Dual sensor measurement shows that temperature outperforms pH as an early sign of aerobic deterioration in maize silage

High quality silage containing abundant lactic acid is a critical component of ruminant diets in many parts of the world. Silage deterioration, a result of aerobic metabolism (including utilization of lactic acid) during storage and feed-out, reduces the nutritional quality of the silage, and its acceptance by animals. In this study, we introduce a novel non-disruptive dual-sensor method that provides near real-time information on silage aerobic stability, and demonstrates for the first time that in situ silage temperature (Tsi) and pH are both associated with preservation of lactic acid. Aerobic deterioration was evaluated using two sources of maize silage, one treated with a biological additive, at incubation temperatures of 23 and 33 °C. Results showed a time delay between the rise of Tsi and that of pH following aerobic exposure at both incubation temperatures. A 11 to 25% loss of lactic acid occurred when Tsi reached 2 °C above ambient. In contrast, by the time the silage pH had exceeded its initial value by 0.5 units, over 60% of the lactic acid had been metabolized. Although pH is often used as a primary indicator of aerobic deterioration of maize silage, it is clear that Tsi was a more sensitive early indicator. However, the extent of the pH increase was an effective indicator of advanced spoilage and loss of lactic acid due to aerobic metabolism for maize silage.

www.nature.com/scientificreports/ relationships between pH and T si determined in situ following aerobic exposure of maize silage; and (3) evaluate aerobic loss of lactic acid associated with the temporal courses of T si and pH change. Silages treated or untreated with a microbial inoculant, in three replicated bunker silos each, were used to create a robust data set to better examine patterns of T si and pH change after exposure to air, the dual-sensor, and relationships with organic acid content. Comparison of the silage treatments per se was not the study purpose.

Results
Initial information on the two silages. According to a quality classification for maize silage 5 , initial counts of yeast and mold suggest excellent aerobic stability of the additive-treated silage, but reduced aerobic stability in the untreated silage (Table 1). In addition, the amount of acetic acid, and the ratio between lactic and acetic, in the additive-treated silage were not consistent with a maize silage that underwent heterolactic fermentation 1 .
Aerobic responses of T si and pH. The time courses of pH and T si (each course represents the mean of three replicates) from the silages incubated at 23 °C (T c ) (Fig. 1a,b) and 33 °C (Fig. 1c,d) were similar. Increases of T si began earlier for 33 °C silage following exposure to air. Assuming that: Table 1. Chemical and microbial analyses of the two silages at sampling, not treated or treated with a biological additive. Data are means ± SE of n = 3 replicates; WW, wet weight; DM, dry matter; cfu, colonyforming units.  Table 2). The T si at 23 °C increased with a subsequent decline, whereas T si at 33 °C increased to plateau. Generally, there was a lag of 28 to 48 h between the onset of a T si increase and onset of a pH increase following aerobic exposure.
Validation of the in situ measurement of pH. Figure 2 (n = 12 × 2) shows the piecewise relationship between silage pH resulting from water-extraction (i.e., ex situ determination) and direct measurement (i.e., in situ), where paired values of in situ and ex situ were measured from the same instrumented jar/pH-sensor. The in situ and ex situ measurements agreed well, and were consistent with a 1:1 relationship of (i.e., R 2 = 0.914, RMSE = 0.054, P < 0.01) over the pH range of 3.5 to 4 (Fig. 2a, initial period) and (R 2 = 0.931, RMSE = 0.092, and P < 0.01) over the pH range 6.5 to 7.5 (Fig. 2b, final period).
A general comparison between the additional jars with ex situ and in situ determinations using the instrumented jars over the process of experiment is in Fig. 3 (n = 120). In contrast to Fig. 2, the higher R 2 of 0.985 and higher RMSE (0.187) is likely due to more data. Nevertheless, both Figs. 2 and 3 demonstrate that in situ pH measurements from 3.6 to 7.4 ( Fig. 1) were accurate, and that the dual-sensor technique suited the mini-silos.
