A new statistical approach for assessing similarity of species composition with incidence and abundance data

The classic Jaccard and Sørensen indices of compositional similarity (and other indices that depend upon the same variables) are notoriously sensitive to sample size, especially for assemblages with numerous rare species. Further, because these indices are based solely on presence–absence data, accurate estimators for them are unattainable. We provide a probabilistic derivation for the classic, incidence‐based forms of these indices and extend this approach to formulate new Jaccard‐type or Sørensen‐type indices based on species abundance data. We then propose estimators for these indices that include the effect of unseen shared species, based on either (replicated) incidence‐ or abundance‐based sample data. In sampling simulations, these new estimators prove to be considerably less biased than classic indices when a substantial proportion of species are missing from samples. Based on species‐rich empirical datasets, we show how incorporating the effect of unseen shared species not only increases accuracy but also can change the interpretation of results.     

A statistical approach to estimate soil ciliate diversity and distribution based on data from five continents

A total of 359 soil samples collected from five continents (Africa, Asia, Australia, Europe and South America) were investigated for the presence/absence of soil ciliate species. Merging records by species identity, we have compiled a master data list (species by sample matrix). In the list, a total of 964 soil ciliate species (644 described and 320 undescribed) are recorded. The species distributions within the 359 samples and across the five continents are examined. The frequency distribution of the species over samples is used for global diversity estimation. A statistical ACE (abundance‐based coverage estimation) model which links observed data to unseen species is applied to assess regional and global soil ciliate species richness. The model, whose reliability was tested by its power to predict the number of new species in additional samples from Africa, may resolve the controversial issue on global species diversity of soil ciliates. Although an accurate point estimate is not feasible due to severe undersampling, the statistical model enables us to obtain a minimum regional diversity and global species diversity. A consistent finding over all five continents is that at least half of the species diversity is still undiscovered. Our model also yields a global soil ciliate diversity of at least 1900 species, that is, doubles the number of currently known species, and thus diversity is relatively high. This is consistent with the finding of Foissner, who used a probability‐based method. Soil ciliate distributions between continent pairs are analyzed by adjusted abundance‐based similarity/overlap indices. These new indices account for the effect of unseen species and also reduce the bias generated by undersampling. The adjusted abundance‐based Jaccard (or Sørensen) index shows that there is about 30% (18% for Sørensen) dis‐similarity between any two continents, supporting the moderate endemicity model. The results are discussed with respect to protist species distribution, that is, whether they are cosmopolitan or of restricted distribution.     

Diatoms as indicators: The influences of experimental nitrogen enrichment on diatom assemblages in sub-Arctic streams

Indices using diatoms are widely used to assess water quality, but are usually constructed from field correlations and not tested through rigorous experimentation. We tested experimentally the performance of the Sørensen and the Shannon indices, and the trophic diatom index (TDI). Nitrogen was naturally limiting in the eight remote sub-Arctic streams used and we measured the effects of experimental nitrogen enrichment on diatom assemblages. Diatom densities increased significantly in the enriched reaches but there was no significant difference in invertebrate density between control and treatment reaches. Grazing effects were thus controlled for. Diversity within streams (Shannon index) was significantly reduced by nutrient addition but the Sørensen index did not change. The trophic diatom index (TDI), which is presumed to reflect nutrient concentration, was not influenced by nutrient addition and generally the values were low in both control and treatment reaches. Densities of the diatom genera Achnanthes and Gomphonema increased significantly with enrichment while those of Nitzschia and Fragilaria decreased significantly. Less abundant diatom species, which collectively constituted around 40% in relative abundance in the control reaches, were around 15–18% in treatment reaches. Growth forms were altered by the nutrients. Diatoms attached by mucilage pads were more abundant in treated reaches compared with control reaches. Motile diatoms became scarcer. The size of diatom species was unaffected by nutrient enrichment. This study showed that it is important to test experimentally indices that are developed for particular habitats before using them elsewhere.     

