Forage yield, competition and economic benefit of intercropping cactus and millet with mulch in a semi-arid environment

This study aimed to quantify forage yield, economic performance, biological efficiency and competitive ability in cactus intercropped with millet, compared with their monocrops, both with and without mulch, under irrigation. The experiment was conducted over two years in semi-arid of Brazil. The experimental design was of randomised blocks with four replications and six treatments: monocropped cactus without mulch; monocropped cactus with mulch; monocropped millet without mulch; monocropped millet with mulch and cactus intercropped with millet, with and without mulch. The individual yields of fresh (270 t ha−1) and dry (23 t ha−1) matter in the cactus were not affected by the mulch or by intercropping. Mulching improved the individual yields of fresh (69.7 t ha−1) and dry (23.4 t ha−1) matter in the monocropped millet. Total dry matter was greater in the intercropping systems, both with (32.8 t ha−1) and without (31.4 t ha−1) mulching, being the average monetary advantage index equal to 8 404 BRL ha−1. The cactus-millet configuration, irrespective of the use of mulch, but irrigated in dry environments or during dry periods of the year, is more advantageous than the monocrops, because it promotes gains in production and food diversity for meeting the demand of the herd.

There are several studies with monocropped forage cactus under rainfed conditions (i.e. Silva et al. 2014aSilva et al. , 2015, and less with irrigation (Pereira et al. 2015;Morais et al. 2017; Barbosa et al. 2018). Furthermore, not much research evaluating intercropping cactus forage with other crops has been done, whether under rainfed conditions (cactus-cotton, Silva et al. 2013;cactus-sorghum, Farias et al. 2000) or irrigated (cactus-sorghum, Diniz et al. 2017;Lima et al. 2018aLima et al. , 2018bJardim et al. 2021b). Examples of using mulch in cultivation of forage cactus include studies by Carvalho et al. (2017), Amorim et al. (2017), Lopes et al. (2019) and Souza et al. (2021). For millet, in monocropped rainfed system, there is the study carried out by Santos et al. (2017), whereas Lira et al. (2020) evaluated the millet under irrigation. The pearl millet-cowpea intercropping was reported by Nelson et al. (2018). However, no study combined the use of mulch and intercropping forage cactus and millet.
We hypothesised that the use of cactus-millet intercropping has productive advantages, compared with monocropping of these species, and that the use of mulches reduces the impacts of the intercropping on crop yields, helping to decrease seasonality of forage availability in semi-arid environments and to improve the economic return to producers. For this reason, the aims of this study was: (i) quantify the yield of the forage cactus and millet in different cultivation systems, (ii) demonstrate the biological efficiency and competitive ability of the intercropping system, and (iii) evaluate the economic performance of the cactus-millet intercrop, in order to suggest the most appropriate cultivation system for a sustainable increase in forage production.

Site of the experiment
The experiment was conducted at the International Reference Centre for Agrometeorological Studies of the Cactus and other Forage Plants, at the Serra Talhada Academic Unit of the Federal Rural University of Pernambuco in Serra Talhada, Pernambuco, Brazil (7º56′20″ S, 38º17′31″ W, at an altitude of 499 m). According to the Köppen classification, the climate in the region is type BShw', characterised as semi-arid, with an average annual rainfall of 642 mm and a rainy season concentrated mostly between the summer and autumn seasons, an air temperature of 24.8 ºC, relative humidity of 63% and an accumulated atmospheric demand of 1 800 mm (Pereira et al. 2015). The soil in the area was classified as a typic eutrophic Ta Haplic Cambisol, as per criteria proposed by Embrapa (2013). The physical and chemical properties of the soil are shown in Table 1.

