Get ready for my Banana poster...
Updated: Nov 1, 2019
Today is CLESCon at the University of Exeter!
I am presenting a poster and also giving a talk today - as my talk is not going to be about the meta-analysis itself, I am uploading the poster and a bit of a description just in case someone misses my very colourful poster with a banana plant in front of it!
Please keep in mind - I have only officially started my PhD in October!
While biocontrols are generally viewed as an environmentally friendly alternative to chemical controls of crop diseases and pests, it may be called a ‘common perception’ that biocontrols are expensive and unreliable in the field. These descriptions may only be myths (Harman, 2000). Bale et al.’s (2008) showed that biocontrols have greater success ratios (1:20 compared to 1:200,000) and lower developmental costs ($2 million/product compared to $180 million/product) compared to chemical controls.
While there are complicated issues with biocontrol commercialization, to begin with (e.g. scaling up production, registration, shelf life) (Fravel, 2005), it is high time now to review publications to critically evaluate how researchers have been testing the effect of biocontrols and if some experiments are more effective due to experimental designs (e.g. duration of experiments, inoculation methods) and general approaches (e.g. using a mixture of biocontrols) which need to be considered for future experiments.
In general, researchers isolate fungi or bacteria from healthy soils or use well-known biocontrol isolates against plant diseases. Scientists then culture and screen the isolates for biocontrol-potential qualities in vitro (e.g. predation, parasitism, antibiosis, antibiosis, antagonism) (Pliego et al., 2011). Isolates of interest are trialled under greenhouse conditions or detached leaf or fruit essays, where either plant health (e.g. height, weight) or disease-related measurements (e.g. disease incidence or growth) are recorded. Finally, the isolates trialed under field conditions and their efficiency against disease is established. These approaches are similar to finding pest biocontrols (e.g. entomopathogens, parasites) but focusing on increasing pest mortality or decreasing pest reproduction.
Most publications report statistically highly significant results on biocontrols reducing disease severity, so the question is no longer “Do biocontrols have an effect on diseases and pests?” but “What is the magnitude of the effect of biocontrols?” (Madden et al., 2015). Multiple studies investigating the effect of biocontrols allow us to identify different sources of variation (e.g. experimental setups, experiment length, biocontrol types) in the outcomes of experiments across studies with the use of meta-analysis. Meta-analysis uses a set of statistical tools to synthesize effect sizes (i.e. outcomes) across different publications to examine the general patterns of experiments’ outcomes (Gurevitch et al., 2018).
Some of the main groups of biocontrols against plant diseases and pests are i) endophytes, ii) plant growth-promoting rhizobacteria (PGPR), iii) mycorrhizal fungi and iv) entomopathogens. Endophytes are microorganisms (i.e. fungi and bacteria) that spend at least part of their life cycles inside plants without causing any damage (Hardoim et al., 2015). PGPR can be endophytes but can also be found in bulk soil and rhizosphere which directly (e.g. biofertilisation or root growth stimulation) or indirectly promoting plant growth (e.g. antibiosis or induction of plant resistance) (Lugtenberg and Kamilova, 2009). Mycorrhizal fungi have been important symbiotic partners of plants for approximately 450 million years, helping with nutrient uptake in exchange for plant carbohydrates (Bonfante and Anca, 2009). Entomopathogen is a general term for any disease-causing organism (e.g. viruses, bacteria, fungi, mycoparasite) in arthropods (Fuxa, 1987) which can also be endophytes and PGPR (Vega et al., 2009). Some of the biocontrol mechanisms against plant diseases and pests include antibiosis, synthesis of lytic or detoxifying enzymes, competition for space on roots with detrimental microorganisms, activation of plant defence mechanisms and improving plant nutrition.
The term ‘banana’ is used for genomic clusters of Musa spp., including dessert bananas (genome AAA) and cooking bananas (ABB) and plantains (genomes AAB) (D’Hont et al., 2012; Heslop-Harrison and Schwarzacher, 2007). Bananas can be diploid, triploid and tetraploid which arose from the hybridisation of two haploid species, Musa acuminata Colla (AA) and Musa balbisiana Colla (BB). Banana breeding for disease resistance is challenged by the perennial nature of bananas, low female and male fertility and introducing new traits into a hybrid is extremely difficult (Dale et al., 2017b).
