Outlier detection methods are still effective even using virtual species created with the probabilistic approach

Liu et al. (Journal of Biogeography, 2018, 45 :164–176) presented an approach to detect outliers in species distribution data by developing virtual species created using the threshold approach. Meynard et al. (Journal of biogeography, 2019, 46 :2141–2144) raised concerns about this approach stating that ‘using a probabilistic approach … may significantly change results’. Here we provide a new series of simulations using the two approaches and demonstrate that the outlier detection approach based on pseudo species distribution models was still effective when using the probabilistic approach, although the detection rate was lower than when using the threshold approach.     

Testing methods in species distribution modelling using virtual species: what have we learnt and what are we missing?

Species distribution models (SDMs) have become one of the major predictive tools in ecology. However, multiple methodological choices are required during the modelling process, some of which may have a large impact on forecasting results. In this context, virtual species, i.e. the use of simulations involving a fictitious species for which we have perfect knowledge of its occurrence–environment relationships and other relevant characteristics, have become increasingly popular to test SDMs. This approach provides for a simple virtual ecologist framework under which to test model properties, as well as the effects of the different methodological choices, and allows teasing out the effects of targeted factors with great certainty. This simplification is therefore very useful in setting up modelling standards and best practice principles. As a result, numerous virtual species studies have been published over the last decade. The topics covered include differences in performance between statistical models, effects of sample size, choice of threshold values, methods to generate pseudo‐absences for presence‐only data, among many others. These simulations have therefore already made a great contribution to setting best modelling practices in SDMs. Recent software developments have greatly facilitated the simulation of virtual species, with at least three different packages published to that effect. However, the simulation procedure has not been homogeneous, which introduces some subtleties in the interpretation of results, as well as differences across simulation packages. Here we 1) review the main contributions of the virtual species approach in the SDM literature; 2) compare the major virtual species simulation approaches and software packages; and 3) propose a set of recommendations for best simulation practices in future virtual species studies in the context of SDMs.     

The effect of positional error on fine scale species distribution models increases for specialist species

Species occurrences inherently include positional error. Such error can be problematic for species distribution models (SDMs), especially those based on fine-resolution environmental data. It has been suggested that there could be a link between the influence of positional error and the width of the species ecological niche. Although positional errors in species occurrence data may imply serious limitations, especially for modelling species with narrow ecological niche, it has never been thoroughly explored. We used a virtual species approach to assess the effects of the positional error on fine-scale SDMs for species with environmental niches of different widths. We simulated three virtual species with varying niche breadth, from specialist to generalist. The true distribution of these virtual species was then altered by introducing different levels of positional error (from 5 to 500 m). We built generalized linear models and MaxEnt models using the distribution of the three virtual species (unaltered and altered) and a combination of environmental data at 5 m resolution. The models’ performance and niche overlap were compared to assess the effect of positional error with varying niche breadth in the geographical and environmental space. The positional error negatively impacted performance and niche overlap metrics. The amplitude of the influence of positional error depended on the species niche, with models for specialist species being more affected than those for generalist species. The positional error had the same effect on both modelling techniques. Finally, increasing sample size did not mitigate the negative influence of positional error. We showed that fine-scale SDMs are considerably affected by positional error, even when such error is low. Therefore, where new surveys are undertaken, we recommend paying attention to data collection techniques to minimize the positional error in occurrence data and thus to avoid its negative effect on SDMs, especially when studying specialist species.     

Keystone species and food webs

Different species are of different importance in maintaining ecosystem functions in natural communities. Quantitative approaches are needed to identify unusually important or influential, ‘keystone’ species particularly for conservation purposes. Since the importance of some species may largely be the consequence of their rich interaction structure, one possible quantitative approach to identify the most influential species is to study their position in the network of interspecific interactions. In this paper, I discuss the role of network analysis (and centrality indices in particular) in this process and present a new and simple approach to characterizing the interaction structures of each species in a complex network. Understanding the linkage between structure and dynamics is a condition to test the results of topological studies, I briefly overview our current knowledge on this issue. The study of key nodes in networks has become an increasingly general interest in several disciplines: I will discuss some parallels. Finally, I will argue that conservation biology needs to devote more attention to identify and conserve keystone species and relatively less attention to rarity.     