Temporal variation of pH versus lactic and acetic acids. Profiles of pH versus lactic and acetic acids over time in silages incubated at 23 °C and 33 °C are present in Fig. 4. For those incubated at 23 °C, the lactic acid content began to decline at 40 h (control) or 70 h (treated silages) and continued to decline to very low levels. In contrast, lactic acids of these silages incubated at 33 °C decreased earlier and faster, reflecting the influence of the incubation temperature on microbial metabolic activity. The lactic acid contents in all samples were initially Table 2. Final (i.e., 168 h of aerobic incubation) chemical and microbial analyses of the two silages, not treated or treated with a biological additive. WW, wet weight; DM, dry matter; cfu, colony-forming units. *, P < 0.05; **, P < 0.01; NS, P > 0.05.  www.nature.com/scientificreports/ higher than those of acetic acid which, in the silages incubated at 23 °C, increased between 24 and 48 h, then declined to low levels. In general, the patterns in Fig. 4 show that most lactic acid in each sample was metabolized before the pH began to increase.
Estimating relative loss of lactic acid (RLLA). Figures 1 and 4 suggest that the relative loss of lactic acid in these maize silages can be estimated from the combined information in the time courses of T si and pH. In the initial 2 to 3 days, lactic acid declined but remained the primary organic acid (Fig. 4), although it is likely that buffering capacity of the silages mitigated a pH change despite the decline of lactic acid. This allowed RLLA to be estimated as: ΔT si = T si -T c as an index of its loss (Fig. 5). The relationships of RLLA (0-60%) and ΔT si (0 to 6-10 °C), on the left side of Fig. 5, were linear (R 2 ≥ 0.901, P < 0.01). After T si stabilized (Fig. 1), RLLA could be estimated from the quadratic increase in pH (R 2 ≥ 0.781, P < 0.01; right side of Fig. 5). The non-linear relationship between pH and RLLA was likely due to changes in the acetic acid content. In the final stage, contents of lactic and acetic acids were similar (Fig. 4). Thus, for these maize silages over the range of 0 to 60% loss of lactic acid, ΔT si was a useful predictive tool. However, as RLLA became ≥ 60%, estimation based on pH became more appropriate, albeit increasingly uncertain due to the non-linear relationship between pH and RLLA (right side of Fig. 5).

Discussion
The slower changes in pH versus T si over time of incubation after air exposure may reflect buffering capacity of the silage 4,10 . Since buffering is a normal property of plant material 26 , a time lag is inherent but likely situationally specific. During the lactic acid generating fermentation phase, a delayed pH decline has been observed [27][28][29] . However, effects of buffering on this lag time relative to the pH increase after silo opening have not been adequately addressed, although a potential role of buffering in aerobic stability was previously suggested based upon limited experimental support 1,30 . Our results, measured in situ, provide the first characterization of the time lag between T si and pH in response to aerobic exposure, and demonstrate that a delayed pH increase during feed-out (i.e., de-acidification) is affected by the concentration of lactic acid and buffering capacity of the silage mass.
In a study where maize silage was exposed to air 31 , the authors noted that yeast counts increased from 3 to 6.5 (log 10 cfu/g) while pH remained < 4 in the initial 130 h. Thereafter, pH increased rapidly to 6.3 by 20 h. The authors 31 attributed this outcome to buffering capacity, and presented numerous models associated temporal counts of yeast and mold to predict buffering capacity 26 . Tables 1 and 2, show that silage yeast counts increased at both incubation temperatures, and the only slight increase of acetic acid at 23 °C between 24 to 48 h (Fig. 4a) could be related to the microbial profile of yeasts over time.
pH has been regarded as an indicator/inhibitor of microbial activity/growth in silage 1,22 , with a lower pH reflecting stronger suppression to microbial activity 30,32 . However, simultaneous measurements of T si and pH (Fig. 1) during the early unstable period of silage after exposure to air demonstrate the immediate onset of aerobic activity based on the T si increase. That pH remained low for an additional 1 to 2 days is likely due to the buffering capacity of the maize silage. Overall, this suggests that pH may be an inadequate measure of microbial activity when silage is in the early aerobically unstable phase after silo opening 31 .
