INTRODUCTION
By the year 2050, growth in the global population and shifts in diet may require an associated 70% increase in global food production. Demand for meat (and to lesser extent for milk) is directly correlated with per capita real income, and is increasing at an even higher rate, particularly in developing nations (Tilman et al., Reference Tilman, Balzer, Hill and Befort2011). Current efforts have focused on the intensification of livestock systems in developed countries and greater land clearing (extensification) in developing nations. If this trend is to continue, an estimated one billion ha of land would need to be cleared globally by 2050, representing a 30% increase over current pasture area. This number could be decreased to 0.2 billion ha if policy-makers and research efforts instead focus on moderate intensification of existing agricultural systems in under-yielding regions (Tilman et al., Reference Tilman, Balzer, Hill and Befort2011). Thus, transfer of high-yielding technologies to existing production areas may substantially reduce environmental impacts, while satisfying the global food demand.
Aside from providing 25% of protein consumed worldwide, appropriately managed livestock systems have been shown to support diverse ecosystem services, including water-flow regulation and erosion control, climate regulation as well as soil biodiversity conservation (Fisher et al., Reference Fisher, Rao, Ayarza, Lascano, Sanz, Thomas and Vera1994; Lavelle et al., Reference Lavelle, Rodríguez, Arguello, Bernal, Botero, Chaparro, Gómez, Gutiérrez, Hurtado, Loaiza, Rodríguez, Sanabria, Velásquez and Fonte2014; Montenegro et al., Reference Montenegro, Barrantes and Dilorenzo2016). However, a large portion of livestock systems are based on low-yielding forage crops and apply practices that contribute to environmental degradation and high greenhouse gas emissions (Herrero et al., Reference Herrero, Havlík, Valin, Notenbaert, Rufino, Thornton and Obersteiner2013). In Central America, pastures dominated by Hyparrhenia rufa (locally known as Jaragua) were introduced to Pacific parts of the region several decades ago and are typically managed with fire to stimulate regrowth at the end of the dry season. Pasture burning has been shown to contribute to soil degradation and when not closely monitored can impact forested areas that support a range of landscape level ecosystem services (Steinfeld et al., Reference Steinfeld, Wassenaar and Jutzi2006). Such pastures are widespread throughout the region and their relatively low biomass yields suggest considerable room for improvement. Given the pervasiveness of cattle production in Central America and globally, there is great potential for more productive forages and management practices to enhance sustainability of these regions (Rao et al., Reference Rao, Peters, Castro, Schultze-Kraft, White, Fisher, Miles, Lascano, Blümmel, Bungenstab, Tapasco, Hyman, Bolliger, Paul, Hoek, Maass, Tiemann, Cuchillo, Douxchamps, Villanueva, Rincón, Ayarza, Rosenstock, Subbarao, Arango, Cardoso, Worthington, Chirinda, Notenbaert, Jenet, Schmidt, Vivas, Lefroy, Fahrney, Guimarães, Tohme, Cook, Herrero, Chocón, Searchinger and Rudel2015).
In the last 20 years, a number of improved grasses have been developed and made commercially available with the aim of increasing forage productivity (Argel et al., Reference Argel, Miles, Guiot García, Cuadrado Capella and Lascano2007; Miles et al., Reference Miles, do Valle, Rao, Euclides, Moser, Burson and Sollenberger2004; Pizarro et al., Reference Pizarro, Hare, Mutimura and Changjun2013; Rao et al., Reference Rao, Peters, Castro, Schultze-Kraft, White, Fisher, Miles, Lascano, Blümmel, Bungenstab, Tapasco, Hyman, Bolliger, Paul, Hoek, Maass, Tiemann, Cuchillo, Douxchamps, Villanueva, Rincón, Ayarza, Rosenstock, Subbarao, Arango, Cardoso, Worthington, Chirinda, Notenbaert, Jenet, Schmidt, Vivas, Lefroy, Fahrney, Guimarães, Tohme, Cook, Herrero, Chocón, Searchinger and Rudel2015). Many of these (e.g., Brachiaria sp.) are adapted to sub-optimal environments (i.e., pests, drought or waterlogging prone areas). Legumes have also been considered as potential forage crops, and in addition to their benefits, such as N-fixation and contribution to soil nutrient cycling, legumes also produce high quality feed. Similar to work on grasses, research efforts have focused on identifying and selecting legumes that are adapted to acidic soils with low to moderate fertility. Promising legumes include those in the genera Vigna and Canavalia. For instance, Canavalia brasiliensis performs well in areas with extended dry seasons (Peters et al., Reference Peters, Franco, Schmidt and Hincapié2010).
