Removal of solids and chemical oxygen demand in poultry litter anaerobic digestion with different inocula

 Population growth has contributed to increasing poultry production, entailing a high waste loading, mainly poultry litter. One of the alternatives to treat such residues is anaerobic digestion, in which digester startup and generated-digestate quality are related to the material to be digested and to operation conditions, wherein inoculum use is one of the factors. This study therefore aimed to investigate how digestates, such as inocula, influence poultry litter (PL) anaerobic digestion, as well the reduction of solids and chemical oxygen demand (COD). For this, two inocula (bovine and swine digestates) were tested in the digestion process. The inocula were added at loads of 0.67, 1.00 and 1.67 gVS.L-1day-1. A split-plot design was developed and data underwent analysis of variance with means compared by the Tukey's test at 5% significance. Concerning bovine and swine inocula, it was concluded that both are indicated in the process. However, swine inoculum is better indicated because it had a better removal of total solids (TS), volatile solids (VS) and COD.

Rev. Ambient. Água vol. 15 n. 2, e2469 -Taubaté 2020 It is worthy to note that different inocula lead to different biogas production, with significant effect of the inoculum on the Biochemical Methane Potential (BMP) (Vrieze et al., 2015). Another paper (Rico et al., 2017) shows the influences of inocula on the anaerobic digestion of pig slurry in UASB reactor.
Therefore, this technology must be optimized to reach maximum benefits (Gyenge et al., 2014). Understanding the microbial community and its role in the process is critical to the process control (Shen and Zhu, 2016).
Given this background, this study tested two types of inocula, bovine and swine, in the anaerobic digestion of poultry litter, quantifying reductions of solids and COD.

Substrate and Inocula
The substrate consisted of poultry litter gathered from a farm in the city of Quatro Pontes -PR, Brazil. Table 1 shows the characterization of the material with values of hydrogenation potential (pH), total solids (TS), fixed solids (FS), volatile solids (VS), moisture, N and chemical oxygen demand (COD). Prior to biodigestion, residue underwent a pre-treating process with 8 mm mesh sieving to increase its contact surface and facilitate anaerobic digestion. As inoculum (IN), two types of fresh material were used: bovine (B) and swine (S), both collected from field biodigesters and kept at room temperature for characterization. The previous table shows the physicochemical characterization of both inocula.

Poultry litter digestion
The trial was carried out at the Laboratory of Bioreactors of the Research Group on Water Resources and Environmental Sanitation of the Western Paraná State University.
Two biodigesters built in polyvinyl chloride (0.60 x 0.10 m) were operated. The equipment had an inlet port and a digestate outlet arranged horizontally, with 4 L useful volume, as shown in Figure 1. The two biodigesters were inoculated with 30% of their useful volume and 70% with a mixture of water and chicken litter at a ratio of 50 grams of litter for each mixture liter. Twentyfour hours after inoculation, removal of oxygen traces was completed and daily feeding was started, in a semi-continuous flow model, applying loads of 0.67 (L1), 1.00 (L2) and 1.67 (L3) gVS.L -1 .day -1 , at a concentration of 2% volatile solids. These loads were defined according to previous work (Alcantara et al., 2016).
The operating times for each load were 74, 69, and 66 days, respectively, including stabilization time, monitored by the ratios VA/TA (Volatile Acidity/Total Alkalinity) and IA/PA (Intermediate Alkalinity/Partial Alkalinity). The temperature was room temperature, daily monitored.
The volume of biogas and methane produced was measured with a Mariotte device filled with sodium hydroxide (NaOH) solution to dissolve carbon dioxide (CO2) (Bertin et al., 2004;Buitrón and Carvajal, 2010;Scoma et al., 2013). However, these results are shown and discussed in another paper.

