Three counties in the lower Ribeira River valley, all of them in the State of São Paulo were selected for this study: Eldorado, Sete Barras and Registro (Fig. 1). The climate in Sete Barras and Registro is rainy tropical (Af-type according to Köppen), with mean rainfall greater than 60 mm in the driest month. In Eldorado, climate is Köppen Am tropical, with rainfall less than 60 mm in the driest month. The average annual rainfall for all counties ranges from 1500 to 1600 mm (CEPAGRI 2018; CIIAGRO 2018), with the highest concentration of rains occurring from January to March. The mean annual temperature ranges from 23.9 to 24.3 °C, with the lowest temperature (13 °C) in July and highest (34.2 °C) in February. Soils in the valley originate from sedimentary, metabasic and amphibolic rocks (Oliveira et al. 2002), with high natural fertility (calcium, magnesium, potassium, and phosphorus content) and high organic matter levels, due to seasonal river floods that deposit alluvial material. Soil texture varies from loam to clay. The areas chosen in the three counties are characterized by smaller watersheds that flow into the Ribeira River with banana crops on the high ground level and Atlantic forest sites (control sites) in advanced stages of regeneration close to the Ribeira River. General characteristics of the areas are given in Table 1.

Figure 1. 

Location of the counties sampled in the Ribeira de Iguape River valley, São Paulo State, Brazil.

Earthworms were collected using an adaptation of the standard sampling method proposed by the Tropical Soil Biology and Fertility (TSBF) Programme (Anderson and Ingram 1993). In each area 10 samples (25 × 25 cm square to 20 cm depth) were taken, divided into 2 equally-numbered transects with samples every 20 m. Distance between transects was ca 10 m. Earthworms were hand-sorted from the soil in the field and fixed in 80% alcohol. In the laboratory, earthworms were identified to species or family level (juveniles) using taxonomic keys (Michaelsen 1900; Righi 1990; Blakemore 2002). The material was deposited in the Fritz Müller Oligochaete collection (COFM) at Embrapa Forestry in Colombo, Brazil. The earthworm data obtained were used to determine the total species abundances (no. individuals and fresh mass m^-2) and richness, per site and land use (banana, forest).

Both the common and scientific names of banana were used for a bibliographic search online using the keywords for bananas and plantains in English, Portuguese, French and Spanish: Musa (genus), Musa acuminata, Musa balbisiana, banana, banane, banano, plátano and plantain. These were then crossed with the common names of earthworms in these languages: earthworms, minhoca, oligochaeta, oligoqueta, vers de terre and lombriz de tierra. Online scientific databases Web of Science, Science Direct, Scielo, google academic and the Base de Dados de Teses e Dissertações (BDTD – Thesis and Dissertation Database) of Brazil were consulted. All the resulting publications were consulted and those containing data on earthworm abundance (density and/or biomass) or species identification were selected and these data extracted, as well as information on sampling sites (counties, countries, management practices of the plantations). Earthworm species were separated into different families and into native or exotic to the region of occurrence, and species richness per site and for each group (native, exotic), when available. Although we treated bananas and plantains separately when possible, for most of the analysis we considered them together, since not all publications provided details regarding the types of bananas cultivated, and even plantains are often called ‘bananas.’ Details on the species and management data obtained and presented in this paper are available for download online from the open access repository Mendeley Data at http://dx.doi.org/10.17632/p8ywsnj8c5.1 (Cremonesi et al. 2020).

Only two earthworm species belonging to two families (Rhinodrilidae, Megascolecidae) were found at the six sampling sites in the three counties (Table 2): Pontoscolex (Pontoscolex) corethrurus and Amynthas gracilis, both considered peregrine/exotic in southern Brazil (Brown et al. 2006). Pontoscolex corethrurus may have originated in the Guyana shield area (Righi 1984), and A. gracilis may be native to China (Blakemore 2002). The former species was found living in all sites, while the latter was found in both banana plantations and native forest in Sete Barras and in banana plantations in Registro. At the other sites, only juveniles of the Megascolecidae family were found. These were most likely A. gracilis as well, but could not be identified to species level. Maximum richness found per site was similar in banana crops and Atlantic forest fragments (two spp. in each land use), but with some variation between sites (Table 2).

