3.1. Phytosociological Parameters of Weeds
In the first phytosociological survey, 962 individuals were recorded, distributed into 16 species, belonging to nine botanical families. The Poaceae family was the most representative, with five species, followed by Cyperaceae, Fabaceae, and Verbenaceae families, each one with two species (
Table 1). The importance of the Poaceae family for crops grown in the country has already been observed in other studies involving weed plants in crops in the Amazon region, e.g., Fontes et al. [
17], Da Gama et al. [
18], Damasceno [
19], De Almeida et al. [
20], Dos Santos [
21], Miléo et al. [
22], and Albertino et al. [
23].
Concerning classes, there was a balance between the number of monocotyledon and dicotyledon species, each one with eight species. However, there was a predominance of monocotyledons in the first year (83.05%), mainly due to the high number of individuals of the Axonopus affinis and Paspalum multicaule species, which together represented more than 70% of the individuals found.
Concerning species, the highest IVI was achieved by
A.
affinis (84.81), mainly due to the total number of individuals and high relative density. This is a perennial, stoloniferous, creeping, fast-growing grass species, tolerant to cutting and trampling, which is considered difficult to control due to the high rooting capacity of its stolons [
24].
This species was also identified by Miléo et al. [
22] in a cassava cultivated field in the state of Amazonas, where a high IVI for
A.
affinis was found. Likewise, the study conducted by Da Gama et al. [
18] showed that a species of the genus
Axonopus achieved the highest IVI in guarana culture.
At the end of the first year of the experiment, changes in the floristic composition of the weed plants were observed, with the emergence of new species, while other species disappeared (
Table 2).
Weed plants were not found in the herbicide-treated area, differently from the treatment using mechanical control, where the presence of some weed species was recorded, although there was a reduction of 80% in the total number of individuals compared with the treatment with no weed control.
Among the cover crops,
B. ruziziensis was the one that exhibited the smallest number of weed species (4). This cover crop also reduced the total number of individuals by nearly 75% when compared to the treatment with no control, mainly due to its rapid establishment and because it grows in clumps, which increases its competitive ability. These results are in agreement with those found by Soares et al. [
3] and Da Gama et al. [
18], where this species exhibited excellent soil coverage and good weed suppression, with potential for use as cover crops in the Amazon region.
Regarding
C. ensiformis, this cover crop exhibited the greatest number of weed species (10), probably due to the upright, determined, and initially slow growth of this legume, which may have favored the emergence of new weed species [
25]. Yet, it provided a 29% reduction in the total number of weeds, an intermediate percentage between that found for
B. ruziziensis and
M. pruriens.
Mucuna pruriens was the cover crop with the lowest reduction of the total number of individuals (12%) compared with the treatment with no weed control, which may be due to the climbing growth habit of this species, allowing more space available in the soil for weed germination and development [
3].
In the phytosociological survey conducted in the total area, prior to the installation of the second experiment, 924 individuals distributed in 19 species were found, 9 monocotyledons and 10 dicotyledons. The Poaceae family was the most abundant, with five species, followed by the Cyperaceae and Euphorbiaceae families, each one with three species (
Table 3).
Differently from what was found in the first year, there was predominance of dicotyledonous individuals in the second year (67.32%), mainly due to the great number of M. pudica individuals, which corresponded to approximately 44% of the individuals identified in this survey.
Despite the changes in the floristic composition of the weeding community, most of the individuals identified before the installation of the second experiment had already been recorded in the weed survey conducted in the first year. Among the five weed species with the highest IVI in the surveys carried out before installation of the experiments in both years were M. pudica, A. affinis, and P. multicaule.
Mimosa pudica stands out from the other species for being the one with the highest IVI in the second year (80.89). It is a perennial, herbaceous, or slightly woody, thorny weed plant, with sensitive leaves, prostrate growth habit and propagation by seeds [
26]. It is considered a very rustic plant, with good development in soils with low nutrient availability, producing seeds that are capable of germinating under water and saline stress conditions, being indicated for recovery of degraded areas [
27].
The mechanical control of this species is difficult because of the thorns and woody roots and also because of the high seed production. Many seeds can remain in the seedbanks in the soil and cause long periods of reinfestation. Because it is very common in the Amazon region,
M. pudica was already found in diverse studies on weed plants in regional cultivated areas, such as those by Dos Santos [
21], Alves Albuquerque et al. [
28], and Albertino et al. [
23], and has already been considered one of the most important weeds in cassava crop fields [
22] and cowpea cultivation [
29].
