Intercropping Cassava and Legumes for Local Food Security in Florida and Puerto Rico

Weeding Experiment

Legume species and weeding treatment both had significant effects on weed biomass at harvest and a significant interaction—plots that received full weeding had less weed biomass than unweeded plots, but this effect was most pronounced in soybean and sole cassava plots (Figure 1). Cowpea and peanut plots had low weed biomass regardless of weeding treatment, comparable to fully weeded soybean or sole cassava plots. The ability of cowpea and peanut to suppress weed growth may be due in part to their greater height: at 8 weeks after planting, cowpea, peanut, and soybean plants had an average height of 64.6 cm, 41.4 cm, and 36.5 cm, respectively, all higher than the average cassava height of 21.2 cm. Linear regression of weed biomass and legume height showed that weed biomass decreases as legume height increases, and that this effect is stronger in unweeded plots than in weeded plots (Figure 2).

Figure 1. Effect of legume and weeding regime on weed biomass at harvest. Error bars are standard error, bars sharing the same letter are not significantly different (p > 0.05).

Figure 2. Effect of legume height and weeding regime on weed biomass at harvest.

Legume yield varied by year: in 2022 pod yield of cowpea (489 kg/ha) was higher than peanut (107 kg/ha) and soybean (169 kg/ha), whereas in 2023 pod yield of peanut (631 kg/ha) and soybean (596 kg/ha) were higher than cowpea (217 kg/ha) (Figure 3). Total crop biomass in each system (including intercropping with cassava) followed a similar pattern (Figure 4). Crop biomass proportion (crop biomass/weed biomass + crop biomass) was calculated to reflect the efficiency of each legume and weeding regime combination in increasing the proportion of plant biomass not found in weeds. Sole cassava (regardless of weeding regime) and unweeded soybean had significantly lower crop biomass proportions than all other combinations, with less than 20% of biomass as crops in the former group and more than 50% of biomass as crops in the latter group (Figure 5). Fully weeded peanut plots had the highest crop biomass proportion at 0.76, i.e. 76% of biomass as crops and 24% as weeds. Cassava yield was higher in weeded plots (627 kg/ha) than in unweeded plots (209 kg/ha), but this difference was not statistically significant (p = 0.108). Yields of sole and intercropped cassava and legumes in each weeding regime are shown in Table 1, along with their associated partial LERs and the total LER for each system. Average LER was 2.57 for weeded systems and 3.56 for unweeded systems.

Figure 3. Legume yields in 2022 and 2023. Box plots sharing the same letter do not have significantly different means (p > 0.05)

Figure 4. Total crop biomass in 2022 and 2023. Box plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 5. Effect of legume and weeding regime on crop biomass proportion (total crop biomass divided by total plant biomass, including weeds). Error bars are standard error, bars sharing the same letter are not significantly different (p > 0.05).

Table 1. LER of weeded and unweeded cassava-legume intercropping systems in 2023. LER is the sum of partial LER (PLER) of each component crop.

                          Legume yield (kg/ha)                                                                                                                Cassava yield (kg/ha)

The results of the weeding experiment support the use of intercropped legumes as a weed management technique in cassava-legume intercropping. The capacity of legumes to suppress weeds was correlated with their vigor and height: plots with cowpea and peanut, which were taller than soybean and qualitatively more vigorous, had less weed biomass than plots with soybean. Unweeded soybean had the highest weed biomass, and soybean correspondingly received a greater benefit from weeding than cowpea and peanut plots.

The slow early growth of cassava makes it particularly susceptible to weed pressure, and weed biomass was high in sole cassava plots, even when the full weeding treatment was applied. Intercropping cowpea and peanut reduced weed biomass to an extent comparable to full weeding, and the combination of peanut and full weeding produced the lowest weed biomass. Legume and cassava yields were not significantly affected by intercropping, suggesting that the use of intercropped legumes for weed management does not have an undue yield penalty. On the contrary, high LER values in these systems indicated a yield benefit of cassava-legume intercropping, particularly in the unweeded treatments (average LER = 3.56 vs 2.57 in weeded treatments). This result supports the Stress Gradient Hypothesis (SGH), which predicts that the removal of a biotic stress (e.g. weed pressure, removed by weeding) shifts plant interactions toward competition (i.e. lower LER). These results also suggest that the yield benefit of cassava-legume intercropping is due in part to the ability of intercropped legumes to suppress weeds in the first few months after planting.

