Introduction
Biological control using wasps and parasitoids is an important method to reduce economic losses caused by L. coffeella (David-Rueda et al., Reference David-Rueda, Constantino, Cecilia Montoya, Ortega, Nancy Gil and Benavides-Machado2016; Tomazella et al., Reference Tomazella, Jacques, Lira and Silveira2018; Medeiros et al., Reference Medeiros, Grandinete, Manning, Harper, Cutler, Tyedmers, Righi and Ribeiro2019a, Reference Medeiros, Martello, Almeida, Mengual, Harper, Grandinete, Metzger, Righi and Ribeiro2019b; Rezende et al., Reference Rezende, Venzon, Santos, Cardoso and Janssen2021; Rosado et al., Reference Rosado, Araújo, Pallini and Venzon2021; Venzon, Reference Venzon2021). It has been increasingly used in recent years with insecticide application, causing problems such as the selection of resistant populations and biological imbalances (Fragoso et al., Reference Fragoso, Guedes, Picanço and Zambolim2002; Vega et al., Reference Vega, Posada, Infante and Pimentel2007; Costa et al., Reference Costa, Fernandes, Alves, da Silva, Visôtto and Trdan2016; Leite et al., Reference Leite, Dos Santos, Resende-Silva, da Costa, Moreira, Lemos, Guedes and Castellani2020a, Reference Leite, dos Santos, da Costa, Moreira, Guedes and Castellani2021, Reference Leite, Guedes, da Costa, Colmenarez, Matsumoto, Dos Santos, Coelho, Moreira and Castellani2022; Rocha et al., Reference Rocha, Silva, da Silva, Cruz and Fernandes2022). At least 28 species of parasitoids from the families Eulophidae and Braconidae (Hymenoptera) (Parra and Reis, Reference Parra and Reis2013; David-Rueda et al., Reference David-Rueda, Constantino, Cecilia Montoya, Ortega, Nancy Gil and Benavides-Machado2016) have been associated with the coffee leaf miner. The most important species include Closterocerus coffeellae Ihering, Reference Ihering1914; Proacrias coffeae Ihering, Reference Ihering1914; Cirrospilus neotropicus Diez & Fidalgo, 2003; and Horismenus aeneicolis Ashmead, 1904 (Eulophidae); Stiropius reticulatus Penteado-Dias, Reference Penteado-Dias1999, and Orgilus niger Penteado-Dias, Reference Penteado-Dias1999 (Braconidae) (Parra et al., Reference Parra, Gonçalves, Gravena and Marconato1977; Penteado-Dias, Reference Penteado-Dias1999; Melo et al., Reference Melo, Castellani, Nascimento, Menezes Junior, Ferreira and Lemos2007; Lomelí-Flores et al., Reference Lomelí-Flores, Barrera and Bernal2009, Tango et al., Reference Tango, Fernandes, Paz, Lara and Perioto2014; Marques et al., Reference Marques, Fernandes, Morais, Haddi and Silveira2022; Calderón-Arroyo et al., Reference Calderón-Arroyo, Togni, Pantoja, Saenz and Venzon2023).
Parasitoids are responsible for about 15 and 30% of the coffee leaf miner biological control (Parra and Reis, Reference Parra and Reis2013) and up to over 50% in organic crops (Ecole et al., Reference Ecole, Moraes and Vilela2010). The coffee leaf miner control potential by parasitoids is underestimated due to the lack of knowledge on the interactions between natural enemies and the coffee leaf miner, in addition to the aspects that may affect trophic relationships (Reis et al., Reference Reis, Souza and Vilela2000). Among these aspects, climate factors directly affect the populations of L. coffeella, which are favoured in hot and dry seasons at low altitudes (Tuelher et al., Reference Tuelher, de Oliveira, Guedes and Magalhães2003; Pereira et al., Reference Pereira, Picanço, Bacci, Della Lucia, Silva and Fernandes2007a, Reference Pereira, Picanço, Bacci, Crespo and Guedes2007b; Lomelí-Flores et al., Reference Lomelí-Flores, Barrera and Bernal2010; Dantas et al., Reference Dantas, Motta, Vidal, Nascimento, Bilio, Pupe, Veiga, Carvalho, Lopes, Rocha, Silva, Pujol-Luz and Albuquerque2021). Intensive use of non-selective insecticides increases selection pressure on pest populations that rapidly recolonise the habitat. Meanwhile, their natural enemies take longer to reestablish their populations since a large proportion of individuals is usually eliminated (Fragoso et al., Reference Fragoso, Jusselino-Filho, Guedes and Proque2001; Perfecto et al., Reference Perfecto, Vandermeer and Philpott2010; Fernandes et al., Reference Fernandes, Silva, Gorri, Pucci and Silva2014; Harelimana et al., Reference Harelimana, Rukazambuga and Hance2022).
