposure to air; t t 9 d after start of exposure to air; J l 6 d after sian of exposure to air. Stability of silage. A definition of the stability of silage was made for the purpose of comparing the experimental and simulated results. It was considered that a silage was stable until a 0-5 unit increase in the pH was observed. The experimental and simulated results were compared on the basis of the duration of the aerobic exposure preceding the rise in pH by 0-5 units, as shown in Table 2. Inhibition of yeasts by organic acids. Equation 4 resulting from the regression of experimental data collected on the growth of yeast in liquid media was found to provide too stringent inhibition of yeasts in silage. In order to have the yeast flourish in the simulations to the same degree as in the experiments, the level of inhibition had to be reduced in the simulations. The simulated results presented here were achieved by reducing the calculated concentration of undissociated acetic acid by a factor of two. Prediction of oxygen consumption and carbon dioxide production. Only one study could be found that reported results of the consumption of oxygen during deterioration and included enough biochemical data to run a reliable simulation. The experimental and simulated results are compared in Table 3. The production of carbon dioxide was compared against measurements taken in two studies as shown in Table 3. predicted from the release of energy from the oxidation of the substrates by the microorganisms was found to be higher than normally observed in silage. Discussion Organisms causing deterioration The stability of a silage as predicted by the model is largely dependent on the initial numbers of yeasts and the level of organic acids. The competition between the yeasts and the acetic bacteria depends on the DM of the silage and the concentration of the fermentation acids. The model shows that in maize, deterioration is caused by the growth of acetic acid bacteria when the initial yeast counts are low (2 logarithmic units (LU) g " ' ) . Both yeasts and acetic acid bacteria play a role when the silage contains a relatively low concentration of acetic acid ( < 6 g kg ~ ') and the initial population of yeasts is above 3 LU g" '. Silages with a large yeast population (eg. > 5 LU g " ' ) upon exposure to air, will be subject to deterioration by these organisms although the numbers of acetic acid bacteria will increase as well. Simulation of the deterioration of grass silage suggests that the dominant organisms are yeasts; however, the model predicts an active role for the acetic acid bacteria in lower DM silages. Buffering of silage Healing of silage. The heat production predicted by the model for the simulated silages ranged between 1-03 and 2-31 MJ (kg D M ) " ' for the maize silages and between 0-94 and 2-19 M J {kg DM)" ' for the grass silages over the first 7-9 d of aerobic exposure. The temperature The pH increase, corresponding to the consumption of organic acids, is well predicted by the model in the range of buffering of lactic, acetic and butyric acid, namely pH 3-8-5-0 (Figure 1). However, once .this range has been exceeded the simulated pH rises more quickly A model of changes during aerobic deterioration of silage and attains a higher final value than observed in experiments. The equations for buffering capacity used were estimated for silage in the pH range of 4 to 6 and are, therefore, probably not valid at higher pH values. It is likely that compounds other than ammonia, as used in the model, are active in buffering silage above pH 5. 