Worldwide, the trade in coffee (Coffea arabica, C. canephora) is rising, driven by its increasing consumption as a beverage, including in Africa (International Coffee Organization, 2022). This global increase in demand is largely supported by rapidly expanding plantations in Asia (Krishnan, 2017). So, with increasing demand, why is Africa’s coffee production declining? In Africa, coffee supports the livelihoods of some 10.9 million African farmers, who produce 15.7 million bags per crop year, and which accounts for approximately 13% of global production (International Coffee Organization, 2015). Across East Africa, a general decline in coffee production is attributed to various factors, including loss of land to real estate. But for remaining plantations, pests and diseases rank high among the reasons for production losses (Gitonga and Townsend, 2019). Plant-parasitic nematodes, however, are rarely indicated as a threat, or as a cause for concern even. There is in general, an almost complete lack of knowledge of nematodes as pests of coffee within the coffee sector in East Africa (Souza, 2008; Coyne et al., 2018; Villain et al., 2018). This is despite being considered amongst the most serious production constraints of coffee in other parts of the world, such as in Latin America and Asia (Souza, 2008; Villain et al., 2018).

In Africa, this declining trend in coffee production is particularly evident in Kenya, where coffee has traditionally been one of the country’s most important export crops (Coffee Research Institute - Kenya Agricultural and Livestock Research Organization, 2022). Although demand for Kenyan coffee remains high on the global market, due to the high quality of the Arabica type primarily grown, investments in new plantations are limited, with yields sliding in the ageing plantations. With perennial crops, pest and disease issues often build up gradually over time, especially nematode pests (Campos and Silva, 2008). The hallmark of nematode infection in mature plantations is the gradual suppression of growth and productivity over time, without necessarily causing plant death. Nematodes are especially aggressive on transplanted young seedlings, while on mature plants they can induce nutrient deficiencies, defoliation, stunting and ultimately impact yield (Villain et al., 2018). Above-ground symptoms of nematode damage, however, are notoriously difficult to diagnose, as they are otherwise indistinguishable from general plant malaise and stunting due to low soil fertility, nutrient deficiencies or low water availability and drought susceptibility (Coyne et al., 2018). Considering the damage that nematodes are known to cause on coffee in Latin America and Asia, the likelihood that they are also imposing a sizable damage on Kenyan coffee production is high.

Globally, root-knot nematodes (RKN; Meloidogyne spp.) are viewed as the number one nematode pest across crops (Jones et al., 2013) and considered as the greatest biotic threat to agriculture in sub-Saharan Africa (Coyne et al., 2018). On coffee, they are also the most commonly occurring and are generally regarded as the most important nematode pest (Villain et al., 2018).

There are several RKN species that infect coffee, as well as other genera of plant-parasitic nematodes (Humphreys-Pereira et al., 2014; Nzesya et al., 2014; Janssen et al., 2016; Gitonga, 2020). Among the coffee-parasitic RKN species, 17 have been recognized worldwide: M. exigua, M. Africana, M. arabicida, M. arenaria, M. coffeicola, M. decalineata, M. hapla, M. incognita, M. inornate, M. izalcoensis, M. javanica, M. kikuyensis, M. konaensis, M. mayaguensis, M. megadora, M. oteigae and M. paranaensis (Carneiro and Cofcewicz, 2008). Nematologists and extensionists must identify the species of parasitic nematodes infesting a coffee field at the species level in order to determine the appropriate management strategies, especially with regards to which resistance sources should be employed (Campos and Silva, 2008).

In East Africa, the information on species distribution and incidence in the region is scant and erratic, with most records documenting their detection only. Data on the levels of damage nematodes cause to coffee in the region, or their economic importance, is effectively non-existent and yet to be determined. Coffee yield losses due to nematode infection is based mostly on data from Latin America and documented as between 10 to 35%, depending on species (Barriga, 1976; Bertrand et al., 1997; López-Lima et al., 2015), but with some indications of wholescale plantation destruction in Brazil (Lordello, 1984). Considering the perennial nature of coffee, however, accurately estimating the economic losses due to nematodes is difficult. In addition, loss estimates should also consider the impact of infection on plantation longevity and the replant cost of non-performing trees.

The management of soil-borne pests and diseases on perennial crops presents numerous challenges. Once established, they become extremely difficult to eliminate, including nematodes. Reducing infection through the use of clean, healthy, resistant and protected seedlings is therefore an advised nematode management strategy, and a policy which is being actively promoted in Latin America and Asia (Villain et al., 2018). In Africa, with an almost total ignorance of coffee nematode pests, few if any management practices are observed or followed. With limited recognition and costly or unsuitable management options available, such as total renovation of aged plantations, the outlook does not look positive. Smallholder farmers also represent a high proportion of coffee farmers in the region, who are less able or flexible to incorporate nematode management practices.