Furthermore, the characteristics of the agroecosystem landscape directly affect natural enemies (Iverson et al., Reference Iverson, Gonthier, Pak, Ennis, Burnham, Perfecto, Rodriguez and Vandermeer2019; Stüber et al., Reference Stüber, Tack, Zewdie, Mendesil, Shimales, Ayalew, Nemomissa, Sjögren, Vesterinen, Wezel and Hylander2021). They include biodiversity reservoirs of beneficial species, which disperse to coffee plantations exploiting resources in time and space, regulating the growth of pest populations (Vandermeer et al., Reference Vandermeer, Perfecto and Schellhorn2010; Aristizábal and Metzger, Reference Aristizábal and Metzger2019; Medeiros et al., Reference Medeiros, Grandinete, Manning, Harper, Cutler, Tyedmers, Righi and Ribeiro2019a, Reference Medeiros, Martello, Almeida, Mengual, Harper, Grandinete, Metzger, Righi and Ribeiro2019b; Hohlenwerger et al., Reference Hohlenwerger, Tambosi and Metzger2022). However, this information is restricted to the role of predatory wasps and does not measure the advantages of the coffee leaf miner parasitoids.
Thus, current hypotheses cover that coffee leaf miner infestations, natural parasitism, and parasitoid diversity increase with forest cover and the presence of natural landscape habitats that provided resources and space for natural parasitism to occur in coffee plantations. In addition, both coffee leaf miner and parasitoid populations would be influenced by climatic conditions. This study aimed to investigate the effects of climate factors, intensity of insecticide use, and multiscale landscape on natural parasitism, parasitoid diversity, and infestation of L. coffeella on coffee plantations in the Planalto region, Bahia, Brazil.
Materials and methods
Study area
This study was conducted at commercial coffee plantations in the Planalto region. The experimental period ranged from December 2020 to November 2021.
Six coffee plantations were selected for sampling, two belonging to the municipality of Barra do Choça (BCH1 and BCH2) (Y: 14°55′11.54″S; X: 40°35′53.81″W and Y: 14°50′26.59″S; X: 40°31′17.5″W), three located in Mucugê (MUC1 – Y: 13°5′48.68″S; X: 41°26′58.39″W, MUC2 – Y: 13°7′30.04″S; X: 41°29′34.43″W, and MUC3 – Y: 13°6′52.59″S; X: 41°27′32.66″W) in the Chapada Diamantina region, and one crop in Vitória da Conquista (VDC1) (Y: 15°0′2.28″S; X: 40°45′55.5″W) (fig. 1).
Sampling sites in coffee plantations in the Planalto region, Bahia, Brazil. 1 = VDC1 (Vitória da Conquista); 2 = BCH1 (Barra do Choça); 3 = BCH2 (Barra do Choça); 4 = MUC1 (Mucugê); 5 = MUC2 (Mucugê); 6 = MUC3 (Mucugê). Four repetitions (sampling units) comprising 30 leaf collection points were established at each site. The buffers represent the estimated local landscape within a 500, 1000, 1500, 2000, and 3000 m radius of each coffee plantation. Coordinate System: SIRGAS 2000.
The coffee plantations in Barra do Choça and Vitória da Conquista were 34.5 km from each other, which were, in turn, 222 and 226 km from the coffee plantations in Mucugê.
The coffee plantations located in the municipalities of Barra do Choça and Vitória da Conquista are inserted in a transition area between the Atlantic Forest and Caatinga biomes (CONAB, 2022). Their vegetation belong to the Semideciduous Seasonal Forest type, popularly known as ‘Mata de cipó’ (SEI, 2015). According to Köppen's classification, the climate is type Cwb, also classified as sub-humid to dry by Thornthwaite (SEI, 1998). The two coffee plantations in Barra do Choça and the one in Vitória da Conquista were planted with the Catuaí 144 cultivar, which is more than 30 years old.
The Caatinga biome predominates in the Chapada Diamantina, where the municipality of Mucugê is located, and the vegetation typology is shrubby Caatinga (SEI, 2015). The climate of Mucugê is classified as Am by Köppen and semi-arid by Thornthwaite (SEI, 2015). The coffee plantations were planted with the Topázio and Catuaí 144 cultivars around 18 years ago. The large territorial extent of the Planalto coffee region reflects different contexts of the coffee tree and coffee leaf miner management, climate, and landscape structure, especially insecticide use to manage the pest (Leite et al., Reference Leite, Guedes, Santos, Costa, Moreira, Matsumoto, Lemos and Castellani2020b).
The coffee plantations in Barra do Choça and Vitória da Conquista are rainfed, while those in Mucugê are fully irrigated through a centre pivot (MUC1 and MUC3) and drip irrigation (MUC2).
The history of insecticide use is different according to each coffee farm (table S1 see Supplementary Material), for example, VDC1 (Vitória da Conquista) has not used insecticides to control the coffee leaf miner for about 15 years.
Natural parasitism assessment and parasitoid species identification
We installed four rectangular sampling units of 2.7 hectares each (fig. S1 of the supplementary material) in each coffee plantation selected, each comprising 30 sampling points. The collection points were equally spaced at 30 m, each corresponding to a set of the four closest coffee plants within a 5-m radius.