163 acid {0-5°/ii DM) in some silages (see Table 3, Woolford et al., 1979) could explain the greater stability observed in these silages than was predicted by the model. The inhibition characteristics of butyric acid on silage microorganisms should be further investigated and eventually included in future attempts to model aerobic deterioration. Role of water-soluble carbohydrates The use of residual water-soluble carbohydrates (WSC) as a substrate for growth of microorganisms in aerobic deterioration was not included in the formulation of the model. The growth of acetic bacteria is not likely to be affected by the concentration of WSC since sugars are not a carbon source of preference (De Ley and Schell, 1958; De Ley et al., 1984; Ftuckiger and Ettlinger, 1977). Indeed there is evidence to suggest that the presence of ethanol, acetate and other metabolites actually inhibit the respiration of glucose by Acetobacter aceti (Huber-Frohli. 1982). In contrast the yeasts found in silage have been shown to grow on glucose as well as, or better, than on lactate (Middelhoven and Franzen, 1986). The level of WSC does generally decrease during aerobic deterioration, with higher losses being observed in silages that have higher initial WSC levels (Ohyama et al.. 1975; Woolford et a!., 1979). However, deterioration does not seem to be dependent on the WSC content since silages with very low levels are often very unstable, as can be seen from Table 2 (see Middelhoven and van Baalen, 1988), while silages with high residual WSC are often rather stable (see Woolford et al.. 1979). The role of residual sugars in the growth of organisms remains unclear emphasizing the need for more research into the substrates utilized by microorganisms during aerobic growth on silage. Role of butyric acid The butyric acid content of the silages was taken into account insofar as it affected the buffering capacity of the silage. The inhibition of microorganisms by the acid was not considered. Silages high in butyric acid are often very stable upon exposure to air (Ohyama et al., 1975). Butyric acid was shown by Gross and Beck (1970) to inhibit lactic acid assimilation by yeast at pH 4-0. The relatively high level of butyric Heating of silage The levels of energy released as predicted by the model agree with the values of 1 •53-2-64 MJ (kg D M ) " ' measured by Hara et al. (1979) for maize silage. However, the temperature increase, as simulated by the model for the energy liberated, had to be reduced to produce realistic peak silage temperatures. That all the energy released by these reactions is not translated into a rise in temperature is not surprising. A portion of the energy is used in cell growth and maintenance while the remainder will be released as heat. The systems in which the experimental results were collected were not perfectly insulated so that a large fraction of energy was lost to the surroundings by convection, conduction and evaporation. The actual energy that goes directly to raising the temperature of the silage was estimated to be about 10% of the total available. Conclusions The onset of aerobic deterioration could be simulated by modelling the major microbiological and chemical processes. This result shows that the present understanding of aerobic deterioration allows a quantitative description. The model is applicable to loosened maize and grass silage, as found in the front of silages, in silage removed before feeding and in many laboratory tests for aerobic deterioration. To make the model fit for complete farm silos it has to be coupled to a model describing the physical processes including gas and heat transportation in the silo mass. Acknowledgments We are grateful to Dr G. Pahlow, Institut fur Griinland und Futterpflanzenforschung, Braunschweig, FRO for his permission to use unpublished data. 164 M. G. Courtin and S. F. Spoelstra The authors thank Mr J. A. C. van Beers, Instituut voor Veevoedingsonderzoek (IVVO), Lelystad, The Netherlands for his technical assistance and Dr A. G. Meiering, School of Engineering, University of Guelph, Canada for stimulating discussions. This project was funded in part by the Programmacommissie Landbouwbiotechnologie (PcLB) in Wageningen, The Netherlands. References ABADIE M . (1962) Associaiion dc Candida mycoderma Reei Loddcr et d'Acetobacter xytinum Brown dans la fermentaiion acctique dcs infusion de ihe. [The relarionship between Candida myoderma Rees Lodder and Aceiobacter xylinum Brown in tea fungus|. Annaies des Sciences Naturettez. Botanique et Bioiogie Vegetate. 