**4. Discussion**

#### *4.1. Influence of IFP on Maize Cropping*

The following discussion will clarify the suitability of each IFP and offer suggestions for maize intercropping.

The single IFP provided a flower abundance for different periods of time and offer a food source for different insect species. This can make a major contribution for biodiversity promotion. With a flowering start before maize anthesis, alfalfa is an interesting partner to increase the flower abundance in maize crops. The literature showed that alfalfa is pollinated by wild bees, provides a pollen source for 29 species of wild bees and is a food resource for bumblebees [45–47]. Therefore, maize–alfalfa intercropping creates a flower/food abundance over a very long period. Yellow sweet clover per se is a pollen source for six wild bee species [46], which would have made this IFP interesting for biodiversity conservation. However, the results showed that yellow sweet clover did not flower at any location during the two years of the experiment. Literature showed that yellow sweet clover is not able to flower when shaded [48]. Additionally, this species has single- and two-year genotypes; the single year genotypes will already flower in the year of sowing and the two-year genotypes will only flower in the second year [49]. Since sweet clover is a plant species that has hardly been researched in breeding, no guarantee can be provided which flowering genotype or which mixing ratio the seed contains. The earliest flower abundance after sowing was supplied by common vetch, even earlier than alfalfa. The flowering period ends at about the same time as maize anthesis starts. Common vetch is pollinated by wild and honeybees, bumblebees, and is also a food resource for bumblebees [45–47]. Nasturtium, with a long flowering period which continued after the silage maize harvest, provides an interesting aspect for biodiversity. It will flower until the first frost or until soil tillage [50]. Therefore, it might be a habitat over a long time. A maize–nasturtium crop flowers 50 DAS, two weeks before the flowering of a sole maize crop. Nasturtium is pollinated by honey and wild bees, and also *Syrphidae* used it as a host [51]. In the region of maize origin, the traditional cropping system for maize was a combined cultivation of maize with summer squashes, common beans and others, the so called MILPA system. The summer squashes are pollinated by honey and wild bees and tolerate some shading [45,52], which makes them interesting for maize intercropping for biodiversity reasons. Flowering starts quite late, 1–2 weeks before maize anthesis, but as already observed for nasturtium, it continues even after harvest, until frost or soil tillage. Intercropping of maize and common beans is already done in practice. In 2019, 4000 ha were cultivated in Germany [53]. Common beans as the other main partner in the MILPA system are mainly self-pollinating, although insect pollination can enhance the seed yield [54–56]. Wild bees, butterflies, flies and bugs were observed to pollinate common beans [57]. The two-year experiments showed that common beans tend to a quite late flowering start. Mostly they started flowering after the maize started. Both mixtures (MM1, MM2) combined the mentioned flower abundance characteristics of two di fferent IFP; summer squash II and common bean I (MM1), common vetch and common bean I (MM2). Especially for MM1 this leads to a prolonged flower abundance after end of maize anthesis by the squash, MM2 leads to an early flower abundance due to the common vetch.