Despite the potential of these improved forages in tropical conditions, adoption by local producers has been limited. According to Rao et al. (Reference Rao, Peters, Castro, Schultze-Kraft, White, Fisher, Miles, Lascano, Blümmel, Bungenstab, Tapasco, Hyman, Bolliger, Paul, Hoek, Maass, Tiemann, Cuchillo, Douxchamps, Villanueva, Rincón, Ayarza, Rosenstock, Subbarao, Arango, Cardoso, Worthington, Chirinda, Notenbaert, Jenet, Schmidt, Vivas, Lefroy, Fahrney, Guimarães, Tohme, Cook, Herrero, Chocón, Searchinger and Rudel2015), the main constraints to the adoption of legumes and grasses have been the susceptibility to diseases and pests, the lack of clear management recommendations, seed availability and unrealistic expectations of farmers for rapid and dramatic increases in production. Another possible limitation is that development organizations often view legumes solely as cover crops and green manure, when their potential uses as feed may be much more attractive to land managers (Douxchamps et al., Reference Douxchamps, Rao, Peters, Van Der Hoek, Schmidt, Martens, Polania, Mena, Binder and Schöll2014; Kebede et al., Reference Kebede, Assefa, Feyissa and Mengistu2016). High spatial heterogeneity within and between farms can also act as a barrier to improved forage adoption by smallholders, since the optimal types and arrangements of grasses or legumes can vary widely depending on different niches within farms and landscapes, thus greatly complicating the selection process (Paul et al., Reference Paul, Muhimuzi, Bacigale, Wimba, Chiuri, Amzati and Maass2016).
The participation of local farmers in selection, adaptation and dissemination processes has been shown to increase the adoption of new innovations (Peters et al., Reference Peters, Lascano, Roothaert and De Haan2003; Pretty, Reference Pretty1995). Under this approach, farmers play an active role in the development of practices and contribute intimate knowledge of their farming systems as well as provide the social, economic and cultural context that often determines feasibility of adoption. A case study by Stür et al. (Reference Stür, Horne, Gabunada, Phengsavanh and Kerridge2002) in Southeast Asia emphasized the wide range of constraints, opportunities, and goals that are considered in farmer decision-making. Aside from high forage yields, farmers valued easy to cut herbage, fast regrowth after harvesting, low competition with adjacent crops, and ease of collection and transportation of plant material. Overall, farmers were more likely to adopt varieties that best met these locally valued criteria. In fact, Horne and Stür (Reference Horne and Stür1997) suggest that researchers may often focus on completely different forage evaluation criteria (e.g., live weight gain) than those that are the most valued by smallholders (e.g., risk management and labour constraints).
In this study, on-farm trials were conducted to evaluate the establishment and potential productivity of five improved grasses (with H. rufa as a control) and five forage legumes across seven different locations in the Dry Corridor of Central America (western Honduras and northern El Salvador), a region with a prolonged dry season that lasts five to six months. Results from the multi-site trials were combined with participatory evaluation by local producers and technicians to identify the most adapted and favourable cultivars in the region. Along with formal assessment of biomass production, diverse stakeholders were involved in a hands-on evaluation of improved grass and legume cultivars in order to understand the selection criteria that are of greatest concern to farmers and to identify the most viable options for adoption and scaling. We hypothesized that the improved grass options would outperform the widely distributed H. rufa across all sites, but that the best producing species and those mostly highly evaluated by farmers would vary according to the unique environmental contexts of each site. For legume species, we hypothesized that at least one species would perform well across local conditions and receive strong evaluations from participating farmers.
MATERIALS AND METHODS
Study site and experimental design
This study was carried out in a part of the Dry Corridor of Central America, specifically in the Chalatenango department in El Salvador and Lempira department in Honduras. Due to their close proximity (Supplementary Figure S1, available online at https://doi.org/10.1017/S0014479718000364), the sites share a similar climate and soil properties. Both areas are characterized by mountainous topography and annual crops and pastures dispersed throughout sub-humid tropical forest. Soils, generally shallow and rocky, are largely dominated by Entisols and Inceptisols (Fonte et al., Reference Fonte, Barrios and Six2010; Kearney et al., Reference Kearney, Coops, Chan, Fonte, Siles and Smukler2017). Average monthly temperature varies between 22 and 27°C and average annual precipitation is 1500 mm, with at least 90% of rainfall occurring between May and November. Economic activity in both Lempira and Chalatenango is focused on agriculture, specifically maize (Zea mays L.), sorghum (Sorghum bicolor L.) for grain and forage, and beans (Phaseolus vulgaris L.). Cattle production is becoming increasingly more important in the region, particularly in Chalatenango.