Similar to the first year of the experiment, there were changes in the floristic composition of the weeding community at the end of the second year. However,
M. pudica was the most important species, both in the first survey and in the final one, considering the highest values found for all parameters assessed (
Table 4).
Concerning the population dynamics of the weed plants, the main differences observed were in terms of the number of species and total number of individuals, and it was not possible to identify control patterns among the weed species and the assessed treatments, with predominance of some species in most treatments, especially M. pudica, P. multicaule, and C. glandulosus.
For the total number of individuals in the second year, the cover crops with B. ruziziensis, C. ensiformis, and M. pruriens provided reductions of 55, 60, and 35%, respectively, when compared with the treatment with no weed control. With respect to the quantity of weed species, B. ruziziensis was again the cover crop with the lowest quantity (4), which underlines its competitive ability.
The variations observed in the weed species and in the phytosociological parameters assessed in the surveys conducted at the beginning and end of this study corroborate the understanding of the floristic composition of the weed plants as a natural dynamic, fluid process, where agricultural practices, soil management system, and the cultural practices adopted can promote significant changes [
30].
In general, the weed species identified in the present surveys were recorded in other studies and are commonly found in the Amazon region and are well adapted to the regional conditions. From this perspective, more in-depth studies on the ecology and behavior of these species with respect to different methods of control are necessary for the development of sustainable control strategies.
3.2. Cassava Yield
The summary of variance of the yield components of cassava intercropping with different cover crops is shown in
Table 5. Regarding the weed management practice, there was significance for root fresh matter, root dry matter, and shoot dry matter, and no difference for plant height and diameter of the stem base.
With respect to yield, the treatment with chemical control achieved the best results (29.23 t ha
−1), followed by mechanical control (21.79 t ha
−1) and
B.
ruzizienis (17.60 t ha
−1) and
C.
ensiformis (15.47 t ha
−1) cover crops. The lowest yields were observed in the treatments with
M.
pruriens (13.77 t ha
−1) and with no weed control (13.60 t ha
−1) (
Table 6).
Although cassava is recognized as rustic plant, fresh root yield was severely affected by weed interference. The treatment with no weed control exhibited more than 50% yield loss compared with the treatment with chemical control, which shows the importance of weed management in cassava yield. Fontes et al. [
31] studied the periods of interference of weeds with cassava cultivar
Manteiga and found that the coexistence of the culture with weeds throughout the crop cycle reduced yields by 96%. According to these authors, this cultivar has low competitive ability against weeds in Amazonas dry land.
The treatment with chemical weed control showed the highest values of yield, root dry matter and shoot dry matter values, probably due to the formation of a uniform straw bed on the soil surface after herbicide application, which may have worked as a physical barrier, helping to keep water in the soil and preventing the emergence of new weeds. The hand-hoeing weed treatment may have suffered the effects of the weed interference found in the treatment, as well as of lower water availability due to the greater soil exposure, which favors water loss to the atmosphere. Although the hoeing treatment achieved lower yields compared with the chemical control, both treatments did not show differences with respect to root dry matter.
In this regard, other studies evaluating weed control in cassava crop fields recorded higher yields in treatments with herbicide applications, compared with treatments using hand weed control [
32,
33]. Different results were found by Fontes et al. [
17], who did not observe differences in cassava cultivar Manteiga yields with herbicide and hand weeding treatments.
With respect to the cover crops, B. ruziziensis and C. ensiformis showed a moderate degree of interference with cassava, providing yield values similar to the ones achieved with the mechanical control due to the growth habit of these plants, which is compatible with the culture, and reduced quantity of weed species, resulting in less interference when compared with the treatment with no weed control. It should be emphasized that future studies evaluating planting time, density, spacing, and adequate intercropping time, can contribute to raise cassava yields and reduce the potential of weed interference on this crop.
Mucuna pruriens did not show differences from the treatment with no weed control for any of the assessed parameters, indicating intercropping incompatibility with cassava, mainly as a function of its climbing growth habit. However, as it is a rustic legume and of rapid establishment, it has potential for use as green manure or mulch, if cultivated in an area not intercropped with cassava, or even in rotation with this culture.