In addition to the practical difficulties of nematode management in coffee fields, sustainability aspects also need to be taken into consideration. Nematicides are often a first line of defense, especially in more commercial farming systems. Biological control options, however, offer environmentally sensitive alternatives, especially under organically oriented systems. And in regard to this, there is increasing interest in the use of biologically based products for pest and disease management and improving coffee production (Alwora and Gichuru, 2014). Various microbial antagonists have been identified for use on coffee, such as the bacterium Pasteuria penetrans (Sharma and Lordello, 1992; Maximiniano et al., 2001) and egg parasitic fungi (Naves and Campos, 1991; Ribeiro and Campos, 1993). Trichoderma spp. is the most studied biocontrol fungi and has been extensively investigated and utilized for their capacity to compete and parasitize phytopathogens as well as to mitigate unfavorable growth conditions (Poveda, 2021). Purpureocillium lilacinum (=Paecillomyces lilacinus) is also a well-recognized fungal biocontrol agent for use against root-knot nematodes (Holland et al., 1999; Kiewnick and Sikora, 2006; Heidari and Olia, 2016). On coffee, Campos and Campos (1997) demonstrated the potential of P. lilacinum against Meloidogyne exigua on seedlings in Brazil. Additionally, Trichoderma harzianum (Arita et al., 2020; Zinger et al., 2021) and P. lilacinum (Arita et al., 2020) have been evaluated for efficacy against RKN on coffee in Brazil. Although they were ineffective against the nematodes, P. lilacinum promoted coffee plant growth (Arita et al., 2020). In Kenya, both P. lilacinum and Trichoderma asperellum are registered for use against nematodes and are commercially available, where they have been proven effective against RKN on pineapple (Kiriga et al., 2018) and T. asperellum against Radopholus similis on banana (Kisaakye et al., 2023).

Given the current scarcity of suitable or sustainable nematode management options amongst a rising global demand for nematode control, there is much current interest to identify options including biologically based alternatives to synthetic chemicals (Campos and Silva, 2008; Coyne et al., 2018). However, how the long-term and intensive application of biological products to perennial crops impacts on the soil food web and soil ecosystem functions has been little studied. Some studies have demonstrated the positive effect of Trichoderma spp. on enhancing the species richness and evenness of microbial distribution (Alexander and Phin, 2014) as well as the improved availability of nutrients for microbial consumption and ultimately promotion of plant health (Cai et al., 2015; Carvalhais et al., 2015). Conversely, some studies have indicated that application of microbial products disrupts the microbial community structure in the rhizosphere leading to impaired nutrient absorption (Ros et al., 2017; Halifu et al., 2019). Free-living nematodes have long been recognized as useful bioindicators of soil health due to their high abundance, rapid response to new resources and food specificity (Yeates et al., 1993). Since the development of nematode community indices (Bongers, 1990; Ferris et al., 2001; Ferris and Bongers, 2009), attention towards the use of nematodes as a quantitative measure to evaluate the impact of management practices on soil health (Okada and Harada, 2007) and measure the ecological impact (Krashevska et al., 2019) has increased.

This current study evaluated the application of two commercial products based on P. lilacinum (IMPEDE^®, RealIPM, Kenya) and T. asperellum (SUSTAIN^®, RealIPM, Kenya) over an 18-month period on Kenyan Arabica coffee suffering from Meloidogyne infection. The study focused on in-field application to established coffee fields of various ages. The objectives were to evaluate the efficacy of T. asperellum and P. lilacinum to suppress field densities of plant-parasitic nematodes, and to evaluate these products for their impact on the soil food web and soil health using nematode community assemblages as a basis.

