Weed diversity in tomato crops in the mountainous region of South Kivu, DR Congo

Description of the subject. Despite the importance of tomato in Central Africa, little is known about the weed flora occurring in tomato crops in this part of the world. Objectives. We aimed to provide a comprehensive description of the weed flora found in tomato crops in the mountainous region of South Kivu, DR Congo. Method. Floristic inventories were made in four cultivation zones of the Kabare territory, South Kivu. Results. In total, 79 species (belonging to 64 genera and 31 families) were recorded. The most species-rich families were Asteraceae and Poaceae. Therophytes and geophytes were the dominant life forms. Weed diversity was not significantly different between sites but species composition varied considerably, likely in relation to soil type. Based on the frequency and the abundance, the most problematic weeds were Commelina benghalensis, Galinsoga quadriradiata, Commelina diffusa, Galinsoga parviflora, Bidens pilosa, Oxalis debilis and Digitaria velutina. Conclusions. The weed flora in tomato fields of the mountainous regions of DR Congo is diversified, but only a few cosmopolitan species are problematic.


INTRODUCTION
Tomato (Solanum lycopersicum L.) is a widely grown vegetable in Central Africa, where it provides important income to smallholder farmers (Chausse et al., 2012).
Tomato yields in DR Congo are low (ca 7.5 ton . ha -1 ) as compared to world leading countries such as China or India (FAOSTAT, 2019). This is attributable to several factors, notably weed competition and pests damages. Through competition for water, light, and nutrients, weeds are a major yield-limiting constraint (Nichols et al., 2015). In contrast to other pests, they are ubiquitous in agricultural systems. Adigun (2005) attributed a 40 to 60% decrease in tomato fruit yield to weed pressure in Nigeria. Likewise, Amare et al. (2015) recorded ca 90% yield loss caused by early weed competition in Ethiopia. In Central Africa, where hand-weeding is the dominant weed management method, smallholder farmers spend up to 50% of agricultural activities to control weeds (Ellis-Jones et al., 1993). Several studies (e.g. Le Bourgeois, 1993;Noba et al., 2004) emphasized the need for a better knowledge of weed infestation as a key step for setting up an efficient management strategy. However, no study to date has been carried out on weed pressure in vegetable crops, especially tomato, in the mountainous regions of Democratic Republic of the Congo. The main goal of this research was to describe the weed community occurring in tomato crop and to identify the most problematic, i.e. the most frequent and abundant ones. Our results can be used to provide guidance for weed management strategy in this part of the world.

Study sites
This study was carried from March to June 2019 in four cultivation zones (Figure 1) that are representative of the soil variation in South Kivu. According to Wils et al. (1986), the mountainous region of Kivu can be divided into two soil types: volcanic soils, usually highly productive; and ferralitic sedimentary soils, with low to medium fertility. The cultivation zones of Luhihi (altitude ca 1,695 m) and Bugorhe (altitude ca 1,712 m) are implemented on volcanic soils, whereas Mudaka (altitude ca 1,588 m) and Bushwira (altitude ca 1,646 m) are implemented on ferralitic soils. The distances between cultivation zones varied between 3.9 km (Bushwira-Mudaka) and 18.7 km (Bushwira-Luhihi). The study area is characterized by a tropical mountain climate with a long rainy season (from September to May) and a short dry season (June to August). Mean daily temperature ranges from 17 to 19.5 °C. Annual rainfall is ca 1,400 mm.

Field records
A preliminary study was achieved to determine the minimum quadrat size according to Gillet (2000), and 4 m 2 was the minimum area retained. On-field floristic inventories were then conducted in 10 tomato fields (all planted in March 2019) in each four cultivation zones, one month after tomato plantation/sowing and before the fields were weeded. In each field, three 2 m x 2 m quadrats were placed randomly and all plant species were recorded. The three quadrats per field were pooled prior to analyses. Cover was assessed using Braun-Blanquet (1952) scale, as a proxy for abundance. Complementary samplings were made 2 and 3 months after tomato planting, in the same quadrats, to take into account new species. Tomato was implemented after another culture in 30 out of the 40 fields. The most common previous crops were beans, maize and onions. Continuous tomato cultivation was more common in Mudaka (5 fields) compared to Bushwira (1 field), Luhihi (2 fields) and Bugorhe (2 fields).