The study was conducted from August 2014 to October 2015 at seven research sites in the region, the majority of which contained both improved grass and legume trials (Table 1). The experimental sites were located on land of local cattle producers with interest in evaluating and planting the grass varieties and legume forage options. Five grasses were tested: Brachiaria brizantha CIAT 6780 (cv. Marandu), B. brizantha CIAT 26110 (cv. Toledo), Brachiaria decumbens CIAT 606 (cv. Basilisk), Brachiara hybrid (CIAT 36087; B. ruziziensis x B. decumbens x B. brizantha cv. Mulato II), Megathyrsus maximus CIAT 6962 (cv. Mombasa; previously known as Panicum maximum, cv. Mombasa). These were compared to H. rufa as a control since this is the most commonly grown grass species in the region and likely serves as a benchmark against which new grasses would be evaluated. Five species of legumes were also evaluated: Canavalia ensiformis L., C. brasiliensis (CIAT 17009), Vigna unguiculata (cowpea), Cajanus cajan (pigeon pea) and Vigna radiata (mung bean) as supplementary protein fodder. Improved grasses and forage legumes were selected based on their performance at other sites with similar environmental conditions, farmer interest, and local seed availability (particularly in the case of legumes). All materials were tested using a randomized complete block design, with all treatments established in 4 × 4 m2 plots, and each treatment present in four replicate blocks at each experimental site.
Site locations and select soil characteristics for improved pasture and forage legume trials in El Salvador and Honduras.
*Soil texture evaluated by hand at this site, so precise numbers were not obtained.
Soil texture was determined by hydrometer method, pH using a ratio of soil to water of 2.5:1, soil organic matter (SOM) by Walkley and Black, and available P and K were evaluated using a Mehlich-3 extraction method.
Grass plots were established in August 2014 under no-till management. Rows were spaced at 50 cm with 30 cm spacing between holes and five to eight seeds per hole. Fertilizer (43 kg N ha−1 and 23 kg P ha−1) was applied in rows to the soil surface when plants were approximately 15 cm in height. Legumes were also established in August 2014, as per recommendations provided by Peters et al. (Reference Peters, Franco, Schmidt and Hincapié2010) and without fertilization. Briefly, C. ensiformis and C. brasiliensis were planted in rows spaced 50 cm apart and 30 cm spacing between holes and two seeds per hole. V. unguiculata and V. radiata were planted in rows spaced 50 cm apart with 20 cm between holes containing three seeds of V. unguiculata and 10 cm between holes containing three seeds of V. radiata. Rows of C. cajan were spaced at 1 m with 30 cm between holes, each containing 4 seeds.
Soil analyses
Baseline soils (0–20 cm) were sampled prior to the start of the experiment by collecting five sub-samples per site to form one composite sample for analysis. Upon collection, soils were air-dried and passed through a 2 mm sieve for analysis of soil texture (hydrometer method), pH using a ratio of soil to water of 2.5:1, soil organic matter (SOM; Walkley and Black), available phosphorus (P) and potassium (K) using Mehlich-3 extraction at the CENTA (Centro Nacional de Tecnología Agropecuaria y Forestal) laboratory in El Salvador.
Evaluation of forage and seed production
Biomass yield was measured at 90 days after planting (November 2014) for the grass trials at each of the seven sites to evaluate establishment. Grasses were cut to 15 cm from the soil surface in the entire plot, while a 2 × 2 m2 sub-plot in the centre of each experimental plot was used for evaluation of biomass production to avoid edge effects. In order to assess the productivity and regrowth potential in the dry season, biomass production during the six month dry season was evaluated in two sites in Honduras (San Jose and Tenango) at the start of the wet season (May 2015). Additionally, a sub-set of the trials, three sites in Honduras and El Salvador, were re-evaluated at key time points in the subsequent wet season. In July and September of 2015, pastures at these three sites were uniformly cut and left to recuperate for approximately 60 days before sampling.