Subsampling Herbarium Collections to Assess Geographic Diversity Gradients: A Case Study with Endemic Orchidaceae and Rubiaceae in Cameroon

We compiled herbarium specimen data to provide an improved characterization of geographic patterns of diversity using indices of species diversity and floristic similarity based on rarefaction principles. A dataset of 3650 georeferenced plant specimens belonging to Orchidaceae and Rubiaceae endemic to Atlantic Central Africa was assembled to assess species composition per half‐degree or one‐degree grid cells. Local diversity was measured by the expected number of species (S_k) per grid cell found in subsamples of increasing size and compared with raw species richness (S_R). A nearly unbiased estimator of the effective number of species per grid cell was also used, allowing quantification of ratios of ‘true diversity’ between grid cells. Species turnover was measured using a presence/absence‐based similarity index (Sørensen) and an abundance‐based index that corrects for sampling bias (NNESS). Our results confirm that the coastal region of Cameroon is more diverse in endemic species than those more inland. The southern part of this coastal forest is, however, as diverse as the more intensively inventoried northern part, and should also be recognized as an important center of endemism. A strong congruence between Sørensen and NNESS similarity matrices lead to similar delimitations of floristic units. Hence, heterogeneous sampling seems to confer more bias when measuring patterns of local diversity using raw species richness than species turnover using Sørensen index. Overall, we argue that subsampling methods represent a useful way to assess diversity gradients using herbarium specimens while correcting for heterogeneous sampling effort. Abstract in French is available in the online version of this article.     

Modified TWINSPAN classification in which the hierarchy respects cluster heterogeneity

Aim: To propose a modification of the TWINSPAN algorithm that enables production of divisive classifications that better respect the structure of the data. Methods: The proposed modification combines the classical TWINSPAN algorithm with analysis of heterogeneity of the clusters prior to each division. Four different heterogeneity measures are involved: Whittaker's beta, total inertia, average Sørensen dissimilarity and average Jaccard dissimilarity. Their performance was evaluated using empirical vegetation datasets with different numbers of plots and different levels of heterogeneity. Results: While the classical TWINSPAN algorithm divides each cluster coming from the previous division step, the modified algorithm divides only the most heterogeneous cluster in each step. The four tested heterogeneity measures may produce identical or very similar results. However, average Jaccard and Sørensen dissimilarities may reach extreme values in clusters of small size and may produce classifications with a highly unbalanced cluster size. Conclusions: The proposed modification does not alter the logic of the TWINSPAN classification, but it may change the hierarchy of divisions in the final classification. Thus, unsubstantiated divisions of homogeneous clusters are prevented, and classifications with any number of terminal clusters can be created, which increases the flexibility of TWINSPAN.     

On the fauna of Scathophagidae (Diptera) of Yakutia

The fauna of the family Scathophagidae of Yakutia, including 44 species of 15 genera, was studied. 23 species are recorded from the territory for the first time. The Arctic and subarctic zones of Yakutia have been still poorly studied, and additional species are found there. The scathophagid fauna of Yakutia was compared with the faunas of the neighboring regions using the dendrogram similarity method based on calculation of the Czekanowski–Sørensen coefficient.     

Cost‐efficiency of Subsampling Protocols to Evaluate Oribatid‐Mite Communities in an Amazonian Savanna

Sampling oribatid mites in large areas using conventional methods is expensive, time‐consuming, and this constrains their use in environmental monitoring programs. We used samples collected in 38 plots of 3.75 ha spread over 30,000 ha in an Amazonian savanna to evaluate the reduction in costs and person‐hours in sampling and sorting and to elaborate cost‐effective protocols. Ten samples per plot were collected and extracted using a Berlese‐Tullgren apparatus. In the laboratory, samples were reduced to 50, 25, 12.5, and 6.25 percent of the initial content. Field‐effort reduction was estimated by reducing the number of subsamples per plot. Dissimilarity matrices were generated using Bray–Curtis, Sørensen, and Chao–Sørensen indices. Correlations between each reduced‐effort dissimilarity matrix and 100 or 50 percent sorting were used as an index of how much information was retained in reduced‐effort sampling, and could still be used in multivariate analyses. The effects of most predictor variables on mite composition were detected in data based on every level of sample reduction. The intensive sampling was insufficient to reveal the full oribatid‐mite fauna in the savanna; as more plots were sampled, more species were recorded. Our data indicate subsampling protocols for biodiversity assessment of oribatid mites in savanna that increase field and laboratory efficiency, and optimize both taxonomic and ecological aspects of the investigation.     