Treatments, experimental design and crop management
The treatments were arranged in a 3 × 2 factorial scheme, being three production systems (exclusive cactus, exclusive millet and intercropping) and two growing conditions (with and without mulch), resulting in six treatments: monocropped cactus without mulch (CNM); monocropped cactus with mulch (CWM); monocropped millet without mulch (MNM); monocropped millet with mulch (MWM) and cactus intercropped with millet, with (CMWM) and without mulch (CMNM).
The experiment had a randomised block design, with four replications. The treatments were distributed in 24 experimental units, consisting of 4 cultivation lines (i.e. for both cultures) with 4 meters in length each, totalling an area of 25.6 m 2 . Each work plot was formed by the two central rows, except one plant at each end, with an area equal to 11.52 m 2 .
For this, on 1 July 2016, the clone of the forage cactus Orelha de Elefante Mexicana -OEM -(Opuntia stricta (Haw.) Haw.), was planted in a spacing of 1.6 × 0.2 m (density equal to 31 250 plants ha −1 ). Before planting, the initial preparation of the soil was carried out, with ploughing and harrowing, and, subsequently, furrows were opened in the north-south direction for the planting of cladodes (known as basal cladodes), planted with 50% of their lengths buried. A mulch of current grass (Urochloa mosambicensis) (17.6 t ha −1 dry weight) was added on 5 January 2017. The mulch was not replaced over time. The millet (Pennisetum glaucum [L.] R. Br.) was introduced on 12 February 2017, when the forage cactus was already showing first-order cladodes (cladodes that arise from the basal cladode), to avoid any impact from the intercropping system on the initial growth of the crop and, consequently, on its establishment. The millet was sown manually in furrows approximately 0.03 m deep, with a distance of 1.6 m between the rows. In the intercropping system, millet was sown at a lateral distance of 0.4 m from each row of forage cactus (Figure 1). After 15 days of seedling emergence, thinning was carried out, leaving 20 plants per linear meter (initial density of 125 000 plants ha −1 ). The intercropping arrangement adopted was that of single rows (i.e. for each row of forage cactus, one row of millet).
The experimental trial lasted two years from the planting of forage cactus (1 July 2016), harvested on 22 June 2018. During this period, six cycles of millet were cultivated. Three cycles were conducted with the cultivar BRS-1501, where the first cycle (plant) took place between 12 February and 25 April 2017; the second cycle, referring to the first regrowth of the crop, was completed on 21 June 2017. For the third cycle, where the crop was in its second regrowth, the harvest was carried out on 5 September 2017. On 15 September 2017, the fourth cycle began, with the sowing of the cultivar IPA Bulk-1-BF, which lasted until 18 December 2017 (i.e. plant cycle). The second harvest, referring to the fifth cycle (i.e. first regrowth of cv. IPA Bulk-1-BF) was carried out on 2 March 2018; whereas the sixth and final cycle (second regrowth) was completed on 3 May 2018. Both cultivars were harvested when the plants were in the dough stage (Ullah et al. 2017) when they are commonly used for livestock feed .
Initially, we chose to use the BRS-1501 cultivar, because it has the good biomass production capacity, tolerance to water deficit and good regrowth potential. However, this cultivar was developed for the Midwest, Southeast and South regions of Brazil. On average, the plants of this cultivar reach 1.80 m in height. The cycle is medium, with flowering 50 days after sowing, reaching maturity at 120 days. Later, the cultivar IPA Bulk-1-BF was used, because it was developed specifically for semi-arid regions of the state of Pernambuco, with tolerance to water deficit, salinity and sandy soils with low fertility. On average, the IPA Bulk-1-BF cultivar can reach 2.20 m in height. Its cycle is precocious, with flowering between 50 and 60 days, and harvesting can be carried out between 45 and 70 days after sowing Ullah et al. 2017;Dias-Martins et al. 2018).
The experiment was irrigated three times a week on alternate days, by a pressurised drip system operating at a flow rate of 1.25 l h −1 , a pressure of 100 kPa and a uniformity coefficient of 93%. The drips were spaced 0.40 m apart. Irrigation management was based on 120% of the crop evapotranspiration (ET C ) of the forage cactus, with the aim of meeting the demand of both crops. The ET C was calculated as the product of the reference evapotranspiration (ET 0 ) and the crop coefficient (Kc) of 0.52, as recommended for the cactus (Queiroz et al. 2016). In turn, the ET 0 was estimated using the Penman-Monteith method (Allen et al. 1998). To calculate the ET 0 , daily meteorological data were used from an automatic weather station of the National Institute of Meteorology, close to the experimental area. The rainfall data were also collected from the same weather station. From the ET C and rainfall data, sequential water balance was estimated to ascertain the water availability in the soil over time (Rolim et al. 1998). It was considered an available soil water capacity equal to 50 mm. This routine was performed only for the cultivation system of the monocropped forage cactus without the use of mulch, because there are no values for crop coefficients for the other systems evaluated here. The water used for irrigation came from an artesian well located in the experimental area of the institution, and had an electrical conductivity of 1.62 dS m −1 , pH of 6.84, Na + of 168.66 mg l −1 and K + of 28.17 mg l −1 , and was characterised as brackish.
During the study, fertilisation was carried out based on the nutritional requirements of forage cactus, considered the main crop in the system. On two occasions (January and July 2017) 525 kg ha −1 (73.5 kg N ha −1 , 94.5 kg K 2 O ha −1 and 84.0 kg S ha −1 ) were applied based on the formulation 14-00-18 + 16S, following the recommendation of Diniz et al. (2017). Whenever necessary, manual weeding was carried out to control invasive plants in the experimental area.

