Influence of Anaerobic Digestion Processes on the Germination of Weed Seeds

Anaerobic digestion (AD) produces bioenergy and bio-fertilizer. However, weed seeds can contaminate the substrates for biogas production. This work assesses seed viability of four weeds species origin from Germany and China under AD conditions at 37, 42 and 52 °C with different retention times. All investigated seeds were inactivated after 12 h at 52 °C, 128 h at 42 °C and 512 h at 37 °C under AD treatment. Inactivation duration varying between species and a global test showed that the countries of seeds origin have a significant effect on inactivation times. For two species, by contrast, short treatment times increased the germination rate which increases the risk of weed dispersal. An anoxic water bath treatment with identical temperatures resulted in longer deactivation times, indicating that other factors also play a role in the weed seeds elimination. The results indicate that the risk of weed seed spreading through digestate application can be minimized by an adequate retention time.


Introduction
The potential contribution of biogas to renewable energy generation from various waste and residues is substantial, both in terms of energy supply and in terms of the greenhouse gas emission reduction (Scarlat et al. 2018). Anaerobic digestion (AD) not only offers an opportunity for the generation of renewable energy, but also for nutrient recovery via the production of digestates, which can be used as crop fertilizer (Westerman et al. 2012b). However, the substrates for biogas production, namely livestock manure, energy crops (i.e. maize, whole-crop cereal silage, grass silage, cereal grains, and sugar beets), and agricultural processing waste (FNR 2010), can be contaminated with pathogens and weed seeds. If these survive through the AD process, the application of such contaminated digestates entails the risk of spreading new and invasive species (Baute et al. 2016;Johansen et al. 2013;Westerman et al. 2012b), which can lead to a reduction of crop yield, as they aggressively compete for water, nutrients and sunlight.
In Germany, the implementation of the German Renewable Energy Act (Act on Granting Priority to Renewable Energy Sources, known by its German acronym "EEG") has made a great contribution to the development of the biogas sector (BMWi 2017; Hahn et al. 2014). Up to the end of 2018, there were about 9494 biogas plants, with a total installed electricity capacity of 4.8 GW. Renewable resources are widely used as main substrates for biogas production, accounting for 48.9% of the total substrate input in biogas plants in 2016. Among the renewable resources, maize silage and grass silage are the most dominant substrates, accounting for 69 and 14% of the total mass related substrate input, respectively (Daniel-Gromke et al. 2018;. As one of the largest agricultural countries in the world, China has made efforts to develop its biogas sector. By the end of 2015, investment by the central and local governments supported the completion of 110,975 biogas projects of various types. However, nearly 180 million tons of crop straws, e.g. from maize, rice, wheat, legumes or potato, are not used properly but are burned in the field directly, causing severe air pollution (NDRC and MOA 2017). To avoid being hit hard by smog, China, through the 13th Five-year Plan for National Rural Biogas Development, has set up targets to increase the biogas production capacity for the application of 8.64 million tons of crop straws in order to replace 1.14 million tons of chemical fertilizer by producing 26.5 million tons of digestate by 2020 (NDRC and MOA 2017). By these measures, the application of biomass for biogas generation, as well as the digestate application, will put the agricultural sector at a higher risk if weed seeds, pathogens and other impurities are not well treated. Lately in China, the sizes of biogas plants tend to be larger. Link-ing multiple farms, they normally deliver various types of biomass for feedstock and collect a share of the digestates to spread in their fields as fertilizer. Obviously, this practice increases the risk of spreading harmful organisms (Drosg et al. 2015;FNR 2010;Johansen et al. 2013;Philipp et al. 2005).