Monitoring pH, Ammoniacal Nitrogen and Ratios VA/TA and IA/PA
The pH was determined using a benchtop pH meter (Model 3MP) and ammonia nitrogen concentration was monitored following the 4500-NH3 titration method (APHA, 1998).
The values for the ratios VA/TA and IA/PA were measured through titration, observing the threshold values reported by Méndes-Acosta et al. (2010) and Ripley et al. (1986).
The total alkalinity in anaerobic systems is determined by titration of the sample to pH 4.30, measuring the buffer capacity due to bicarbonate and volatile acids, defined by Ripley et al. (1986) as partial alkalinity (PA) and intermediate alkalinity (IA) respectively. Ripley et al. (1986) established that the ratio IA/PA with values greater than 0.30 indicates the occurrence of disturbances in the process of anaerobic digestion.
According to Méndez-Acosta et al. (2010), the VA/TA ratio should be between 0.10 and 0.35, to assure the stability of anaerobic digestion.

Removal of solids, COD Removal and characterization of digestate and sludge
Measurements of TS, FS, and VS were performed based on method 2540 of APHA (2005). In this case, COD quantification was made following method 5220D of APHA (2005). For soluble COD, samples were filtered, using the blank test prepared with distilled water.
To determine the micro (Zn, Fe, Cu and Mn) and macronutrients (K, N, Mg, Na, Ca) of the digestate, a nitro-perchloric digestion (3:1) of the samples was performed and an aliquot was taken for reading with atomic absorption equipment.

Statistics
Statistical analyses were performed following a split-plot design, wherein inoculum factor was the main plot (with two levels: B and S). The secondary plot consisted of feeding load factor (with three levels: L1, L2, and L3). The analyses were performed using SISVAR software v. 5.3 (Ferreira, 2010). As response variables, both ratios VA/TA and IA/PA were considered, as well as removal rates of total solids, total volatile solids, total fixed solids, COD total and COD soluble.
Data presenting normality by the Kolmogorov-Smirnov's test and homoscedasticity by the Bartlett or Levene's test had the variances submitted to analysis of variance, at 5% significance. Tukey's test was used for mean comparison at a 5% significance level. neutrality, as shown in Figure 2A. The highest mean values of pH occurred at loads of 1.00 and 1.67 gVS.L -1 .day -1 , for bovine and swine inocula, respectively. These results corroborate the findings of Onofre et al. (2015), who evaluated the pH of the biofertilizer from poultry litter and pointed out that the pH ranged from 6.6 to 8.0, and determined that the anaerobic digestion process was suited.