Biomass of the legumes was measured when 50% of the experimental plots had reached flowering stage at each site. Half of the plants from each plot were cut to the soil surface for estimation of biomass (with the exception of pigeon pea, which was cut to a height of 60 cm to allow for potential regrowth, a unique attribute of this species; Rusinamhodzi et al., Reference Rusinamhodzi, Makoko and Sariah2017). Dry biomass was determined for each species after oven-drying samples at 60 °C. The other half of each plot was left intact to determine days to maturation and seed production potential. Seed yield was reported at a moisture content of 13%. These plots were not re-evaluated after the first harvest since not all of the species tested have the ability to regenerate successfully after cutting.
Participatory evaluation of forage materials
Approximately 60 days after planting, participatory workshops were held at three of the study sites, but involved cattle producers from all of the experimental sites. Producers first worked together with project staff to define a set of key criteria for assessing grasses and legume forage crops (Hernández, Reference Hernández Romero2007). The four main criteria included: growth, soil cover, foliage colour (all estimated visually) and perceived palatability or lusciousness (assessed by smell and texture; Table 2). These criteria were then ranked by the producers (1–10) to develop a weight of the relative importance of each to be used in the final calculation of an overall score for each grass and legume material tested. Following this discussion, six groups of 3–4 producers were formed and asked to closely observe the materials growing in all of the replicate blocks at the experimental site. Each material (grasses and legumes) was then ranked on a scale of 1–5 for each criterion (1 – poor; 2 – fair; 3 – good; 4 – very good; 5 – excellent) and scores were tallied to provide an overall weighted measure of producer acceptance. The participatory evaluation carried out here sought not only to capture farmer perceptions of the genetic materials tested, but also to facilitate dissemination of these materials and engage in preliminary training of cattle producers and local technicians.
Criteria and importance levels defined by local producers to evaluate the quality of each evaluated species of grass or legume.
Data analysis
Comparison of dry biomass production for each trial and sampling time were analysed using ANOVA. Natural log transformations were applied as necessary (mainly for the grass production data) to meet the assumptions of ANOVA (i.e., normality, homogeneity of variance). A preliminary analysis was conducted in which data across sites were analysed together, with treatment considered a main effect and both sites and blocks treated as random variables. Significant interactions between site and treatment indicated that treatment effects were better evaluated on a site-by-site basis, with only forage species and block (treated as a random variable) included in the model for each site. Tukey's honest significant difference was used to determine differences between treatments. Results from participatory evaluations by local producers were analysed with a non-parametric Kruskal–Wallis test. All statistical analysis was carried out using the software INFOSTAT and significant differences reported at the P < 0.05 level.
RESULTS
Biomass production
At the first sampling, 90 days after planting, there was considerable variation in initial grass biomass production between sites; one site in particular (San Jose) presented the highest biomass production with twice the value observed at the other sites. Overall, B. decumbens, M. maximus, B. brizantha (Marandu) and B. brizantha (Toledo) generally produced more biomass than the Brachiaria hybrid (Mulato II) and H. rufa across all sites, although the most productive grass varied across sites (Table 3). For example, in both San Jose and San Lorenzo, Honduras, B. decumbens was the most productive, with more than four times higher biomass than H. rufa. In Tenango and Upatoro, B. brizantha (Toledo) was the most productive, having five times greater biomass than H. rufa at the Upatoro site. While in Comalapa and Chalatenango, El Salvador, M. maximus was the highest yielding grass cultivar, producing significantly more than H. rufa at both sites. While never being the most productive at any particular site, B. brizantha (Marandu) was consistently high yielding across all sites showing the highest stability value in biomass production during establishment.
Mean forage production of six grasses at seven sites in Honduras and El Salvador.
a Cultivar abbreviations: Mar, Marandu; Tol, Toledo; B. hybrid, Brachiara hybrid CIAT 36087.
†ng= Seed did not germinate.
Samples were cut at a height of 15 cm above soil surface on the following times: 90 days after planting (Nov 2014), just after the dry season (May 4–15, 2015) and during the wet season, after a ~ 60 day recovery period (July 13–24, 2015 and September 8–18, 2015). Values in italics to the right of each mean represent the standard error of the four blocks tested at each site. Means with a common letter are not significantly different according Tukey's Test. P-values for treatment comparisons at each site are presented below each set of means (ns, not significant at P < 0.05).