Competition Pots

Analysis of cassava and legume height, yield, and biomass did not reveal significant interactions between the effects of above-ground separation and below-ground separation. Table 2 shows the main effects of above and below-ground separation, and all the statistically significant effects were caused by above-ground competition. In cassava-peanut competition pots, peanut height was significantly lower in the presence of above-ground competition (Table 2). Peanut biomass, cassava height, and cassava biomass were also lower in the presence of above-ground competition, but these effects were not significant.

Table 2. Effect of above-ground and below-ground competition on peanut and cassava. P values calculated from one-way ANOVA. Cowpea height listed as 60 cm because all cowpea plants reached the maximum height of the competition plots.

Cowpea, unlike peanut and cassava, was more productive in the presence of above-ground competition, with higher biomass and yields (Table 2). Cassava biomass was significantly lower in the presence of above-ground competition (2.1 g isolated vs 1.2 g competing, p = 0.042). The cassava biomass proportion (cassava biomass/cowpea biomass + cassava biomass) was not affected by below-ground competition, but was significantly higher with above-ground separation, where cassava did not have to compete with cowpea for light (Figure 6). Above-ground competition with cowpea did not significantly affect cassava height, but was associated with more spindly cassava stems, with less weight per unit height (Figure 7).

Figure 6. Effect of above-ground and below-ground competition on cassava biomass proportion (total cassava biomass/total pot biomass). Data from cassava-cowpea competition pots. Horizontal lines represent population means. No significant interaction between above-ground and below-ground effects.

Figure 7. Effect of above-ground and below-ground competition on cassava stem stoutness (g dry biomass/cm stem length). Data from cassava-cowpea competition pots. Horizontal lines represent population means. No significant interaction between above-ground and below-ground effects.

Competition pot experiments showed that cowpea outcompetes cassava while peanut and cassava are more competitively balanced. Above-ground competition in the cassava-peanut system led to lower heights and biomass in both peanut and cassava. Above-ground competition in the cassava-cowpea system, on the other hand, led to lower cassava biomass but higher cowpea biomass and yield. Competition with cowpea also induced a spindly phenotype in cassava; cassava plants competing above-ground with cowpea had significantly less biomass per unit height, likely in an effort to grow beyond the dense cowpea canopy (Craine and Dybzinski, 2013). The competitive vigor of cowpea must be managed to prevent undue reduction of cassava growth in cassava-cowpea intercropping systems; for example, planting cassava earlier than cowpea may allow it to reach a more competitive height before cowpea canopy closure, and lower cowpea density or less vigorous cowpea varieties can be used to the same effect (Cenpukdee and Fukai, 1992a).

In both cassava-cowpea and cassava-peanut competition pots, above-ground competition was more impactful than below-ground competition. This suggests that competition for light, more than competition for water or nutrients, drives interactions between cassava and legumes in this environment (the effect of environment should not be ignored; this experiment was irrigated daily, making competition for water unlikely) (McPhee and Aarssen, 2001). Observation of the competition pots supports this hypothesis: in pots with above-ground separation, plants from both sides grew up against the hardware cloth barriers, and in pots without above-ground separation, the cassava and legume canopies mingled, mostly due to cowpea and peanut leaves and stems growing into the cassava halves of the pots. Cassava and peanut were similar in height while cowpea grew much taller, accessing a greater portion of light and significantly reducing cassava biomass when given the opportunity to overshadow cassava.