Over a two-year period, the application of T. asperellum and P. lilacinum products demonstrated that even over this relatively short period, RKN densities, and consequently damage, could be suppressed on established Arabica coffee plantations. Some positive changes in the soil food web structure were also observed following the application of these biocontrol agents. How sustainable this is, in terms of reversing such damage, yet needs to be determined, as does the economic viability of such a practice. But this study demonstrates the potential for the use of biocontrol agents on perennial crops. It took up to one year before we detected endophytic colonization of the fungi in coffee roots, which was longer than we anticipated. However, this could have been affected by tree age, some of which are approaching 100 years, while unfavorable weather could have affected fungal establishment. For example, excess rainfall at the beginning of the study (March 2020 = 241 mm and April 2020 = 370 mm) could have washed out the applications, while a following long dry period between June to September in 2020 with less than 10 days of rain during the four months, without irrigation, may have restricted penetration of the drench to lower depths. Abiotic factors are well known to influence the colonization of biocontrol agents in soil and roots (Timper, 2011). In a similar study on mature coffee trees in Brazil, Zinger et al. (2021), evaluated the efficacy of T. harzianum against RKN. Unlike our finding, a significant reduction in M. incognita was observed in roots after just two months of the application, although no record of fungal colonization in roots was provided. On the other hand, Arita et al. (2020) reported a poor efficacy of P. lilacinum and T. harzianum to suppress M. paranaensis on Arabica coffee in Brazil. The differences in these findings might derive from the different application dosage, methods, RKN species and tested cultivars. Moreover, the use of organic amendments and thus the soil condition can also affect biocontrol colonization, which are often recommended to enhance biocontrol efficacy (Garbeva et al., 2004; Chilosi et al., 2020). In the current study, assessment of the various nematode based indices showed that the soils in which the trials were situated were all generally depleted and stressed, and consequently not ideal for biocontrol establishment. For example, low soil microbial diversity in coffee fields were associated negatively with soil functions and reduced coffee production (Zhao et al., 2018). Application of organic amendments may therefore help in enhancing the efficacy and speed of activity of the biocontrol agents.

The biocontrol agents, T. asperellum and P. lilacinum, are both common soil inhabitants and suppress RKN in a similar manner, which is mainly through the parasitism of nematode eggs and J2, although Trichoderma spp. also produces metabolites, which are nematicidal (Sharon et al., 2001). And yet, in our study, P. lilacinum reduced the RKN root population densities more rapidly than T. asperellum but then disappeared from soil and roots faster, once the application was stopped. The persistence of biocontrol agents in soil or roots mainly depends on how well they can multiply, where various external factors affect their ability to proliferate, such as edaphic factors, the microbial community structure, and the abundance of the host organisms (Abd-Elgawad and Askary, 2020). It is possible that P. lilacinum failed to adapt in the soil and root habitats and thus could not multiply efficiently in the fields of our study. It was not clear from our study as to which factors specifically prevented the longer-term establishment of P. lilacinum but for it to proliferate it would likely demand a higher frequency application in order to maintain a similar control efficacy of RKN that T. asperellum showed.

The nematode community assay in our study revealed different responses of the nematode fauna to the application of T. asperellum and P. lilacinum. As both products are fungal based, an increase of fungivores for both treatments was anticipated; however, an increase in fungivores was only obvious for P. lilacinum. Furthermore, of the two dominant fungivores, Aphelenchoides and Aphelenchus, only Aphelenchus was responsive to the P. lilacinum application, which may imply that P. lilacinum is a preferable food source to some genera than others. The increase in fungivores by the P. lilacinum application is not necessarily an ideal change to the soil health conditions, however, as the decomposition channel shifts from bacterial to fungal, which results in a slower decomposition process. Nonetheless, it would be important to understand the mechanism behind this phenomenon towards optimizing P. lilacinum application. An increased abundance of omnivores, dominated by Dorylaimid, and an improvement in maturity indices (i.e. maturity index, maturity index 2-5 and sigma maturity index), Structure index and Shannon index following T. asperellum application indicated less stressed, less disturbed and more stable environments with higher diversity (Ferris and Bongers, 2009). The enhanced diversity richness with T. asperellum applications also reflects previous studies (e.g. Alexander and Phin, 2014). Despite the more rapid suppression of RKN in roots by P. lilacinum, the nematode community assay revealed that T. asperellum had more positive impacts on soil health conditions, as demonstrated through the various indices.

The PCA was useful in highlighting the difference in nematode assemblages between the treatments as well as a temporal effect on the genera associated with the treatments. A few observations in the PCA supported the observation above, such as the shift of Aphelenchus spp. early in the study from a weak association across treatments to a strong association with the P. lilacinum treatment and the strong association of Meloidogyne in the control later in the study. On the other hand, the PCA revealed perspectives that were not clear from the generalized linear mixed models. For example, the genus Mesorhabditis, which is a c-p 1 group nematode, was initially associated with T. asperellum but later showed a strong association with P. lilacinum. Nematodes in c-p 1 group are indicative of food-enriched conditions (Bongers and Bongers, 1998), and that the application of P. lilacinum may enrich the environmental quality. Furthermore, Rotylenchulus showed a strong association with P. lilacinum later in the study, which was not confirmed using generalized linear mixed models. The analysis of the nematode community at the genus level, also highlighted the distinguished assemblage associated with the T. asperellum treatment, which showed less association with the most abundant nematode genus of this study. The evaluation of the impact of treatments at the genus, or even at species level, can be more insightful to better understand the soil ecosystem as opposed to analysis based on feeding type (Brussaard et al., 2004). The analysis of the nematode community data at multiple levels therefore facilitates an integrated understanding of the soil health conditions and should be considered in similar studies.