Two leaves per plagiotropic branch located in the upper third of the coffee tree were randomly collected monthly from each sampling point, numbering six coffee leaves per point, 240 leaves per sampling unit, and 960 leaves per coffee plantation. The two leaves were collected at the third or fourth plagiotropic branch from the apical end of the branch (Melo et al., Reference Melo, Castellani, Nascimento, Menezes Junior, Ferreira and Lemos2007; Rosado et al., Reference Rosado, Araújo, Pallini and Venzon2021).
We opened the mines and recorded the presence of live caterpillars of the coffee leaf miner and pupae, larvae, and exuvia of the parasitoids and parasitised coffee leaf miner caterpillars.
All parasitised caterpillars, parasitoid pupae and larvae were placed in 2 ml microcentrifuge tubes (Eppendorf®). These tubes were stored in a controlled environment (25 ± 3°C T; 65 ± 5% RH; and 12-h photophase) until the emergence of the adults. Then, the adults were fixed in 95°GL alcohol for preservation and forwarded for identification.
We have identified parasitoids of the families Braconidae and Eulophidae at the generic and specific levels (Ihering, Reference Ihering1914; Schauff et al., Reference Schauff, LaSalle, Coote, Gibson, Huber and Woolley1997; Wharton et al., Reference Wharton, Marsh and Sharkey1997; Penteado-Dias, Reference Penteado-Dias1999). Part of the material was deposited in the Invertebrate Collection of the National Institute of Amazonian Research, Manaus, Amazonas, Brazil (INPA) (Y: 3°5′28.25″W; X: 59°59′21.08″S).
Parasitism percentages were determined by the equation: TPN (%) = [(No. of emerged parasitoids/Total number of parasitoids collected) × 100]. Infestations were identified by the equation: ML (%) = [(No. of intact mined leaves/total number of leaves with mines) × 100] (Melo et al., Reference Melo, Raetano, Nery, Cardoso, Moreira, Leite, Jesus, Silva and Castellani2019).
Climate data collection
Throughout the experimental period (December 2020 to November 2021), the following variables were recorded daily in the coffee plantations: mean, maximum, and minimum air temperature (°C), relative humidity (%), rainfall (mm), and wind speed (m s−1).
The data were stored in databases of weather stations (Vantage Pro2 Davis Instruments, Hayward, California, USA, and TFA Nexus, Reicholzheim, Germany) located at the MUC1 property in Mucugê and BCH2 in Barra do Choça. We obtained the climate data for Vitória da Conquista from the database of the Integrated Environmental Data System (SINDA) of the National Institute for Space Research (INPE) (INPE, 2022a).
Insecticide use intensity assessment
We calculated the insecticide application frequency index (Gravesen, Reference Gravesen2003; Bakker et al., Reference Bakker, Van der Werf and Bianchi2021) and gathered information on the insecticides used on the crops during the study period through questionnaires to the landowners (table S1 in the supplementary material). The following equation determined the insecticide application frequency index:
where AD is the applied dose of the commercial product, and MD is the maximum dose of the commercial product recommended by the manufacturer.
Landscape metrics
Each of the six coffee plantations was georeferenced for characterising the local landscape using a GPS receiver (Gramin eTrex 20, Garmin International, Inc., Kansas, USA). Based on the coordinates surveyed, we mapped the coffee plantations to determine land use and vegetation cover classes using high-resolution images from the CBERS 04A satellite (2-m panchromatic spatial resolution) (INPE, 2022b).
The images were processed in ArcGIS 10.8 (ESRI – Environmental System Research Institute, California, USA) at a 1:5000 scale. The area around each set of coffee plantations was assessed within five landscape radii: 500, 1000, 1500, 2000 and 3000 m (Aristizábal and Metzger, Reference Aristizábal and Metzger2019; Medeiros et al., Reference Medeiros, Grandinete, Manning, Harper, Cutler, Tyedmers, Righi and Ribeiro2019a) (fig. 1). Each land cover determined in the buffers through image processing was vectorised for forest cover estimation, edge density, and Shannon index of landscape diversity (McGarigal and Marks, Reference McGarigal and Marks1995). The metrics were estimated using the Patch Analyst for ArcGIS tool (Rempel et al., Reference Rempel, Kaukinen and Carr2012).
Statistical analyses
The data were analysed using generalised linear models (GLM) with a Poisson error distribution to verify variations in natural parasitism and infestations as a function of the wet and dry seasons in the region. The model selection criterion was based on the smallest values according to the Akaike's information criterion, corrected for small samples (AICc) (Burnham and Anderson, Reference Burnham and Anderson2004; Medeiros et al., Reference Medeiros, Grandinete, Manning, Harper, Cutler, Tyedmers, Righi and Ribeiro2019a).
We calculated the faunal richness (fig. S2, supplementary material) and Shannon–Wiener diversity indexes (H′) (Krebs, Reference Krebs and Krebs2014), which were analysed along with parasitism and infestation rates, climatic factors (maximum, mean, minimum temperature, relative humidity, rainfall, and wind speed), insecticide use intensity (application frequency index) by Canonical Correlation Analysis, with the PROC CANCOR procedure on the SAS Software (Statistical Analysis System) (SAS Institute, 2011).