Most of the individuals collected (76% of the total) were of P. corethrurus, representing 29% of the total abundance in banana crop sites and 46% in Atlantic forest fragments (Fig. 2). Amynthas gracilis, although not occurring in all areas, accounted for 12% of all individuals sampled, of which 10% were found in banana crops but only 2% in Atlantic forests. Rhinodrilidae juveniles represented only 2% of the earthworms found, and occurred only in the Atlantic forest, while Megascolecidae juveniles represented 10% of all earthworms, and were often found in Atlantic forest fragments. Both species are widespread in Brazil (Brown et al. 2006), especially in agricultural and disturbed ecosystems, and display relatively high tolerance to a range of abiotic/biotic conditions, which have allowed these species to spread throughout most of the tropics and subtropics worldwide (Brown et al. 2006; González et al. 2006; Taheri et al. 2018). They have also been recommended as indicators of soil quality in agroecosystems and of disturbance in natural landscapes (Nunes et al. 2007; Fernandes et al. 2010).

Figure 2. 

Frequency of earthworm species (% of total individuals collected) in each ecosystem sampled (A) in Atlantic Forest (AF) and banana plantations (BP) and by counties (B).

The predominance of P. corethrurus in both native forest and banana plantations of the Ribeira River valley indicate that non-native species have extensively colonized disturbed soils of this region. Nonetheless, this potentially widespread occurrence of exotics should be further evaluated both regionally and nationally, in order to better determine the extent of this phenomenon as well as its possible causes.

Mean overall abundance and biomass of earthworms found in the three sites studied here (6 to 13 indiv. m^-2 and 2.5 to 9 g m^-2) tended to be quite low compared with others observed overall in Brazil (21 to 459 indiv. m^-2 and 3.1 to 177.4 g m^-2; see Table 3). At sites near the Ribeira River valley in the neighboring state of Paraná (Römbke et al. 2009; Maschio et al. 2010), and within the Ribeira River watershed in the nearby Turvo River valley (a tributary of the Ribeira River; Brown et al. 2009), both abundance and biomass were generally much higher (Table 3), even though the predominant earthworm species was the same (P. corethrurus). This is probably due to the less intensive and more traditional agroforestry management practices used in these sites, including slashing and mulching, as well as the presence of other trees, particularly atmospheric N₂-fixing leguminous trees, and the absence of or lower pesticide use (Brown et al. 2009; Römbke et al. 2009). These practices may benefit earthworm populations, particularly P. corethrurus, as observed comparing a mulched and non-mulched plantation in Antonina, where earthworm abundance was ~13 times higher with mulching (Maschio et al. 2010). Reasons for the lower values found in the Ribeira River valley sites may be due to the more intensive management practices typical of commercial banana plantations in the region, including insecticide and nematicide applications, which may reduce earthworm populations (Clermont-Dauphin et al. 2004).

More than 70 studies were found from 28 countries with data on earthworms in banana and plantain fields (Tables 3, 4, 5). Of these studies, 49 had species data (Table 4; see also full dataset in Cremonesi et al. 2020), coming from ≥ 210 sites (Table 5), of which most were in the Spanish Canary Islands (N = 77), mainly due to the intensive sampling efforts of Talavera in Tenerife (Talavera 1992a). Interestingly, two of the major banana-producing countries in terms of area were not represented (Tanzania, Rwanda), and in China (another important producer), only one study reported earthworms from a single site (Sun et al. 2012). Plantain banana fields were sampled in only 22 locations (10% of total) in four countries (Colombia, Cameroon, Ivory Coast and Ecuador; Tondoh 2007; Norgrove et al. 2011; Avilés 2017; Feijoo et al. 2018), and involved traditional management practices, rather than conventional cultivation. Most of the fields evaluated were banana plantations, and only in Ecuador were mixed banana/plantain fields evaluated (Avilés 2017).

Overall, ≥104 earthworm species from 10 earthworm families were recorded from banana/plantain fields worldwide, of which around 61 (59%) were native and 43 exotic to the sampling sites (Table 5). Estimating these numbers is difficult due to insufficient taxonomic resolution in some samples, as well as the uncertain origin of some widespread anthropochores (peregrines transported by humans), particularly in the Megascolecidae family (Blakemore 2002). Highest species richness (27) was observed overall in India, where most of the species found were native (74%). High proportions of native species were also observed in Ivory Coast, Madagascar, and Uganda (possibly 100%) as well as Cameroon (75%), but were lower in Mexico (58%) and Colombia (53%). In these countries, many of the plantations were managed more traditionally, or using agroforestry, although the low number of sampling sites may also be responsible for these high values, particularly in the former countries. In fact, agroforestry systems had a total of 22 species from nine sites, while conventional production systems had only nine species from 13 sites. Nonetheless, because not enough information was provided in the publications on management practices (not reported in ≥150 sites; Table 4), the role of less intensive banana production systems in maintaining native earthworm populations must still be further evaluated.