Bayesian species identification under the multispecies coalescent provides significant improvements to DNA barcoding analyses

DNA barcoding methods use a single locus (usually the mitochondrial COI gene) to assign unidentified specimens to known species in a library based on a genetic distance threshold that distinguishes between‐species divergence from within‐species diversity. Recently developed species delimitation methods based on the multispecies coalescent (MSC) model offer an alternative approach to individual assignment using either single‐locus or multiloci sequence data. Here, we use simulations to demonstrate three features of an MSC method implemented in the program bpp . First, we show that with one locus, MSC can accurately assign individuals to species without the need for arbitrarily determined distance thresholds (as required for barcoding methods). We provide an example in which no single threshold or barcoding gap exists that can be used to assign all specimens without incurring high error rates. Second, we show that bpp can identify cryptic species that may be misidentified as a single species within the library, potentially improving the accuracy of barcoding libraries. Third, we show that taxon rarity does not present any particular problems for species assignments using bpp and that accurate assignments can be achieved even when only one or a few loci are available. Thus, concerns that have been raised that MSC methods may have problems analysing rare taxa (singletons) are unfounded. Currently, barcoding methods enjoy a huge computational advantage over MSC methods and may be the only approach feasible for massively large data sets, but MSC methods may offer a more stringent test for species that are tentatively assigned by barcoding.     

Deriving site-specific species pools from large databases

Defining the species pool of a community is crucial for many types of ecological analyses, providing a foundation to metacommunity, null modelling or dark diversity frameworks. It is a challenge to derive the species pool empirically from large and heterogeneous databases. Here, we propose a method to define a site-specific species pool (SSSP), i.e. the probabilistic set of species that may co-occur with the species of a target community. Using large databases with geo-referenced records that comprise full plant community surveys, our approach characterizes each site by its own species pool without requiring a pre-defined habitat classification. We calculate the probabilities of each species in the database to occur in the target community using Beals’ index of sociological favourability, and then build sample-based rarefaction curves from neighbouring records with similar species composition to estimate the asymptotic species pool size. A corresponding number of species is then selected among the species having the highest occurrence probability, thus defining both size and composition of the species pool. We tested the robustness of this approach by comparing SSSPs obtained with different spatial extents and dissimilarity thresholds, fitting different models to the rarefaction curves, and comparing the results obtained when using Beals co-occurrence probabilities or presence/absence data. As an example application, we calculated the SSSPs for all calcareous grassland records in the German Vegetation Reference Database, and show how our method could be used to 1) produce grain-dependent estimations of species richness across plots, 2) derive scalable maps of species richness and 3) define the full list of species composing the SSSP of each target site. By deriving the species pool exclusively from community characteristics, the SSSP framework presented here provides a robust approach to bridge biodiversity estimations across spatial scales.     

Detecting outliers in species distribution data

Aim

Species distribution data play a pivotal role in the study of ecology, evolution, biogeography and biodiversity conservation. Although large amounts of location data are available and accessible from public databases, data quality remains problematic. Of the potential sources of error, positional errors are critical for spatial applications, particularly where these errors place observations beyond the environmental or geographical range of species. These outliers need to be identified, checked and removed to improve data quality and minimize the impact on subsequent analyses. Manually checking all species records within large multispecies datasets is prohibitively costly. This work investigates algorithms that may assist in the efficient vetting of outliers in such large datasets.

Location

We used real, spatially explicit environmental data derived from the western part of Victoria, Australia, and simulated species distributions within this same region.

Methods

By adapting species distribution modelling (SDM), we developed a pseudo‐SDM approach for detecting outliers in species distribution data, which was implemented with random forest (RF) and support vector machine (SVM) resulting in two new methods: RF_pdSDM and SVM_pdSDM. Using virtual species, we compared eight existing multivariate outlier detection methods with these two new methods under various conditions.