RESULTS AND DISCUSSION
Although no variations in pH were observed, the presence of ammonia was also monitored throughout the process, and the highest concentration of ammonia (280.51 mg L -1 ) occurred with a load of 1.67 gVS. L -1 day -1 swine inoculum. According to Chen et al. (2008), ammonia concentrations greater than 1000 mg.L -1 were harmful to the process. Since the presented value is below the concentration threshold reported in the literature, it was concluded that there was no effect of ammonia on anaerobic digestion. It should be pointed out that the temporal scale shown in Figure 2A is different from the scales of the other figures because the pH monitoring was registered daily, a fact that did not occur with the monitoring of the ratios VA/TA and IA/PA, solids and COD removals ( Figure  2B and 2C).
According to Chernicharo (2007), the monitoring of the partial or bicarbonate alkalinity is more useful than the pH monitoring due to scales characteristics: while the pH scale is logarithmic, the scale for alkalinity is linear. It is fundamental to maintain a low concentration of volatile fatty acids and pH in the interval 6.6 to 7.4 (Lahav and Morgan, 2004).
The main buffer in anaerobic digesters is the ion bicarbonate, at pH from 5.3 to 7.3, expressed in mg of CaCO3 L -1 , varying with the waste type (Lahav and Morgan, 2004).
According to Ripley et al. (1986), IA/PA values greater than 0.30 may indicate instability problems in digestion. Here, it was observed that IA/PA values for a load of 0.67 gVS.L -1 .day -1 exceeded 0.30; however, decreasing over time and remaining stable for the other applied loads. Figure 2B denotes that the highest mean values of IA/PA were observed for loads of 1.67 and 0.67 gVS.L -1 .day -1 , respectively for bovine and swine inocula; however, these values are still below the recommendation of Ripley et al. (1986).
Martín-Gonzalez et al. (2013) asserted a possible process stability for values different from 0.3 due to variations in the characteristics of each effluent.
Values for the ratio VA/TA were similar for both inocula, presenting slightly higher values in a load of 0.67 gVS.L -1 .day -1 , as seen in Figure 2C, ranging from 0.252 to 0.228 for bovine and swine inocula, respectively. Table 2 shows the statistical breakdown of VA/TA ratio means. It can be observed that, by fixing the load factor and evaluating its variation in all inoculum levels, a significant difference occurred between loads. The highest values occurred for the loads of 0.67 and 1.67 gVS.L -1 .day -1 , considered statistically equal for bovine inoculum. By fixing the inoculum factor, there was no statistically significant difference between loading levels. Means followed by the same uppercase letters within columns (inoculum) and lower case within lines (loads) do not differ from each other by the Tukey's test at 5% significance.
The greatest stability time was observed for a load of 1.00 gVS.L -1 .day -1 ; however, at 0.67 gVS.L -1 .day -1 , the highest means were reached. Nevertheless, these values are within the optimum range for operating levels. The results corroborate those of Kuczman et al. (2011), who evaluated biogas production from cassava wastewater with increasing organic loads and feeding volumes. The microbial adaptation was analyzed by monitoring VA/TA ratio, which remained within a range between 0.14 and 0.30.
It is important to note that the operational conditions of the process affect the response of the indicators (Li et al., 2014), indicating the unbalance between acid production and consumption and warning about a possible digestion failure.
The temperature mean values were from 16°C to 25°C and values below 20°C occurred only for a load of 1.00 gVS. L -1 day -1 , caused by climatic interference. From the mean and standard deviation, the variation in load was 2.1°C and, as reported by Deublein and Steinhauser (2011), methanogenic Archaea are able to change rapidly ideally but the temperature should not vary abruptly by ± 2°C.