Residue Experiment

Legumes were generally unaffected by intercropping in this experiment. Legume yields were significantly higher in 2022 than in 2023, but there was no significant difference in yields between sole legumes and intercropped legumes (Figure 8). Legume heights at maturity were significantly different between species (p < 0.0001), with cowpea the tallest (71.3 cm), followed by peanut (53.4 cm) and soybean (38.6). Decomposition of legume residues on the soil surface after legume harvest is shown in Figure 9. Decomposition was most rapid in the first 30 days after harvest, reaching ~40% of initial biomass, then continued gradually to 50.3% of peanut residues and 62.5% of cowpea residues at cassava harvest. Based on these decomposition rates, an estimated 3864 kg/ha of cowpea residue and 6442 kg/ha of peanut residue had decomposed in residue-retained plots by the time of cassava harvest in 2022, and 2027 kg/ha cowpea residue and 785 kg/ha peanut residue decomposed by cassava harvest in 2023. Using previously determined rates of N content in legume residues (Kermah et al., 2018), these decomposed biomass quantities can be estimated to contain 98 kg N/ha, 189 kg N/ha, 52 kg N/ha, and 23 kg N/ha, respectively.

Figure 8. Sole and intercropped legume yields in 2022 and 2023. Box plots include yields of all legumes (n = 24 for each year/system combination). Box plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 9. Decomposition of cowpea and peanut residues determined by litterbags. Final timepoints (84 days after harvest for peanut and 123 days after harvest for cowpea) were taken at cassava harvest. Error bars represent 95% confidence intervals.

Cassava height at harvest was greater in 2022 (119.3 cm) than in 2023 (66.1 cm), and intercropped legume species had a significant effect on cassava height in 2022, with cowpea suppressing cassava height (98.4 cm) compared to peanut (120.3 cm), soybean (124.6 cm) and sole cassava (129.3 cm) (Figure 10). Intercropping with cowpea was also associated with higher cassava mortality—only 60% of cassava plants intercropped with cowpea survived to harvest, compared with 82% with peanut, 90% with soybean, and 91% with sole cassava (Figure 11). Cassava yield in 2022 ranged from 2866 kg/ha intercropped with cowpea to 10321 as a sole crop, whereas in 2023, cassava yields were consistently low, ranging from 1233 kg/ha intercropped with soybean to 1658 kg/ha with peanut (Figure 12). Figure 13 shows that while intercropped cassava yielded significantly less than sole cassava in 2022, intercropping produced no difference in cassava yield in 2023. Residue management did not significantly affect cassava yield.

Figure 10. Effect of legume intercrop on cassava height in 2022 and 2023. Legume intercrop “None” indicates sole cassava. Each point represents one cassava plant at harvest. Violin plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 11. Effect of legume intercrop on cassava mortality. Legume intercrop “None” indicates sole cassava. Each point represents one plot. Violin plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 12. Effect of legume intercrop on cassava yield in 2022 and 2023. Legume intercrop “None” indicates sole cassava. Box plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 13. Sole and intercropped cassava yields in 2022 and 2023. Box plots sharing the same letter do not have significantly different means (p > 0.05).

Yields of sole and intercropped cassava and legumes are shown in Table 3, along with the corresponding partial LERs and overall LER of each system in each year. LER was > 1 in all systems, indicating a yield benefit of cassava-legume intercropping that ranged from mild (with cowpea in 2022, LER = 1.21) to dramatic (with peanut in 2023, LER = 2.71) (Figure 14). Residue management did not significantly affect overall LER, but cassava partial LER in 2023 was significantly higher with residues removed (Figure 15). To test the prediction of the Stress Gradient Hypothesis that plant interactions shift towards facilitation in more stressful environments, linear regressions of LER with sole cassava yield and sole legume yield were conducted and found significant negative relationships between sole yield and LER (Figure 16). Similar negative correlations were found between LER and sole cassava residue, sole legume residue, and overall sole crop biomass (Figure 17). There was no significant relationship between LER and intercropped crop yields or biomass.