Decomposition coefficient of the mulch and forage yield of the crops
The decomposition coefficient (k, d −1 ) was calculated by fitting the exponential equation to the dry matter data of the remaining mulch (DMRM, t ha −1 ) as a function of the number of days after adding the mulch (DAM) (DMRM = 17.6 exp (−k DAM) ). The model was used to estimate the mulch decomposition rate, average half-life (t 1/2 = ln(2)/k, in days), the time required to decompose 50% of the initial dry matter, and time at which 5% of the dry mass remained (t 5/100 = 3/k, in days).
At the end of each millet cycle, the number of plants was counted in two linear metres of the two central rows of each plot. Because of the uniformity of the plants present in the plot, ten representative plants were selected, cut at 0.10 m from the soil, and weighed on an electronic scale to extrapolate the crop yield. Of these plants, three were fractionated (green leaves, dead leaves, stem, panicle and grain), packed in paper bags, weighed on a semi-analytical balance to obtain the fresh weight and placed in a forced circulation oven at 55 °C to constant weight (dry weight).
The forage cactus was harvested on 22 June 2018. All the plants in the working plots were collected and weighed on an electronic balance, leaving only the basal and primary cladodes in the field. At that time, the number of plants per plot was quantified to obtain the final plant density. Three representative cladodes were weighed (fresh weight), fractionated, packed in paper bags and placed in a forced ventilation oven at 55 °C to constant weight (dry weight).

Indices of biological efficiency
For more details on biological efficiency and competitive ability indices, see Appendix 1. The indices of biological efficiency of the cactus-millet intercrop (land equivalency ratio, LER, area time equivalent ratio, ATER, land equivalent coefficient, LEC and system productivity index, SPI) were Treat.: Treatment. Ds: bulk density. Dp: particle density. Pt: total porosity. St: total sand. S: silt. EC: electrical conductivity of the saturation extract. SB: sum of bases. CEC: cation exchange capacity. V: base saturation.
where, LER m = partial land equivalency ratio of millet, calculated as LER m = Y Y m mc , being Y m and Y mc the yields of monocropped and intercropped millet, respectively, and LER c = partial land equivalency ratio of cactus, calculated as LER c = Y Y c cc , with Y c and Y cc representing the yields of monocropped and intercropped cactus, respectively. When LER > 1 the intercropping system has a productive advantage over the monocropped system, whereas if LER = 1 there is no productive advantage, and if LER < 1 there is a disadvantage to adopting the intercropping system (Yilmaz et al. 2014).
where, LER c and LER m = partial land equivalency ratio of the cactus and millet crops, respectively, t c and t m = length of the cactus and millet cycle in days, respectively, t cm = total time of the intercropping system in days. If ATER > 1 the intercropping system has a productive advantage, if ATER = 1 there is no advantage, and if ATER < 1 there is a disadvantage to intercropping (Diniz et al. 2017). In this case, t c and T cm = 722 days (from planting to harvesting the cactus), and t m = 418 days (suming up all millet cycles).
where, if LEC > 0.25, there is a productive advantage to the intercropping system, because the minimum production coefficient is 25% (Diniz et al. 2017).
where the SPI equates the yield of the millet to that of the forage cactus (Sadeghpour et al. 2013).