Results

The two new methods based on the pseudo‐SDM approach had higher true skill statistic (TSS) values than other approaches, with TSS values always exceeding 0. More than 70% of the true outliers in datasets for species with a low and intermediate prevalence can be identified by checking 10% of the data points with the highest outlier scores.

Main conclusions

Pseudo‐SDM‐based methods were more effective than other outlier detection methods. However, this outlier detection procedure can only be considered as a screening tool, and putative outliers must be examined by experts to determine whether they are actual errors or important records within an inherently biased set of data.     

Scaling down distribution maps from atlas data: a test of different approaches with virtual species

Aim Conservation managers are increasingly looking for modelled projections of species distributions to inform management strategies; however, the coarse resolution of available data usually compromises their helpfulness. The aim of this paper is to delineate and test different approaches for converting coarse‐grain occurrence data into high‐resolution predictions, and to clarify the conceptual circumstances affecting the accuracy of downscaled models. Location We used environmental data from a real landscape, southern Africa, and simulated species distributions within this landscape. Methods We built 10 virtual species at a resolution of 5 arcmin, and for each species we simulated atlas range maps at four decreasing resolutions (15, 30, 60, 120 arcmin). We tested the ability of three downscaling strategies to produce high‐resolution predictions using two modelling techniques: generalized linear models and generalized boosted models. We calibrated reference models with high‐resolution data and we compared the relative reduction of predictive performance in the downscaled models by using a null model approach. We also estimated the applicability of downscaling procedures to different situations by using distribution data for Mediterranean reptiles. Results All reference models achieved high performance measures. For all strategies, we observed a reduction of predictive performance proportional to the degree of downscaling. The differences in evaluation indices between reference models and downscaled projections obtained from atlases at 15 and 30 arcmin were never statistically significant. The accuracy of projections scaled down from 60 arcmin largely depended on the combination of approach and algorithm adopted. Projections scaled down from 120 arcmin gave misleading results in all cases. Main conclusions Moderate levels of downscaling allow for reasonably accurate results, regardless of the technique used. The most general effect of scaling down coarse‐grain data is the reduction of model specificity. The models can successfully delineate a species’ environmental association up until a 12‐fold downscaling, although with an increasing approximation that causes the overestimation of true distributions. We suggest appropriate procedures to mitigate the commission error introduced by downscaling at intermediate levels (approximately 12‐fold). Reductions of grain size > 12‐fold are discouraged.     

The significance of assemblage-level thinning for species richness

1 A unimodal relationship between species richness and primary productivity is commonly reported. To explain this pattern, the mechanisms proposed in the many hypotheses are generally complex and almost all are without a strong empirical foundation. Here we evaluate the role of self-thinning in plant assemblages: assemblage-level thinning.
2 We developed a simple two-parameter model of species richness that predicts that plant species richness will be determined by a unimodal relationship between total plant density and above-ground biomass. This model provides a very narrowly defined set of testable quantitative predictions, and thus is the first falsifiable model of assemblage-level thinning. We fit this model to the species richness–above-ground biomass data from 14 empirical studies that are often cited as evidence of a general diversity–productivity relationship. In addition, we compared our model to two other models, one more flexible and one more constrained than our own.
3 We found that our model of species richness explained a substantial and statistically significant portion of the species richness observed in 11 of the 14 empirical studies of species richness–biomass patterns. Therefore, given the conservative nature of our model, and the number of previously published data sets explained by this model, we argue that assemblage-level thinning not only provides a viable and exceedingly parsimonious explanation, but may also be a widespread phenomenon.     

virtualspecies,an R package to generate virtual species distributions

virtualspecies is a freely available package for R designed to generate virtual species distributions, a procedure increasingly used in ecology to improve species distribution models. This package combines the existing methodological approaches with the objective of generating virtual species distributions with increased ecological realism. The package includes 1) generating the probability of occurrence of a virtual species from a spatial set of environmental conditions (i.e. environmental suitability), with two different approaches; 2) converting the environmental suitability into presence–absence with a probabilistic approach; 3) introducing dispersal limitations in the realised virtual species distributions and 4) sampling occurrences with different biases in the sampling procedure. The package was designed to be extremely flexible, to allow users to simulate their own defined species–environment relationships, as well as to provide a fine control over every simulation parameter. The package also includes a function to generate random virtual species distributions. We provide a simple example in this paper showing how increasing ecological realism of the virtual species impacts the predictive performance of species distribution models. We expect that this new package will be valuable to researchers willing to test techniques and protocols of species distribution models as well as various biogeographical hypotheses.     

Unpacking the species conundrum: philosophy,practice and a way forward

The history of ecology and evolutionary biology is rife with attempts to define and delimit species. However, there has been confusion between concepts and criteria, which has led to discussion, debate, and conflict, eventually leading to lack of consistency in delimitation. Here, we provide a broad review of species concepts, a clarification of category versus concept, an account of the general lineage concept (GLC), and finally a way forward for species discovery and delimitation. Historically, species were considered as varieties bound together by reproduction. After over 200 years of uncertainty, Mayr attempted to bring coherence to the definition of species through the biological species concept (BSC). This has, however, received much criticism, and the last half century has spawned at least 20 other concepts. A central philosophical problem is that concepts treat species as ‘individuals’ while the criteria for categorization treats them as ‘classes’. While not getting away from this problem entirely, the GLC attempts to provide a framework where lineage divergence is influenced by a number of different factors (and correlated to different traits) which relate to the different species concepts. We also introduce an ‘inclusive’ probabilistic approach for understanding and delimiting species. Finally, we provide a Wallacean (geography related) approach to the Linnaean problem of identifying and delimiting species, particularly for cases of allopatric divergence, and map this to the GLC. Going one step further, we take a morphometric terrain approach to visualizing and understanding differences between lineages. In summary, we argue that while generalized frameworks may work well for concepts of what species are, plurality and ‘inclusive’ probabilistic approaches may work best for delimitation.     

The effect of species models on estimates of within-lineage variation in integration

Species may be modeled as comprised of individuals, populations or a virtual code. A virtual code can be understood as general potential that appears as actualization within specific environmental, both internal and external, contexts. These general potentials form a capacity to network that allows potentials to be expressed and offers robustness through its interconnections. In the present work, the degree of within-lineage variation in integration was not strongly model-dependent. However, the relationships among model-dependent estimates of such variation and within-lineage phyletic variation were not equal. The strongest relationship was between within-lineage variation in integration, when species were modeled as a virtual code, and within-lineage phyletic variation. The second strongest, and only other statistically significant relationship, was between variation in integration when species were modeled as a virtual code and as a collection of populations. The last result argues for a strong ontogenetic and micro-environmental effect on the expression of features in an individual. If species were a virtual code they would evolve by incorporation of all attributes, ontogenetic, environmental and genetic into that code until it becomes unstable and bifurcates. Species as a virtual code, an approach that explicitly incorporates developmental change into evolution, is a non-material representation of species as a complex information system, incorporating, if we refer to mathematical analysis, both the real and the imaginary. If one wished to stress the material, this study could be seen as empirical documentation of species as information systems.     

On the selection of thresholds for predicting species occurrence with presence‐only data

Presence‐only data present challenges for selecting thresholds to transform species distribution modeling results into binary outputs. In this article, we compare two recently published threshold selection methods (maxSSS and maxF_pb) and examine the effectiveness of the threshold‐based prevalence estimation approach. Six virtual species with varying prevalence were simulated within a real landscape in southeastern Australia. Presence‐only models were built with DOMAIN, generalized linear model, Maxent, and Random Forest. Thresholds were selected with two methods maxSSS and maxF_pb with four presence‐only datasets with different ratios of the number of known presences to the number of random points (KP–RP_ratio). Sensitivity, specificity, true skill statistic, and F measure were used to evaluate the performance of the results. Species prevalence was estimated as the ratio of the number of predicted presences to the total number of points in the evaluation dataset. Thresholds selected with maxF_pb varied as the KP–RP_ratio of the threshold selection datasets changed. Datasets with the KP–RP_ratio around 1 generally produced better results than scores distant from 1. Results produced by We conclude that maxF_pb had specificity too low for very common species using Random Forest and Maxent models. In contrast, maxSSS produced consistent results whichever dataset was used. The estimation of prevalence was almost always biased, and the bias was very large for DOMAIN and Random Forest predictions. We conclude that maxF_pb is affected by the KP–RP_ratio of the threshold selection datasets, but maxSSS is almost unaffected by this ratio. Unbiased estimations of prevalence are difficult to be determined using the threshold‐based approach.     

A simple method to estimate the probable distribution of species

Species distribution models (SDMs) are broadly used to predict species distributions from available presence data. However, SDMs results have been criticized for several reasons mainly related to two basic characteristics of most SDMs: 1) general lack of reliable species absence information, 2) the frequent use of an arbitrary geographical extent (GE) or accessible area of the species. These impediments have motivated us to generate a procedure called niche of occurrence (NOO). NOO provides the probable distribution of species (realized niche) relying solely on partial information about presence of species. It operates within a natural geographical extent delimited by available observations and avoids using misleading thresholds to obtain binary presence–absence estimations when the species prevalence is unknown. In this study the main characteristics of NOO are presented, comparing its performance with other recognized and more complex SDMs by using virtual species to avoid the omnipresent error sources of real data sets.     

Integrative species delimitation reveals cryptic diversity in the southern Appalachian Antrodiaetus unicolor (Araneae: Antrodiaetidae) species complex

Although species delimitation can be highly contentious, the development of reliable methods to accurately ascertain species boundaries is an imperative step in cataloguing and describing Earth's quickly disappearing biodiversity. Spider species delimitation remains largely based on morphological characters; however, many mygalomorph spider populations are morphologically indistinguishable from each other yet have considerable molecular divergence. The focus of our study, the Antrodiaetus unicolor species complex containing two sympatric species, exhibits this pattern of relative morphological stasis with considerable genetic divergence across its distribution. A past study using two molecular markers, COI and 28S, revealed that A. unicolor is paraphyletic with respect to A. microunicolor. To better investigate species boundaries in the complex, we implement the cohesion species concept and use multiple lines of evidence for testing genetic exchangeability and ecological interchangeability. Our integrative approach includes extensively sampling homologous loci across the genome using a RADseq approach (3RAD), assessing population structure across their geographic range using multiple genetic clustering analyses that include structure , principal components analysis and a recently developed unsupervised machine learning approach (Variational Autoencoder). We evaluate ecological similarity by using large‐scale ecological data for niche‐based distribution modelling. Based on our analyses, we conclude that this complex has at least one additional species as well as confirm species delimitations based on previous less comprehensive approaches. Our study demonstrates the efficacy of genomic‐scale data for recognizing cryptic species, suggesting that species delimitation with one data type, whether one mitochondrial gene or morphology, may underestimate true species diversity in morphologically homogenous taxa with low vagility.     

Identification and differentiation of Heterotardigrada and Eutardigrada species by riboprinting

In the last decades, the number of known tardigrade species has considerably increased to more than 960 species with new ones being discovered every year. However, the study of tardigrade species presents a general problem which is frequently encountered during the work with invertebrates: small size and remarkable degrees of phenotypic plasticity may sometimes not permit a definite identification of the species. In this investigation we have used riboprinting, a tool to study rDNA sequence variation, in order to distinguish tardigrade species from each other. The method combines a restriction site variation approach of ribotyping with amplified DNAs. In eight investigated species of heterotardigrades and eutardigrades we have amplified the genes for the small subunit ribosomal RNA (SSU; 18S) and subsequently sequenced the genes. Virtual riboprints were used for identification of restriction sites from ten already published 18S rDNA sequences and seven new 18S rDNA sequences. On the basis of the obtained sequences, diagnostic restriction fragment patterns can be predicted with only 11 restriction enzymes. The virtual digestion confirmed the obtained restriction fragment patterns and restriction sites of all amplified and digested tardigrade DNAs. We show that the variation in positions and number of restriction sites obtained by standard restriction fragment analysis on agarose gels can be used successfully for taxonomic identification at different taxonomic levels. The simple restriction fragment analysis provides a fast and convenient method of molecular barcoding for species identification in tardigrades.     

Is species selection dependent upon emergent characters?

The architects of punctuated equilibrium and species selection as well as more recent workers (Vrba) have narrowed the original formulation of species selection and made it dependent upon so-called emergent characters. One criticism of this narrow version is the dearth of emergent characters with a consequent diminution in the robustness of species selection as an important evolutionary process. We argue that monomorphic species characters may at times be the focus of selection and that under these circumstances selection at the organism level is by-passed due to the absence of critical variance. Selection therefore shifts to the species level where variability reemerges in a clade. The absence of critical variance among organisms prevents effect macroevolution from operating. If species-wide properties are important in macroevolutionary processes, as we contend, systematists should pay more attention to their elucidation.     

The negative ecological impacts of a globally introduced species decrease with time since introduction

While there is a long‐history of biological invasions and their ecological impacts have been widely demonstrated across taxa and ecosystems, our knowledge on the temporal dynamic of these impacts remains extremely limited. Using a meta‐analytic approach, we investigated how the ecological impacts of non‐native brown trout (Salmo trutta), a model species with a 170‐year‐long and well‐documented history of intentional introductions across the globe, vary with time since introduction. We first observed significant negative ecological impacts immediately after the species introduction. Second, we found that the negative ecological impacts decrease with time since introduction and that the average ecological impacts become nonsignificant more than one century after introduction. This pattern was consistent across other ecological contexts (i.e., geographical location, levels of biological organization, and methodological approach). However, overall negative ecological impacts were more pronounced at the individual and population levels and in experimental studies. While the mechanisms leading to this decrease remain to be determined, our results indicate that rapid response of native organisms (e.g. adaptation, but also local extinction) may play an important role in this dynamic. Changes in native species traits and local extinction can have important conservation implications. Therefore, we argue that the decline of the negative ecological impacts over time should not be used as an argument to neglect the negative impacts of biological invasions.     

Presence‐only versus presence–absence data in species composition determinant analyses

Aim Studying relationships between species and their physical environment requires species distribution data, ideally based on presence–absence (P–A) data derived from surveys. Such data are limited in their spatial extent. Presence‐only (P‐O) data are considered inappropriate for such analyses. Our aim was to evaluate whether such data may be used when considering a multitude of species over a large spatial extent, in order to analyse the relationships between environmental factors and species composition. Location The study was conducted in virtual space. However, geographic origin of the data used is the contiguous USA. Methods We created distribution maps for 50 virtual species based on actual environmental conditions in the study. Sampling locations were based on true observations from the Global Biodiversity Information Facility. We produced P–A data by selecting ∼1000 random locations and recorded the presence/absence of all species. We produced two P‐O data sets. Full P‐O set was produced by sampling the species in locations of true occurrences of species. Partial P‐O was a subset of full P‐O data set matching the size of the P–A data set. For each data set, we recorded the environmental variables at the same locations. We used CCA to evaluate the amount of variance in species composition explained by each variable. We evaluated the bias in the data set by calculating the deviation of average values of the environmental variables in sampled locations compared to the entire area. Results P–A and P‐O data sets were similar in terms of the amount of variance explained by the different environmental variables. We found sizable environmental and spatial bias in the P‐O data set, compared to the entire study area. Main conclusions Our results suggest that although P‐O data from collections contain bias, the multitude of species, and thus the relatively large amount of information in the data, allow the use of P‐O data for analysing environmental determinants of species composition.     

Modelling cross-hybridization on phylogenetic DNA microarrays increases the detection power of closely related species

DNA microarrays are a popular technique for the detection of microorganisms. Several approaches using specific oligomers targeting one or a few marker genes for each species have been proposed. