Cover Crop And Weed Management In An Inter-seeded Cover Crop Plus Reduced-Rate Herbicide System In Wide-Row Vegetables

Tomato trial

Tomato yield from the herbicide treatments were up to 10 tons ha^‑1 higher than the weed-free hand-weeded control (66 tons ha^‑) in 2014. In 2015 and 2016, the control plots (33 tons ha^‑1 in 2015 and 101 tons ha^‑1 in 2016) had the highest yields. While in 2015, tomato yield from the herbicide treatments (16 to 27 tons ha^‑1) were statistically comparable to the control, tomato yield from all the herbicide treatments in 2016 (58 to 66 tons ha^‑1) were lower than the control (Figure 3).

The incidence of disease was high in 2015 since it was a very wet year. Presence of living mulches in these conditions would have made the tomato plants more vulnerable to disease. However, in 2015, tomato yield increased with increasing living mulch biomass, which means that more the vigorous living mulch stands were beneficial to tomato yields (Figure 4). So, reductions in tomato yield were likely not caused by living mulches.

During this year, increasing order of treatments in terms of tomato yield was almost identical to the decreasing order of treatments in terms of weed biomass (Figure 5).

This is probably an indication that among the mixture of living mulches and weeds, the weeds alone were responsible for adversely affecting tomato yields. If the tomato crop had been affected by the competition from both living mulches and weeds, then there would have been a negative relationship between tomato yield and the combined biomass of living mulch and weeds. But, this was not the case. The combined biomass of living mulch and weeds had no (positive or negative) relationship at all with tomato yield. This provides further corroboration to the earlier inference that it was solely the weeds, and not the living mulch, that adversely affected tomato yields.

In 2015, the living mulch generated 2.1 to 2.8 tons ha^‑1 of biomass in the herbicide treatments, while the untreated cover crop check with no herbicide applications recorded 2.7 tons ha^‑1 of living mulch biomass (Figure 6).

This indicates that the herbicide treatments did not really affect living mulch biomass production. On the other hand, weed biomass ranged from 0.43 to 1.33 tons ha^‑1 in the herbicide treatments, whereas, weed biomass in the weedy check was 2.64 tons ha^‑1. Hence, all herbicide treatments were able to considerably suppress weeds. The fact that the herbicide applications considerably reduced only weed biomass, and not living mulch biomass, is indicative that, as expected, the herbicide applications were capable of disproportionately targeting the weeds.

Surprisingly, there was no relationship between living mulch biomass and weed biomass (Figure 7).

Increase in living mulch biomass/vigor was expected to decrease weed biomass. This however, suggests that herbicide applications played an important role in weed control. The range of weed biomass in the herbicide treatments was similar in 2015 and 2016. But, unlike in 2015, tomato yield decreased with increasing living mulch biomass and weed biomass in 2016 (Figure 8).

The summer of 2016 was very dry, and although irrigation was provided, the reduction in tomato yield was likely to be a consequence of competition for water. The positive relationship between tomato yield and living mulch biomass in 2015 could then, have been possible due to lack of living mulch-tomato competition for water under the wet conditions.

In 2014, living mulch biomass negatively impacted tomato yield (Figure 9). However, scarcity of water was not likely to be a cause for this since, even though the summer of 2014 was relatively cold, soil moisture conditions were adequate. The reason for this relationship is likely due to the unexpected positive relationship between weed cover and living mulch biomass during 2014 (Figure 10). Weed pressure was higher in plots with more vigorous living mulch stands. Furthermore, tomato yield decreased with increasing weed pressure (Figure 11). So, like in 2015, it is likely that only the weeds adversely affected the tomato crop, even though there was a negative relationship between tomato yield and living mulch biomass, which was created due to the positive association of living mulch biomass with weed pressure.

The reason for this anomalous positive relationship between living mulch biomass and weed pressure was probably because the living mulch was not able to adequately suppress the weeds by itself, without sufficient assistance from the herbicide applications. Herbicide treatments in 2014 did not consist of different herbicide types, but only two applications of one herbicide at a fixed rate. So, in plots where weaker herbicide applications were used, living mulch vigor was not affected much, and larger amounts of living mulch biomass was produced. However, at the same time, the weeds were also affected to a lesser degree, causing their presence to extend too, due to insufficient suppression from the living mulch. In those plots where stronger herbicide applications were used, both the living mulch and the weeds were affected more, leading also to greater reductions in living mulch biomass. But, in this case, tomato yield was improved. The herbicide treatment, after they were modified in 2015 in order to asymmetrically target weeds, were able to prevent any relationship between living mulch biomass and weed biomass in 2015. In 2016, perhaps due to the extremely dry conditions leading to excessive competition for water, the weed biomass decreased with increasing living mulch biomass.

In 2016, due to scarcity of water, it was likely that living mulches had another adverse impact on tomato yield. More vigorous living mulch stands (which produce greater amounts of biomass) remove more water from the soil, thereby making it comparatively drier. In tomato, dry soil conditions can lead to calcium deficiency in the plants because optimum uptake of this nutrient requires sufficient soil moisture. This can lead to yield losses from bottom rot disease. During this year, concentration of calcium in tomato leaf tissue was found to decrease with increasing living mulch biomass, which supports the hypothesis that more vigorous living mulch stands affected the tomato crop more through removal of higher quantities of water from the soil. These results substantiate the necessity for adequate soil moisture conditions in living mulch systems.

In 2016, tomato yield from the untreated cover crop check (29 tons ha^‑1) was lower than even the weedy check (37 tons ha^‑1). In comparison, tomato yield from the control was 101 tons ha^‑1 and in the herbicide treatments, tomato yield ranged from 58 to 66 tons ha^‑1. Since weed biomass from the untreated check (2.1 tons ha^‑1) was much lower than from the weedy check (9 tons ha^‑1), weed pressure was likely not responsible for this outcome. However, the untreated check was very different from the other treatments in terms of living mulch height (116 cm). Height of weeds was not measured, but they were shorter than the living mulch in the untreated check. Tomato yield decreased with increasing living mulch height. When tomato yield from the treatments were adjusted for living mulch height before running the analyses, the untreated living mulch check projected a 44% increase in yield. The herbicide treatment reduced living mulch height by 16 to 49 cm (14 to 42% reduction, respectively). Reduction in living mulch height was expected to be an important function of the herbicide applications, and these results support our hypothesis.

The negative impact of excessive living mulch height on tomato yield was observed in 2014 also, but not in 2015. In 2015, there was no relationship between tomato yield and height of the living mulch stand. This could perhaps, be because the only major competitive stress on the tomato crop from the living mulch was probably shading caused by excessive living mulch height. Whereas in 2014 and 2016, the negative relationship between tomato yield and living mulch biomass (absent in 2015) suggest that during these years, in addition to competition for light, competition for water was also likely significant. Therefore, the combination of the competition for both these factors might have caused reductions in tomato yield in 2014 and 2016, as opposed to 2015.

In 2015, during mid-August, living mulch height in the metribuzin + halosulfuron treatment (105 cm) was not very different from that in the untreated check (114 cm), but all other treatments had much lower living mulch height (74 to 85 cm). After the end of tomato harvests, living mulch in the metribuzin + halosulfuron treatment (131 cm) was taller than even the living mulch from the untreated plot (121 cm). The most likely explanation for the presence of taller living mulch plants in herbicide-treated plots relative to the untreated check, is that, both metribuzin and halosulfuron being (at the applied rates) weaker herbicides, did not affect the larger living mulch so much as the smaller (at time of application) weeds. This possibly provided the living mulch a less competitive environment in this treatment, compared with the untreated check. The pattern was similar in 2016, where the height of living mulch two weeks before harvest, were not very different between the metribuzin + halosulfuron (100 cm) treatment and the untreated check (116 cm). Living mulch heights in the other herbicide treatment were again, much lower (67 to 81 cm). It is, as discussed previously, an important function of the herbicide applications, to reduce living mulch height. But, for the treatment to be successful, this must be achieved without significant losses in living mulch biomass or cover. Such losses in biomass or cover can lead to increase in weed pressure, and as discussed earlier, weeds have more capacity to adversely affect the crop, than the living mulch.