Response of saprotrophic microfungi degrading the fulvic fraction of soil organic matter to different N fertilization intensities,different plant species cover and elevated atmospheric CO<Subscript>2</Subscript> concentration

The response of the cenosis composition of soil saprotrophic microfungi able to utilize the fulvic fraction of soil organic matter to increased concentration of atmospheric carbon dioxide, plant species cover quality and different levels of nitrogen fertilization was determined under field conditions in a free-air carbon dioxide enrichment experiment. Twenty-nine species of microfungi were isolated from the tested soil. The effects of CO₂ enrichment and plant species cover were not significant. Nitrogen fertilization was identified as the only significant factor inducing changes in the abundance of soil microorganisms. This was reflected in a relatively low value of quantitative Sørensen similarity index on comparing fertilized and unfertilized treatments and in 2-way ANOVA of total CFU counts. Some differences were observed in species diversity between the two variants of all treatments. No association between microfungi and the factors under study was found by using the Monte Carlo Permutation test in redundancy analysis.     

Pairwise null model analyses of temporal patterns of bird assemblages contradict the assumptions of competition theory

Compensatory dynamics assume inverse patterns of population dynamics of species with similar ecological resource requirements (temporal segregation). The objective of this study was to test this hypothesis on temporal samples (10–57 years) of 19 breeding bird assemblages of various habitats. We used presence/absence null model (SIM2) in combination with the C-score and Sørensen indices. The C-score index estimates the average number of checkerboards for two species, while the Sørensen index measures the qualitative similarity of co-occurrence between two species in a time series. We used pairwise null model analysis to select significant species pairs based on three selection criteria: the standard confidence interval criterion, conservative empirical Bayes mean based criterion and confidence limit based criterion. Altogether, 21 402 species pairs were analysed. The SIM2 algorithm detected from 157 to 7 segregated pairs depending on the selection criterion. The number of significant negative pairs with possible biological significance (foraging guild membership, predator–prey interactions) was far lower and represented approximately 0.0–0.3% (4–65) of pairs in a matrix. Indeed, the number of detected negative associations depended on the selection criterion. Moreover, the number of segregated pairs was negatively related to the area of the census plots and fill of the species matrix. Our results underline the minor importance of interspecific competitive interactions in temporal patterns of bird assemblages. Instead, we suggest that stochastic factors, climate or heterospecific social information may lead to more or less synchronous dynamics of bird pairs.     

Estimating similarity of communities: a parametric approach to spatio‐temporal analysis of species diversity

Several stochastic models with environmental noise generate spatio‐temporal Gaussian fields of log densities for the species in a community. Combinations of such models for many species often lead to lognormal species abundance distributions. In spatio‐temporal analysis it is often realistic to assume that the same species are expected to occur at different times and/or locations because extinctions are rare events. Spatial and temporal β‐diversity can then be analyzed by studying pairs of communities at different times or locations defined by a bivariate lognormal species abundance model in which a single correlation occurs. This correlation, which is a measure of similarity between two communities, can be estimated from samples even if the sampling intensities vary and are unknown, using the bivariate Poisson lognormal distribution. The estimators are approximately unbiased, although each specific correlation may be rather uncertain when the sampling effort is low with only a small fraction of the species represented in the samples. An important characteristic of this community correlation is that it relates to the classical Jaccard‐ or the Sørensen‐indices of similarity based on the number of species present or absent in two communities. However, these indices calculated from samples of species in a community do not necessarily reflect similarity of the communities because the observed number of species depends strongly on the sampling intensities. Thus, we propose that our community correlation should be considered as an alternative to these indices when comparing similarity of communities. We illustrate the application of the correlation method by computing the similarity between temperate bird communities.     