Since being proposed by Ferris (2010), the functional metabolic footprint is gaining increasing recognition for the description of food web form and function, and to evaluate the impact of land use (Krashevska et al., 2019), farming systems (Briar et al., 2011; Ilieva-Makulec et al., 2016) and management options (Gupta et al., 2019) on food web conditions. In our study, the analysis was not sensitive enough to capture any differences between treatments. All three treatments fell into the quadrant that indicates stressed conditions, while the temporal factor failed to influence the food web conditions over the two-year period. It is possible that this is due to the high abundance of fungivores, as well as c-p 2 nematodes across our coffee fields. For example, the two most abundant nematode families, Cephalobidae and Aphelenchoididae, are c-p 2 group bacterial scavengers and fungal feeders (Bongers and Bongers, 1998). These general opportunists can quickly adapt to and are associated with stressed conditions (Ferris et al., 2001). The high ratio of fungivores to bacterivores also indicated a fungal based slower decomposition channel rather than bacterial. This thereby reflects depleted, poor quality nutrient availability, resulting in a low Enrichment index value (Ferris et al., 2001). An increased abundance of c-p 3 and c-p 4 group nematodes was observed, however, after fungal colonization was recorded in roots for T. asperellum. This indicates that the shift in the food web condition following application of such biocontrol agents is gradual and requires longer-term monitoring to better appreciate the influence of biocontrol applications, especially when dealing with stressed, degraded soils.

In respect to plant-parasitic nematodes, the most commonly occurring and dominant genus was Meloidogyne, which was identified solely as M. hapla. No other Meloidogyne species was detected from any of the coffee fields in the study. Meloidogyne hapla has been detected in Kenya on other vegetable crops (Karuri et al., 2017; Chitambo et al., 2018) and has also been identified from neighboring coffee estates (Saikai, personal observation). Among the Meloidogyne species reported on coffee globally, M. exigua, M. incognita, and M. paranaensis are considered the most important (Villain et al., 2018); M. africana and M. decalineata were reported as important from a study in Tanzania (Bridge, 1984). All the fields in our current study were heavily infested with M. hapla with significant root damage observed, unlike the reports of Villain et al. (2013). It has been recorded from coffee elsewhere in the region, such as Democratic Republic of the Congo (Villain et al., 2018) and Tanzania (Bridge, 1984; dos Santos et al., 2019), with some occasional reports from higher altitudes in Latin America (Villain et al., 2013) but has generally been considered to be of limited economic importance on coffee (Lordello, 1982; Bridge, 1984). It is, however, reported to be of significant importance on coffee in Hawaii (Handoo et al., 2005). The current study indicates that M. hapla is likely of significant importance on coffee, at least in Kenya, although no studies have so far been conducted to determine this. During our study, RKN soil densities were not necessarily reflective of the actual disease pressure observed on the roots, with which there was a poor correlation between the RKN soil densities and RKN root densities. We often recovered low or even zero RKN densities from soil, even when high RKN densities were observed in roots, and vice versa. Consequently, this might be associated with weather conditions to some extent, but there is also potential for some improvements in sampling and nematode extraction techniques. This is particularly relevant as the modified Baermann plate method relies heavily on nematode mobility. Nonetheless, based on our study, assessment of the RKN infection in roots was preferable to soil, to better appreciate pest pressure.

Globally, RKN is considered as the most significant nematode threat to the coffee production (Villain et al., 2018) but there is relatively little knowledge on the distribution of species or damage levels in Africa (Coyne et al., 2018). Bridge (1984) estimated a ~20% loss in yield on coffee due to nematode pests, mainly RKN, in Tanzania, while a survey in Ethiopia (Mekete et al., 2008) found only a limited occurrence of RKN on Ethiopian coffee. Some concern regarding RKN damage to coffee in Kenya has been raised (Nzesya et al., 2014) but in general, there is little understanding of the issue across the coffee sector. In our study, the mean total abundance of herbivores, including the root hair feeders, was just 7%, whereas Nzesya et al. (2014) reported a 64% total abundance of herbivores in their samples. Despite some extensive root galling damage observed in our fields, recovered nematodes from samples were at times relatively low. In general, poorly managed coffee farms tend to be more affected by RKN, compared to well-managed commercial farms (Nzesya et al., 2014).