(ECHCG), known as barnyardgrass, the yield of maize was reduced to 18% of that of the weed-free control. Amaranthus retroflexus L. (AMARE), known as redroot pigweed, at a density of 4 plants/m row length reduced yield of maize by 5% (Knezevic et al. 1994). Chenopodium album L. (CHEAL), known as lamb's quarters or fat-hen, tested at a density of 277 plants/m 2 , resulted in maize yield losses of up to 58% (Sibugaz and Bandeen 1980). Keller et al. (2014) found that over time in south-western Germany, there was an increasing frequency of summer annual weeds, among which are two species considered to be the most abundant weeds in European maize fields: CHEAL and ECHCG. In China, ECHCG is the most dominant weed in paddy rice farmland, while in maize production, the crops are inhibited by Digitaria sanguinalis (L.) Scop. (DIGSA), AMARE, CHEAL and other species (Su and Ahrens 1997;Zhang 2003). Therefore, assessing weed seed survivability after AD is important for helping to determine the level of risk associated with using digestate.
For germination to occur, many dormant seeds need only to be hydrated under conditions that support metabolism, e.g. a suitable temperature and the presence of oxygen. In anaerobic digesters, the viability of seed declines tremendously over time. Temperature was found to be the most important factor influencing seed survival; lag phase was shorter and declined faster with increasing temperatures (Westerman and Gerowitt 2013). For example, Katovich et al. (2014), found that CHEAL could still have a 12% germination rate after 20 days of AD at 37°C, while AMARE had only a 1% germination rate under the same conditions. The research conducted by Leonhardt et al. (2010) under mesophilic conditions (35°C) showed that AMARE and ECHCG were totally inactivated after three days. CHEAL was more robust and took seven days to reduce the germination rate to 2.5%. The authors also found that AMARE, ECHCG and CHEAL were completely inactivated in one day under thermophilic conditions (50°C). In an experiment by Schrade et al. (2003), at thermophilic conditions (52-55°C), all investigated eight species were completely inactivated in 24 h, while under mesophilic conditions (35-37°C,) it took 2 days to inactivate one species, one week for two species and three weeks for CHEAL. Interestingly, after the first three days of treatment, the CHEAL germination rate increased when compared with its control, then dropped after one week, and was completely inacti- vated after three weeks (Schrade et al. 2003). Similarly, an AD reactor at 50°C inactivated ten weed species in one day, an additional four species in three days and another five species in one week (Leonhardt et al. 2010). Johansen et al. (2013) also found that all examined seven weed species lost their ability to germinate at 55°C after less than two days staying in the fermenters, and at mesophilic conditions (37°C), three out of seven species had completely lost germination ability, while three species still maintained low levels (about 1%) of germination ability after one week. CHEAL survived one week at substantial levels (7%) although after 11 days, germination ability was totally lost. These researches also indicated that the speed of germination of surviving seeds decreased with increasing exposure time to AD in experimental bioreactors due to the absence of oxygen (Westerman et al. 2012a(Westerman et al. , 2012b. Depending on the operational temperature, the hydraulic retention time (HRT) of co-digestion biogas plants using energy crops and livestock manure is usually long. Based on the German Biogas Measurement Program II (FNR 2009), HRT for the 61 measured biogas plants varied between 29 to 289 days. Ruile et al. (2015) found 21 German agricultural biogas plants with HRT varying between 20 to 200 days. However, in continuous flow, stirred digesters, it is possible that fractions of feedstock and the impurities contained in them find a short cut through the digester (AL Seadi and Lukehurst 2012), which may cause a risk if the digestate is applied to the agricultural field directly after flowing out from digesters.
The purpose of this study was to assess the ability of seeds from selected weed species to survive the AD processes and to evaluate the risks of digestate application as crop fertiliser. In contrast to previous studies, short retention times were also considered. Therefore, the influence of temperature and retention time in AD systems on the germination rate was investigated. Furthermore, the same weed species from different regions (Germany and China) were tested to evaluate the effect of the origin of the weed seeds on the germination rate. To test for possible AD process effects apart from temperature, absence of oxygen and treatment time, an anoxic water bath treatment with corresponding temperatures was applied to one weed species.