Figure 14. Effect of legume intercrop on cassava-legume LER in 2022 and 2023. Dotted green line at ln(LER) = 0 indicates additivity; ln(LER) > 0 indicates a yield benefit of intercropping. Box plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 15. Effect of legume residue management on cassava partial LER in 2022 and 2023. Cassava partial LER calculated as intercropped cassava yield/sole cassava yield. Box plots sharing the same letter do not have significantly different means (p > 0.05).

Figure 16. Linear regression of LER with sole cassava and legume yields. Shaded areas represent 95% confidence regions for linear fit. Intercropping has a yield benefit where ln(LER) > 0.

Figure 17. Linear regression of LER with sole cassava and legume residue production and total sole crop biomass. Residue was defined as dry mass of leaves, stems, and other non-yield plant parts. Total sole crop biomass was defined as cumulative biomass of sole cassava and legume plots. Shaded areas represent 95% confidence regions for linear fit. Intercropping has a yield benefit where ln(LER) > 0.

Table 3. LER of cassava-legume intercropping systems by legume intercrop and legume residue management in 2022 and 2023. LER is the sum of partial LER (PLER) of each component crop.

                                         Legume yield (kg/ha)                                                                 Cassava yield (kg/ha)                                                                                                

All cassava-legume systems in this experiment had LER > 1 in both 2022 and 2023, indicating a consistent yield benefit from cassava-legume intercropping. (Table 3) Legumes yields were not significantly affected by intercropping, suggesting that the legumes were highly complementary with cassava or that the legumes were competitively dominant over cassava. Data from cassava intercropped with cowpea and peanut in 2022 support the latter option: cassava yields were lower in intercropped than sole, and cassava intercropped with cowpea was significantly shorter and had a lower survival rate than sole cassava. In most cassava-legume intercropping systems, cassava is dominant (Leihner, 1983; Mutsaers et al., 1993). The dominance of legumes in this experiment (cowpea in particular) could be due to slow cassava growth rates in response to nutrient stress (El-Sharkawy, 2004), as the study was carried out on sandy soil with no inputs of fertilizer. Cowpea had the greatest negative effect on cassava growth and yield, possibly due to cowpea being the tallest legume, and qualitatively the most vigorous.

Litterbag data showed that cowpea and peanut residues decomposed quickly enough for roughly half of residue nutrients to enter the soil in the time between legume harvest and cassava harvest. There was no major difference in decomposition rates between cowpea and peanut residues, but since cowpea was harvested earlier in the season, nutrients from cowpea residues may have been available to cassava sooner. Nutrients from decomposing residues are generally immobilized by soil microbes for a time depending in part on the C:N ratio of the residues, so these nutrients are not immediately available to plants. A previous study of cowpea and peanut residues in the context of cassava intercropping found that cowpea and peanut residue nutrients were immobilized for 30-70 days and 30-45 days after application, respectively, after which net mineralization occurred; however, this data was from an incubation of legume residues mixed with soil, not in field conditions (Gnahoua et al., 2016). Decomposition of legume residues can be sped up by incorporation of the residues into the soil (Sanchez, 2019), but this is not practical in cassava-legume intercropping as the digging required for residue incorporation would likely damage cassava storage roots. In addition to potential nutrient benefits from decomposition, legume residues may also affect cassava by decreasing soil evaporation, harboring microbes and insects, and lowering soil temperature.

Removal of residues did not directly affect cassava yield but had a significant positive impact in 2023 on cassava partial LER—the ratio of intercropped cassava yield to sole cassava yield. Correspondingly, retaining legume residues increased yields of sole cassava more than it increased yields of intercropped cassava. This could be because sole cassava plots, which contained no legume biomass prior to the addition of peanut residues, were limited by nutrient availability, while intercropped cassava plots were limited by competition with legumes. However, the impacts of legume residue removal and retention are inconsistent between 2022 and 2023, and further study is needed to unpick the costs and benefits of legume residue retention in cassava-legume intercropping. For the purpose of farmer decision-making, the yield benefit of retaining legume residues must be balanced against the benefit of removing residues for use as animal feed or for sale; the labor cost of residue removal must also be taken into account.