Evaluating Functional Diversity: Missing Trait Data and the Importance of Species Abundance Structure and Data Transformation

Abstract

Functional diversity (FD) is an important component of biodiversity that quantifies the difference in functional traits between organisms. However, FD studies are often limited by the availability of trait data and FD indices are sensitive to data gaps. The distribution of species abundance and trait data, and its transformation, may further affect the accuracy of indices when data is incomplete. Using an existing approach, we simulated the effects of missing trait data by gradually removing data from a plant, an ant and a bird community dataset (12, 59, and 8 plots containing 62, 297 and 238 species respectively). We ranked plots by FD values calculated from full datasets and then from our increasingly incomplete datasets and compared the ranking between the original and virtually reduced datasets to assess the accuracy of FD indices when used on datasets with increasingly missing data. Finally, we tested the accuracy of FD indices with and without data transformation, and the effect of missing trait data per plot or per the whole pool of species. FD indices became less accurate as the amount of missing data increased, with the loss of accuracy depending on the index. But, where transformation improved the normality of the trait data, FD values from incomplete datasets were more accurate than before transformation. The distribution of data and its transformation are therefore as important as data completeness and can even mitigate the effect of missing data. Since the effect of missing trait values pool-wise or plot-wise depends on the data distribution, the method should be decided case by case. Data distribution and data transformation should be given more careful consideration when designing, analysing and interpreting FD studies, especially where trait data are missing. To this end, we provide the R package "traitor" to facilitate assessments of missing trait data.

Fig 1. Flow diagram of the consecutive methodological steps.

Upper left corner—in each plot species are ordered by their relative abundance and FD index is calculated for each plot of a community. Upper right corner—0.5% of the species relative abundance is removed in consecutive steps, starting with the least abundant species and FD index is then calculated again for each plot at each reduction step. Upper middle columns—plots are ranked based on the values of the FD index and the ranks of original data and data at each reduction step are correlated. Figure in the middle—regression slopes from fitting the linear model represent the robustness of FD index to missing trait data; in this example FD index is (A) less robust and (B) more robust to missing trait data (example RaoQ on head length of ants).

Fig 2. Plot-wise and pool-wise trait data thresholds.

Schematic figure depicting plot-wise and pool-wise scenarios for setting the thresholds for trait data sampling. (A) species from all plots make up the pool of species; (B) species can be ordered by their abundance in each plot or in the whole pool; (C) the least abundant species in the whole pool of species are removed until reaching the desired threshold for trait sampling; (D) the least abundant species in each plot are removed until reaching the desired threshold for trait sampling.

Fig 3. Effect of sampling scenario on FD index sensitivity.

Barplots showing the results of linear mixed effects model, specifically the effect of the two sampling scenarios on the sensitivity of indices for three different types of organisms. The more negative the regression slope, the more sensitive the particular index is to missing trait information. The error bars denote the 95% confidence intervals. (A) plant community (n = 12 plots), (B) ant community (n = 58 plots), and (C) bird community (n = 8 plots).

Fig 4. Effect of abundance transformation on FD index sensitivity.

Barplots showing the results of linear mixed effects models, specifically the effect of the abundance transformation on the slopes for the three different types of organisms. The more negative the regression slope, the more sensitive the particular index is to missing trait information. The error bars denote the 95% confidence intervals. (A) plant community (n = 12 plots), (B) ant community (n = 58 plots), and (C) bird community (n = 8 plots). The right panels depict dominance-diversity curves for the respective organism dataset before and after log-transformation.

Fig 5. Effect of sampling method and abundance transformation on FD index sensitivity.

Barplot depicting the results of linear mixed effects models, specifically the interaction between abundance transformation and the different abundance measures used in plant ecology (all three abundance measures were used for the same plant dataset in order to make their effects comparable). The effect of down-weighting the dominant species by log-transformation of their abundance was most pronounced in the biomass abundance measure. When log transformed, all three sampling methods have a very similar effect on the sensitivity of indices to missing trait data. Error bars denote the 95% confidence intervals.

Fig 6. Effect of trait transformation on FD index sensitivity.

The effect of trait transformation on the improvement in slope (transformed—untransformed trait data)—the bigger the improvement in slope, the more robust the index becomes to missing trait data (y axis). The right panels illustrate the different improvements in trait skewness, depicting examples of trait distribution before and after transformation, which correspond to the x axis of the main figure (matching colours).