Weed Seeds Selection
Four common, resistant, and globally distributed weed species among maize fields and grasslands both in Germany (DE) and China (CN) with different characteristics were chosen for these experiments. Weed species were referred to their EPPO codes (EPPO 2019). The selection of these species was based, firstly, on their occurrence in farmland, such as maize and grassland, both in Germany and China. Secondly, the selection of species was based on the seed availability in the laboratory. Table 1 summarizes the details on the seeds, EPPO codes, country of origin and year of receipt where the year of receipt does not necessarily mean the year of harvesting. All German seeds were provided by the Institute of Phytomedicine, University of Hohenheim. Chinese seeds were obtained from Shandong Sino-agri Union Biotechnology Co., Ltd, China.

Laboratory Fermenter
The anaerobic digestion experiment was performed in two 400 L liquid volume capacity horizontal digesters each located at the State Institute of Agricultural Technology and Bioenergy at the University of Hohenheim, Stuttgart ( Fig. 1) (Stockl and Oechsner 2012). Each digester is equipped with one central stirring mixer, one feeding pipe, one outlet pipe for the substrates, and four perforated tubes specifically designed for the seed unit exposure to the fermentation liquid (anaerobic conditions). The digesters were heated by warm water pumped from a thermostatcontrolled water bath (SE, Julabo, Seelbach, Germany) and circulating inside a water jacket. Four temperature sensors (EBI 40 TC-01, Ebro Electronic, Kirchheim, Germany) were placed to measure the temperature in each of the four tubes in an interval of 15 min. For measuring the gas quality, double-beam NDIR gas sensors for CH4 and CO2 (Advanced Gasmitter, Pronova, Berlin, Germany) were employed.

Anaerobic Digestion Operation Conditions
In the present study, three different temperatures (37, 42 and 52°C) were applied, of which 37°C is more commonly used in Chinese biogas plants and 42°C is commonly used in Germany (FNR 2010). The thermophilic temperature (52°C) is also often applied in Germany but not as frequently in China.
The digesters were inoculated with fermentation slurry from the secondary fermenter of the full-scale research biogas plant "Unterer Lindenhof", University of Hohenheim. Substrates, namely maize silage with dry matter (DM) content of 38.12 ± 1.04% fresh matter (FM); and organic dry matter (ODM) content of 36.71 ± 0.98% FM and slurry with DM of 6.71 ± 1.39% FM and ODM of 4.56 ± 0.98% FM (all data reported as averages ± standard deviation wherever possible) were collected and stored in the cooling room (4 ± 1°C) in the laboratory of the State Institute of Agricultural Technology and Bioenergy. Digesters were fed 2 kg of maize silage and 4 L of slurry per day, similar to practical conditions, which led to a daily organic loading rate of approximately 2.5 kg ODM/(m 3 · d). Before the start of the experiment, two weeks of initial operation served to stabilize the AD process in each digester.
The substrates and fermentation liquid samples were analysed for DM and ODM content twice per week. DM and ODM concentrations were determined in accordance with DIN EN 12880 and DIN EN 12879 (German Institute for Standardization 2001a, 2001b). The pH values of fermentation liquid in the digesters were measured using a digital pH meter (pH 330, WTW, Weilheim, Germany) once per day by taking samples from the digestate outlet. The content of volatile fatty acids (VFA) was determined by capillary gas chromatography (GC) (type CP3800 with an FID-detector, capillary column WCOT Fused Silica, Agilent Technologies Germany GmbH, Böblingen, Germany), as detailed in Haag et al. (2015) and Lindner et al. (2016). The ratio of volatile organic acids (VOA) and to-tal inorganic carbon (TIC) (VOA/TIC), was obtained by determining acetic acid equivalents of the fermentation slurry through an automatic titration system (785 DMP Titrino-Metrohm Filderstadt) after the samples were centrifuged in a refrigerated centrifuge at 5000rpm (Z323K, Hermle-Labortechnik, Wehingen, Germany). A Kjedahl apparatus (Vapodest ® 50s, Gerhard Analytic System, Königswinter, Germany) was used for the measurement of total Kjeldahl nitrogen (TKN) and ammonium (NH4 + ) (Chen et al. 2014;Lemmer and Krümpel 2017;Ruile et al. 2015). A gas measuring unit automatically analysed the biogas content of CH4 and CO2 (D-AGM Plus, Sensors Europe, Germany) every two hours (Nägele et al. 2017).