The metribuzin + rimsulfuron treatment, in both 2015 and 2016, reduced living mulch height without compromising living mulch cover or density. In 2016, living mulch cover in this treatment was 76%, which was not very different from the untreated check (82%). It was desirable that the herbicide applications not cause mortality among the living mulch plants, because even small gaps in the living mulch stand can become an epicenter for heavy weed infestation. However, in 2016, all herbicide treatments had lower weed density and lower weed biomass than the weedy check. The untreated living mulch check was similar to both the herbicide treatments and the weedy check in terms of weed density, but was lower than only the weedy check in terms of weed biomass. This might suggest that living mulch-herbicide combinations were capable of causing weed mortality during a dry year, but herbicide applications were essential. In case of weed biomass however, the living mulch was probably capable of reducing weed biomass by itself during a dry year. Comparing these results in 2016 with those in 2015, it is likely that the dry conditions in 2016 as opposed to the wet conditions in 2015, likely exposed and intensified the negative impact of the treatments on weeds.

Weed suppression was better in treatments with an initial application of metribuzin. In 2016, the metribuzin + rimsulfuron (0.44 tons ha^‑1) and the metribuzin + halosulfuron (0.3 tons ha^‑1) treatments reduced weed biomass by 95 and 97%, respectively, compared with the weedy check (9 tons ha^‑1). The untreated check (without herbicide applications) (2.1 tons ha^‑1), reduced weed biomass by 76%. In 2016, weed cover in metribuzin + rimsulfuron (9%) and metribuzin + halosulfuron (4%) reduced weed cover by 91 and 96%, respectively, compared with the weedy check (98%). The untreated check reduced weed cover by 81%, and herbicide applications reduced weed cover by an additional 15%. This indicates that, by including herbicide applications, weed cover was reduced five times more, than by using only the untreated check. In 2015 also, the metribuzin + rimsulfuron treatment (0.67 tons ha^‑1) was capable of considerably reducing weed biomass, with 75% reduction compared with the weedy check (2.64 tons ha^‑1).

During the development of the herbicide treatments, a first application of a Type 1 (stronger post-emergent activity at selected rate) herbicide was expected to be more effective in minimizing living mulch-cash crop competition, thereby resulting in better tomato/crop yields. This is because the smaller size of the living mulch at the time of this application will cause more severe injury and extend recovery time. Therefore, there would be a less competitive environment for the vegetable crop. However, weed suppression will be compromised if the first application is of a Type 1 herbicide, due to prolonged loss in living mulch vigor, and poor residual activity of Type 2 herbicides. On the other hand, a first application of a Type 2 herbicide like metribuzin will not only provide some residual activity against weed emergence, but also affect the living mulch less severely, leading to quicker recovery from herbicide damage. This would help the living mulch to establish well and suppress weeds better. The subsequent second application with a Type 1 herbicide would also affect the living mulch less than the weeds due to the former’s larger size. The disadvantage to this was expected to be greater living mulch-vegetable crop competition.

This was confirmed in case of the living mulch stand. In 2015, living mulch cover, height, density and biomass at the end of the season was 86 and 77%, 131 and 112 cm, 114 and 130 plants m^‑2, and 2.8 and 2.3 tons ha^‑1, respectively, in the metribuzin + halosulfuron and metribuzin + rimsulfuron treatments. However, these parameters were, in comparison, 70 and 55%, 86 and 88 cm, 95 and 94 plants m^‑2, and 2.1 and 2.4 tons ha^‑1, respectively, in rimsulfuron + metribuzin and fomesafen + metribuzin, respectively. The disparity was greater in 2016, where living mulch cover, height density and biomass in metribuzin + halosulfuron and metribuzin + rimsulfuron were 94 and 85%, 100 and 81 cm, 146 and 143 plants m^‑2, and 2.5 and 1.4 tons ha^‑1, respectively, compared to 78 and 40%, 75 and 67 cm, 133 and 121 plants m^‑2, and 1.3 and 0.7 tons ha^‑1, respectively in rimsulfuron + metribuzin and fomesafen + metribuzin, respectively. This greater disparity could be, as described earlier, a result of different degrees of water stress during drought conditions in 2016 and wet conditions in 2015.

Even though the greater reduction in living mulch height by the first application of a Type 1 herbicide was desirable, considerable corresponding differences in tomato yield was not remarkable in 2015 or 2016. As discussed earlier, this was probably because the tomato yield was affected disproportionately more by weeds. In all years, tomato yield decreased with increasing weed biomass and weed cover. But, following modification of herbicide treatments after 2014, no relationship existed between living mulch height and weed biomass. This indicates that greater reductions in living mulch height by rimsulfuron + metribuzin and fomesafen + metribuzin treatments were not necessarily associated with greater weed suppression. So, the potential benefits to the tomato crop from reduced competition for light by shorter living mulch were likely nullified by weed pressure.

