On 19 June 2014, we conducted a two-tiered search (through that date) on the Web of Science Core Collection, CAB International, MEDLINE, Biological Abstracts, FSTA (Food Science and Technology Abstracts), and Zoological Record databases, using the ISI Web of Science search tool. We located 239,571 unique publications with the search terms: cotton OR Gossypium. A search of these records using the term “cover crop” resulted in 424 publications, composed of refereed articles, conference proceedings, research reports, and bulletins. With examination of these 424 eligible publications, 320 were excluded because they met our exclusion criteria: means for cover crop or no-cover crop treatments were not included, cotton yield or weed growth were not reported, article was a duplicate, article did not contain primary data (review or book), or they were not obtainable using interlibrary loan services (five articles). We did not include intercropping (cover crops grown simultaneously with cotton) studies, nor did we include studies that used weed count as the response variable. For the weed biomass effect size (ES), if an experiment included both weed and weed-free fallow no-cover-crop controls, we used the weed fallow no-cover-crop control in our analysis. If an experiment included herbicides applied over all treatments in season, we excluded the weed biomass ES but included the cotton biomass ES. We identified 104 articles that met our screening criteria (a full citation list and details of primary studies are provided in the supplemental material). Papers spanned 48 yr and were in English and Portuguese languages.
Treatment means and number of replications (sample sizes) were collected for each study. For publications reporting means for more than one no-cover-crop (control) treatment in a nonfactorial experiment, we used the no-cover-crop control that most closely approximated the cover crop treatment. If replications were given as a range, we used the smallest value. For studies that did not report number of replications, we used n = 1 unless LSD or SEs were provided, in which case we used n = 2. If data were provided in graphical form, means were extracted using WebPlotDigitizer (Rogatgi, 2011).
Multiple treatment combinations from one article were treated as independent studies (also referred to as trials or paired observations in the meta-analysis literature) and represented individual units in the meta-analysis. For example, Ashworth et al. (2018) and Li et al. (2013) examined the effects of two cover crop species over 3 yr, resulting in six studies from that article for lint yield ES. Vasilakoglou et al. (2011) studied control of three weed genera by four varieties of one cover crop species, resulting in 12 studies for the weed control ES. Although, the use of multiple studies from one publication has the disadvantage of increasing the dependence among studies that are assumed to be independent (Gurevitch and Hedges, 1999), the greater number of studies maximizes the meta-analysis’ statistical power (Lajeuness and Forbes, 2003). This approach has been used often in agricultural and plant biology meta-analyses (Mayerhofer et al., 2013; McGrath and Lobell, 2013; Ferraretto and Shaver, 2015). Therefore, we derived 1117 studies from 104 articles. As in prior meta-analyses (Ashworth et al., 2018; Mayerhofer et al., 2013), we used the final time point in the meta-analysis for studies that included data for multiple time points in one season. One exception was weed control, as an article used in this meta-analysis reported means that were recorded at three time points during the season (Norsworthy et al., 2010). Considering that each year of an experiment provides varying growing conditions only weakly correlated with other years (repeated measures across years is not needed in our experience), we considered each year as an independent study in the meta-analysis.
- Meta analysis of cotton yield and weed suppression moderator data xlsx
To systematically evaluate cover crop effects on cotton yield and weed...MD5:
|Release Date|| |
Ag Data Commons
|Contact Name|| |
|Public Access Level|| |
|Program Code|| |
005:040 - Department of Agriculture - National Research
|Bureau Code|| |
005:18 - Agricultural Research Service