Management intensity affects insect pests and natural pest control on Arabica coffee in its native range

The species diversity and functional structural complexity associated with agroforestry systems provides natural pest control in the form of suppression levels by enemies without human intervention (Karp et al., 2013; Vandermeer & Perfecto, 2010). This is apparent, for example, when comparing to production a single or few annual crops open fields (Pumariño 2015; Rosati 2021). As such, has potential both conserve biodiversity provide an income smallholder farmers (Cerda 2017; Wei 2018). However, comes many shapes shades, we lack insights into how management intensity affects systems, especially along broad gradients Schroth 2000). A promising area study effect centre origin crop, where exist that start crop growing shade forest end more intensively managed plantations (Zewdie 2020). More intensive such as coffee cacao often decreases spatial heterogeneity (Geeraert 2019; Steffan-Dewenter 2007), less are characterized lower levels, example through higher density (Clough 2011). Studies on may shed light sustainability intensification, possibly identify thresholds intensification leads abrupt changes levels. From perspective, it also important know whether patterns uncovered consistent across seasons, similar different species. have generally found sites (Bianchi 2006; Dalin 2009). For showed damage leaf miner Leucoptera coffeella decreased, richness abundance parasitoid wasps increased, increasing cover (Medeiros 2019). Likewise, cocoa increased agroforests, whereas predators native trees (Bisseleua 2013). these relatively short, further be gained from investigation impact broader gradient, which might range its environment commercial plantation. predominant pests frequently differ among years seasons (Teodoro 2009), perspective thus one season reflective those other show responses actions. Despite this, studies focused and/or (Asfaw Bagny Beilhe 2020; Liebig Landscape features, elevation cover, shaping At landscape scale, insect increase decrease Jonsson 2015). local canopy highly sites, strong microclimate thereby affect organismal physiology plant-insect-natural enemy interactions (Jamieson 2012). Management can (Diehl Martin Rusch beneficial insects like parasitoids abundant than (Jonsson Medeiros Whitehouse Several direct indirect mechanisms underlie this including use pesticides scarcity nectar plants, alternative hosts, overwintering shelters Tscharntke 2007). examined plants their (Chen Such important, (i) knowledge indigenous help sustainably manage range, (ii) coevolved pest's component biodiversity, (iii) specialized treasure trove future agricultural scientists globe searching new classic biocontrol (Escobar-Ramírez Arabica southwestern highlands Ethiopia, Mount Marsabit Kenya Boma plateau Sudan (Anthony 2002; Tesfaye 2014). In involves intensities, coffee, wild little no management, semi-forest traditional way filling gaps seedlings, removal slashing competing thinning canopy, semi-plantation systematically planted pruned, but still traditionally (iv) plantation using modern agronomic practices, heavy pruning, application fertilizers, frequent weeding herbicides (Hundera Labouisse 2008; Teketay, 1999). co-exist mosaic fashion unique experimental setting examine control. Understanding practices allows design implementation maintain enhance survival, reproduction efficacy (i.e. conservation biocontrol), contribute Based reduction habitat quality enemies, expected all major increase, parasitism decrease, intensity. Given previous reports species-specific differences life-histories niches, season, shade. was conducted Gomma Gera districts Jimma zone Ethiopia (7°37?–7°56?N 36°13?