2, 76i-800. ARIES V.. CHENEV P . A . and MOSSEL D.A.A. (1982) Ecological studies on the occurrence of bacteria utilizing laciic acid at pH values below 4-5. Journal of Applied Bacteriology. 52, 345-351. BARRV T.N., Dt MENSA M.E.. WEBB P , R . and PARLE J . N . (1980) Some observations on aerobic deterioration in untreated silages and in silages made with Formaldehydecontaining additives. Journal of the Science of Food and Agriculture. 31, 133-146. BECK T . and CROSS F . (1964) Die Ursachen der unterschiedlichen Haltbarkeii von Garfutter. |The causes of the differences in the keeping properties of silage|. Das Wirtschaftseigene Futier. 10, 298-312. CORNISH-BowOEN A. (1976) Principles of Enzyme Kinetics. London: Butterworth. CRAWSHAW R . , THORNE D . M . and LLEWELYN R . H . (1980) The effects of formic and propionic acids on the aerobic deterioration of grass silage in laboratory units. Journal of the Science of Food and Agriculture, 31. 685-694. DANIEL P., HONIG H . . WEJSE F . and ZIMMER E . (1970) Wirkung von Propionsaiire bei der Griinfuttersilieurung. [The action of propionic acid in the ensilage of green fodderj. Das Wirtschaftseigene Futter. 16, 239-252. DEINUM B . and KNOPPERS J. (1979) The growth of maize in the cool temperate climate of the Netherlands: Effect of grain filling on production of dry matter and on chemical composition and nutritive value, Netherlands Journal of Agricultural Science. 27. 116-130. DE LSY and SCHELL J. (1958) Oxidation of several substrates by Acetobacter aceti. Journal of of Bacteriology, 77, 445-451. t>E LEV J . and SCHELL L {1959) Studies on the metabolism of Aceiobacter peroxydans. II. The enzymic mechanism of lactate metabolism. Biochimica el Biophysica Acia, 35, 1S4-I65. DE LEY J . . SWINGS J. and GOSSELE F . (1984) Genus L Acetobacter Beijerinck 1898. In: Murray, R.G.E. (ed.) Bergey's Manual of Systematic Bacteriology Volumel. pp. 268-274. London: Williams and Wilkins. DtviES C, (1972) Action de I'eihanol sur I'oxydation de I'acetate par Acetobacter rancens. |The action of ethanol on the oxidation of acetate by Acetobaaer rancens\. Biochimie, 54, 1073-1083, DIVIES C , and Dupuy P. (1969) Oxydation de I'acetate par Acetobacter rancens. [Oxidation of acetate by/Iceto/wcfer rancens\. Annaies de Technologie Agricole, 18, 339-350. DUPUY P. and MAUGENET J. (1963) Metabolisme de I'acide lactique par Acetobacter rancens. |Metabolism of lactic acid by Aceiobacter rancens. Annaies de Technologie Agricote. 12, 5 - 14. EKLUND T. (1983) The antimicrobial effect of dissociated and undissociated acid at different pH levels. Journal of Applied Bacteriology, 54, 383-389. FLUCKICEH J . and ETTLINGER L . (1977) Glucose metabolism in Acetobacter aceti. Archives of Microbiology. 114, 183-187. FoNTAN C.F. and CHIRIFE J. (1981) The evaluation of water activity in aqueous solution from freezing point depression. Journal of Food Technology. 16, 21-30. GREENHILL W . L . (1964a) The buffering capacity of pasture plants with special reference to ensilage. Australian Journal of Agricultural Research. 15, 511 -519. CREENHILL W.L, (1964b) Plant juices in relation to silage fermentation. III. Effect of water activity of juice. Journal of Ihe British Grassland Society. 19. 336-339. GREENHILL W.L, (1964c) Plant juices in relation to milage fermentation, 1. The role of the juice. Journal of the British Grassland Society. 19, 30-37. CROSS F , and BECK T H (1970) Untersuchtingen uber die Hemmung von aeroben Abbauprozessen durch Propionsaure bei der Auslagerung von Carfutter, [Investigations into the prevention of aerobic deterioration after unloading of silage by propionic acid[. Das Wirtschaflseigene Futter. 16, 1-14. HARA S . . ITOH M . and OHyA.MA Y. (1979) Aerobic deterioration of silages and changes in temperature, gas metabolism, heat production and microflora. Japanese Journal of Zootechnical Science. 