We used generalised linear mixed models (GLMM) with a Gaussian distribution to analyse the data on the effects of agroecosystem landscape on biological control services (natural parasitism and parasitoid species diversity) and coffee leaf miner infestation using (Zuur et al., Reference Zuur, Ieno, Walker, Saveliev and Smith2009; Aristizábal and Metzger, Reference Aristizábal and Metzger2019; Hohlenwerger et al., Reference Hohlenwerger, Tambosi and Metzger2022). All analyses were performed using the ‘glmer’ function of the ‘lme4’ (Bates et al., Reference Bates, Mächler, Bolker and Walker2015) and ‘emmeans’ (Lenth, Reference Lenth2020) packages.
We used the hypothesis-based modelling to identify the predictive factors that affect the biological variables of insect populations at landscape scales. According to these approaches, the models were ranked and selected for testing the hypotheses for each biological variable (table S2, supplementary material). We built and tested models for each response variable, considering the explanatory variables as fixed effects in the model and a null model represented by the absence of effects (y ~ 1). We added the experimental unit and collection site variables to the models as random effects.
Model selection was based on the Akaike's information criterion, with correction for small samples (AICc) (P < 0.05). The ΔAICc parameters were considered for ranking the models (Burnham and Anderson, Reference Burnham and Anderson2004). The ΔAICc values represent the difference between the best model and each model tested. Therefore, the models whose ΔAICc values were less than or equal to 2.0 were selected (P < 0.05) for being adequate to explain the effects of the predictor variables on the dependents. The estimation of these parameters was performed by the ‘AICctab’ function of the ‘bbmle’ package (Bolker, Reference Bolker2010). All statistical analyses were run on the R 4.0.4 Lost Library Book software (R Development Core Team, 2020).
Results
Coffee leaf miner natural parasitism and infestation rates
We collected 2662 parasitoids from the Eulophidae and Braconidae families (table 1). The eulophid species found included C. coffeellae Ihering, Reference Ihering1914; P. coffeae Ihering, Reference Ihering1914; Neochrysocharis sp. 1; Neochrysocharis sp. 2; Cirrospilus sp.; Horismenus sp.; and Zagrammosoma sp. The braconids collected included S. reticulatus Penteado-Dias, Reference Penteado-Dias1999, and Stiropius sp.1.
Parasitoid communities of Leucoptera coffeella on coffee farms in the Planalto region, Bahia, Brazil
VDC1, Vitória da Conquista; BCH1, Barra do Choça; BCH2, Barra do Choça; MUC1, Mucugê; MUC2, Mucugê; MUC3, Mucugê.
a Constancy and Dominance, according to Basha et al. (Reference Basha, Mostafa and Eldeeb2021), representing the presence of species as: W – in more than 50% of the collections; Y – from 25 to 50% of the collections; and Z – in less than 25% of the collections.
b D: dominant species with a frequency greater than 1/S, where S is the total number of species and nd is the non-dominant species with a frequency lower than 1/S.
Species richness was sufficient regarding the number of samples taken (see fig. S2 of the supplementary material), varying according to the coffee plant studied, with the lowest values found for MUC1 and the highest for BCH2. C. coffeellae and P. coffeae were constant and dominant in all sampled coffee plantations, representing more than 84% of the total parasitoids collected. The coffee plantations with the highest richness and diversity of parasitoids were BCH2 and VDC1. Abundance was higher in BCH2 and MUC3.
Coffee leaf miner natural parasitism and infestations varied in the dry and wet seasons (figs 2 and 3), with higher rates and temperatures during the wettest months (fig. 3) (see fig. S3 in the supplementary material). In this period, coffee leaf miner infestations exceeded the threshold of control established for the region, which is 20% infestation in areas with higher temperatures and 30% in regions with mild temperatures (Souza and Reis, Reference Souza and Reis2000).
Temporal variation of natural parasitism and infestations of the coffee leaf miner in six coffee plantations in the Planalto region, Bahia, Brazil, from December 2020 to November 2021. The blue areas indicate the wet season in the region. The white areas represent the dry season. The red dashed lines indicate the threshold of control adopted for the leaf miner in the region. Properties: (a) VDC1 = Vitória da Conquista; (b) BCH1 = Barra do Choça; (c) BCH2 = Barra do Choça; (d) MUC1 = Mucugê; (e) MUC2 = Mucugê; (f) MUC3 = Mucugê.
The relationships between coffee leaf miner infestations, forest cover, and local landscape diversity are explained by the best model predictors. Blue areas represent 95% confidence intervals. 500 (a, b), 1000 (c), 1500, 2000 (d), 1500 (e), and 3000 (f). The GLMM models were selected according to the ΔAICc ≤ 2.0 values (table 3).