–36°39?E; Figure 1a). receives rainfall 1480 2150 mm, main rainy between June September. Mean daily minimum maximum temperatures 12°C 28°C, respectively. ranges 1400 2500 m a.s.l. remnants forests areas agriculture. region Coffea arabica L., understory shrub unmanaged well plantations. Improved varieties resistance berry disease Colletotrichum kahawae sometimes used common plantations, improved not known exhibit cross-resistance (or cross-susceptibility) against (Labouisse 2008). Coffee cash source region. attacked diverse group (Le Pelley, 1973). Three lepidopteran very our area. blotch caffeina (Washbourn) [family Lyonetiidae] oviposits eggs rows 1 13 (Notley, 1956), larvae create distinct mine while feeding gregariously upper side (Figure 1b). serpentine Cryphiomystis aletreuta (Meyrick) Gracillaridae] mines leaf, leaves distinctly imprint feeds snake-like manner 1c). skeletonizer Leucoplema dohertyi (Warren) Epiplemidae] free-feeding larva underside behind epidermis veins 1d). Finally, there several feed lepidopterans, slugs snails 1e; Abedeta Crowe, 2004). borer, berries regions world, (Mendesil area, most problematic L. caffeina. 2011), only occasional predation ants birds. countries related challenge, predatory wasps, ants, birds bats reported predate (Androcioli 2018; De la Mora Fernandes 2010; Librán-Embid 2017). While community been described detail African (Godfray Hassell, 1989; Notley, carried out none gradient assess center crop's Ethiopia. eastern Africa, genus plant species, particularly high 1956; Washbourn, 1940). exact past distribution unknown, scientific record living memory invaded likely hundreds, if thousands, Permits field were granted Tea Authority (No. BS/8075/8–96/83). No ethical approval required. total sixty distributed four selected districts, distance 0.5 km 1a; see Zewdie 2020 detailed information site selection). Following Gole al. (2002), (2008) Hundera (2013), classified forest, semi-forest, systems. Forest had gap-filling planting pruning systematic, following herbicides, though insecticides S2; Within each site, 30 × plot established, 16 shrubs intersections 10 gridlines S3). survey three herbivores wet (July–August 2018) dry (January–March randomly two branches (as based being representative typical size structure shrub) plot. On branches, counted number leaves, damaged, separately miner, herbivores. Values pooled obtain site-level averages analyses. We do note pest-induced drop, from, C. (Dantas 2021; Waller result underestimation herbivory recorded GPS, estimated five pictures taken above ImageJ software v. 1.50i (Schneider 2012; details). To (forest, plantation). 2018, collected 100 mined site. These infested subset medium large unbroken surface. densities low system, could collect enough meaningful comparison, excluded Leaves stored plastic boxes (25 6 cm) sheets toilet paper absorb moisture prevent fungal growth. Boxes laboratory away sunlight, they kept 35 days, host emergence ended. then calculated proportion emerged divided summed adult moths parasitoids, henceforth referred rate. families Braconidae, Encyrtidae, Eulophidae Ichneumonidae morphological characteristics (e.g. Askew, 1968; Ubaidillah, 2006). link empirical perception farmers, interviewed sub-managers system. avoid leading questions influence education sub-managers, asked during larger interview diseases, phrased ‘Did any pathogen change over last decades, so, become problematic?’ ‘What challenges threats you faced your recent years?’ noted type perceived problem. All interviews same persons (Biruk Ayalew Dinkissa Beche). Statistical analyses R v 3.6.3. Mixed-effects models fitted functions lmer glmer lme4 package (Bates 2015), logit function glm base (R Core Team, evaluated model fit diagnostics sjPlot DHARMa packages (Lüdecke, 2020), tested significance Anova car (Fox Weisberg, 2019), compared emmeans (Lenth, modelled average at level (the herbivores) linear mixed Gaussian identity function. Variance inflation factor <5 models, indicating problems multi-collinearity (James account sampling plots included random identity. two-way variables. differed (Table S1), ran season-specific package. scaled predictor variables zero mean unit variance standardized regression coefficients. moran.randtest adespatial test autocorrelation residuals (Diniz-Filho 2003). Spatial significant (p < 0.05), exception weak pattern = 0.03). system rate sum hosts generalized mixed-effects binomial link, accounted overdispersion did include due replicates analysis. composition affected adonis2 vegan (Oksanen complemented individual above, counts response variable Poisson link. probability farmer reporting problem least types. inspected 1499 ± 295 (mean SD) per percentage varied 8.2% 3.8% 38.9% 27.9% remarkable seasonal dynamics. times (12.0 vs. 4.4%, respectively), seven (0.9 6.7%). incidence (38.0% 39.7%, damaged decreased 30.6% 25.2% season. leave detectable 2, Table S1). infestation significantly 2A,C,D). weaker trend apparent 2B). elevation, 3, 1). shade, either S1, total, reared 2332 majority belonging Encyrtidae (66.1%), (24.7%) Braconidae (8.0%; 4). clear (?2 68.6, df p-value 0.001), 4a; S2). (F2,9 3.19, 0.01). 