According to Madembo et al. [
34], although intercropped systems with cover crops may reduce crop yields, the use of these plants can be a viable alternative, especially for small farmers, being necessary to investigate crop arrangements capable of increasing the weed suppression potential and reduce interference with the culture.
Despite the importance of optimal production rates, the stability, sustainability of crops, producers’ health, preservation, and maintenance of the Amazon ecosystem are factors that must be considered when choosing the best weed management system. It is worth noting that a sustainable production system, when properly employed, adds value to the end product, and can even compensate for any yield losses.
3.3. Soil Chemical Properties
With respect to the chemical properties of soil, despite the variations observed in the two growing seasons, higher pH, K, Ca, Mg, and organic matter contents were found for the cover crops compared to the chemical and mechanical control treatments (
Table 7).
The favorable pH value for growing cassava in the North region of Brazil ranges from 5.5 to 7, and 6.5 is the ideal pH [
35]. The cover crops contributed to raise the pH value, being closer to the ideal pH for the culture, in comparison to the chemical and mechanical treatments. The effects of the cover crops on the soil pH values are still not fully understood, and contradictory findings are reported in the literature, sometimes with higher values, sometimes with lower values [
36,
37]. Such variations seem to be mainly related to the biochemical compositions of the cover crops, the soil characteristics, the environmental conditions, and the type of management used.
Regarding organic matter, the cover crops achieved higher values compared to chemical and mechanical weed controls. Higher concentrations of organic matter in soil when using cover crops have been mainly attributed to the incorporation of plant residues, the reduction of the mineralization rate by adopting conservative practices, and less loss of organic matter caused by erosion [
38,
39]. According to Oliveira et al. [
40], agricultural crops that do not use conservation practices tend to reduce the contents of organic matter in soil, especially in the topsoil. Low contents of organic matter tend to diminish the availability of nutrients, such as K, Ca, and Mg, causing more dependence on chemical fertilizers [
41,
42].
In the first year,
B. ruziziensis was the species that exhibited the highest contents of K, Ca, Mg, and organic matter and the lowest potential acidity. Because it is a species that produces great amount of biomass, with capacity to uptake nutrients from the deepest layers in soil [
43], the cutting and rapid degradation of vegetable residues that this species has have contributed to the improved level of chemical properties observed in the upper layers of soil. Similar results were found by Ensinas et al. [
44], who studied the effects of some cover crops on the chemical properties of soil, including
B. ruziziensis, and found that the cover crops provided improved K and Mg contents. Arf et al. [
45] reported that
B. ruziziensis and
C. ensiformis provided higher contents of P and K in the soil, and Demir and Işık [
37], when they studied the influence of cover crops on the soil quality, observed that the cover crops provided increased K and Mg contents, when compared with the treatments with chemical and mechanical weed control.
It should be noted that
B. ruzizienis was the cover crop that exhibited the lowest P contents in the soil in the two years assessed. Studies involving brachiaria species and phosphorus availability in the soil reported that, contrary to expectations,
B. ruziziensis reduced P content in the soil, while exhibiting a higher P concentration in plant tissues [
46,
47].
The chemical control treatment achieved the best P contents in the soil in both years of assessment. Considering that there was fertilization of soil with this nutrient, following recommendation of Dias et al. [
11], it is possible that the higher P contents found in this treatment were due to the absence of weeds or cover crops capable of extracting this nutrient from the soil and better conservation of the soil surface provided by crop residues (straw). According to Magolbo [
48], because it participates in the synthesis of starch in the plants, it is expected that P supply in adequate amounts can increase the plant growth and cassava yields. So, greater P contents in soil are highly desirable for cassava, and many studies demonstrate that this is a crop responsive to phosphate fertilization [
48,
49,
50].
As for Zn contents, no differences were observed between the treatments in the first year, but
B. ruziziensis and
M. pruriens exhibited higher contents of this micronutrient in the second year. Due to the variations found, future studies are necessary to assess the real impact of these cover crops on Zn contents in soil, considering that this is an essential micronutrient for plant growth, and its deficiency in soil represents a global concern, especially in tropical soils, being considered the micronutrient that most commonly limits cassava production [
51,
52].
Finally, although an improvement of the chemical properties of soil is quite desirable, the mere conservation of the parameters initially observed already represents a great advance under the perspective of sustainable cropping systems, considering that conventional cultivation practices, dependent on intense use of fertilizers and pesticides, may cause degradation of the soil properties in the medium and long term.