For practical usage of these biocontrol agents by farmers in the region, the financial and economic viability needs to be established. The frequency of application in our study was conducted to initially establish the fungal antagonists in the soil through a high frequency of application and then to maintain them. More in-depth assessment is required to determine effective and economical establishment rates, and for different soil types. We estimate an approximate product cost of $23.43 USD per tree over the two years of study, plus labor fee. At current 2022 rates, farmers receive approximately $5.00 USD for grade AA dried beans per kilogram. To break even therefore, a yield improvement of approximately 4.7 kg (28 kg fresh cherries) of the best quality dry beans per tree over two years is needed. However, the additional merits of using these biocontrol products would be useful to assess, such as an impact on other pests and diseases and long-term soil health improvement that would improve nutrient use efficiency, for example, due to improved root development and feeder root mass. During sampling of roots towards the latter part of the study, more prolific feeder root growth was visually observed in the biocontrol treatment plots than the control, indicating their influence on root recovery and development. This proved difficult to validate empirically but should be looked at in future studies. Application regimes could also be developed to optimize the application timing, such as targeted application to coincide with the period of root flush, to reduce the frequency. We were unable to measure yield in the current study, as the pruning program/cycle rendered yield data of little value. Instead, yield needs to be monitored over a longer time period, such as five years and a mean per year calculated, to negate the influence of the pruning program. Considering the observed damage caused by RKN alone, it was apparent that root efficiency was significantly compromised and not functioning effectively to access nutrients or water. Neither replacing existing coffee trees with new seedlings nor applying synthetic chemical pesticides are attractive strategies. However, our study has demonstrated that biocontrol agents provide an alternative option for managing nematode diseases on mature and established coffee farms.

Reversing or arresting even, long-term nematode damage inflicted on the root systems of perennial crops is not readily addressed, nor easily implemented. The decades-long manifestation of nematode pests, such as RKN on mature coffee plantations, has major implications to productivity and the economic viability of aging plantations. Replacing these plantations with new plants has huge financial implications, while the blanket treatment of the soil with synthetic pesticides has major environmental and cost considerations. This current study therefore assessed the potential of drenching biocontrol agents on to established, mature coffee fields in Kenya, all of which were affected by nematode infection and damage. Our study has shown that two locally available biocontrol agents, Trichoderma asperellum and Purpureocillium lilacinum, can effectively suppress RKN infection on mature coffee plants and improve soil health. These biocontrol agents offer a potential sustainable alternative to chemical inputs for coffee farmers in the region. During a two-year period, the drenching application of T. asperellum and P. lilacinum products showed a significant reduction in the population density of Meloidogyne hapla in roots of established Arabica coffee plantations. Additionally, some improvements were observed in the soil food web structure after the application of these biocontrol agents, even over this relatively short period. Our study showed that P. lilacinum was more effective than T. asperellum in reducing root population densities of M. hapla, although it disappeared from the soil and roots faster once the application was stopped. However, despite the more rapid suppression of M. hapla by P. lilacinum, the nematode community assay revealed that T. asperellum had a more positive impact on soil health conditions, as demonstrated through various nematode based indices. Contrary to previous reports in other regions, our current study suggests that M. hapla may be significant in coffee cultivation, particularly in Kenya, although no previous studies have investigated its importance in this context.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

KS designed the experiment as the project lead. KS implemented and managed the field trials and the sample processing in collaboration with CO and ES. The data collection, especially for the nematode community data, was conducted by KS. KS analyzed the data. The manuscript was drafted by KS. CO contributed to the experimental design and assisted the implementation and management of the field trial with KS. CO counted and identified the plant-parasitic nematodes. ES assisted the implementation and management of the field trials and the sample processing. SN supported the implementation and management of the field trials as well as the fungal isolation from soil and roots. RM provided the biocontrol products evaluated in the study. SH supervised the lead author at icipe and provided the research facilities. SH contributed to draft the article and approved the version submitted to the journal. DC supervised the lead author at IITA and provided the research facilities. DC contributed to design of the experiment, draft the article and approved the version submitted to the journal. All authors contributed to the article and approved the submitted version.