Data analysis
The "infestation diagram" approach (Le Bourgeois, 1993;Lebreton & Le Bourgeois, 2005) was used to identify the most problematic weed species, i.e. the most frequent and abundant ones: general majors species are very ubiquitous (RF > 50%) and abundant (cover > 14%); general species are also frequent (RF > 50%) but less abundant (cover < 14%). Other species are considered less problematic. Before calculation, Braun-Blanquet values were transformed in % using the abundance (AD) index table of Gillet (2000): 0.03% (r); 0.03% (+); 3% (1); 14% (2); 32% (3); 57% (4) and 90% (5). Then, the abundance average (AD av) was calculated using the following formula: AD av = ∑AD/AF with AF = absolute frequency; ∑AD = sum of abundance index of a species in all quadrats. We estimated α-diversity through species richness and performed a one-way ANOVA to evaluate the effect of cultivation zones. A two-dimensional non-metric multi-dimensional scaling (NMDS) ordination was run on the Bray-Curtis similarity matrix (based on species abundance) with all plots, using the vegan package (Oksanen et al., 2013). Analyses were performed using R statistical software (R Development Core Team 2010).
The NMDS plot (Figure 3) showed an overall separation between the weed flora of the cultivation zones on ferralitic sedimentary soils (Mudaka and Bushwira) vs volcanic soils (Luhihi and Bugorhe).

DISCUSSION
The number of species identified during this study (79 species) may seem low as compared to the flora of DR Congo, but it is much higher than those reported by Mushagalusa et al. (2014) in corn crops (46 species) and Miderho et al. (2018) in coffee trees (18 species), in mountainous regions of DR Congo. The most species-rich families recorded (Asteraceae and Poaceae) belong to the most ubiquitous weed families in agrosystems (Zimdahl, 2007;Jauzein, 2011). In regularly disturbed ecosystems such as cultivated fields, therophytes usually dominate (Noba et al., 2004;Reboud et al., 2009), which was clearly the case in our study. However, a significant proportion of species were geophytes, essentially due to the dominance of cultivated plants (e.g. Allium cepa L., Musa spp.,    The analysis is based on 40 vegetation surveysl'analyse est basée sur 40 relevés de végétation; distance : Bray-Curtis; K = 2; stress = 0.26 -0,26; refer to table 1 for the species codesles codes sont expliqués dans le tableau 1.

Figure 2. Diagram of weed infestation in tomato crops -Diagramme d'infestation des espèces adventices dans les cultures de tomates.
Built using the AD av (Abundance average) and the relative frequency (RF). Refer to table 1 for interpreting the EPPO codes of weed speciesles codes EPPO sont expliqués dans le tableau 1. Digitaria abyssinica, C. dactylon and C. sumatrensis are considered as indicators of degraded soils characterized by nitrogen and phosphorus deficiency (Ngongo & Lunze, 2000). In our case, these species were mainly found on ferralitic soils. On the other hand, G. parviflora, B. pilosa, G. quadriradiata, C. benghalensis and O. debilis dominate deep and fertile soils (Ngongo & Lunze, 2000). They were mainly found in volcanic soils. However, other factors like the altitude or the parcel cultivation history may influence weed composition (Travlos et al., 2018). The data provided in this research was collected during one cultivation season. According to Fried et al. (2008), the weed flora can vary according to inter-annual climatic variation. Despite typically low inter-annual climatic variation in the study area, a longer-term study might have yielded a better picture of the weed diversity in tomato crops in the mountainous region of South Kivu.

CONCLUSIONS
The study showed that the weed flora of tomato fields is diversified in the mountainous regions of Central Africa. Therophytes and geophytes are the dominant life forms. Cultivation zone show variable weed composition, notably in problematic weeds. Soil characteristics seem to be the main factor responsible for that pattern. Galinsoga quadriradiata, C. benghalensis, C. diffusa, G. parviflora, B. pilosa, O. debilis, A. hybridus and D. velutina are the most problematic species.