Indices of competitive ability
The indices of competitive ability of the cactus-millet intercrop (coefficient of relative density (K), aggressiveness (A), actual loss or gain in yield (ALGY) and competitiveness ratio (CR)), were obtained as per Equations 5, 6, 7 and 8 (Sadeghpour et al. 2013;Diniz et al. 2017): where, X cm = proportion of the forage cactus (20%, 31 250 plants ha −1 ) intercropped with millet (80%, 125 000 plants ha −1 ); X mc = proportion of millet intercropped with the cactus. If the product of the two coefficients, i.e. K = (K cm K mc ) > 1, there is a productive advantage to the intercropping system, compared with the monocropped system, if K = 1 there is no productive advantage, and if K < 1 there is a disadvantage to using the intercrop. When K cm > K mc , it indicates that the forage cactus is far more competitive than the millet (Sadeghpour et al. 2013).
where, when A cm = 0, both crops are equally competitive, whereas when A cm > 0 the cactus is dominant over the millet and when A cm < 0 the millet is dominant over the cactus (Sadeghpour et al. 2013). The same calculation is used for the secondary crop (A mc ).
where, if ALGY is positive (ALGY > 0) it suggests an advantage of the intercropping system over the monocropped system, whereas if ALGY is negative (ALGY < 0) it suggests a disadvantage of the intercropping system (Diniz et al. 2017).
where, if CR c < 1, there is a positive benefit to intercropping and the species can be grown together, whereas if CR c > 1 the intercrop is more competitive, compared with the monocrop, and intercropping is not indicated (Sadeghpour et al. 2013). This same interpretation is applied to the millet (CRm).

Economic benefits
The economic analysis was made based on the monetary advantage index (MAI), expressed in BRL ha −1 . The LER was used to calculate the MAI, as shown in Equation 9 (Ghosh 2004).
where, NR = Net revenue of the intercropping system (BRL ha −1 ). The greater the MAI, the more profitable the intercropping system.

Statistical analysis
All the data were submitted to the tests of normality and homoscedasticity followed by an analysis of variance by F-test (p < 0.05). Where the results were significant, the mean values were compared using Tukey's test at 5% probability. Each analysis was carried out using the R Core Team software (2018).

Environmental variables
Throughout the complete cactus cycle (July 2016 to June 2018), the total cumulative rainfall was 1 069.2 mm, and the ET 0 was 3 620.69 mm. The high magnitude of ET 0 is as a result of the high intensity of solar radiation (Figure 2), because the study region has low cloud cover and is located at low altitude. Of these totals, 982.6 mm of rain and 2 417.49 mm of ET 0 occurred after January 2017, when the irrigation events, the addition of mulch and the planting of millet began. The total amount of water applied via irrigation was 855.32 mm, and with the rainfall, totalled 76% of the ET 0 . During the trial, there were two predominant rainy periods: February to April 2017 and February to April 2018, with some irregular rainfall events occurring from May to July 2017. The greatest atmospheric demand was from September to December 2017.

Rate of decomposition of the mulch
The monocropped millet showed a lower decomposition coefficient for the mulch (k = 0.0039 d −1 ) (Figure 3), resulting in a final DMRM value of 2.9 t ha −1 , equivalent to 17% of the initial amount. On the other hand, k for the systems that included the cactus (monocropped and intercropped with millet) were similar, suggesting greater decomposition of the mulch and less dry matter remaining at the end of the cycle. The monocropped cactus, monocropped millet and the intercrop showed a half-life (t 1/2 ) of 147, 178 and 141 days.

Forage yield
There was no significant interaction (p > 0.05) between cropping systems and mulch levels for the variables in Table 2 (FM c , FM t , DM c , DM t , DMC c , DMC m and DMC t ).
As for the individual yields of fresh matter and dry matter of millet, there was a significant interaction between the factors (p < 0.05) ( Table 3). The individual yields of fresh and dry matter of the cactus were not influenced by the addition of mulch or the intercrop with millet ( Table 2). The analysis of variance (ANOVA) of the total fresh matter and total dry matter yields, and the total dry matter content, is presented in Supplementary  Supplementary Table S2. The millet showed a reduction of more than 50% in the individual production parameters, because of the intercropping (Table 3). The mulch increased the yield of fresh (+42%) and dry (+36%) matter in the monocropped millet, compared with the monocropped millet without mulch. The mulch had no effect on the dry matter content of the millet (Table 2).
There was a significant interaction (p < 0.05) between cropping systems and mulch levels only in the first, fifth and sixth cycles (Figure 4). At the beginning, after adding the mulch and planting the millet (between January and February 2017), when the level of mulch was still high (~17.2 t ha −1 ) and the cladode area index of the cactus was low (data not shown), it can be seen from Figure 4, that the millet had the highest yields. As expected, there was a reduction in productivity for the two subsequent regrowth cycles, but even after replanting the millet in September 2017, the yield was seen to be low, compared with the first millet cycle; the reason was as result of competition from the forage cactus. The ANOVA of the individual dry matter yields of the six millet cycles is shown in Supplementary Table S3.     Equal uppercase letters across mulch levels, and equal lowercase letters across cropping systems do not differ by Tukey's test at 5% probability. The standard deviation is shown above the bars. For more details, see Supplementary Table S3 For total forage fresh matter (Table 2), it was found that adopting the cactus-millet intercrop did not significantly increase productivity (p > 0.05). On the other hand, in terms of dry matter, a significant increase in production was seen, irrespective of the presence of mulch.