The cumulative grass biomass production during the dry season, measured at the beginning of the wet season in May 2015, showed a dramatic decrease, considering that measurements represent production across a total of six months. This measurement in San Jose and in Tenango demonstrated a similar trend to that observed in the initial biomass measurement; B. decumbens produced three times more biomass than H. rufa in San Jose, and B. brizantha (Toledo) yielded the highest in Tenango (but was not significantly different from B. brizantha (Marandu), M. maximus or B. decumbens). While B. decumbens continued to produce the most biomass during the wet season in San Jose, significant differences were only encountered for the September 2015 sampling date. While all varieties continued to produce better than H. rufa in Comalapa and Upatoro during the July and September 2015 sampling dates, these differences were not significant (Table 3).
For the legumes, C. ensiformis and C. brasiliensis generally demonstrated the highest biomass production (except for the Isleta site in Honduras). V. unguiculata and V. radiata tended to produce less biomass, but reached their flowering stage in a much shorter period of time (Table 4). While biomass production of C. cajan was high in Upatoro, its yields were highly variable across sites, even failing to germinate in two sites. C. ensiformis, C. brasiliensis and C. cajan required about double the amount of time to reach flowering than did V. unguiculata and V. radiata. A comparison of biomass production on a per day basis showed no significant difference between species, with the exception of C. ensiformis which in Comalapa was superior to all other species except C. brasiliensis (Table 4). V. unguiculata and V. radiata were the only species to produce seed in all sites in which they were established, while C. ensiformis produced seed in three of the sites, C. brasiliensis in two of the sites, and C. cajan produced seed only in Upatoro.
Mean biomass production at flowering, days to maturity and seed production (kg ha−1) of five species of legumes across five sites in Honduras and El Salvador.
* ng= seed did not germinate; nse= was not established due to lack of seed.
†nm= did not mature within period of observation (220 days after planting).
Values in italics to the right of each mean represent the standard error of the four blocks tested at each site. Means with a common letter are not significantly different according Tukey's Test. P-values for treatment comparisons at each site are presented below each set of means (ns, not significant at P < 0.05).
Participatory evaluation of materials
As a general trend, B. decumbens, B. brizantha (Marandú), M. maximus and B. brizantha (Toledo) were the pastures most favoured by local livestock producers (Table 5). The soil cover provided by B. decumbens was particularly desirable and the volume of biomass produced by M. maximus also received high rankings. Conversely, the B. hybrid (Mulato II) and H. rufa indicated low acceptance in terms of growth and soil cover and overall quality.
Participatory evaluation of forage materials at three farmer workshops.
* Mar, Marandu; Tol, Toledo; B. hybrid, Brachiara hybrid CIAT 36087.
†Average of scent and texture rankings.
Criteria defined and evaluated by farmers on a scale of 1 to 5, where 5 is the highest ranking. A weighted average was calculated taking into consideration the producer-determined weight or importance of each criterion. Means with a common letter are not significantly different. P-values for treatment comparisons at each site are presented below each set of means (ns, not significant at P < 0.05).
Examining the sites individually, B. decumbens was ranked the highest by producers at the San José (Honduras) site, predominantly due to its soil cover, growth, and colour. In Chalatenango, all species except B. brizantha (Toledo) scored higher than the native control H. rufa. M. maximus was favoured due to its rapid growth, while B. decumbens once again received high rankings due to the soil cover it provides. In Comalapa, all species received higher rankings than H. rufa, but none were clearly favoured by producers.
For the legumes tested, C. ensiformis was the highest ranked by producers across all sites, primarily due to its growth, soil cover and colour. V. unguiculata and C. cajan scored well among producers in terms of the perceived palatability (lusciousness). In San José, soil cover provided by V. unguiculata was also noted among farmers, being ranked as favourably as C. ensiformis. Similarly, in Comalapa C. ensiformis and V. unguiculata were favoured along with C. brasiliensis for all criteria. In Chalatenango, there was no significant difference between species, but C. ensiformis was rated higher on average than the other species.
DISCUSSION
Forage production and adaptability across experimental sites
The grasses evaluated in this study demonstrated establishment and early biomass production within the expected range for these species (Peters et al., Reference Peters, Franco, Schmidt and Hincapié2010; Pizarro et al., Reference Pizarro, Hare, Mutimura and Changjun2013), thus suggesting that most of the improved materials were appropriately selected for the biophysical conditions studied here. Forage yields of improved varieties were generally higher than the H. rufa (Jaragua) control at the first sampling and in the dry season (at least for the two sites considered), but in the following wet season (July through September) this trend was less pronounced. This may be related to the short evaluation interval (~60 days) under lower than average rainfall conditions. The relatively low biomass production of the Brachiaria hybrid (Mulato II) was surprising and possibly due to the generally low soil fertility across all sites. Although Mulato II was developed to address low P availability and pH, as well as high aluminium toxicity (Argel et al., Reference Argel, Miles, Guiot García and Lascano2005), the poor fertility of soils at these sites may be unique and related more to high sand content, than issues such as aluminium toxicity, but more research is needed. With the exception of Mulato II, all of the improved grasses evaluated in the study appear to be viable options for the replacement of H. rufa due to their high forage yields and general acceptance by local producers. Nonetheless, it is important to note that the pastures tested here were grown under recommended management techniques that are often not or inadequately applied by farmers due to lack of knowledge or resources, including labour.