The pH time courses were dominated by variations in lactic acid during the initial period (Fig. 4) since its contents were higher than those of acetic acid. However, contents of these acids later converged (Fig. 4). Even with similar concentrations, it was likely lactic acid which was the primary pH determinants because the pKa of lactic acid (3.86), its acid dissociation constant, coincided with the lowest pH value (3.6, Fig. 1), whereas the pKa value of acetic acid is higher (4.75) reflecting it being a ten times weaker acid than lactic acid 30,33,34 . Thus, as the weaker acid, acetic had more undissociated molecules in silage water when the pH was < 4.75 34 . In addition, as acetic acid is more volatile than lactic 2,29,30 , and thus more acetic acid would have been lost by volatilization. Thus, pH is a useful indicator of aerobic loss of lactic acid (right sides of Fig. 5), but only after buffering capacity is exceeded.
Using this novel dual-measurement technique we found that pH is not as effective T si as an earlier aerobic marker of maize silage spoilage, but is effective at longer times of air exposure. Indeed, both T si and pH have been suggested as indicators of oxidative degradation 5,10,13,21,35-38 . The (ΔT si = + 2 °C) as the threshold of aerobic stability www.nature.com/scientificreports/ has been widely accepted 1,7,8 , while other studies have suggested a pH-based threshold of aerobic deterioration when pH exceeds the initial value by 0.5 units 17,39-41 . Our case study suggests that: ΔT si = + 2 °C, represents a 11 to 25% loss of lactic acid from these maize silages whereas: ΔT si = + 3 °C, also a commonly accepted threshold of aerobic deterioration, represents 18 to 35% loss. In contrast, at ΔpH = 0.5 unit, over 60% of lactic acid had been lost, thereby further supporting the suggestion 21 that pH is a less sensitive indicator of aerobic deterioration than is T si for maize silage. However, for farm silos where ambient temperature fluctuates diurnally 5,10,21 , this may interfere with determining the threshold of aerobic stability. As the chemical definition of pH is independent of ambient temperature, this may be an additional advantage for outdoor investigation. Extrapolation of our outcomes with maize silage to grass and/or legume silages should be done with care as maize silage has a relatively low acid buffer capacities 42-44 for a silage. Consequently, silage with higher acid buffer capacity may exhibit different pH time delays relative to T si in response to air exposure and subsequent aerobic deterioration.

Conclusions
Time courses of T si and pH simultaneously measured in maize silages in situ using a novel dual-sensor documented lags of pH change relative to those in T si at two incubation temperatures. These lags, likely due to ongoing losses of lactic acid and silage buffering capacity that delayed the pH rise following aerobic exposure, demonstrate that pH is a less sensitive indicator than T si of short time aerobic deterioration of maize silage. However, pH is an effective indicator of advanced spoilage and loss of lactic acid due to aerobic metabolism.
Our results refer to a case observation using the dual-sensor tool that is different from ex situ method. We presented the comparison/discussion of the in situ and ex situ data and outcomes recommend this simpler and faster in situ method for mini-silo studies. As all maize samples measured were extracted non-destructively from farm silos, this technique is promising for farm level use, although a protective shield for the glass electrode of pH sensor is necessary. The stronger conclusion that the temperature outperforms pH as an early sign of aerobic loss of various silage depends on extending experiments, associated with multiple affecting factors in future study.
To facilitate sampling from the silos, while minimizing air entry, a metallic coring device (inner diam. 9.5 cm, length 25 cm; Fig. 6a) was constructed with the same inner diameter as the glass jars (1.5 L, inner diam. 10 cm, depth 20 cm; Fig. 6a) into which the samples were placed for deterioration assessment. The sampling area (1.8 m × 1.4 m) was located in the center of each silo. Prior to sampling from the bunker, 20 cm of the surface of the exposed face was removed. Sampled silage was immediately packed to a density of 220 kg/m 3 dry matter (DM) in glass jars which were tightly sealed with aluminum lids (diam. 10 cm, thickness 1 cm), rubber O-rings and four elastic clamps (Fig. 6a).