Seed Unit Preparation
Prior to the experiment, seeds were sterilised in a 70 vol.% ethanol solution for 1.5 min, rinsed in distilled water three times, and stored in a refrigerator at 4°C after air drying at room temperature. Only seeds that appeared viable were used. Visually damaged seeds or immature seeds were removed. A total of 100 seeds of each species were carefully counted and placed on 5 * 5cm 2 fabric material and sewed into a bag tightly to avoid any seed loss in the digesters. Different colours of the sewing thread (polyester) corresponded to different seed species to ensure accurate identification. After preparation of all the seed bags, four replicates of each species were attached to one end of a wood stick (80 * 2 * 0.5 cm). Seed bags were randomized on each stick by using the RAND function in EXCEL, conspecific German and Chinese seed bags were placed in pairs at the same depth on the stick to ensure equal AD liquid exposure. Each wood stick with seed bags is called a seed unit, which represents a single treatment with one seed exposure time.
Once the digesters reached stable operation, four seed units were inserted into each digester tube and all seed bags were entirely in contact with the AD liquid. After seed unit insertion, a rubber stopper was placed in each tube to ensure both complete wood stick submersion and total seed bag exposure to anaerobic conditions. Depending on the operating temperature, different treatments and sampling dates and frequencies, ranging from short periods to long periods were tested. For the experiment at 37°C, eight retention times, namely 16,24,32,64,128,256,336 and 512 h were chosen. To German DIGSA, an additional treatment of eight hours was added. Each seed species was subjected to eight retention times, namely 8, 16,24,32,64,128,256 and 336 h at 42°C, and six retention times (2, 4, 8, 12, 16 and 24 h) at 52°C. As control, 100 seeds per treatment were put aside untreated for subsequent germination tests.
After taking the sample, the seed units were retrieved and quickly flushed with tap water. After the seed bags were removed from the sticks, the seeds from each bag were counted to check if any of the seeds were lost in the digesters. None of the seed bags was broken and all 100 seeds were recovered from each bag for the germination tests.

Water Bath Treatment
To investigate the effect of anaerobic conditions without the influence of microorganisms and chemical environment of the biogas digester, a water bath treatment with seeds of German DIGSA (DIGSA-DE) was additionally conducted. The preparation of seed bags and the germination test were the same as for AD treatment. DIGSA-DE seed bags were placed in glass syringes (100 ml) with 70 ml distilled water (pH = 8.24) in a motor-driven rotor, which is located in an incubation chamber in the Hohenheim Biogas Yield Test laboratory of the State Institute of Agricultural Technology and Bioenergy (Helffrich and Oechsner 2003;Mittweg et al. 2012). In total, three incubation chambers were used for temperatures of 37 ± 0.5°C, 42 ± 0.5°C, and 52 ± 0.5°C, with identical number of treatment times as in the AD treatments. As with the AD experiments, each treatment was conducted with four replicates, and seed bags were transferred to the germination test afterwards.