Choosing the best similarity index when performing fuzzy set ordination on binary data

Abstract. Fuzzy set ordination (FSO) may be used with either abundance data or binary (presence/absence) data. FSO requires a similarity index that returns values between 0 and 1. Many indices will do so, but their suitability for FSO has not been tested. Nine binary indices were evaluated in this study. Simulated plant community data sets were generated with COMPAS; they contained five levels of β‐diversity, two levels of qualitative noise, and two sampling arrangements (regular or random) along one gradient. Indices were evaluated with rank and linear correlations between the apparent ecological gradient positions generated by FSO and actual gradient positions; the abilities of the best‐performing indices to minimize the curlover effect were also compared. All indices performed best at intermediate levels of β‐diversity and with regular sampling. Five indices had consistently higher rank and linear correlations (Baroni‐Urbani & Buser, Jaccard, Kulczynski, Ochiai and Sørensen), whereas four were consistently lower (Faith, Russell & Rao, Rogers & Tanimoto and Simple Matching). There were no significant differences in curlover among the five best indices. A step‐across algorithm, a flexible shortest path adjustment, improved correlations and reduced curlover for the five best indices at higher β‐diversity levels. We recommend that one of the five best‐performing similarity indices be used with FSO on binary data; a flexible shortest path adjustment should also be employed at higher β‐diversities when possible.     

Broad Scale Distribution of Ferns and Lycophytes along Environmental Gradients in Central and Northern Amazonia,Brazil

Establishing which factors determine species distributions is of major relevance for practical applications such as conservation planning. The Amazonian lowlands exhibit considerable internal heterogeneity that is not apparent in existing vegetation maps. We used ferns as a model group to study patterns in plant species distributions and community composition at regional and landscape scales. Fern species composition and environmental data were collected in 109 plots of 250 × 2 m distributed among four sites in Brazilian Amazonia. Interplot distances varied from 1 to ca 670 km. When floristically heterogeneous datasets were analyzed, the use of an extended Sørensen dissimilarity index rather than the traditional Sørensen index improved model fit and made interpretation of the results easier. Major factors associated with species composition varied among sites, difference in cation concentration was a strong predictor of floristic dissimilarity in those sites with pronounced heterogeneity in cation concentration. Difference in clay content was the most relevant variable in sites with uniform cation concentrations. In every case, environmental differences were invariably better than geographic distances in predicting species compositional differences. Our results are consistent with the ideas that: (1) the relative predictive capacity of the explanatory variables depend on the relative lengths of the observed gradients; and (2) environmental gradients can be hierarchically structured such that gradients occur inside gradients. Therefore, site‐specific relationships among variables can mask the bigger picture and make it more difficult to unravel the factors structuring plant communities in Amazonia.     

Does phylogenetic distance aid in detecting environmental gradients related to species composition?

Questions: How should we evaluate the success of new distance measures combining community abundance and phylogenetic information? How do we interpret ordinations using these metrics? Methods: We generated synthetic data along a known environmental gradient with two hypothetical underlying phylogenetic structures: niche phylogenetically conserved or dispersed along a gradient. We also examined tree species composition associated with gradients in elevation and longitude in Oregon, USA. NMS ordinations of plots in species space from phylogenetic (PD) and Sørensen distance (SD) matrices allowed comparison of the use of PD in different scenarios. Results: PD caused plots to cluster based on the clades that they contained, reducing stress with the synthetic data but not with the real example. Phylogenetic distance highlighted clades related to gradients when these were associated. When phylogeny was not conserved along a gradient, that gradient was less strong. Regardless of phylogenetic conservation, NMS using SD consistently extracted the strongest gradients in species composition. Conclusions: The success of PD should be evaluated on how well it extracts gradients in species composition and allows community ecologists to determine which gradients are partially explained by phylogeny and not based on its ability to reduce ordination stress. PD ordinations can help community ecologists interpret niche conservation but may obscure gradients related to species composition when niches are not conserved along the gradient of interest at the scale of the study.     