An important aspect in maize intercropping is a system adapted weed control. Weed compete with maize (a water requiring crop) for nutrients and water, especially in dry years. In most tropic and sub-tropic countries, where intercropping in smallholder farm systems is the main agricultural practice, weed managemen<sup>t</sup> is done by hand. However, if maize intercropping should take place on larger scales for a high economic performance, a weed control by common agricultural techniques must be possible. If weed control cannot be done, weeds form a major competition for maize, especially at sites with a high weed infestation. This has been shown for several IFP. Intercropping maize with alfalfa or sweet yellow clover leads to a weed share of 12.5% and 10.8% at ET, mainly consisting of *C. album*. ET is a typical maize location. Redwitz and Gerowitt [58] showed that fields that had, in previous years, a high share of maize in the crop rotation have an increased potential for infestation with *C. album*. The chemical weed treatment under conventional conditions consisted of a reduced amount of *Pendimethalin.* According to the manufacturer, the practical application rate of *Pendimethalin* should not be less than 3.5 L ha−<sup>1</sup> (455 g a.i. <sup>L</sup>−1) for an adequate *C. album* control in both pre- and post-emergence. It is also highly recommended that the application should take place not later than the three-leaf stage of the broadleaved weeds; most e ffective is the pre-emergence application [59]. Additionally, at TH the weed potential was not that high as in ET. Under organic conditions hoeing covered the alfalfa and the sweet yellow clover with soil, which the small plants did not tolerate. Very few or none of the plants managed to grow from the heaped-up soil. The opposite was observed for maize–common vetch intercropping, where an adequate weed managemen<sup>t</sup> is only possible under organic management. The hoeing in FAK worked well and the common vetch showed no di fficulties growing out of the heaped-up soil, otherwise than alfalfa or yellow sweet clover. Under conventional management, especially at ET with its high weed infestation potential, weed managemen<sup>t</sup> was di fficult. In the first experimental year at ET, maize–common vetch plots had a high infestation with *C. album*. Although the same herbicide application as for maize–alfalfa and maize–yellow sweet clover took place in the second year in post-emergence, weeds could not be controlled successfully, for the reasons stated above. At TH, with a low infestation potential, the weed control used worked well. The advantage of nasturtium over alfalfa, yellow sweet clover and common vetch is the availability of an adapted weed managemen<sup>t</sup> by a pre-emergence application of *Pendimethalin*, which controlled the weed infestation better, than a post-emergence application. The reduced amount of active ingredient can be a disadvantage in this intercropping system, especially on sites with a high weed potential. Under organic weed management, the nasturtium showed no negative e ffect when buried under heaped-up soil during mechanical weed management. Afterwards, the nasturtium plants grew up from the heaped-up soil. Plant protection in maize–squash intercropping is practicable under organic conditions, while the chemical plant protection is challenging. Most of the registered maize herbicides, which are also allowed in squashes, can only be applied as inter-row application, which requires special equipment. For common beans, mechanical weed control or the application with *Pendimethalin* and *Dimethenamid*-P enabled an e ffective weed control.

To achieve high DMY, a good maize development and growth must be enabled. An important growth parameter for high DMY is the final plant height of the maize [60–64]. Intercropping of maize with alfalfa mostly showed no di fferences in plant height, except for TH in 2019. Studies from Canada also showed no change in plant height by intercropping maize with alfalfa compared to a sole maize stand [60]. This was confirmed by the DMY results 2019 in TH. On the other hand, ET showed a reduction in DMY but no change in plant height. This can be attributed to the high weed infestation (due to the reduced amount of active ingredient and late application date) and the resulting competition by a high weed share [65]. For grain maize–yellow sweet clover intercropping, Abdin et al. [60] showed that no reduction in plant height is expected. This was confirmed by our results, only TH 2019 forms an outlier with a significantly reduced maize plant height. In maize–yellow sweet clover, plant height was not a ffected by IFP, but the same e ffect on weeds was observed as mentioned above under maize–alfalfa intercropping. The non-significant change in plant height and DMY also make nasturtium an interesting partner for intercropping in maize. Also, no changes in DMY compared to sole maize cropping were found for intercropping maize with common beans. Experiments from Northern Germany also showed no di fferences in DMY [35]. In Great Britain, no significant di fferences in DMY between maize (10 plants m<sup>−</sup>2) and maize–bean (7.5:5 plants m<sup>−</sup>2) were observed [31]. In contradiction to the four previously mentioned intercropping treatments, intercropping with common vetch showed reduction in the final plant height of maize, depending on the year. The reduction in plant height was not significant in the first year at the locations in our study. In the second-year, significant height reductions for maize plants intercropped with common vetch were observed at TH and FAK. Also, the DMY was reduced. Common vetch is a plant which leaches allelopathic substances during decomposition of its biomass [66]. In 2018 it was observed at all locations that, after a short time period, the vetch was infested with mildew and died. Root excretions and leaching's (vanillin acid, p-coumaric acid, ferulic acid) of above-ground common vetch biomass is shown to inhibit the germination of wheat, as well as plant growth and development [66]. Their study concluded that wheat produced more root biomass than above ground biomass, due to the allelopathic substances. This change in above/below biomass ratio is a reaction which should promote plant growth. Since these e ffects has also been observed in rice [67], it is obvious that such e ffects also might occur in maize, which, like wheat and rice, belongs to the Poaceae. The squashes showed di fferent growth behavior and influence on DMY, which indicates a grea<sup>t</sup> variability in the species, which could be