Germination Tests
Seeds were germinated following an ISTA test standard (ISTA 2018) at the Institute of Phytomedicine, University of Hohenheim. Before starting the experiments, preparatory germination tests were conducted to identify the viability of seeds, proper growing media, and duration of the germination test.
Following the ISTA standard, four replicates of 100 seeds per treatment were placed on Ø90 mm petri dishes, 4-5 ml distilled water for DIGSA and ECHCG, and 0.05% GA for AMARE were added by using a pipette (D-5000, Transfer-pette S, BRAND, Wertheim, Germany) on Ø90 mm filter paper (MN 615 Ø90 mm, MACHEREY-NAGEL, Düren, Germany). After placing the seeds evenly on the filter paper, petri dishes were sealed with plastic paraffin film (4 IN.X 250 FT. ROLL, Parafilm M, USA) to ensure uniform moisture conditions. For the germination test of CHEAL, 0.05% of agar (Agar-agar, Kobe 1, Carl Roth, Karlsruhe, Germany) was used as growing media instead of solution. Petri dishes were then placed in a germination incubator (BINDER, Germany) randomly in each level under uniform lighting and temperature, namely12 h day time (DT, 6 am-18 pm) at 22 ± 1°C, 12 h night time (NT, 18 pm-6 am) at 15 ± 1°C. The first counting was conducted after 7 days and the second counting was conducted after 14 days. Seeds were considered germinated when the emergent radicle had reached a length of 1 mm (Guo and Al-Khatib 2003). Seeds that germinated were counted and removed.

Calculations and Statistical Analysis
The records of DM, ODM, VFA, VOA/TIC, NH4-N, TKN, gas qualities and pH-value were saved in a database, averages and standard deviations (SD) of three or more replicates were calculated in Excel.
The analysis of the impact of retention time on germination rate was based on the assumption that the conditions in the digesters are homogenized. In order to verify this assumption, temperature recordings (recorded in an interval of 15 min) taken at each of four tubes of a digester were used as the response, and a mixed model was fitted with a fixed effect for tube and time and an AR(1) model with nugget for residual error in SAS. This covariance structure accounts for serial correlation among observations on the same tube. For comparison, a model with independent residual error was also fitted, showing that the AR(1) model provided a better fit based on the Akaike information criterion (AIC).
Original germinated seeds were counted and recorded. For the evaluation of the germination tests, all seed germination rates were normalized so that the viability was 100% for the controls (Baute et al. 2016), as described by Eqs. 1 and 2. The germination rate of seeds was computed as: Germination rate of seeds.%/ = N 1 + N 2 Total number of seeds 100 (1) where N1 is the number of germinated seeds at the first counting, N2 is the number of germinated seeds at the second counting. The normalized germination rate was defined as: (2) Statistical analyses of the germination rate were performed by using the software R Studio. Boxplots were obtained to display the distribution of data (germination rate as per treatment of each species) and the full range of variation (from min to max).
To analyse the significance between treatments and country of origin, a generalized linear model with binomial distribution and identity link was used. This analysis accounts for heterogeneity of variance between treatments. For each species temperature treatment, we fitted a two-way model with qualitative factors country of origin and retention time. For each country of origin, pairwise comparisons among retention time means were performed by using Wald-type t-tests (α = 0.05). A global test was also performed for differences among countries of origin based on the nested model with effects for retention time and country of origin nested within retention time. Significance (α = 0.05) of the latter effect indicates global differences between countries of origin. These analyses were performed using the GENMOD and GLIMMIX procedures of SAS (Piepho 2018).

Digester Stability
Parameters in Table 2 characterize the performance in the digesters during the entire experimental period. The VOA/TIC value among all digesters did not exceed 0.3 and stabled around 0.2 at any working temperatures during the operation, which means digesters in this study were characterized by a stable performance with working temperatures (Chen et al. 2014). In terms of homogeneity of temperatures in the digesters, the F-test for factor tube was not significant (α = 0.05) under AD treatment at the temperature of 37°C. For digesters under 42 and 52°C, the largest estimated difference was 0.6°C, which is small but significant. The temperature difference was mainly caused by the feeding of substrates, which was stored in the cooling room (4 ± 1°C). It is concluded from these results that conditions within digesters can be considered homogeneous for most practical purposes.