The substantial variability observed in top performing forages across sites highlights the need to consider site-specific conditions when making pasture recommendations to cattle producers in the region. For example, B. decumbens, which demonstrated a great capacity for soil coverage and relatively high yields across all sites could be an appropriate choice on degraded soils or soils that are highly susceptible to erosion (Peters et al., Reference Peters, Franco, Schmidt and Hincapié2010; Shriar, Reference Shriar2007). Meanwhile, M. maximus (Mombasa) demonstrated a high growth potential and high forage yields in most sites, but should not be recommended for use in degraded soils or on steep slopes due to its relatively high nutrient demand and tendency to grow in bunches and thus provide poor soil cover (Hare et al., Reference Hare, Phengphet, Songsiri and Sutin2015). Mulato II has been the grass most highly promoted in El Salvador by government institutions (possibly due to higher forage quality, including crude protein content), but was found in this study to be low yielding on sub-optimal soils and in the environmental conditions of Central America's Dry Corridor. In another study carried out in Africa involving different Brachiaria grasses, B. brizantha cv. Toledo and B. decumbens presented higher biomass production compared to Mulato II in low rainfall regions (Mutimura and Everson, Reference Mutimura and Everson2012). Additionally, other trials established in the Dry Corridor in Nicaragua (not published data) suggest lower, or at best similar, performance of Mulato II compared to B. brizantha (Marandu and Toledo) or M. maximus (cv. Mombasa). When considering all grasses tested here, poor management and/or poorly adapted recommendations may explain, in part, the low adoption rates observed in the region and this clearly illustrates the importance of site-specific evaluation.
B. brizantha (Marandu and Toledo) were relatively productive across all sites and thus appear to be resilient to soils of varying fertility and environmental conditions. B. brizantha (Toledo) has also demonstrated relative tolerance to flooding (Cardoso et al., Reference Cardoso, De La Cruz Jiménez and Rao2014), which may explain its superior biomass production in Upatoro, where topography of the site and high organic matter content suggest seasonal waterlogging. Such resilience can contribute substantially to risk reduction and should therefore be considered in addition to productivity when making local recommendations. The use of more adaptable forages, along with their diversification in forage-based production systems reduces reliance on a single species that may be susceptible to particular abiotic stresses or host-specific diseases. It should, however, be noted that diversified systems are inherently more complex and require greater knowledge and/or labour to manage. Additionally, it should be noted that many of the grasses tested here typically grow for many years (Peters et al., Reference Peters, Franco, Schmidt and Hincapié2010) and results from this study may better reflect potential establishment and early production, rather than long-term productivity. While other participatory forage evaluations have noted the value of early growth in influencing adoption rates (Stür et al., Reference Stür, Horne, Gabunada, Phengsavanh and Kerridge2002), long-term productivity is essential for the success of forage cultivars and cannot be ignored. Still, a certain level of caution is warranted in extrapolating these results to a longer time interval.
Biomass yields of the legumes were also generally within the expected range and these species are therefore likely to be suitable for the study region. The Canavalia and Vigna species also demonstrated greater regional adaptability in their full development and capacity to produce seeds even in management conditions not suited for seed production (Peters et al., Reference Peters, Franco, Schmidt and Hincapié2010). This is an important consideration for forage types (e.g., legumes) with seeds that are particularly expensive or difficult to obtain from local markets. It is recommended to rotate Vigna spp. with other forage crops, such as maize or sorghum, as this genus is reportedly susceptible to common bean pests (Katunga et al., Reference Katunga, Muhigwa, Kashala, Ipungu, Nyongombe, Maass and Peters2014). We note that only one growth cycle for legumes was considered for data collection in this study. It is important to recognize that pigeon pea, for example, can provide several harvests per year and C. brasiliensis can regenerate three times during its biannual life cycle (Costa et al., Reference Costa, Soares, Townsend, Pereira, Magalhães and Rodrigues2013; Douxchamps et al., Reference Douxchamps, Rao, Peters, Van Der Hoek, Schmidt, Martens, Polania, Mena, Binder and Schöll2014). Taking into account multiple harvests per year would likely lead to added production benefits for farmers and therefore may increase the desirability of these legumes.