In situ silage pH sensor. The study required validation that the pH sensor used is suitable for liquids and maize silage with typically high water content (> 600 g/kg) under high compaction/density (500 to 600 kg/ www.nature.com/scientificreports/ m 3 wet weight). Since maize silage is a H 2 O-rich porous material 45 , its dense compaction enables good contact between the pH-electrode and the water phase in the silage. The lids of the glass jars were perforated with five holes (Fig. 6b) 36 , all mini-silos/jars were covered with insulated sleeves throughout (Fig. 6b). Prior to and after the experiment, all pH sensors of the instrumented jars were calibrated with standard calibration liquids: HI 70,004 (pH = 4.01) and HI 70,007 (pH = 7.01), (Hanna Instruments, Inc. Woonsocket, USA). The pH tip and a thermocouple (1 mm diameter, T-type) were inserted together into the sample at the same depths (8 cm) in each jar. Both sensors are connected to a data-logger (ALMEMO-2890-9, AHLBORN GmbH, Germany) with data recorded at 10 min intervals over the 168 h experimental cycles. Each instrumented jar, and 10 additional jars (1.5 L) as a group all containing the same maize silage and packed to the same density. These additional jars served as sampling points during aerobic deterioration measures. During each sampling (15 h interval beginning 24 h after ensilage), a subsample of 150 g was removed from 8 cm behind the face in these additional jars, to same depth as the in situ pH and thermocouple sensors. A 100 g subsample was sealed with a vacuumizer (Boss Mini-Max, Helmut Boss Verpackungsmaschinen KG, Bad Homburg, Germany) to remove air, frozen and shipped on ice to a commercial laboratory for chemical and microbial analyses within 24 h. The remaining 50 g subsample was divided equally, each with 225 g deionized water, to determine pH ex situ, which used for evaluation of in situ the measurements from the instrumented jars.
Statistical analysis. Experimental data were analyzed using IBM SPSS v25.0 (IBM Co., Armonk, NY, USA). Linear regression, curve fitting and fitting errors were evaluated using coefficient of determination (R 2 ), significance (P) and root mean square error (RMSE). A T-test was conducted to determine effect of T c (i.e., incubated at 23 or 33 °C) on chemical and microbial composition at the end of the experiment for final-data processing.
Chemical analysis. Dry matter was measured by drying at 60 °C for 48 h in a forced-air oven 39 . Buffering capacity (BC) was determined by the method of the literature 46 . Acids (i.e., lactic, acetic, butyric) and ethanol were determined using high-performance liquid chromatography (LC-2010AHT, Shimadzu Corp., Kyoto, Japan), with an integrated UV-detector. Water soluble carbohydrate (WSC) content was determined enzymatically 47 .

Microbial analysis.
According to the method of the literature 3 , 30 g of silage was suspended in 270 ml of ¼-strength ringer solution (2.25 g/ l NaCl, 0.105 g/l KCl, 0.06 g/l CaCl2, 0.05 g/l NaHCO3) (Merck, Darmstadt, Germany) and homogenized in a mixer for one minute. From this suspension, total bacterial counts were analyzed on plate-count agar (5.0 g/l enzymatic digest of casein, 2.5 g/l yeast extract, 1.0 g/l glucose, 15 g/l agar, pH = 7.0) (Merck, Darmstadt, Germany) after aerobic incubation at 30 °C for 2 days. Lactic acid bacteria were quantified on MRS agar (Merck, Darmstadt, Germany) after anaerobic cultivation for 3 days at 30 °C. Yeasts and molds were detected using yeast extract glucose chloramphenicol (YGC)-agar (5.0 g/l yeast extract, 20.0 g/l glucose, 0.1 g/l chloramphenicol, 14.9 g/l agar, pH = 6.6) (Merck, Darmstadt, Germany) after incubation at 25 °C for 3 days. www.nature.com/scientificreports/ acknowledge the assistance of Dr. Peter Robinson (University of California at Davis) for his thoughtful comments on an earlier draft of the paper, and on the English text.