Removal of TS and VS
The highest means of TS removal were related to the load 0.67 gVS.L -1 .day -1 , with swine inoculum (59.46%), as shown in Figure 3A.
In addition, bovine inoculum promoted increase in TS removal as loads were raised, with the highest value at the load of 1.67 gVS.L -1 .day -1 (57.36%).
Rev. Ambient. Água vol. 15 n. 2, e2469 -Taubaté 2020  Table 3 displays the statistical breakdown of the means for TS removal. It is shown that by fixing load factor there was a significant difference as inoculum level increased, with the removal by bovine inoculum statistically equal for 0.67 and 1.00 gVS.L -1 .day -1 . The greatest value was observed for 1.67 gVS.L -1 .day -1 (57.36%). Yet for swine inoculum, loads of 0.67 and 1.00 gVS.L -1 .day -1 were statistically similar to each other. On the other hand, loads of 0.67 and 1.67 gVS.L -1 .day -1 differed from each other at 5% significance, with the largest removal observed for 0.67 gVS.L -1 .day -1 (59.46%).
By fixing the inoculum factor, and assessing it according to the loads, there was a significant difference between both inocula at 0.67 gVS.L -1 .day -1 , being the greatest TS removal for the swine one (59.46%). Means followed by the same uppercase letters within columns (inoculum) and lower case within lines (loads) do not differ from each other by the Tukey's test at 5% significance.
The highest VS removal values referred to a load of 1.67 gVS.L -1 .day -1 using bovine inoculum, as shown in Figure 3B. Additionally, there was an increase in VS removal as loads increased, with bovine inoculum, being the greatest mean for a load of 1.67 gVS.L -1 .day -1 (78.21%). Table 4 displays the statistical breakdown of the means for VS removal. It is shown that by fixing load factor there was significant difference as inoculum level increased, with the greatest removal for bovine inoculum at a load of 1.67 gVS.L -1 .day -1 (78.21%). Regarding swine inoculum, loads of 0.67 and 1.00 gVS.L -1 .day -1 are statistically equal, as well as 0.67 and 1.67 gVS.L -1 .day -1 . Conversely, loads of 1.00 and 1.67 gVS.L -1 .day -1 differed at the 5% significance from each other, with the greatest removal for swine inoculum at a load of 1.67 gVS.L -1 .day -1 (71.71%).
By fixing the inoculum factor and assessing it according to the loads, there was a significant difference between both inocula at 0.67 and 1.67 gVS.L -1 .day -1 . The greatest VS removal was found for swine inoculum at 0.67 gVS.L -1 .day -1 (67.85%), and for bovine one at 1.67 gVS.L -1 .day -1 (78.21%). Means followed by the same uppercase letters within columns (inoculum) and lower case within lines (loads) do not differ from each other by the Tukey's test at 5% significance.
The highest removal means, 59.46% (0.67 gVS.L -1 .day -1 ) and 78.21% (1.67 gVS.L -1 .day -1 ), were found for TS and VS, respectively. These findings corroborate Orrico Júnior et al. (2010), who evaluated the technical feasibility of anaerobic digestion of pre-composted poultry litter and carcasses. These authors noted the anaerobic digestion efficiency in degrading resistant compounds, such as chicken litter. Despite the sharp reductions in VS, around 44.05%, this value was lower than that obtained here, evidencing microbial action in the inoculum.
This way, the results encountered here are in agreement with those of Silva et al. (2013), who concluded that anaerobic digestion is an efficient procedure to treat organic residues. They also ascertained that biofertilizer addition not only favored system performance regarding pH, but also influenced biogas production and reduced total solids by 40.85%.

Removal of COD (total and soluble)
The major means of CODtotal removal were registered at loads of 0.67 and 1.00 gVS.L -1 .day -1 , using both swine and bovine inocula, respectively, as shown in Figure 4A. As seen in Figure 4B, the load of 1.00 gVS.L -1 .day -1 provided the major means of CODsoluble removal. Since data regarding CODtotal removal showed no normality, results did not undergo variance analysis. Table 5 presents the statistical breakdown of means for CODsoluble as a function of inoculum and loads. When load factor is fixed, it is noteworthy seeing a significant difference of loads using bovine inoculum, with the emphasis placed at the load of 1.00 gVS.L -1 .day -1 , which provided the highest mean (61.31%). For swine inoculum, there was a significant difference among loads, with the highest removal reached by 1.00 gVS.L -1 .day -1 (53.29%), being statistically equal at 5% significance for a load of 0.67 gVS.L -1 .day -1 (48.91%).
When fixing the inoculum factor and evaluating it in relation to the load levels, we could note a statistically significant difference between both inocula at 0.67 and 1.67 gVS.L -1 .day -1 .
The results found in this study are superior to those reported by Blanco et al. (2014), who studied anaerobic digestion of dairy cattle manure with the addition of poultry litter. These authors reported a low COD removal efficiency, on average 36%, which resulted in a low stability for the biofertilizer. Means followed by the same uppercase letters within columns (inoculum) and lower case within lines (loads) do not differ from each other by the Tukey's test at 5% significance.
The results obtained corroborate Lynch et al. (2013), who presented a review detailing advances in the three main alternative disposal routes for poultry litter, which are composting, anaerobic digestion, and direct burning. These authors concluded that such technologies open up opportunities to marketing both energy and nutrients in poultry litter, mainly for anaerobic digestion, once operation conditions are monitored.
It is known that the presence of inoculum influences the methanogenic activity and can present significant differences for different inoculum sources (Vrieze et al., 2015;Rico et al., 2017). In this study, it is observed that different types of inoculum may also influence the removal of COD and solids.