High species richness was also detected overall in Spain (25), mainly due to the higher sampling effort involving a large number of sites in the Canary Islands. However, all of the species encountered on the islands offshore of Africa were exotic, their introduction having been stimulated over centuries of human colonization bringing in exotic soils and crops (Talavera 2007, 2011). The Caribbean islands had few species (5), despite a large sampling effort, and many sites were dominated by P. corethrurus (Burac et al. 2018). In Brazil, Costa Rica, Martinique, Jamaica, Bermuda, the Seychelles, Taiwan, Malaysia, and China, all the earthworm species encountered were exotic (Table 5). The continent with the highest number of species recorded was Africa (50), of which 40% were native. In Asia, 35 species were recorded, with a higher proportion of natives (66%). In North and South America, around 50% of the species found were native, but these were mainly due to the higher number of natives observed in Colombian (Feijoo et al. 2018) and Mexican (Geissen et al. 2009; Huerta et al. 2013) plantations.

Species richness in individual banana/plantain fields was measured in 166 of the 210 sites, and was generally very low, with an overall mean of 2.7 species per site worldwide, of which less than one (0.5) was native and 2.1 were exotic (full dataset in Cremonesi et al. 2020). Absolute richness in an individual plantation was highest in the banana plantations in Cameroon (Norgrove et al. 2011), where 8 and 10 species were found (Table 4), most of them native. The only other place with such high richness was a plantation in Gomera Island (Laguna de Santiago), where 9 species were found (Talavera 2007), although all of them were exotic. In West Tripura, up to 7 species were found in a banana plantation (Dhar and Chaudhuri 2018), but most plantations in the world had less than 3 species (~70% of sites), and the highest proportion was of sites with only 1 species (~30% of sites).

There was a clear positive relationship between the number of sites sampled in each country and the total number of species encountered (r = 0.7, p< 0.01), particularly for exotic (r = 0.78, p < 0.01) species (Fig. 3A). Although also positive, this relationship was not significant for native species. Nonetheless, the species accumulation curve for native species for all sampling sites in the world revealed a steep slope, that contrasts with the flattened-out accumulation curves for total and exotic species (Fig. 3B). This indicates that greater sampling efforts, particularly in more low-input production systems, especially in tropical countries with high earthworm biodiversity such as Ecuador (no studies with earthworms identified yet), Brazil and Colombia (Brown and James 2007; Feijoo 2007; Zicsi 2007) will certainly increase the number of species known from banana/plantain fields. Greater sampling efforts are also needed in other tropical countries with important plantain/banana production (FAO 2018), particularly when intercropped or in agroforestry systems (Norgrove et al. 2011; Norgrove and Hauser 2014), and where mostly native earthworm species may inhabit these fields, such as seen for Cameroon, Uganda and Ivory Coast. This phenomenon may likely also be applicable to other Western, Central and Eastern African countries, as well as many other Asian and Pacific countries, but the paucity of available data impedes further speculation.

Figure 3. 

A Relationship between species richness (total, native, and exotic species) and the number of sampling sites in each world country (data from Table 5) and B Species accumulation curves for total, native and exotic species, depending on the number of sampling sites across the world. Linear regression equations and the value and significance (p value, with ** indicating p< 0.01) of the Pearson correlation coefficient (r) are provided in (A).

Of the over 100 species found in banana and plantain fields worldwide, most belonged to the Megascolecidae (22%), Lumbricidae (17%) and Acanthodrilidae (16%) families (Cremonesi et al. 2020). These widespread exotic and often invasive species are found throughout the tropics and subtropics, and include several Amynthas and Metaphire spp. (Blakemore 2002). The most consistently recorded megascolecids were A. gracilis (6% of all records), Amynthas morrisi (Beddard, 1892) (5%), Pithemera bicincta (Perrier, 1875) (4%) and Metaphire californica (Kingerg, 1867), Perionyx excavatus Perrier, 1872 and Polypheretima elongata (Perrier, 1872) (all with 2% each) (Cremonesi et al. 2020). These megascolecids were found in over 15 countries, and were especially frequent in the Canary Islands. All of the lumbricids reported were exotic, and mainly found in the Canary and Madeira Islands (Spain, Portugal), with Aporrectodea rosea (Savigny, 1826) and Eisenia andrei Bouché, 1972 (both with ~4%) and Bimastos rubidus (Savigny, 1826) (3%) being the most frequently reported. Various octochaetid Dichogaster spp. of the Benhamiinae subfamily, i.e., Dichogaster (Diplothecodrilus) affinis (Michaelsen, 1890), D. (D.) bolaui (Michaelsen, 1891) and D. (D.) saliens (Beddard, 1893) (all with around 2% each) and the acanthodrilinae Microscolex spp., i.e., Microscolex phosphoreus (Dugés, 1837) with 3% and M. dubius (Fletcher, 1887) with 2% of records, were the most reported acanthodrilids. The Dichogaster spp. were found in 11 countries, mainly in Latin America and the Canary Islands, while the Microscolex spp. were found only in the Canary Island banana plantations. Similarly, the ocnerodrilid Ocnerodrilus occidentalis Eisen, 1878 with 6% of all records, was found in three countries (Brazil, Portugal, Spain), but most frequently in the Canary Islands.