Peak infestations occurred twice. The first peak occurred between December 2020 and March 2021. The second peak occurred between September and November 2021 for BCH1 and BCH2 (fig. 2b and c), with percentages ranging from 7.18 to 42.90%. For the VDC1 coffee plantation, the peak infestation occurred from March to June 2021, with an average of 24.15%. We found an increase in the number of mined leaves starting in October 2021 (fig. 2a).
Two infestation peaks occurred from December 2020 to January 2021 and from September to November 2021 on the MUC1, MUC2, and MUC3 properties (fig. 2d–f).
The species C. coffeellae and P. coffeae showed the highest parasitism percentages among the parasitoids collected in all the coffee plants studied (fig. 2). For VDC1, C. coffeellae and P. coffeae had maximum parasitism rates (4.75 to 24.30% and 3.21 to 20.73%, respectively). However, peaks occurred at the end of the wet season for P. coffeae and at the end of the dry season for C. coffeellae. Mean parasitism rates for Neochrysocharis sp.1, Neochrysocharis sp.2, Cirrospilus sp., S. reticulatus, and Horismenus sp. ranged from 0.40 to 5.88%.
Maximum parasitism rates of C. coffeellae and P. coffeae for BCH1 and BCH2 occurred during the wet season (March 2021) (fig. 2). In MUC3, P. coffeae and C. coffeellae parasitism were higher during the wet season. In MUC2, parasitism by P. coffeae occurred in the wet season (February 2021) and by C. coffeellae in the dry season (September 2021). For the other coffee plantations studied, the parasitism rates for the other parasitoids did not exceed 20%, with rates ranging from 0.31 to 10.12%.
Climatic factors and insecticides effects on the infestations of coffee leaf miners and natural parasitism
The canonical correlation analysis indicated that the first two canonical axes were significant (table 2), showing the effects of climatic factors and the use of insecticides on crops, both for infestations and for natural parasitism of the coffee leaf miner. The first canonical axis for coffee leaf miner infestations explained 97% of the variation, while for natural parasitism, about 99% of all variation was explained by the first canonical axis. The values of the absolute canonical coefficient reveal that the average temperature and precipitation were strongly correlated with infestations; however, positively for temperature and negatively for precipitation (canonical axis 1, table 2). As to the natural parasitism of coffee leaf miners, the variables precipitation and insecticide showed the strongest negative correlations in the two canonical axes (table 2). The other climatic variables showed moderate and weak correlation values (table 2) for infestations and natural parasitism.
Canonical correlation between the climatic factors, insecticides, and the populations of coffee leaf miner and their natural parasitism in Planalto coffee region, Bahia, Brazil
Landscape effects on the coffee leaf miner infestations, natural parasitism, and species diversity
The coffee plantations studied varied regarding landscape contexts (see table S3 in the supplementary material). The significant models for each landscape scale had the lowest AICc value, more than 2 AICc and lower than the null model (table 3).
The GLMM models selected to explain the effects of landscape on coffee leaf miner infestation and natural parasitism
Model selection criteria = ΔAICc ≤ 2.0. The signs in parentheses indicate the direction of the correlation found for the model, where (+) indicates a positive correlation and (−) indicates a negative correlation.
The difference in AICc values from the lowest value is indicated by ΔAICc.
The coffee leaf miner infestations were significantly favoured by increases in forest cover (table 3; Fig. 3a, c, e and f) in all landscape scales, in addition to a greater diversity of the agroecosystem landscape at 2000 m (table 3; Fig. 3d). In a 500 m of radius, the coffee leaf miner infestations increased with the landscape diversity (table 3; Fig. 3b).
The relationship between natural parasitism and edge density (table 3; Fig. 4a, d, g, i and g) in all landscape scales was significant and positive. Greater landscape diversification in 500, 1500, and 3000 m scales (table 3; Fig. 4b, e and k) led to reductions in the pest's natural parasitism. Forest cover and increased natural parasitism were verified at all landscape scales (fig. 4c, f, h and l), except for 500 m since the predictor model was not significant.
The relationships between the natural parasitism of the coffee leaf miner, edge density, local landscape diversity, and forest cover are explained by the best model predictors. Blue areas represent 95% confidence intervals. 500 (a, b), 1000 (c, d), 1500 (e, f, g), 2000 (h, i), and 3000 (j, k, l). The GLMM models were selected according to the ΔAICc ≤ 2.0 values (table 3).
Regarding species diversity, we found that increases in forest cover generated greater diversity (table 3; Fig. 5a, d, g, j and m) in all landscape scales. We observed the same effect for edge density (table 3, fig. 5b, e, h, k and n); however, we found a diversity decrease with landscape diversity increase (table 3; Fig. 5c, f, i and l). For 500 m of radius landscape, the predictor model was not significant.
The relationships between parasitoid species diversity of the coffee leaf miner, forest cover, edge density, and local landscape diversity are explained by the best model predictors. Blue areas represent 95% confidence intervals. 500 (a, b), 1000 (c, d, e), 1500 (f, g, h), 2000 (i, j, k), and 3000 (l, m, n). The GLMM models were selected according to the ΔAICc ≤ 2.0 values (table 3).