Implications and recommendations for scaling
The improvement of pasture management and genetic resources in the region would be an important advancement for the productivity and sustainability of livestock systems (Rao et al., Reference Rao, Peters, Castro, Schultze-Kraft, White, Fisher, Miles, Lascano, Blümmel, Bungenstab, Tapasco, Hyman, Bolliger, Paul, Hoek, Maass, Tiemann, Cuchillo, Douxchamps, Villanueva, Rincón, Ayarza, Rosenstock, Subbarao, Arango, Cardoso, Worthington, Chirinda, Notenbaert, Jenet, Schmidt, Vivas, Lefroy, Fahrney, Guimarães, Tohme, Cook, Herrero, Chocón, Searchinger and Rudel2015). Based on the data provided here, incorporating improved grasses and legumes as forage crops could lead to a two- or three-fold increase in forage production per unit area, which allows for higher stocking rates, assuming adequate management. Many improved forage crops also have a higher nutritional quality, with protein contents up to double that of natural pastures (Kebede et al., Reference Kebede, Assefa, Feyissa and Mengistu2016; Peters et al., Reference Peters, Franco, Schmidt and Hincapié2010). Still, benefits extend beyond higher yields and improved nutritional content. Increased soil coverage associated with the improved pastures could help mitigate erosion, suppress weeds and contribute to C sequestration through the extensive root production associated with improved grasses (Fisher et al., Reference Fisher, Rao, Ayarza, Lascano, Sanz, Thomas and Vera1994; Lemaire et al., Reference Lemaire, Alan, Carvalho and Benoît2014). Improved forages have also been shown to increase the nutritional balance of livestock feed and reduce methane emissions associated with cattle production (Montenegro et al., Reference Montenegro, Barrantes and Dilorenzo2016), while forage legumes in particular can contribute to soil fertility through the fixation of atmospheric N.
To achieve the full benefits of the improved pastures, a change in management practices must accompany the change in genetic material. This region is characterized by relatively low soil fertility and a prolonged dry season, thus grazing schemes should be designed through collaboration between producers and technicians and include rotational grazing to achieve greater efficiency of grazing areas (Peters et al., Reference Peters, Lascano, Roothaert and De Haan2003; Rouquette, Reference Rouquette2015). This co-design of pasture systems also needs to consider climate change and the associated increase in drought intensity, as well as explore the suitability of multiple options (e.g., silage). Additionally, the moderate shade tolerance of improved grasses permits increasing tree density in pastures and the potential to obtain the additional benefits through implementation of agroforestry systems (Peri et al., Reference Peri, Dube, Varella, Peri, Dube and Varella2016).
The favourable response of farmers towards legume species should not be ignored in future efforts to improve livestock-based systems for meat and/or dairy production. While legume adoption as cover crops has not been as high as anticipated, legumes have a wide range of other uses that could provide additional economic benefit to farmers (Kebede et al., Reference Kebede, Assefa, Feyissa and Mengistu2016). For example, legumes could potentially be intercropped with annual crops or pastures, used for human consumption, planted in designated areas as protein banks for cut and carry management and also contribute to silage production (Costa et al., Reference Costa, Soares, Townsend, Pereira, Magalhães and Rodrigues2013; Lima-Orozco et al., Reference Lima-Orozco, Van Daele, Álvarez-Hernández and Fievez2016). Although ranked highly in both agronomic and participatory evaluations, some toxicity issues suggest that some caution should be exercised with the use of C. ensiformis as animal feed. To the contrary C. brasiliensis has been used as forage and green manure in smallholder crop-livestock system of the Nicaraguan hillsides. In these systems, C. brasiliensis is intercropped with maize and during the dry season the maize-Canavalia plots are grazed, allowing the animals to consume the maize stover and the green C. brasiliensis biomass (Douxchamps et al., Reference Douxchamps, Frossard, Uehlinger, Rao, Van Der Hoek, Mena, Schmidt and Oberson2012). Silage could be of particular importance in this region since it is already a widely utilized in parts of the region and offers great potential to meet livestock needs during the dry season when high quality forage is scarce. However, the use of silage and/or cut-and-carry systems depends on the ability of land managers, especially smallholders, to protect land from grazing. More research is needed regarding the nutritional quality of legumes as fodder silage and costs of utilizing legumes vs. traditional maize silage (Reiber et al., Reference Reiber, Schultze-Kraft, Peters, Lentes and Hoffman2010). We suspect that improved familiarity of these legumes and efforts to better integrate them with a systems perspective could further improve perception of legumes and facilitate future adoption. We also note that increased focus on diary production, which typically has more frequent and faster revenue return than beef systems, could improve the ability of smallholders to invest in improved forages.