Indices of competitiveness and economic benefits
The mulch promoted no changes in the indices of biological efficiency (LER, ATER, LEC and SPI), competitive ability (K, A, ALGY and CR) or the MAI (Tables 4 and 5). The ANOVA of the biological efficiency and competitive ability indices is presented in Supplementary Tables S4 and S5, respectively.
The cactus-millet intercrop showed good results in terms of biological efficiency (Table 4). Despite LER c and LER m both having values of <1, indicating that both crops had a productive disadvantage under the intercropping system, the values for LER (LER c + LER m ) were greater than one (on average, 1.42), showing that the intercropping system was more efficient in the use of land than the monocrops.  The ATER also had results >1, and the LEC was greater than 25%. The SPI had a mean value of 34.3, which is expressed in t ha −1 dry matter. Positive values were found for the MAI (on average, 8 404 BRL ha −1 ), suggesting an economic advantage of the intercropping system over the monocrop systems of cactus and millet.
For the competitive ability of the intercrop, the indices showed that the forage cactus was dominant over the millet (Table 5). K values in the cactus were negative, whereas K values in the millet were less than 1.0. Positive results for aggressiveness in the cactus (A c ) and negative results in the millet (A m ), as well as values greater than 1.0 for CR c , and lower values for CR m , demonstrate the dominance of the cactus over the millet in the intercrop. In terms of productivity, ALGY c was positive and superior to ALGY m , and in general, the total value for ALGY was positive, showing an advantage for the intercropping system over the monocrops.