Discussion
We have introduced relevant evidence on the effect of landscape, climate, and insecticide use on ecosystem services provided by coffee leaf miner parasitoids at landscape scales in coffee farms. Direct negative effects occurred on the ecosystem services provided by natural enemies, primarily due to rainfall, insecticide use intensity, and landscape diversity.
The parasitoid community associated with L. coffeella presents high diversity and wide geographic species distribution (Lomelí-Flores et al., Reference Lomelí-Flores, Barrera and Bernal2009; David-Rueda et al., Reference David-Rueda, Constantino, Cecilia Montoya, Ortega, Nancy Gil and Benavides-Machado2016). The Eulophidae family, the most abundant and richest parasitoid species, predominates throughout Neotropical America, mainly in countries such as Mexico, Colombia, Costa Rica, and Brazil (Lomelí-Flores et al., Reference Lomelí-Flores, Barrera and Bernal2010; David-Rueda et al., Reference David-Rueda, Constantino, Cecilia Montoya, Ortega, Nancy Gil and Benavides-Machado2016).
Among the 28 species of coffee leaf miner parasitoids that occur in Brazil, C. coffeellae and P. coffeae are the most abundant in coffee plantations in the states of São Paulo, Minas Gerais, Paraná, and Bahia. These species were the most abundant in the coffee plantations of the Planalto region, where coffee is grown through different management systems, including pesticide use. Thus, they are a promising species to be bred and released in coffee plantations, in addition to their wide geographical distribution (Melo et al., Reference Melo, Castellani, Nascimento, Menezes Junior, Ferreira and Lemos2007).
Growing conditions, especially pesticide use, led to different richness values among the studied properties. Some eulophid species, such as C. coffeellae and P. coffeae, showed ecological plasticity and were present in sites with insecticide selection pressure (Melo et al., Reference Melo, Castellani, Nascimento, Menezes Junior, Ferreira and Lemos2007). Thus, they could exhibit higher parasitism rates than other species, such as braconids, which were not collected on farms with higher insecticide use. The hypothesis of competition between the species can also be considered since eulophid have the habit of parasitising coffee leaf miner caterpillars in the early instars. Meanwhile, braconids prefer to parasitise caterpillars in advanced instars, and adult emergence occurs in the chrysalis stage (Melo, Reference Melo2005).
This study recorded the S. reticulatus species for the first time in the state of Bahia. This species had only been recorded in the states of São Paulo (Penteado-Dias, Reference Penteado-Dias1999) and Minas Gerais (Ecole et al., Reference Ecole, Moraes and Vilela2010; Marques et al., Reference Marques, Fernandes, Morais, Haddi and Silveira2022), showing coffee leaf miner parasitism percentages of around 8.8%. Furthermore, we also recorded the Zagrammosoma sp. species for the first time in Bahia, whose occurrence had been reported only in the state of São Paulo (Gravena, Reference Gravena1983). Its participation in controlling the coffee leaf miner in the region was very low, with only one specimen collected in Barra do Choça. However, the species of this genus do not seem to have significant efficiency in controlling the coffee leaf miner, as verified in Colombia, where Zagrammosoma multilineatum Ashmead, 1888 (Hymenoptera: Eulophidae) occurred in low abundance (David-Rueda et al., Reference David-Rueda, Constantino, Cecilia Montoya, Ortega, Nancy Gil and Benavides-Machado2016).
There are controversies regarding parasitoids' role in the coffee leaf miner's biological control in Brazil since parasitism rates are generally considered low (Reis and Souza, Reference Reis and Souza1996; Parra and Reis, Reference Parra and Reis2013). It has been attributed to the constant phytosanitary treatments in coffee plantations and antagonistic interactions with predatory wasps that can prey on the parasitised caterpillars of the coffee leaf miner indiscriminately (Reis et al., Reference Reis, Souza and Vilela2000).
There have been discussions on the effects of climate conditions on infestations and parasitism of the coffee leaf miner, especially temperature and rainfall (Pereira et al., Reference Pereira, Picanço, Bacci, Della Lucia, Silva and Fernandes2007a; Lomelí-Flores et al., Reference Lomelí-Flores, Barrera and Bernal2009; Amaral et al., Reference Amaral, Venzon, Pallini, Lima and De Souza2010; Jaramillo et al., Reference Jaramillo, Garcia-Gonzalez and Rugno2019). Under higher temperature conditions, infestations are favoured and parasitism also increases since species benefit from the availability of hosts and the greater allocation of food resources from plant species that produce nectar and pollen (Rosado et al., Reference Rosado, Araújo, Pallini and Venzon2021; Venzon, Reference Venzon2021).
Rainfall is one of the main factors in the coffee leaf miner's natural mortality. Droplets concentrating on leaves can drown caterpillars inside the mines (Nestel et al., Reference Nestel, Dickschen and Altieri1994; Fernandes et al., Reference Fernandes, Mantovani, Bonfim Neto and Nunes2009) and increase moisture on the plant tissue surface, leading to decreased oviposition. Mortality of pest adults can also occur when directly hit by raindrops (Pereira et al., Reference Pereira, Picanço, Bacci, Della Lucia, Silva and Fernandes2007a, Reference Pereira, Picanço, Bacci, Crespo and Guedes2007b). Since the parasitoids are small, they also reduce their movement during the wet season (Ecole et al., Reference Ecole, Moraes and Vilela2010; Lomelí-Flores et al., Reference Lomelí-Flores, Barrera and Bernal2010). Thus, high humidity and rainfall conditions, either natural or artificial, through irrigation (Custódio et al., Reference Custódio, Moraes, Custódio, Lima, Faria and Gomes2009), affect population dynamics and the correlation between the coffee leaf miner and its natural enemies (Fernandes et al., Reference Fernandes, Mantovani, Bonfim Neto and Nunes2009).
Insecticide use intensity led to reduced pest infestations, natural parasitism, and parasitoid diversity of the coffee leaf miner. Morgan et al. (Reference Morgan, Thébault, Seymour and Van Veen2017) demonstrated that insecticide use directly affects the dependency relationship of parasitoid densities. Therefore, selective insecticides that kill pests without causing negative effects on parasitoids should be searched to allow the conservation of these natural enemies in coffee plantations (Fernandes et al., Reference Fernandes, Silva, Gorri, Pucci and Silva2014; Carvalho et al., Reference Carvalho, Grützmacher, Passos, Oliveira, Souza, Vázquez and Marucci2019). Studies have shown that coffee leaf miner parasitoids are more sensitive to the effects of insecticides than predatory wasps (Gusmão et al., Reference Gusmão, Picanço, Gonring and Moura2000; Carvalho et al., Reference Carvalho, Miranda, Vilela, Moura and Moraes2004; Bacci et al., Reference Bacci, Fernandes, Picanço, Crespo and Campos2006, Picanço et al., Reference Picanço, Oliveira, Fernandes, Martinez, Bacci and Silva2012). Organophosphate insecticides, neonicotinoids, and some pyrethroids present toxicity to parasitoids in coffee plantations (Carvalho et al., Reference Carvalho, Miranda, Vilela, Moura and Moraes2004, Reference Carvalho, Miranda, Moura, Rocha, Reis and Vilela2005).
In the coffee plantations of Mucugê, the use of neonicotinoid, pyrethroid, and organophosphate insecticides (Leite et al., Reference Leite, Guedes, Santos, Costa, Moreira, Matsumoto, Lemos and Castellani2020b) may be associated with the lower diversity of parasitoids found. These results indicate the need for selectivity studies and clarifying the impacts of the compounds on the parasitoids' bioecology. Therefore, the dependence on pesticides in coffee crops should be reduced for MUC1, MUC2, and MUC3. A more sustainable practice would the intercropping with no-cultivated pollen-producing plants between the coffee lines, as they provide food and shelter for parasitoids and other enemies, which are important natural resources to control the coffee leaf miner (Rosado et al., Reference Rosado, Araújo, Pallini and Venzon2021; Calderón-Arroyo et al., Reference Calderón-Arroyo, Togni, Pantoja, Saenz and Venzon2023). In addition, reforestation programmes on surrounding coffee crops can also contribute to the increase in natural parasitism in these areas.
In regenerative coffee farming, the search for diversifying coffee plantations has proven to be an important tool for pest management for conserving ecosystem services provided by parasitoids and predators (Rezende et al., Reference Rezende, Venzon, Perez, Cardoso and Janssen2014; Rosado et al., Reference Rosado, Araújo, Pallini and Venzon2021; Venzon, Reference Venzon2021). Natural vegetation areas, forest remnants, and ecological corridors are true biodiversity deposits of natural enemies in agroecosystems (Librán-Embid et al., Reference Librán-Embid, De Coster and Metzger2017; Villa et al., Reference Villa, Santos, Sousa, Ferreira, da Silva, Patanita, Ortega, Pascual and Pereira2020; Vilchez-Mendoza et al., Reference Vilchez-Mendoza, Romero-Gurdián, Avelino, DeClerck, Bommel, Betbeder, Cilas and Beilhe2022).
According to Medeiros et al. (Reference Medeiros, Grandinete, Manning, Harper, Cutler, Tyedmers, Righi and Ribeiro2019a), natural enemies exploit crop resources under certain temporal and spatial circumstances. Our results showed that the multi-scaling structure of the landscape edge is a spatial factor that favours the natural parasitism of the coffee leaf miner and the diversity and richness of parasitoids in coffee plantations. The resource abundance at the edges of vegetation near coffee plantations allows the parasitoids to explore more than one habitat. Thus, their stay at these sites is favoured and allows for increased parasitism.