50. 549-556. HoNic H. (1975) Umsetzungen und Verluste bei der Nachgarung. [Changes and losses by secondary fermentation[. Das Wirtschaftseigene Futter, 21, 2 5 - 3 2 . HONIG H . and WOOLFORD M . K . (1980) Changes in silage on exposure to air. Occasional Symposium of the British Grassland Society. No- II. Brighton 1979. pp. 7 6 - 8 7 . HuBEH -FROHLI J . (1982) Beitrage zum Clucose-Stoffwechsel in Aceiobacter aceti. [A report of glucose metabolism in Acetobacter aceii\. Ph.D. thesis Eidgenossiche Technische Hochschule Zurich. JoNSsoN A. and PAHt.ow G. (1984) Systematic and biochemica! characterization of yeasts growing in grass siiage inoculated with Lactobacitlus cultures. Animal Research and Development, 20, 7 - 2 2 . JUCKER W, and ETTLINGEH L . (1985) The inhibition of acetate oxidation by ethanol in Acetobacter aceti. Archives of Microbiology. 1-13, 283-289. Ki.M S.U., KIM D.C. and DHURJATI P. (1988) Mathematical modelling for mi.xed culture growth of two bacterial populations with opposite substrate preferences. Biotechnology and Bioengineering. 31, 144-159. LANIGAN C . W . (1963) Silage bacteriology. I, Water activity and temperature relationships of silage strains of Lacrobacillus plantarum. Lactobaciltus brevis and Pediococcus cerevisiae. Australian Journal of Biological Science. 167, 606-615. LEIBENSPERGER R . Y , and PITT R.E. (1987) A model of clostridial dominance in ensilage. Grass and Forage Science. 42. 297-317. LINDGREN S., PETTERSSON K . , KASPERSSON A., JONSSON A. and LiNGVALL P. (1985) Microbial dynamics during aerobic deterioration of silages. Journal of the Science of Food and Agriculture. 36. 765-774. MANKAD T , and BLNCAV H . R , (1988) Model of microbial growth with more than one limiting nutrient. Journal of Biotechnology. 7, 161-166. MCDONALD P. (1981) The Biochemistry of Silage. Chichester: John Wiley and Sons, MEIERINC A.G., CouRTiN M . C . SPOELSTRA S.F,, PAHLOW G . . HoNic H,, SuBOEN R.E, and ZIMMER E . (1988) Fermentation kinetics and toxic gas production of silages. Transactions of the American Society of Agricultural Engineers. 31, 613-621. MIDDELHOVEN W . J . and FRANZEN M . M , (1986) The yeast A model of changes during aerobic deterioration of silage 165 flora of ensiled whole-crop maize. Journal of the Science of Food and Agriculture, 37, 855-361, MmobLHDvtN W J . and VAN BAALE> A . H . M . (1988) Development of the yeast flora of whole-crop maize during ensiling and during subsequent aerobiosis. Journal of the Science of Food and Agriculture. 42, 199-207. MOON N.J. and ELY L , O , (1979) Identification and properties of yeasts associated with the aerobic deterioration of wheat and alfalfa silages. Mycopaihotogica. 69, 153-156, NANHA A., TAMURA A. and NAGAIS- (1984) Synergistic effects of acetic acid and ethanol on the growth of Acetobacter sp. Journal of Fermentation Technology. 62, 501-505. NEAt H.D. SlC. and THOBNIEV J . H . M . (1983) A model of the anaerobic phase of ensiling. Grass and Forage Science. 38, 121-134. N()u\ F,, HAYA^HI K. and MIZLNLM* T , (1982) Influence of pH on inhibitory activity of acetic acid on osmophilic yeasts used in brine fermentation of soy sauce. Applied and Environmental Microbiology. 43, 245-246. NiiMUHA Y,, IwAHARA M. and HONGO M . (1988) Acetic acid production by an electrodialysis fermentation method with a compuierized control system. Applied and Environmental Microbiology. 54, 137-142. OHVAMA Y, and MCDONALD P. (1975) The effect of some additives on aerobic deterioration of silages. Journal of the Science of Food and Agriculture, 26, 941 -948, OHYAMA Y . , HARA S . and MASAKI S. (1981) Effects of dry matter contents of silages and maximum environmental temperature during ensilage on the occurrence of aerobic deterioration. Japanese Journal of Zooiechnical Science. 52, 789-797. PAHIDW G , (1982) Verbesserung der aeroben Stabilitat von Silage durch Impfpraparate. | Improvement of the aerobic stability of silage by inoculants]. Das Wmschaftseigene Futter. 