Biodiversity assessment in incomplete inventories: leaf litter ant communities in several types of Bornean rain forest

Biodiversity assessment of tropical taxa is hampered by their tremendous richness, which leads to large numbers of singletons and incomplete inventories in survey studies. Species estimators can be used for assessment of alpha diversity, but calculation of beta diversity is hampered by pseudo-turnover of species in undersampled plots. To assess the impact of unseen species, we investigated different methods, including an unbiased estimator of Shannon beta diversity that was compared to biased calculations. We studied alpha and beta diversity of a diverse ground ant assemblage from the Southeast Asian island of Borneo in different types of tropical forest: diperocarp forest, alluvial forest, limestone forest and heath forests. Forests varied in plant composition, geology, flooding regimes and other environmental parameters. We tested whether forest types differed in species composition and if species turnover was a function of the distance between plots at different spatial scales. As pseudo-turnover may bias beta diversity we hypothesized a large effect of unseen species reducing beta diversity. We sampled 206 ant species (25% singletons) from ten subfamilies and 55 genera. Diversity partitioning among the four forest types revealed that whereas alpha species richness and alpha Shannon diversity were significantly smaller than expected, beta-diversity for both measurements was significantly higher than expected by chance. This result was confirmed when we used the unbiased estimation of Shannon diversity: while alpha diversity was much higher, beta diversity differed only slightly from biased calculations. Beta diversity as measured with the Chao-Sørensen or Morisita-Horn Index correlated with distance between transects and between sample points, indicating a distance decay of similarity between communities. We conclude that habitat heterogeneity has a high influence on ant diversity and species turnover in tropical sites and that unseen species may have only little impact on calculation of Shannon beta diversity when sampling effort has been high.     

Spatial structure of epiphytic yeast communities on fruits of <Emphasis Type="Italic">Sorbus aucuparia</Emphasis> L.

The subject of this research is epiphytic yeast communities formed on the surface of Sorbus aucuparia. The object is to make quantitative assessment of the yeast communities’ differentiation of the same but distant substratum. Results of the nested ANOVA demonstrated that with increase in distances, there are increases in the variation of total number and relative abundance of the dominant yeast communities. The average similarity between groups of single fruits (Sørensen’s f Similarity Coefficient) regularly decreased with distance. The results demonstrate that the number and structure of separate yeast groups depend not only on ecological factors but also on proximity to other communities. Such aggregation in the distribution of the microorganisms’ species caused by migration and colonial resettlement should be taken into account when analyzing their diversity in natural habitats.     

Effects of habitat age and disturbance intensity on the biodiversity of three trophic levels in Central Kenya

In recent decades, the extent of primary forest in tropical regions has decreased drastically, with concurrent increases in the extent of tropical secondary forest. This has important implications for conservation management. We present novel data on species diversity and composition for three taxa (bats, geometrid moths and plants) in forests at two stages of secondary growth located in the Aberdare Mountains in Central Kenya. We found no significant differences in alpha diversity for any of the sampled groups between forest types. However, we found disturbance‐driven differences of tree and herb community compositions and correlations between tree and moth – and tree and shrub community compositions. Changes in community compositions were more pronounced using an abundance‐based indicator (Bray–Curtis) in comparison with an incidence based (Sørensen). Our results demonstrate that solely working with alpha diversity values can be misleading in conservation planning as they might not reflect compositional changes between habitats. Furthermore, abundance‐based compositional measures appear to be superior to incidence‐based measures for detecting subtle effects of disturbance on biodiversity.     

Coupling impoverishment analysis and partitioning of beta diversity allows a comprehensive description of Odonata biogeography in the Western Mediterranean

Islands host a subset of organisms occurring at their sources, and these assemblages are usually dominated by the most generalistic and dispersive species. In this study, we aim to identify which species are missing on islands and which ecological traits are responsible for differential occurrence. Then, we apply this information to beta diversity analyses. As a study group and area, we selected the Odonata in the Western Mediterranean. Based on the presence/absence of 109 species, we applied a series of analyses at both community and individual species level. The islands of the Balearics, Corsica, Sardinia and Malta are highly impoverished, but Sicily is not. Non-parametric multivariate adaptive regression splines predicted the occurrence of individual species on each island. Principal component analysis recognised differences between damselflies (Zygoptera) and dragonflies (Anisoptera), but members of the two suborders have similar occurrences on islands, and island occurrence is determined mostly by species’ frequencies at source and by their degree of generalism. Island species predicted correctly to occur on islands showed opposite characteristics to species unpredicted to occur and being present. The similarity pattern highlighted by turnover (Simpson index) is clearer than that obtained by non-partitioned beta diversity (Sørensen index). In fact, indicator value analyses revealed more indicator species for the Simpson compared to Sørensen index, and indicator species from islands where unpredicted to occur by impoverishment analysis. This suggests that island species predicted absent determine most of an island’s turnover pattern, thus encompassing fundamental biogeographic information. Due to their absence on nearest sources, they are also at higher risk of extinction, and deserving of special conservation effort.     