Forest cover is positively associated with higher diversity and abundance of natural enemies in coffee agroecosystems (De la Mora et al., Reference De la Mora, García-Ballinas and Philpott2015; Allinne et al., Reference Allinne, Savary and Avelino2016). This study showed that the forest cover favours coffee leaf miner infestations and natural parasitism. However, it is worth noting that the landscapes varied across the regions where coffee was grown. In the proprieties of Mucugê, for example, there is landscape diversification (table S3, supplementary material), where landcover predominates with annual crops such as potatoes, vineyards, and areas of exposed soil prepared for new plantings surrounding the coffee crops. Normally, the pressure exerted by pesticides in these areas is intense and can lead to, in addition to the management of the coffee crop itself, negative impacts such as sublethal effects on populations of natural enemies (Carvalho et al., Reference Carvalho, Miranda, Moura, Rocha, Reis and Vilela2005).
These conditions decrease the natural parasitism and the diversity of parasitoid species of the coffee leaf miner. Moreover, the coffee leaf miner is considered a monophagous pest, attacking exclusively the coffee tree, which indicates that the surrounding landscape impacts their populations at a much lower level than their natural enemies. Thus, in such conditions, their populations decrease mainly due to the use of insecticides, both inside and outside the coffee crops. It has been shown that the simplification of natural areas into agricultural areas is associated with a decrease in biological control in coffee plantations (Righi et al., Reference Righi, Campoe, Bernardes, Lunz, Piedade and Pereira2013; Medeiros et al., Reference Medeiros, Martello, Almeida, Mengual, Harper, Grandinete, Metzger, Righi and Ribeiro2019b).
Further studies should focus on the C. coffeellae and P. coffeae species since their effective participation in the biological control of the coffee leaf miner indicates their great potential for exploitation in rearing and massive release programmes in coffee crops. The occurrence of S. reticulatus and Zagrammosoma sp. expands the species diversity of the parasitoid communities present in coffee plantations in Bahia. However, further studies should address species surveys in other locations and coffee-growing regions in Bahia, such as the Atlantic (South and Far South of the state), and types of coffee management, such as organic, intercropping, and agroforestry systems, among others. In the case of the BCH1 and BCH2 properties, management recommendations to producers would be the cultivation of no-intercropping pollen-producing plants, as well as the use of semiochemicals (Bacca et al., Reference Bacca, Lima, Picanço, Guedes and Viana2006), which could help in monitoring, especially in a season of coffee leaf miner outbreaks. In addition, the reduction of pesticide use is important to avoid the problems of resistant population selection.
Coffee farms may present different conditions for the aforementioned factors, thus affecting population dynamics. Therefore, these scenarios should be further explored based on the effects of the temperature, rainfall, insecticide use, and landscape characteristics in the coffee leaf miner and its parasitoid populations.
We have introduced new insights into the natural parasitism of the coffee leaf miner, in addition to demonstrating that natural parasitism occurs in all seasons. The C. coffeellae and P. coffeae species are highlighted for being abundant in all the studied areas. We report the unprecedented occurrence of S. reticulatus, Neochrysocharis sp. 1, Neocrhysocharis sp. 2, and Zagrammosoma sp. in Bahia. We emphasise the C. coffeellae and P. coffeae as potential species for the development of novel commercial bioproducts that are still not available for coffee crops, but are proven necessary to reduce dependence on pesticides in coffee leaf miner management.
The forest cover and edge density increase the natural parasitism of the coffee leaf miner. The diversity of crops around coffee plantations reduces natural parasitism, which also occurs in coffee areas with greater use of pesticides in coffee plantations. Thus, we recommend that coffee growers pay attention to the role of the forest cover close to the coffee plantations. It enhances the biological control provided by the parasitoids and maintains the diversity of species and their valuable ecosystem service. Furthermore, incorporating cultural techniques with pollen-producing non-cultivated plants may be a more sustainable option to be adopted in coffee crops.
Such an information should be used to develop coffee leaf miner management programmes with parasitoids in the region, adopting conservation practices of the natural landscape and habitat, and reducing the use of insecticides to contribute to the sustainability of coffee production.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007485323000482.
Acknowledgements
We would like to thank the University of Southwestern Bahia (UESB), the Research Support Foundation of the State of Bahia (FAPESB) and the Coordination for the Improvement of Higher Education Personnel (CAPES) for their financial support. We would also like to thank the coffee producers at Progresso, Ceci, Primavera, and Empresa Café Bahia farms for their support in the experiments. Furthermore, we would like to thank Prof. Daniell Rodrigo Rodrigues for identifying the parasitoids and the Institutional Capacity Building Program – CI/CNPq (process no. 312879/2019-9).
Authors’ contributions
Conceptualisation and design of the work – Santos M. P. dos, Castellani M. A., Leite S. A., and Carvalho G. A.; data acquisition and methodology: Santos M. P. dos, Castellani M. A., Santos-Neto B. de M., Cardoso, A. C. P., Santos I. dos and Coelho B. S.; Software and statistical analysis – Santos, M. P.; Species identification – Fernandes D. R. R.; Investigation: Santos M. P. dos and Castellani M. A.; writing: Santos M. P. dos, Castellani M. A., Carvalho G. A., and Fernandes D. R. R.
Competing interest
None.