Participatory evaluation of pasture systems
This study emphasizes the importance of a participatory approach to establish more productive and sustainable livestock production systems in the region. Involvement of local producers informs the assessment of adaptability of new species while increasing the potential of adoption and impact (Horne and Stür, Reference Horne and Stür1997; Peters et al., Reference Peters, Lascano, Roothaert and De Haan2003). The participatory methodology utilized in this study to evaluate forage species proved to be effective, as farmer response closely coincided with the agronomic data that were subsequently collected. Local input allowed the evaluation to extend beyond establishment and early biomass production, including farmers’ criteria such as lusciousness and foliage colour. Farmer evaluations can differ from scientific findings. For example, when ranking perceived palatability (scent and texture), farmers favoured the Brachiaria hybrid (Mulato II), B. brizantha (Marandu), B. decumbens, C. cajan and V. unguiculata, while according to Peters et al. (Reference Peters, Franco, Schmidt and Hincapié2010) B. decumbens is not considered to have high palatability in Central America.
The involvement of farmers in the research process can lead to increased adoption of improved forages. Participating farmers have the opportunity to observe favourable attributes on their own land, such as improved soil coverage of B. decumbens and C. ensiformis, and are more likely to promote these materials amongst their neighbours. As a result, adoption of the improved pastures and legumes within the study area has been widespread following the completion of this research (Smukler et al., Reference Smukler, Barillas, Siles, Garcia, Kearney and Fonte2017). While the findings presented here are encouraging, further experimentation (by farmers and researchers) is needed to better understand the role of inter-annual variability in driving the performance of these improved forage options.
CONCLUSIONS
In the face of rising demand for animal products, sustainability and productivity of smallholder livestock systems must be increased. Four of the five improved grasses – B. brizantha (cv. Marandu), B. brizantha (cv. Toledo), B. decumbens (cv. Basilisk) and M. maximus (cv. Mombasa) – exhibited high production potential and could therefore be considered viable replacements for traditional pastures, (i.e., H. rufa, cv. Jaragua). This suggests important benefits for forage production as well as soil conservation efforts, since H. rufa is typically burned annually and has poor soil cover at the onset of the rainy season. Forage legumes, specifically of the genera Canavalia and Vigna, also showed high regional adaptability. The multiple uses of these forages and their favourable response by farmers should help to inform future research efforts regarding their incorporation into livestock systems. In this study, participatory evaluation appears to be an effective approach for evaluating the performance and potential for adoption of forage crops across sites. This is supported by the fact that farmer evaluations largely agreed with the observed biomass production and their perceptions of forage quality (i.e., lusciousness) will likely be an important factor driving adoption. The materials evaluated here show a great potential for diffusion throughout Central America and similar regions, but additional studies are needed to better understand how inter-annual variability and environmental differences across sites affect not only biomass production, but also the nutritional value of the forage produced. Future research and dissemination efforts should seek to promote optimal management practices and explore the co-design of pasture systems together with researchers, technicians and local land managers. This approach would better facilitate the development and adoption of locally adapted pastures that contribute to the long-term sustainability of tropical livestock systems.
Acknowledgements
We thank Daniel Velasquez and Carlos Romero for their hard work in the management and supervision of the multi-site trials. The community ‘La Montañona’, the Agronomy Department at the University of El Salvador, the AZAMBIO Cooperative, the INFRAMS Institute in Chalatenango, and the Community Technical Institute in Candelaria were integral in research planning and implementation. We'd also like to thank the farmers in El Salvador and Honduras for providing their land and participating in the study. Funding for this work was provided by Research Program on Integrated Systems for the Humid Tropics of CGIAR as well as the United States Agency for International Development (USAID) [award number AID-519-A-12-00002-00]. USAID had no role in the study design, collection, analysis and interpretation of data, writing of the report, nor the decision to submit the article for publication. No other conflicts of interest exist.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S0014479718000364