shown by squash III in FAK 2019. Squash I and II showed no change in maize plant height and DMY, but squash III in FAK reduced the final maize plant height. While squash I and II produced fruit weights of 0.1 and 0.2–0.4 kg respectively, squash III can achieve fruits of 2–4 kg [68]. The production of fruits with a high weight requires nutrients and water, which are no longer available for maize. To the best of our knowledge, there are no experiments on maize–squash intercropping for silage usage. Most maize–squash experiments are done with maize for grain production. A study from traditional smallholder farm systems in Africa showed, that there were no di fferences in grain yield by intercropping with squashes; as long as the sowing rate of squashes meets 20% of the maize sowing rate [69]. Our experiment used 1.6 seeds m<sup>−</sup>2, which equals 20% of the sowing rate of 8 maize seeds m<sup>−</sup>2. Therefore, no changes were expected, which was proved by no changes in DMY. But the study from Mashingaidze et al. [69] also showed, that no di fferences in grain yields can only be achieved in years where a high infestation with *Sphaerotheca fuligniea* took place. The infestation inhibits the formation of fruits. In years without infestation, fruits are formed, and the squash plants compete with maize for nutrients. This leads to the conclusion that larger fruits have a higher competitiveness and, therefore, a greater influence on maize. This is in line with the results of squash III. This squash formed bigger fruits than the other two squashes and showed significant reductions in final DMY of the MSIII plots and yield relevant parameters. Also, the sowing of a mixture of maize and squash is a challenge. The 1000 grain-mass varies widely within the genus Cucurbita. While squash I and II had a 1000-grain mass of 55 g and 27 g, respectively, squash III had 102 g and was suitable for a combined seeding with maize as a mixture.

Prevention of the water bodies due to soil NO3-N leaching by IFP could not be fully proven. Under maize–alfalfa intercropping the soil NO3-N content was significantly reduced in the layer of 30–60 cm depth compared to the control at vegetation end in ET 2018. It could be observed that the alfalfa continued growing after harvest. It is therefore reasonable to assume that the IFP continued absorbing soil NO3-N. Although, nasturtium continues to grow after harvest and provides a flower abundance, it did not show significant di fferences in the soil NO3-N content in the single experimental years or sites, except for a reduction in 30–60 cm at the end of the vegetation period in ET 2018.

The IFP or a 'pollution' by a high weed share can also cause changes in the chemical composition of the harvested material. Most often the CP and CA content were changed. For biogas production and cattle feeding, the following important parameters should also not change to a disadvantageous content: CP (too long decomposition time in biogas plant, in dairy cattle to much CP causes high milk urea concentration, or CP is sometimes degraded to fast and, therefore, has a low utilization), CX (di fficult digestion in biogas plant and dairy cattle stomach), CA (reduction in digestion space of the biogas plant and no substances that can be converted for energy and feeding purpose). An increase in NfE has no negative e ffects, due to the fact that highest methane concentration can be achieved by this parameter and cows can convert most energy from these, respectively. The matrix of correlation coe fficients in ET for 2018 showed that there is a negative correlation of CP on the NfE content (rp = −0.68) (Figure 5). This trend can be explained due to the fact, that NfE are the nitrogen-free extracts. If there is an increase in CP, less amount of the dry matter can consist of NfE. This was proven 2018 in ET where intercropping with common vetch increased the CP contents significantly, while the NfE decreased significantly. Such a trend could also be observed in TH 2019 for mixture I (MM2), which also contains vetch. CP can build high contents of methane, but it is a slowly degradable ingredient in biogas plants. This has been shown by the correlation coe fficients for biogas (rp = −0.76) and methane (rp = −0.47). A high content in CP reduces the biogas contents due to the long retention time. But the higher the content of CP, the more methane could be built. Also, the CA content influences the final yields of biogas and methane. If the content of CA in the biogas substrate is too high, ash can settle at the bottom of the fermenter, reducing the digestion space and thus the yields. Therefore, the matrix showed a significantly negative influence of CA on biogas (rp = −0.90) and methane (rp = −0.75). Results from 2019 in TH verified these findings. An increase in CA by intercropping with common bean I leads to increased CA contents and decreased biogas and methane contents. The nutrition parameters GE, ME and NEL are mainly influenced by CP and CX. CX showed a negative influence on GE (rp = −0.50), ME (rp = −0.99) and NEL (rp = −0.99), CP had a negative influence on ME (rp = −0.50) and NEL (rp = −0.50). This negative influence of CX was also proved, in 2019 in TH, an increase in CX by intercropping with common bean I significantly reduced GE, ME, and NEL due to the low nutritive value of CX.

