A global synthesis and meta-analysis of the environmental heterogeneity effects on the freshwater biodiversity
Data files
Sep 29, 2023 version files 108.18 KB
-
data.csv
-
meta_data.xlsx
-
R_script.R
-
README.md
Abstract
In this study, we aimed to synthesize the global knowledge about the relationship between spatial Environmental Heterogeneity (EH) and freshwater biodiversity (i.e., taxonomic and functional diversity, and their respective α and β components). Through a systematic review, we integrated results from 98 studies, published in 33 different countries, about the role of spatial EH – biodiversity relationship in freshwater ecosystems. Following the PRISMA-ecoevol criteria, we generated a dataset with 276 observations of the effect of EH – biodiversity relationship. For the meta-analysis, we extracted the effect size from 74 studies that effectively tested the effect of EH over any metric of biodiversity. Through this subset, we provided 242 individual effect sizes with their respective variances and sample sizes.
README: A global synthesis and meta-analysis of the environmental heterogeneity effects on the freshwater biodiversity
https://doi.org/10.5061/dryad.g4f4qrfw2
The present database comes from a review and meta-analysis research that aimed to elucidate the overall effect of the environmental heterogeneity over biodiversity in freshwater ecosystems. The alpha and beta facets of taxonomic and functional diversity were considered, as biodiversity metrics. The research covers data from articles published in English between 1975 and 2020 around the world. The R code to run the meta-analysis and bias analysis is also available.
Description of the data and file structure
1. data.csv file - is a database in which each line corresponds to an observation of the relationship between environmental heterogeneity and a biodiversity metric.
2. meta_data.csv file - describes each attribute of data.csv file
3. R_scrip.R file - provides the full code to calculate the z-Fisher and variance\, as effects sizes. Also\, we provide code to run four random-effect meta-analysis for each metric of biodiversity (alpha and beta taxonomic and functional diversity) and mixed-effects meta-analysis models that incorporated moderators to investigate variations on the overall mean effect size of EH on taxonomic alpha diversity.
Methods
Data collection
We searched for studies that investigated the effect of EH over any metric of biological diversity (e.g., richness, taxonomic diversity indexes, functional diversity, or beta diversity). We applied three different approaches of data searching, aiming to detect as many studies as possible and, consequently, avoiding data bias. The first approach consisted of searching for terms in the three search engines ISI Web of Science, Scopus and Scielo, from 1945 to 2020. We used 46 terms that are frequently used to refer to EH (e.g., "habitat* heterogen*" OR "habitat* diversity" OR "habitat* complexity" OR "structural complexity" OR “fractal heterogen*” OR "biotop* heterogen*" OR "environment* complexity"). The second approach consisted of a general search for studies on the Google Scholar webpage, using the four most used terms to refer to EH according to a previous study. In this case, we established a maximum of 677 articles to screen because the number of results was excessive (1,370) and overestimated. The third approach consisted of searching the references of previous reviews and opinion articles that included freshwater ecosystems in the central theme. Through this, we added 29 studies that were not captured by the search engines and Google Scholar. After duplicate exclusion, we obtained 2,381 studies.
To select studies we considered the following criteria: (1) encompassed freshwater ecosystems and their associated organisms; (2) were either observational, experimental or both; (3) considered only spatial EH (temporal EH was not included); (4) explicitly investigated the relationship between EH and any biological metric of taxonomic and functional diversity; (5) were published in a peer-reviewed journal. In the title and abstract screening stage 2,176 studies we excluded, thus a total of 205 studies were fully read and evaluated. After this second screening, we selected 98 studies to be included in our dataset.
Qualitative data extraction
To conduct a systematic integrative review we extracted information from the 98 studies about: 1) geographical location; 2) type of freshwater ecosystem; 3) study type (i.e., observational or experimental); 4) most cited term in the title and/or abstract used to indicate EH (e.g., “habitat heterogeneity”, “environmental complexity”); 5) environmental variables used as proxy of EH (e.g., substrate, flow velocity); 6) methodological approaches to manipulated EH proxies (e.g., qualitative variation, roughness); 7) calculation methods to measure EH (e.g., indices, coefficient of variation); 8) biological group evaluated; 9) zone of communities’ occurrence (i.e., nekton, plankton, benthos or in the riparian meta-ecosystem); 10) type of biological metric considered as a response variable (i.e., functional or taxonomic, and alpha or beta), 11) whether experimental design controlled for the effect of surface area, and 12) the mechanisms invoked by authors to discuss patterns or used to test the relationship between EH and the biological metric response. We only considered the mechanisms that were explicitly cited by authors. Occasionally, studies encompassed more than one taxonomic group, at different occurrence zones, with different EH proxies, and used different methodological approaches and calculation methods. We incorporated this information in separate outcomes from the same study, which generated 276 comparisons or independent registers.
EH proxy denotes the element that authors actually quantified in the ecosystem aiming to evaluate the effect of EH over biological metrics of diversity. During data extraction, we identified several EH proxies used by authors, which were then sorted into 12 categories: aquatic vegetation; artificial substrates; natural substrates; channel morphology; elevation; food resources; land cover; riparian vegetation; water flow; water chemistry; wood debris; and mixed. The mixed category included studies that used more than one category of proxy to manipulate EH in the experiments
The methodological approach denotes the way authors manipulated the EH proxies to create EH gradients or EH levels, aiming to test the effects of spatial EH on a biological metric. In the same way, we created eight categories of methodological approach: density; mixed; presence vs. absence; qualitative variation; quantitative variation; roughness; structure; and size variation.
Quantitative data extraction
For a meta-analytical study, we used 74 studies within our dataset that reported statistical results (e.g., correlation coefficient, mean and deviation values between treatments), variance and sample size which estimated the relationship between EH and any biological metric. When authors reported results in figures, we extracted information using the software ImageJ®. When effects were not provided in the text, table, or figures, we requested data and/or statistical results from the corresponding author of the study.
Every statistical result reported in those studies was converted into a correlation coefficient (r). To estimate the magnitude of the overall effect of EH on taxonomic and functional alpha and beta diversity, we used Fisher’s z-value as the metric of effect size. We derive z values using the formula: z = 0.5 x ln [(1 + r) / (1 - r)]. For each z value, we calculate its respective variance (vz) based on the sample size (n) of the test performed in the study: vz = 1 / (n - 3). Based on z variance and sample size, we calculated the 95% confidence intervals around each z value. Negative values of z represent a negative effect of the EH over the biological metric whereas positive z values represent a positive effect, and z = 0 represents no effect. If confidence intervals did not overlap zero, the estimate of the true effect size is considered significant.
From the 74 studies, we recorded 242 individual effect sizes. We separated effect sizes into four distinct categories according to the type of biological diversity metric considered as the response variable: taxonomic alpha, taxonomic beta, functional alpha and functional beta. Among these, 200 referred to the effect size on taxonomic alpha, 20 on taxonomic beta, 16 on functional alpha and six on functional beta diversity.