Characterization of digestate and sludge
The quality of the digester and its potential for agronomic use depends on many factors. Because of this, each project have to include a specific analysis to determine available nutrient content (Nicoloso et al., 2019).
Regarding the values of micronutrients (Zn, Fe, Cu and Mn) in Table 6, note that the highest mean Zn values are for the 1.00 gVS.L -1 .day -1 load. Fe values refer to the loads 1.67 gVS.L -1 .day -1 and 0.67 gVS.L -1 .day -1 for bovine and swine inoculum, respectively.
It is observed that the highest mean values of K refer to the loads 0.67 gVS.L -1 .day -1 and 1.67 gVS.L -1 .day -1 for bovine and swine inocula, respectively. There is a decrease in the concentration of this element with the use of bovine inoculum and increase with swine inoculum in relation to increased loads.  The highest average values of N refer to the loads of 1.00 gVS.L -1 .day -1 (359.33 mg.L -1 ) and 1.67 gVS.L -1 .day -1 (367.30 mg.L -1 ) for bovine and swine inocula, respectively. In addition, the values of K and N are below the values found in effluent from poultry litter digestion by Singh et al. (2002): potassium (1632 mg.L -1 ) and nitrogen (1570 mg.L -1 ).
The highest average values of Mg and Na refer to the load 0.67 gVS.L -1 .day -1 , in addition to a decrease in the concentration of these elements with the increased loads. The Ca values refer to the loads of 0.67 gVS.L -1 .day -1 and 1.67 gVS.L -1 .day -1 for the use of bovine and swine inoculum, respectively. An increase in calcium (Ca) concentration was also observed with increasing charges for swine inoculum.
The results presented corroborate with Tessaro et al. (2015), which evaluated the digestion with poultry litter by varying the presence of digestate and water, and concluded that the digestate produced presented assimilable macro and micronutrients by vegetables such as nitrogen, phosphorus, potassium, calcium, magnesium, sodium, iron, boron, copper, zinc and manganese. However, it must be taken into consideration that nutrient loss and segregation processes may occur in the biodigester (Nicoloso et al., 2019).
At the end of the experiment, biodigester sludge was quantified, but not presented differences in relation to the volume of sludge generated in the process (1.40 liters), being characterized in relation to nutrients, as presented in Table 7. The bovine and swine sludge presented very close concentration in relation to the nutrients observed and have a higher concentration of Mn, K and Ca. The use of poultry litter as a digestate is economically desirable, since it represents an internal resource of the rural property. It is a waste that contains a high concentration of nutrients; however, it should be considered that fertilizers with higher proportions of organic nutrients in the organic medium and high lignin and fiber contents have lower decomposition rate in the soil and therefore lower release and availability of nutrients to the plants Nicoloso et al., 2019).

CONCLUSIONS
Comparing bovine and swine inocula, it was realized that both could be used in poultry litter digestion because of their stability, as shown by the ratios VA/TA and IA/PA, as well as by the pH values close to neutrality.
Relatively to the TS and VS removals efficiencies, the greatest mean values (59.46 and 78.21%, respectively) occurred with the swine inoculum, at the 0.67 gVS.L -1 .day -1 load and with the bovine inoculum at the 1.67 gVS.L -1 .day -1 load.
It was observed that the greatest mean values of CODtotal removals referred to the 0.67 gVS.L -1 .day -1 load, with the swine inoculum and 1.00 gVS.L -1 .day -1 , with the bovine inoculum.
It is worth mentioning that the choice inoculum depends on the main purpose process (biogas or biofertilizer production). Therefore, in this case, where the goal was biofertilizer production, the use of swine inoculum is recommended, because it resulted in the best TS and COD removals and higher potassium (K) and calcium (Ca) levels in the generated sludge.