Decomposition of the mulch and forage yield of the crops
The use of mulch increased productivity in the monocropped millet only (36%), without affecting the yield of the monocropped cactus or the cactus-millet intercrop, which may be associated with the higher rate of decomposition of the mulch in the latter systems (0.0047 d −1 and 0.0049 d −1 ). These values of k can be considered high, when compared with others studies, which it verified values of k varying by 0.0021 to 0.0041 d −1 using 3.5 and 14.0 t ha −1 of trash in the field (Pimentel et al. 2019), and of 0.0012 and 0.0026 d −1 , when used 3.5 and 21.0 t ha −1 , respectively (Sousa Jr et al. 2017). The value of k depends on mulch quantity, composition, and management, microbial activity, and environmental conditions (Sousa Jr et al. 2017;Pimentel et al. 2019), and the hot Brazilian semi-arid region favors the decomposition rate in irrigated fields (Zhou et al. 2016). For the millet system, the lower rate of decomposition (k = 0.0039 d −1 ) and, consequently, longer half-life (t 1/2 ) (178 days), guarantee the benefits of maintaining the mulch in the production system for a longer period. In this cropping system, the effect of mulch would persist up to 769 days after its deposition in the field (t 5/100 , time at which 5% of the dry mass would remain). In addition, the less aggressiveness, observed in intraspecific competition may have enabled better productive performance compared with the cactus-millet system, where (interspecific) competition is more aggressive (Jardim et al. 2021b(Jardim et al. , 2021cDiniz et al. 2017).
The mulch maintains more moisture in the soil, reduces irrigation time, adds nutrients while decomposing, helps with carbon sequestration and improves the physical properties of the soil. In addition, it has been shown that mulching leads to an increase in the concentration of chlorophyll a and b, and of total carotenoid content in the leaves of plants, and facilitates root growth and nutrient absorption, explaining the increase in productivity (El-mageed et al. 2016(El-mageed et al. , 2018; however, the benefits of mulch vary depending on the amount of mulch, weather conditions, soil texture, length of the test period, and amount of water entering the system (Qin et al. 2015(Qin et al. , 2017. These last factors possibly explain the absence of the mulch effect in the cactus monoculture system. In the present study, the water depth, via irrigation, was applied at short intervals (i.e. watering shift of 1 day), which probably prevented the significant effects of coverage. Furthermore, the lack of any effect on palm productivity may be associated with the high amount of water (irrigation + rain = 1 924.5 mm) received during the experimental period. A similar result was reported by Queiroz et al. (2015) and Araújo Júnior et al. (2021).
The intercrop did not affect the individual yield of the forage cactus (Table 2); this may be related to the crop already having stabilised when the millet was planted. As such, when intercropping the cactus with grasses, first-order cladodes must be present, as in the present study, so that there is no impact on cactus yield. Similar management was employed by Diniz et al. (2017) and Lima et al. (2018a). Diniz et al. (2017) no observed effects significant of the intercropping cactus-sorghum on the yield individual of the forage cactus, however, they found a significant reduction in the production of fresh and dry matter in the sorghum crop. The aggressiveness presented by the OEM cactus is one of the characteristics that favour an excellent adaptation of this clone in intercropping systems (Jardim et al. 2021b(Jardim et al. , 2021c. Whereas for the millet, it was found that intercropping promoted a reduction in crop productivity, especially as the cycles progressed. When the millet was planted, the cactus was already 442 days old, which supposedly reduced the ability of the millet to develop deeper roots (Nelson et al. 2018). Even when planting the new cultivar (IPA Bulk-1-BF) in September 2017, competition with the cactus was found to inhibit the performance of the millet. During this period, the high atmospheric demand, the absence of rain and  Table 5: Coefficient of relative density (K), aggressiveness (A), actual loss or gain in yield (ALGY) and competitiveness ratio (CR) in the cactus-millet intercrop (CM) with and without mulch the irrigation management based on 120% of the cactus evapotranspiration may also have made it difficult for the millet to become established, as a result of the competition for water. Cacti of the genus Opuntia, when subjected to irrigation, have greater adaptive capacity to the environment and plasticity of their photosynthetic metabolism, which favors better performance in the cultivation site (Jardim et al. 2021c). In addition, the rows of both crops were planted in a north-south direction, which affected the radiation intercepted by the millet during the early days of the cycle. Each of these factors hampers the stability and vigour of the millet in the intercropping system . This trend was even more noticeable during the first, second, third and fifth crop cycles (Figure 4), which resulted in a lower total yield for the millet (Table 3). Havilah (2011) states that, although the millet can withstand low fertility and drought, its yield is very sensitive to poor initial establishment, which explains the impact of the intercrop on the productivity of the species. In the configuration millet-cowpea, Nelson et al. (2018) verified a reduction of 55% of the total grain yield of millet in comparison to the monocropped equivalent. In our study, such as the fresh and dry matter yield of the millet was upper in the intercropping with mulch, the reduction was greater (56%), when compared with the intercropping without mulch (42%). However, even with a greater drop in yield, the intercropping system with mulch produced 89% and 103% more fresh and dry matter than the intercropping system without mulch, proving the benefits of the intercropped cactus-millet.
The increase in the total dry matter yield of the intercrop when including the productivity of both crops, shows that this practice really is very important in reducing the seasonality of forage production in regions with limited water, not only in quantitative terms, but also in terms of quality, because it increases food diversity for the animals. Other studies have also noted that, with agricultural crops, intercropping increases the overall productivity of the production system (Sadeghpour et al. 2013;Temesgen et al. 2015;Masvaya et al. 2017), especially the cactus-sorghum configuration in the semi-arid region of Brazil, where the cactus is a widely used resource in animal feed (Diniz et al. 2017;Lima et al. 2018a). Lima et al. (2018a), during the third productive year of forage cactus with two cycles of sorghum, observed increase dry matter yield by 260% of the cultivation system, compared with the forage cactus monocropping. Already, Diniz et al. (2017) cite an increase of 62% during the fourth productive year of the forage cactus also with the cultivation of two cycles of sorghum.