The most commonly encountered earthworm species in banana plantations was P. corethrurus (11%), found in 15 countries, mainly in Latin America, but also in places as far away as South Africa, India, Bangladesh, Malaysia, Philippines and Taiwan. Interestingly, 37 out of 54 sites (69%) that identified earthworm species reported P. corethrurus as dominant in the banana plantations (Table 3).

Although P. corethrurus may affect soil physical properties negatively by increasing soil compaction under some conditions, it can also positively affect biogeochemical processes, microbial activity, plant production, and soil recovery (see review in Taheri et al. 2018). Furthermore, this species is known to reduce plant-parasitic nematode incidence in banana plants (Loranger-Merciris et al. 2012), and has also been known to promote beneficial plant growth-promoting bacteria in the rhizosphere (Braga et al. 2015). Hence, further work is warranted on the potential beneficial impacts of the presence and populations of P. corethrurus on banana plants, particularly considering its widespread distribution and high abundance in some locations (e.g., Costa Rica, Brazil, Guadeloupe, Martinique, Mexico). Several megascolecids such as A. gracilis are also known to affect soil physical and chemical properties in annual cropping systems (e.g., Peixoto and Marochi 1996; Bartz et al. 2010) as well as crop production (Brown et al. 1999), but little is known of their effects on banana plants. The latter statement is also valid for all of the other species most commonly found in banana plantations.

A total of 31 studies performed in 153 sites and 15 countries (Table 5) had quantitative earthworm data (on abundance and/or biomass) taken mainly by hand sorting soil monoliths of variable size (mostly 25 x 25 cm but sometimes larger, e.g., 50 × 50 cm) and occasionally using liquid extraction (e.g., formalin expulsion). Most of the study sites were in Guadeloupe (N = 40, of which 34 were by Clermont-Dauphin et al. (2004) and Colombia (N = 32; Molina and Feijoo 2017).

Overall earthworm abundance ranged from a minimum of 0 (Figueroa 2019) in an Ecuadorian plantation, to a maximum mean of over 1500 indiv. m^-2 in banana plantations in Kwazulu-Natal, South Africa (Dlamini and Haynes 2004). Maximum biomass attained was 453.6 g m^-2 for a site in West Tripura, India (Dhar and Chaudhuri 2018). Interestingly, a large number of sites (>50) had abundance values over 100 indiv. m^-2, which could be considered quite high for earthworm density in annual agricultural crops (Bartz et al. 2013). Nonetheless, bananas are perennials often cultivated over several cropping cycles, allowing for reduced negative effects of soil preparation, and the soils are also often limed to correct pH and fertilized with inorganic fertilizers (mainly N, P and K) to promote soil fertility and banana production. In these conditions, earthworms present find a soil protected from rainfall impact, as well as frequent organic matter additions through the management of the banana trees, particularly where the residues are left on the soil surface. Consequently, their populations can increase rather rapidly over time, as observed by Okwakol (1994) in Uganda (Table 3).

These high earthworm abundances and biomasses may be contributing significantly to soil processes (bioturbation, nutrient cycling) in banana/plantain fields, as biomasses over 17 g m^-2 and above 32 g m^-2 are known to lead to moderate (20–40%) and important (>40%) grain production increases, respectively (Brown et al. 1999). Earthworm-induced improvement of plant health and production includes, e.g., plant-parasitic nematode population control (Lafont et al. 2007; Loranger-Merciris et al. 2012), high stable bioaggregate formation, creation of many galleries in the soil and enhanced nutrient mineralization (Lavelle 1997), all factors that deserve future attention. On the other hand, low earthworm abundance may be an indicator of soil degradation, or the use of inappropriate management practices, such as soil inversion or toxic pesticide use (Demetrio et al. 2019). This type of information could be used to help farmers with their management decisions, such as reduction in nematicide applications that reduce earthworm populations (Clermont-Dauphin et al. 2004).

Finally, 18 of the major banana-producing countries in the world (34 countries with >30,000 ha in production, or >1 Million T bananas produced yr^-1; FAO 2018) were not examined in the present review due to lack of data. Hence, further sampling efforts are needed in order to provide adequate information on earthworm abundance and biodiversity in banana plantations in these countries, and to complement those reported here but with low sample intensity, particularly focusing on the presence of native species and/or large earthworm abundances, and to identify the reasons for these phenomena and their consequences for banana production and biodiversity conservation.