Amaranthus Retroflexus L.
Through AD treatment at 37°C, AMARE-DE and AMARE-CN were inactivated after 64 h (1 ± 1%) and 32 h (4 ± 3%), respectively. However, AMARE-CN was more robust for the first 16 h as its viability was 46 ± 7% as opposed to 8 ± 6% for AMARE-DE (Fig. 2a,b). As shown in Fig. 2c,d, AMARE-CN and AMARE-DE were inactivated after 24 and 16 h at 42°C, respectively. Under mesophilic conditions, AMARE-DE was inactivated quicker than AMARE-CN. Under AD treatment at 52°C, AMARE-CN was inactivated after four hours while AMARE-DE was completely inactivated after two hours (Fig. 2e,f), which is similar to the results found by Knödler (2015), showing that AMARE seeds were totally inactivated after four hours at 52°C. Westerman et al. (2012b) found that after its batch AD test at 37°C, no AMARE seeds survived after 30 days. However, in that research, only one treatment time (30 days) was tested. It could be possible that the actual inactivation time is shorter. A similar experiment with AMARE was conducted by Leonhardt et al. (2010) at an AD mesophilic temperature of 35°C, and it was found that  all AMARE seeds were killed after one week. The fact that the inactivation lag phase was longer than in the present work could be due to the lower temperature used. The authors also found that AMARE seeds were completely inviable after one day at 50°C, indicating that thermophilic conditions jeopardize the weed seeds quicker.
After AD treatment at mesophilic conditions, viability differed between AMARE-CN and AMARE-DE, especially after a short retention time. For example, with 16 h at 37°C and eight hours at 42°C, AMARE-CN still had significant viability, reaching 46 ± 7% and 63 ± 5%, respectively, indicating that the resilience of AMARE to AD will differ depending on its country of origin. However, in thermophilic conditions (52°C), the viability between AMARE-CN and AMARE-DE was not significantly different after two hours under AD treatment, which indicates that temperature plays a key role in causing the inactivation of seeds. The significance test for "species*treatment" indicates that AMARE seeds have global differences between countries of origin. This is supported by Bewley et al. (2013), who showed that even members of the same seed family differed in their habitat and temperature requirements for germination.

Chenopodium album L.
Compared with the other three seed species, CHEAL seeds were the most robust under AD treatment. CHEAL-CN was inactivated after 512 h at 37°C, and inactivation occurred after 128 h under AD conditions for CHEAL-DE. Interestingly, we found that after 16 h, the germination rate of CHEAL-DE increased by 19% while this did not happen to CHEAL-CN (Fig. 3a,b). As shown in Fig. 3c,d, under AD treatment at 42°C, CHEAL-CN germination was entirely inactivated after 128 h while CHEAL-DE was inactivated after 64 h. Under AD treatment at 52°C, both CHEAL-CN and CHEAL-DE were completely inactivated after eight hours (Fig. 3e,f). Johansen et al. (2013) indicated that CHEAL seeds were eliminated after 11 days at 37°C, which closely resembles the results found in the present study for CHEAL-CN. Research conducted by Knödler (2015) showed that CHEAL seeds were totally inactivated after two weeks at 37°C, after one week at 42°C and in one day at 52°C. Furthermore, Leonhardt et al. (2010) found that CHEAL seeds were killed after 21 days at AD thermophilic 35°C and seven days at 42-45°C in a commercial continuous stirred-tank reactor, while completely inactivated at 50°C in one day. This was supported by similar results found in earlier studies (Gansberger et al. 2009;Hahn et al. 2018;Schrade et al. 2003). Katovich et al. (2014) pointed out that in a plug flow reactor under AD treatment at 37°C, CHEAL seeds still had 12% viability after 20 days, suggesting that different AD technology could have an impact on the seeds inactivation (Westerman et al. 2012b). After the AD treatment at 37 and 42°C, CHEAL-DE germination rate dropped down quicker than CHEAL-CN, hence it is higher sensitive under the mesophilic condition. By contrast, CHEAL-CN had a lower germination rate than CHEAL-DE at 52°C, though these species were both inactivated after eight hours. CHEAL-CN is more robust than CHEAL-DE under mesophilic conditions. Among the same treatment, analysis also shows that several replicates have high SD, which can be caused by the impact of thickness of seed coat the selected seeds (Westerman et al. 2012b).