It is possible that the living mulches affected the tomato crop relatively less than the weeds, at least in part because of the geometry of planting and plant architecture of the living mulch. The distance provided between the living mulch and tomato rows, along with the erect stature of sunnhemp plants, could have helped to minimize its disturbance to and competition with the tomato crop. On the other hand, presence of weeds close to or within tomato rows, combined with their comparatively higher degree of interference to the tomato canopy, could be responsible for the stronger (negative) associations between weeds and the cash crop.

Analyses of nutrient concentrations in sunnhemp or tomato tissue have not revealed considerable treatment impacts or patterns yet. In 2016, concentrations of nitrogen, potassium and phosphorous in the tomato tissue were positively correlated to living mulch biomass. However, the relationship between tomato yield and living mulch biomass was negative. So, other factors likely caused more vigorous living mulches to affect tomato yield. It is possible that the greater competition for light and water between the tomato crop and the living mulch in the more vigorous living mulch stands, superseded the benefits from higher nutrient concentrations in the tomato plant. In the inoculated untreated check, the living mulches were unacceptably tall at 141 cm, with corresponding biomass production of 4.7 tons ha^‑1. In the preliminary inoculated trial, overall, the tomato crop was badly affected by the inoculated living mulch because inoculation made the living mulch too vigorous. This was probably because the increased availability of nitrogen gave the inoculated living mulch a boost, which in turn increased its competitiveness in, and uptake of, other nutrients and water also.

Overall, the results from the tomato trial demonstrated that:

1. Lack of dry conditions is required to prevent living mulches from adversely affecting crop yield
2. Herbicide applications may be essential to obtain adequate weed control in living mulch systems
3. In the presence of a combination of living mulch and weeds, crop yield is capable of being affected only by the weeds.

In a cropping system, there are many factors that can be manipulated to reduce herbicide input and so that herbicide applications target the weeds disproportionately more. But once these factors are established, results from these trial show that, it is important that the strength of the herbicide applications are based on achieving adequate weed control, and not on achieving a target living mulch biomass. This is because even vigorous stands of living mulches are relatively more unreliable in the context of crop yield.

In addition to the several benefits to soil, inter-row tillage during the season was eliminated, which in turns translated to reduced cultivation. Pre-emergent herbicide applications were also not used in our trials, and this is a consequential outcome because pre-emergent herbicide applications are made at larger rates than post-emergent applications. They are also more persistent, thus posing a greater risk of leaching and ground water contamination. No additional fertilizer input was used in these experiments to compensate for incorporation of the living mulch. Besides putting no additional economic or environmental stress on the production system, such a cropping system helps to improve the long-term efficiency of fertilizer applications by tying up the fertilizer in the inter-row space in living mulch tissue. The implementation of such a living mulch system was not without difficulties, and many constraints were revealed. But, being a relatively scarcely researched science, more research is required to lay down broad protocols. However, our experiments demonstrated that such a living mulch-herbicide system is a sustainability tool worth considering in fresh market tomato.

Sole cover crop trial

Biomass generated by the living mulch in 2015 and 2016 were similar, even though the two seasons were very different in terms of the amount of precipitation. All herbicide treatments, except fomesafen + metribuzin recorded similar amounts of biomass in 2015 and 2016. This suggests that the herbicide treatments have adequate predictability in their cover crop control between seasons with varying climatic conditions. With regard to weed control, the weed biomass in the weedy check was significantly higher in 2016 (4.2 tons ha^‑1) than in 2015 (2.2 tons ha^‑1). But, weed biomass from the individual herbicide treatments were not considerably different between the two years. This means that the cover crop-herbicide treatment combinations had predictability in their degree of weed control, regardless of the absolute weed pressure during a season.