28. 107-122. Pirr R.E., Muii. R.E. and LEiHfcvspERGER R,Y, (1985) A quantativc model of the ensilage process in lactate silages. Grass und Forage Science. 40, 279-303. RLLS V . H . (1982) The aerobic deterioration of grass silage and its effeci on the water-soluble carbohydrates and the associated heat production. Journal of the Science of Food Agriculture. 33,499-508. Rois K.D. (1975) Estimation of water activity of intermediate moisture foods. Food Technology. 29, 26-34. SHIMI/.U H . , MiVAi K.. MATSLJHISA H . , UASAK; E . and ToMovtuA M. (1977) Effect of ethanol on acetate oxidation by Aceiobacter aceii. European Journal of Applied Microbiology. 143, 283-289, SPotL.stRAS.F.. T^ooNK L. and Si EG A, (1985) Admixing cage layer manure (CLM) improves the aerobic stability of whole-crop maize silage. Proceedings of the Fifth Iniernaiional Symposium on Agricultural Wastes. Chicago. 1985. pp, 159-166, SPOLI STRA S , F . , COUHTIM M , G , and VAN BEERS J , A , C . (1988) SPONHOLZ W , R , and DUTRCCH H , H . (1985) Uber die Herkunft von Gluconsaure, 2- und 5-O)(o-G!uconsaure sowie Glucuronund Galacturonsaure in Mosten tind Weinen, |Origin of gluconic, 2- and 5-Oxo-gluconic, glucuronic and galacturonic acids in musts and wines|. 1^1/15, 24, 31-58, SvtiNGS J, and OE LEY J, (1981) The Genera Gluconobacier and Acetobaaer. In: Starr M,P,, Stolp H,, Truper H.G., Baiows A, and Schlegel H.G. (eds) The Prokaryotes Vol. t. pp, 771-778, Berlin: Springer-Verlag, WEISE F , (1963) Beitrage zur Grundlagenforschung in der Garfutterbereitung, II, Das bakteriologische Bild einer Grassilage, |A report of baiic research in silage making, II. The bacteriological picture of a grass silage|. Landbauforschung Volkenrode. 13, 111-116, WILSON R.K, (1986) Measurement of the energy of activation for the production of lactic acid in grasses made in laboratory silos, Irish Journal of Agricultural Research. 25, 269-272. W(X)LFORt)M.K. and COOK J,E. (1978) A n.ote on the effect on the aerobic deterioration of maize silage of the manipulation of the microfiora by means of antibiotics. Animal Feed Science and Technology. 3, 89-94, WooLFORoM.K, and WILKIE A,C, (1984) Investigations into the role of specific microorganisms in the aerobic deterioration of maize silage. Journal of Agricultural Science Cambridge. 102.97-104. WoouFORD M.K., HoNiG H. and FENLON J.S. (1977) Untersuchungen uber aerobe Umsetzungen in Silage mit Hilfe einer Labortechnik. Teil I: bescreibung und statistische Bewertung der Methode. [5tudies on the aerobic deterioration of silage using a small-scale technique. Part I: a description of the technique and statistical appraisal|. Das Wirtschaftseigene FuIter, 23, 10-22, WOOLFORD M.K.. HONIG H. and FENLON J , S , (1978) Untersuchungen liber den aeroben Abbau in Silage mit einer Labortechnik. Teil 2: mikrobiologische, physikalische und chemische Veranderungen wahrend des aeroben Abbaus von Maissilage. |Studies on the aerobic deterioration in silage with a small-scale technique. Part 2: microbiological, physical and chemical changes during aerobic deterioration of maize silagej. Das Wirtschaftseigene Futter. 24, 125-139, WOOLFORD M.K., HOMG H . and FENLON J.S, (1979) Unter- suchungen uber den aeroben Abbau in Silage mit einer Labormethode. Teil 3: mikrobiologische, physikalische und chemische Veranderungen wahrend des aeroben Abbaus in frischer und angewelkter Grassilage. |Studies on the aerobic deterioration in silage with a small-scale technique. Part 3: microbiological, physical and chemical changes during the aerobic deterioration of fresh-cut and wilted grass silage|. Das Wirtschaftseigene Futter. 25, 158-177. WjtEST R.C, (1987) Handbook of Chemistry and Physics, 68th edn. Florida: CRC Press, Acetic acid bacteria can initiate aerobic deterioration of maize silage. Journal of Agricultural Science; Cambridge. I l l , 127-132. (Received 3 January 1989; revised 6 June 1989}