Banana production and diseases have been reviewed thoroughly previously (Ploetz, 2009; Ploetz et al., 2015; Stover, 1986), so we aim to provide here a quick overview of major banana pests and diseases and report a meta-analysis and open-access database on biocontrol experiments to identify how experimental design affect experimental outcomes.
METHODS (the basics)
Articles reporting experiments using biological controls against banana pests and diseases were accessed to assess if authors have reported standard deviations or standard errors of the means of treatment (sT) and control groups (sC). With standard errors, we extracted treatment means (XT) and control means (XC) along with sample sizes (i.e. replications, NT and NC).
For disease control (e.g. disease incidence or severity), the smaller the disease means (XT), the more effective the biocontrol was compared to the control means (XC). For plant health measurements (e.g. yield, plant height), the higher the treatment means (XT) were, compared to the control means (XC), the more effective the biocontrol was. The absolute mean differences were calculated XC-XT for disease measurements, and XT-XC for plant health measurements. To quantify the effect of biocontrols on disease control or plant health, we calculated Hedges’ d (Hedges, 1994; Rosenberg et al., 2004).
RESULTS (in a nutshell)
In total, we identified 91 publications with 1,093 observations on biocontrol experiments against banana pests and diseases from 26 countries. The top five banana pest and disease were FOC (35.8%), R. similis (10.5%), M. fijiensis (8%), C. sordidus (6.5%) and Banana Bunchy Top Virus (6.4%). Most of the biocontrols were bacteria (52.5%) or fungi (34.5%), while the combination of bacteria with fungi was rarer (6.5%).
Most observations were reported for Bacillus amyloliquefaciens (11.3%), Pseudomonas fluorescens (8.5%), B. subtilis (5.8%), Glomus mosseae (4.6%), non-pathogenic Fusarium oxysporum (3.8%), Beauveria bassiana (3.4%) and Trichoderma harzianum (2.9%). Most observations were from greenhouse (40.5%) and in vitro experiments (33%) while field trials were rarer (22%) and only a few detached leaf or fruit essays were reported (4.6%). Most experiments only used one biocontrol strain (84.6%) while a small proportion used two (12.7%), three (1.4%), four (1%) and five (0.3%) strains.
From the 91 publications, 29 did not report standard errors or variances for 554 observations. From these observations, 37.7% was from in vitro experiments, 30.5% from the greenhouse, 29.5% from the field and 2.3% from detached leaf or fruit essay. These observations could not be included in the quantitative meta-analysis but several of these publications reported large response ratios (80-95% disease control or plant health promotion).
In total, 62 publications with 516 observations reported mean values with standard errors or variances which could be included in the quantitative meta-analysis.
Our meta-analysis found that:
PGPR with mycorrhiza and mixtures of biocontrols generally have greater effects against banana pests and diseases in terms of disease reduction or plant health promotion while endophytes had the smallest effect.
The combination of PGPR and mycorrhiza had the largest effect overall against all banana pests and diseases and especially against FOC.
The biocontrol’s effects were larger in controlled environments and in relatively shorter experiments, when the plants were drenched with biocontrol inoculums, suggesting that greenhouse experiments might be over-optimistic for predicting biocontrol efficiency under field conditions
We had to discard over half of the reported observations from 92 publications due to the lack of standard error or variance reporting. While authors stated that they carried out 3-10 replications of experiments, in many cases, a summary table only contained the mean values and only letters (e.g. a, b, c) indicating statistically significant differences. Some inconsistencies of sample omittance were also observed in publications. Some articles did not report banana cultivars or genotypes or for FOC, did not report if it was Race 1 or TR4.
It is necessary to standardise biocontrol experiments and their reporting for future experiments.
Please feel free to get in touch - let me know what you think of the poster! : )
It was made for a rather informal PhD conference at my own uni, hence the fun-ness of it! Also, watch out for this place, I hope to upload a pre-print of the meta-analysis soon!!!