7.48, 0.02), 12.90, 0.002) 55.6, 0.001) notable relative encyrtid eulophid 4b; 5; Tables S4 S5). Farmers perceive half 5). None interviewees mentioned investigate intensity, origin. Levels highest dynamics, declined weakly influenced matched community. Together, findings suggest does gradually erode control—rather, moving guide decision-making landscapes southwest rapidly increasing, time incentives sustainability. reverse pattern. contrast, seasons. previously, finding matches (Ababulgu, match Lomelí-Flores (2010) abundances Mexico. Brazil, world's largest producer reaches peak (Pereira undetectable why pronounced applied emphasizes should rely data devise plans. higher. confirmed interviews, revealed unimportant rarely caused reached outbreak congruent, cannot rule some awareness pests. Our line garden (Abebe, 1987). particular, disturbed sites. Mendesil (2011) environmental requirement miners. study, antestia bugs borer previously Babin 2018), highlighted varies factors explain difference hand, hand. cause Teodoro mostly unaffected range. When parts realize full grown sun near non-existent. It non-linear, directly coffee. Besides factors, productions explained bottom-up top-down effects. striking cultivars farms largely genetic reservoir kahawae, (nor believed) susceptible pests, uniformity favour adaptation crop. find diseases 2023). reduce vegetation complexity, 2010) effects Schmidt-Jeffris 2022). Matching expectation, supports hypothesis Lower suggested spiders Kone 2014; Philpott Sperber characterization hymenopteran Braconidae. dominance reflects investigations miners Africa (Kerrich, 1969; research stations Interestingly, families, drop sensitive others, unfortunately available validate hypothesis. One interesting direction explore network (van Nouhuys Hanski, 2005). Another would detected agents, (Hoehn 2009; Karp Armbrecht, densities, reduced, highlights pressures below economic threshold farms, losses occur actions retain need taxonomic describe morphologically molecularly, ecological life cycles requirements Ecologically informed even process gives dual benefit sustainable livelihood farmers. Conserving harbours Tamiru Shimales, Esayas Mendesil, Beyene Zewdie, Biruk Ayalew, Kristoffer Hylander Ayco J. M. Tack conceived ideas designed methodology. Shimales data. analysed led writing manuscript. authors contributed critically drafts gave final publication. funded grant Swedish Research Council (2019-04493 A.J.M.T.), Bolin Centre Climate Ethiopian Institute Agricultural (EIAR). thank Moritz Stüber producing map, owners allowing us work plots. conflict interest. Data via Dryad Digital Repository https://doi.org/10.5061/dryad.h9w0vt4nw (Shimales S1. caffeina, aletreuta, Shown p-values ?2-values degrees freedom. Significant > 0.05) shown bold. S2. Pairwise comparisons compared, estimates, standard errors (SE), z-ratios p-values. S3. family. There Ichneumonidae, formal S4. perceives pest. S5. Comparisons complete separation 100% “failures”), case logit(0) ???, results extreme values parameter estimates errors. (a) (b) (c) dohertyi, (d) blue red circles represent solid dashed lines non-significant relationships, photo compilation illustrating systems: management; canopy; managed; herbicides. Photo credits: Zewdie. Schematic overview layout (one site). Each 50 m, intersection inner 10–63 gridlines. Canopy photos locations indicated 64 circles. Please note: publisher responsible content functionality supporting supplied authors. Any queries (other missing content) directed corresponding author article.