Indices of competitiveness and economic benefits
The mean value of 1.42 for the LER, irrespective of the use of mulch, shows that the monocrops would require an additional 42% of land (0.42 ha) for production to equal that of the intercropping system (Sadeghpour et al. 2013;Yilmaz et al. 2014). Studies on land use efficiency in the forage cactus intercropped with other plants are still scarce. Silva et al. (2013), studying different intercropping systems, found a respective LER of 2.58, 2.99 and 2.83 for the cactuscotton-sesame, cactus-cotton-peanut and cactus-cotton configurations. Diniz et al. (2017) obtained a LER of 1.51 for a cactus-sorghum intercrop, but according to the authors, the LER does not consider the length of the crop cycle, which may overestimate the advantage of the intercropping system. This explains the use of the ATER index, which in the present study was 1.20, showing the advantage of the cactus-millet intercrop. This result can be confirmed by the LEC value (0.46), which was greater than 0.25, as suggested by Diniz et al. (2017). Those authors found an ATER of 1.30 and LEC of 0.58, showing advantage the intercropping the cactus-sorghum. In the present study, the SPI, used to equate the millet yield to that of the cactus, was equal to 34.3 t MS ha −1 , which is higher than the yield of the monocropped cactus (23.8 t ha −1 ), confirming the use of this intercropping system for forage production.
In terms of competitive ability, the negative values for Kc and K demonstrate that cactus yield in the intercropping system was higher than in the monocropped system, suggesting intraspecific competition. The A index and the CR index show that the cactus was strongly dominant over the millet. The cactus was also dominant over forage sorghum in an intercropping system conducted by Diniz et al. (2017) and Jardim et al. (2021bJardim et al. ( , 2021c. The positive and higher values for ALGY c , compared with ALGY m indicate that, as an intercrop, millet is more susceptible to a loss in yield than the forage cactus (Yilmaz et al. 2014;Machiani et al. 2018). The MAI demonstrates the economic benefit of the cactus-millet intercrop, compared with the monocrops, showing that the larger LER increases the MAI (Ghosh 2004;Sadeghpour et al. 2013). Lima et al. (2018a) also found a monetary advantage in an irrigated cactussorghum intercrop.

Conclusion
This study was dedicated to analysing the benefits of using adapted species (forage cactus and millet), with the minimal use of brackish water via irrigation together with mulching and intercropping, on forage production, biological efficiency, competitive ability and economic return. Mulching had no influence on yield in the forage cactus, whether as a monocrop or intercrop, whereas it greatly contributed to an increase in the fresh and dry matter of the monocropped millet. Planting the millet only after the first-order cladodes of the forage cactus were established, inhibited the effect of intercropping on the individual yield of the cactus. The cactus-millet intercrop displayed biological efficiency, competitive ability and economic return, demonstrating its sustainability and viability for arid and semi-arid environments with strong rainfall seasonality and with a water regime sufficient for the minimal use of brackish water via irrigation. Additional research is suggested into water depths and irrigation frequencies, other planting guidelines, different planting times for intercropped systems with millet, and different levels of mulch.

Index
Description Interpretation LER Expresses the proportion of land, in hectare, that the monoculture system needs to have the same yield as the intercropping system LER > 1: there is an advantage in using the intercropping. LER = 1: there is no productive advantage. LER < 1: there is a disadvantage in using the intercropping. ATER It is necessary to consider the time factor, especially when crops have different production cycles ATER>1: there is an advantage in using the intercropping. ATER = 1: there is no productive advantage. ATER < 1: there is a disadvantage in using the intercropping. LEC It requires a minimum productive coefficient of 25% LEC > 0.25: there is an advantage in using the intercropping. SPI Is used to match the yield of the secondary crop (millet) with the yield of the main crop (cactus) SPI > dry matter yield of the main crop (cactus) in monoculture: there is an advantage in using the intercropping.

MAI
Expresses the economic viability of the intercropping based on net revenue the higher the index value, the greater the economic advantage.

K
Expresses both the viability of the intercropping, as well as the relative dominance of one crop over another K > 1: there is an advantage in using the intercropping. K = 1: there is no productive advantage. K < 1: there is a disadvantage in using the intercropping. K c > K m : the cactus is dominant over the millet. A Expresses the aggressiveness of one crop over another in a intercropped system A c > 0: cactus is more competitive than millet. A c < 0: millet is more competitive than cactus. A c = 0: both crops are competitive. ALGY Expresses the loss and/or proportional gain in the productivity of the intercropping in relation to monocropped ALGY > 0: there is an advantage in using the intercropping. ALGY < 0: there is a disadvantage in using the intercropping.

CR
Expresses the number of times one crop is more competitive than another CR c < 1: there is positive benefit for intercropping and species can be cultivated together. CR c > 1: intercropping is more competitive compared to monocropped, and intercropping is not indicated. This same interpretation is applied to CRm.
LER: land equivalency ratio. ATER: area time equivalent ratio. LEC: land equivalent coefficient. SPI: system productivity index. MAI: monetary advantage index. K: coefficient of relative density. A: aggressiveness. ALGY: actual loss or gain in yield. CR: competitiveness ratio.