**Figure 5.** Pearson's coefficient of correlation within and across yield and quality parameters at ET in 2018, averaged over the parameters N-Level, seed placement of the intercropped flowering partners (IFP) and IFP. Non-significant coefficients (*p* < 0.05) are crossed out.

Significantly higher CA contents were observed for ET and FAK in 2018 under maize–alfalfa, maize yellow sweet clover and maize–common vetch intercropping. This can be attributed to the high share of weed in the harvested biomass. Most of the weed consisted of *C. album*, which has a CA content of 23.3% [70]. Adedapo et al. [70] also showed that *C. album* has high contents of CX (16.7%) and CA (23.3%). An increase in these two parameters has negative effects on biogas and methane yield, and on the feed quality parameters, GE, ME and NEL. This was proven by the significant reduction in biogas, methane, GE, ME and NEL in ET 2018. High contents of CA can disturb the biogas process by settling down on the bottom of the fermenter and reduce the space for digestion. In addition, CA does not provide a basis that can be converted into energy, neither in biogas plants nor in cattle feeding. These effects were proven by the highly negative correlations between CA and CX on biogas, methane, GE, ME and NEL shown in Figure 5. Biogas and methane formation depend on the composition and biodegradability of the substrates used [71]. CX did not contribute much to methane yield [72].

The increased CP content could also be attributed to the high weed infestation. In ET 2018, the weed covered area was higher than 50%. *C. album* has a high CP content [73,74]. Depending on plant age, *C. album* can accumulate an additional 1.75–5.27% nitrogen, which corresponds to 10.9–32.9% CP [73,74]. The *C. album* will be chopped together with the maize. A study by Sarabi et al. [75] showed that the control of *C. album* in maize is absolutely necessary in order to prevent growth inhibition and prevent yield losses. Especially for common vetch intercropping, root extracts from the vetch can increase the CP content. Aarssen et al. [76] showed in an intercropping trial with *Avena sativa* and common vetch that the vetch root extracts increased the nitrogen content of *Avena sativa*. These changes were also confirmed by the 2019 results of TH and FAK. In FAK, an increase in CP content was measured, which resulted in a reduced NEL. In contrast, a decrease of the NfE content was measured in TH, which corresponds to an increase in CP and also achieved reduced NEL contents.

The above-mentioned studies from northern Germany and Great Britain showed increased CP contents at unchanged DMY. However, our study could not prove an increase in CP content by intercropping with common beans. This could be ascribed to the sown maize:bean ratio and the weather conditions. The studies from northern Germany and Great Britain used a higher bean and a lower maize proportion. While in our study the sown maize:bean ratio of 8:4.5 plants m<sup>−</sup><sup>2</sup> resulted in a percentage share of 64% maize and 36% beans seeds, Fischer et al. [35] used maize:bean ratios of 8:6, resulted in a share of 57% maize and 43% beans and Dawo et al. [31] a 7.5:5 ratio (60% maize and 40% bean) and a 5:5 ratio (50% maize and 50% bean). Most studies which proved a higher CP content sown a higher proportion of beans. An additional reason for the missing CP increase in our study, could be the environmental conditions in Southern Germany. Common beans are sensitive to high temperatures. Growing at 32 ◦C decreases the biomass of *P. vulgaris*, compared to growing at 25 ◦C [77,78]. At full flowering, temperatures above optimum (15–30 ◦C) were shown to a ffect all enzyme activities of the nitrogen metabolism negatively [79]. This leads to the conclusion that the temperatures during flowering in southern Germany can be above the optimum, which influences the nitrogen metabolism and therefore the formation of CP, while in the cooler climates of northern Germany and Great Britain the nitrogen metabolism is less a ffected. This was also shown by Porch and Jahn [80]. They showed that at high day/night temperatures (32/27 ◦C) heat-sensitive *P. vulgaris* genotypes react with an excessive abscission of their reproductive organs. This gives an explanation, why in hot, dry 2018 no e ffect on CP could be observed. Another reason for the lack in increase in CP, even in 2019, may be the proportion of beans in the silage. This should be above 20% [81]. This increase could not be achieved at any site in any year (Tables 4–6). In addition, it must be noted that di fferent varieties of *P. vulgaris* or species of Phaseolus beans were used in the above-mentioned experiments. Depending on the genotype used, beans showed a wide range in CP contents. Celmeli et al. [82] showed that the CP content in both landraces and modern cultivars of common beans varied widely (landraces: 16.5–25.2%, modern cultivars: 19.7–24.3%). Thus, in addition to the seeding rate and growth conditions, the selected bean variety can have a significant influence on the CP content of the harvested material. In 2019, the results from TH showed a decrease in biogas and methane, while CX and CA increased and NfE decreased. The correlation matrix in Figure 5 verifies this finding. High CA contents had a negative correlation on biogas and methane (rp = −0.90 and −0.75), also high CX (rp = −0.89 and −0.89). CA and CX are not or only slowly digestible in the biogas plant, while NfE had a positive correlation on biogas and methane (rp = 0.95 and 0.80), respectively. Therefore, the decreasing NfE content leads to decreasing biogas and methane yields. The reduced contents in GE, ME and NEL also could be explained by these increases in CA and CX. Mixture I showed neither in TH nor in FAK di fferences in DMY or CP. Mt. Pleasant [83] also showed that traditional MILPA systems did not reach higher yields or an increase in CP compared to a sole maize crop. Mixture II (common vetch and common bean II) had significantly lower DMY but higher CP and lower NfE compared to the control. These changes can be ascribed to the mentioned negative e ffects of common vetch intercropping. Decomposing vetch parts inhibit maize growth (=reduced plant height), while the combination of two legume IFP can increasetheCPcontent.Also,thehighweedinfestationincreasedtheCPcontent.

#### *4.2. Requirements on Agricultural Practice and Equipment*

Intercropping of maize and IFP has several challenges, like seeding technique and weed control, which will be discussed in detail in the following contents.

When maize and an IFP are sown together in a single working step as mixture via a single seed precision planter, the morphology of the seeds should be similar, e.g., shape, 1000-grain mass. This enables sowing as a seed mixture. A single-step sowing reduces the demand of fossil fuels, reduces soil compaction and saves machinery and labor costs. However, sowing as a mixture requires that both maize and IFP can deal with the weed management. Adapted weed control is important, otherwise there will be too much competition which will lead to yield losses. As seen by Abdin et al. [44], cover crops were not able to suppress weed development adequately, especially on sites with high weed potential. This has been shown in ET in 2018. Reduced amounts of the recommended plant protection agents for maize were used in order to not harm the IFP. These reduced amounts lead to a high pressure of problematic weeds, such as *C. album* (L.) and *Galinsoga* spp. (RUIZ & PAV.). Small IFP seeds, like alfalfa, yellow sweet clover, common vetch, and squash I and II are technically not suitable for sowing as a mixture. These IFP also did not permit an adequate chemical weed control under the local site conditions. Some IFP's had under mechanical weed control problems by burying during hoeing. The only practicable method could be a sowing of IFP via box spreader during the last mechanical weeding step.

Separate sowing enables on the one hand the establishment of a spatially separated crop stand and on the other hand, it is also possible to establish the plants at di fferent times. Thus, maize is sown first and weed control can be done (chemically or mechanically). Afterward, IFP are sown in the maize crop. However, this delays the flowering of the IFP. In case of a spatially separated establishment of the IFP between the maize rows, no mechanical weed control can take place, if the IFP seed is sown immediately after the maize sowing. In both cases, establishment in a second step is necessary, which costs time and fossil resources. Special machines for maize and IFP sowing in one step are not ye<sup>t</sup> widely used. A box spreader on the maize seed drill also means that no mechanical inter-row weed treatment can take place. It should be clear, when sowing as mixture, the use of a partner additionally reduces the share of maize in the crop, if the proportion of maize in the seeding mixture is not increased. This is caused by the seed separation at the sowing disc of the precision planter. The seeds will randomly ge<sup>t</sup> into the holes of the seed disc. This means that at the adjusted sowing rate (e.g., 10 plants m<sup>−</sup>2), it is not possible to identify exactly the distribution pattern of maize and IFP within the row, this will happen randomly. Since maize forms the major share of biomass for most of the observed IFP treatments, except for the common bean, it must be ensured by a high maize proportion and a low IFP proportion that the maize seed ratio is not "thinned" by the IFP. In general, if an intercropping mixture is sown in one working step, the proportion of the IFP should be kept as high as possible for biological promotion and as low as possible so the maize will not be "thinned".