During 2015, the cover crop treatments provided substantial soil cover (82 to 94%), except fomesafen + metribuzin (21%). Cover crop biomass was not very different between the treatments, except in case of fomesafen + metribuzin, where cover crop biomass was low, relative to the untreated check (7 tons ha^‑1), the herbicide treatments reduced cover crop biomass by 17 (imazethapyr + rimsulfuron, 5.8 tons ha^‑1) to 46% (metribuzin + fomesafen, 3.8 tons ha^‑1). The herbicide-type combinations with a first stronger Type 1 herbicide application, which affected the cover crop more, did not necessarily affect the weeds more too. Weed biomass in Type 2 + Type 1 and Type 2 + Type 2 were 0.06 and 0.14 tons ha^‑1, respectively, while weed biomass in Type 1 + Type 1 and Type 1 + Type 2 were higher at 0.27 and 0.44 tons ha^‑1, respectively.

The herbicide treatments were able to reduce cover crop height by up to 31%, relative to the untreated cover crop check. Greatest reductions in cover crop height resulted from treatments including an application of fomesafen. Density of cover crop stands was not affected by the individual herbicide treatments or the herbicide-type combinations. It was a positive outcome that the herbicide applications did not cause remarkable mortality of the cover crop stands. This is because such gaps in the cover crop stand can lead to weed outbreak. In terms of weed cover, measured in late August, between all the cover crop treatments (2 to 5%) were similar to each other, and all were lower than the weedy check (46%). The cover crop-herbicide treatments were effective in suppressing weeds. In seven out of the 11 cover crop treatments, weed biomass was more than an order of magnitude lower than the weedy check.

Even with the extremely dry conditions in 2016, the cover crops were able to perform very well with irrigation. In late August, cover crop ground cover was considerable in all treatments (89 to 99%), except in fomesafen + metribuzin (15%). In rimsulfuron + metribuzin treatment, cover crops were able to recover from only 27% cover in mid-July and 15% cover in late July, to 89% in late August. With regard to cover crop biomass, metribuzin 0.1 + rimsulfuron 0.007 (5.6 tons ha^‑1), metribuzin 0.05 + rimsulfuron 0.005 (5.6 tons ha^‑1) and metribuzin + halosulfuron (6.1) were numerically higher than the cover crop biomass from the untreated check (5.5 tons ha^‑1). As described previously in the ‘Tomato Trial’, although this was not statistically significant, these might be indicative of the benefits to the cover crop from a less competitive environment provided by the asymmetric targeting of weed by the herbicide applications.

Three out of the four herbicide-type combinations provided sufficient cover crop cover, which indicates that the herbicide applications were effective in preventing excessive losses to soil cover. Living mulch in Type 1 + Type 2 was most likely affected by increased weed pressure, more so than even by the herbicide applications. Weed cover in these plots was 55% in early August, which was lower than the weedy check (98%), but became 74% in late August, which was similar to the weedy check (92%). Due to slow recovery of the cover crop from an initial application of a Type 1 herbicide, when at a young stage, and due to the primarily post-emergent nature of Type 1 herbicides, there is an increased likelihood of an elevation in weed pressure, which can further smother the cover crop.

The untreated cover crop check was able to suppress weeds to some extent both in 2015 and 2016, even without herbicide applications. In 2015, cover crop biomass or cover crop cover did not affect weed biomass. Weed cover also showed no relationship with cover crop biomass. This was likely because of the lack of cover crop-weed competition for water in 2015. In 2016, weed biomass had negative relationship with both cover crop biomass and cover crop cover. The trends were similar in case of weed cover. So, the extent of cover crop cover, and cover crop biomass, decreased weed cover and weed biomass even though herbicide applications tend to decrease cover crop cover and cover crop biomass. These results demonstrate the importance of a healthy living mulch stand in weed suppression, and its contribution towards herbicide reduction.

Even in late July, cover crops in fomesafen + metribuzin and rimsulfuron + metribuzin (both 19 cm) were extremely short. Seven of the cover crop treatments (40 to 51 cm), including the untreated check (44 cm) were taller than these treatments. Reduction of cover crop height is a positive outcome, and is expected from the herbicide applications. But, shorter living mulches might suppress weed poorly. In addition, as discussed previously, if these reductions in cover crop height are associated with, or will be accompanied by, losses in cover crop ground cover or cover crop biomass, then that is a critical loss in the objectives of establishing a living mulch system. Furthermore, if these losses in cover crop height, cover, biomass or vigor, lead to increase in weed pressure, then these reductions in cover crop height are counterproductive because the cash crop is more likely to be adversely affected by weeds than living mulches due to planting design and plant geometry.