Multisite and multispecies measures of overlap,co‐occurrence,and co‐diversity

Several indices measure the association or segregation between two species and the similarity or differentiation between two sets of species. These indices are based on the overlap in the distribution of species (measured with the number of co‐occurrences) or on the overlap in species composition of sites (measured with the number of species that are shared between two sites). This paper shows that when evaluating more than two species the number of overlaps and the number of pairwise co‐occurrences are not equal, as it is the case for two species. Equivalently, when comparing more than two species assemblages, the number of overlaps differ from the number of instances of species sharing by pairs of sites (the ‘co‐diversities’). Considering this distinction, two different types of multispecies and multisite indices can be derived: indices of general overlap and indices of co‐occurrence or co‐diversity. Here I present a complete series of the two types of indices that correspond to the popular Jaccard, Sørensen, and Simpson two‐species or two‐site indices. Indices of general overlap are defined by three parameters (the total number of species, the total number of sites, and the total number of occurrences), whereas indices of co‐occurrence or co‐diversity depend on those parameters plus an additional one that is defined by the values of species richness or range size. Consequently, the two types of indices respond differently to null models, depending on the parameters that are fixed or randomized. All indices correlate well with the mean of the traditional indices calculated pair by pair, and the correspondence is extremely close for the new indices of co‐occurrence and co‐diversity. These properties should be useful in clarifying some of the confusion that exists in the current discussion about the advantages and disadvantages of pairwise vs community‐wide approaches in the analysis of diversity.     

Effects of uncut hay meadow strips on spiders

Standard management of Czech hay meadows consists of machine mowing twice a year, with the first mowing in or before mid-June and the second mowing 40–60 days later. Here, we aimed to analyse the effects of the first year of implementation of the agri-environment scheme (AES), which consisted of allowing 3–10% of permanent grasslands within each land block to remain unmown until at least August 15, on the abundance and diversity of spiders. We conducted the study at 40 paired sampling sites in three meadow types in northern Czechia. Spiders were sampled using pitfall traps and sweeping, and the plant cover and vascular plant species composition were analysed. Nearly all study sites were associated with 100% vegetation cover, and hosted 209 vascular plant species, of which 12 were threatened. The number of vascular plant species exhibited only modest differences between the study sites under standard and AES-prescribed management. We captured a total of 3889 individuals of 103 spider species. The abundance of spiders was three-times higher at sites subjected to AES management. The number of species was similar irrespective of the management applied, but the species composition differed in response to the management (Sørensen index 0.562–0.736). The AES management was associated with higher abundance of common vegetation-dwelling farmland spiders, but epigeic spiders decreased there or were insensitive to AES management. It remains to be investigated whether threatened vegetation-dwelling spiders may benefit from similar AESs at sites of their occurrence, such as in fen meadows or steppes.     

Upward shift of alpine plants increases floristic similarity of mountain summits

Question: Does the upward shift of species and accompanied increase in species richness, induced by climate change, lead to homogenization of Alpine summit vegetation? Location: Bernina region of the Swiss Alps. Methods: Based on a data set from previous literature we expand the analysis from species richness to beta‐diversity and spatial heterogeneity. Species compositions of mountain summits are compared using a two‐component heterogeneity concept including the mean and the variance of Sørensen similarities calculated between the summits. Non‐metric multidimensional scaling is applied to explore developments of single summits in detail. Results: Both heterogeneity components (mean dissimilarity and variance) decrease over time, indicating a trend towards more homogeneous vegetation among Alpine summits. However, the development on single summits is not strictly unidirectional. Conclusions: The upward shift of plant species leads to homogenization of alpine summit regions. Thus, increasing alpha‐diversity is accompanied by decreasing beta‐diversity. Beta‐diversity demands higher recognition by scientists as well as nature conservationists as it detects changes which cannot be described using species richness alone.