Echinochloa crus-galli (L.) Beauv.
For the AD treatment, after 16 h at 37°C and eight hours at 42°C under AD conditions, ECHCG-DE had a similar germination rate as non-treated seeds, and it was inactivated at 64 h under the mesophilic conditions (Fig. 4b,d). As shown in Fig. 4c,d, under AD treatment at 42°C, ECHCG-CN was inactivated after 24 h while ECHCG-DE was inactivated after 32 h. With the AD treatment at 52°C, ECHCG-CN and ECHCG-DE both were completely inactivated after eight hours (Fig. 4e,f). Leonhardt et al. (2010) found that ECHCG seeds exposed to AD mesophilic conditions between 42 and 45°C were inactivated after 3-7 days depending on which AD plant was used and 1 day at a thermophilic temperature of 50°C regardless of the plant. In the experiments carried out by Knödler (2015), ECHCG seeds were totally inactivated after three days at 37°C, two days at 42°C and 1 day at 52°C while after a four hours AD treatment, the germination rate increased by 5% at 52°C. Westerman et al. (2012a) found that in a batch reactor at 37°C, ECHCG seeds have no viability after 30 days. However, Katovich et al. (1993) studied the fate of seeds of eleven weed species after passage through a dairy cow followed by AD process in a commercial batch digestion unit and showed that ECHCG still had significant viability of 36% after 30 days. Such inactivation time is far longer than what has been found in the current work either due to the experimental set up (intervals of treatment time) or the chemical environment in the digesters.
Under AD treatment, ECHCG-DE had much higher viability than ECHCG-CN under the same treatment conditions. This indicates that ECHCG from different countries of origin differed in sensitivity to AD conditions.

Digitaria sanguinalis (L.) Scop.
For the AD treatment at 37°C, both DIGSA-DE and DIGSA-CN were inactivated after 64 h (Fig. 5a,b). Engeli et al. (1993) showed that DIGSA seeds lost viability in nine days undergoing AD treatment at 35°C. For the water bath treatment at 37°C (Fig. 6a), the germination rate of DIGSA-DE remained high up to 89 ± 4% after 64 h and inactivated completely after 256 h, much longer than that in the AD treatment (64 h). Viability in DIGSA-CN decreased slower than in DIGSA-DE, and in general, was more resilient to AD conditions under 37°C.
As shown in Fig. 5c,d, under AD treatment at 42°C, DIGSA-CN and DIGSA-DE were almost completely inactivated after 32 h. The germination rate of DIGSA-CN even increased by 6 ± 9% after an eight hours AD treatment.  With the water bath treatment, DIGSA-DE germination was completely inactivated after 128 h (mean normalized germination rate is 2 ± 2%) (Fig. 6b).
For the AD treatment at 52°C, DIGSA-CN and DIGSA-DE both were inactivated after eight hours. Surprisingly the germination rate of DIGSA-CN slightly increased by  With the water bath treatment at 52°C, DIGSA-DE germination rate increased by 18 ± 8% compared to the control in a two hours anoxic treatment but was inactivated completely after 12 h (no significant difference as after 16 h) (Fig. 6c). A similar experiment was conducted by Hahn et al. (2018) with 11 species at 42°C in a buffer solution (pH = 7) and in an AD reactor for 18 days. They found that there are significant differences in mean inactivation time of four species between buffer solution and AD conditions.
It was also observed that, after eight hours under mesophilic conditions and two hours under thermophilic conditions, the germination rate for DIGSA-CN increased  from 1 to 9% compared to the controls. This might be caused by breaking of dormancy as DIGSA favours warm conditions. However, this situation has not been observed for DIGSA-DE except under the water bath treatment. The inactivation time of DIGSA-DE is much longer than that of the AD treatment for both mesophilic and thermophilic conditions. Therefore, temperature and retention time are likely not the only factors affecting weed seed inactivation. Other factors may also play a role in eliminating weed seeds during the AD. After the AD treatment, germination rate impact differed between DIGSA-CN and DIGSA-DE,

General Discussion
Previous studies showed the AD impact on the viability of weed seeds in a relatively long retention time but lacked short exposure times. However, in continuous flow, stirred digesters, it is possible that fractions of feedstock (and the impurities contained in them) find a short cut through the digester (Seadi and Lukehurst 2012), which may cause a risk if the digestate is applied directly after flowing out from digesters. This work shows that after a short period of 8-16 h exposure, especially at mesophilic conditions, the four selected weed seeds, both from Germany and China, all have a relatively high germination rate, or even an increased one due to the breaking of seed dormancy (i.e. DIGSA-CN, CHEAL-DE). Furthermore, according to the anoxic water bath experiment on DIGSA-DE, the viability of such seeds is significantly higher than under AD conditions up to 128 h at mesophilic conditions (37 and 42°C) and 12 h at thermophilic conditions (52°C) ( Table 3). This indicates that other factors, such as microorganisms and chemical environment, in addition to temperature and retention time, also play a key role in inactivating weed seeds under AD conditions.
According to the revised German Renewable Energy Sources Act (BMWi 2017), in case of installations commissioned after 31 December 2016, and digestate storage facilities constructed after 31 December 2011, the hydraulic retention time in the entire gas-tight system which is connected to a gas consumption device, is prescribed to be at least 150 days. The application of this legal binding clause may significantly reduce the risks of spreading weed seeds on agricultural fields through digestate application.
In biogas plant operation, usually biomass ensilaging and post-digestion storage steps are involved, which can initially or eventually eliminate more weed seeds (Aper et al. 2014;Haag et al. 2015;Westerman et al. 2012b). Furthermore, the experimental design of this study included shorter retention times than that of typical ones in agricultural biogas plants as found by Ruile et al. (2015). In general, the risk of weed spreading is low. However, regulations such as minimum HRT or digestate storage duration times do not currently exist in China, which potentially implies risks if farmers apply digestate directly from the digesters to the fields. Given that the Chinese government will provide significant financial support to the development of the biogas sector without implementing operational guidelines, the risks are likely to increase. China has also increased its input of crop based biomass used as substrate in the biogas plants which will increase the input of weed seeds into the system (Zhao 2018).
According to the results in the present work, loss of seed viability is slower at the mesophilic range than at the thermophilic range. In general, the higher the temperature the Table 3 The significance of differences in germination rate for seeds of German Digitaria sanguinalis (L.) Scop. under anaerobic digestion conditions and anoxic conditions in water Temperature Retention time (h)  2  4  8  12  16  24  32  64  128  256 37°C --*** -*** *** *** *** *** NS 42°C --*** -*** *** *** *** ** NS 52°C *** *** *** * NS NS ---- NS not significant, -not measured *P < 0.05; **P < 0.01; ***P < 0.001 shorter the period required to reach the complete inactivation of seeds. However, for all investigated species, the proportion of viable seeds differed significantly between treatments and countries of origin. Furthermore, the anoxic water bath experiment shows that, at least for DIGSA-DE seeds, apart from temperature and absence of oxygen, other parameters may also play a key role in the inactivation of the seeds, which agrees with results of Schrade et al. (2003) and Hahn et al. (2016). Such parameters could be, for instance, the pH value, the concentration of organic acids and enzymatic effects in the fermentation liquid in contrast to the anoxic water bath. A future study investigating these parameters would be very interesting.

Conclusions
It is concluded that through the AD, besides receiving energy and high-quality digestate, there is an additional benefit of weed seed inactivation. The elimination efficiency towards these investigated weed seeds depends on the combination of temperature, retention time and other factors during AD processes. During the practical operation of biogas plants, special attention should be paid to avoid direct application of digestate after flowing out from digesters due to short-circuit flows. With the typical mesophilic AD temperature, a safe and complete elimination of weed seeds is to be expected within suitable retention times.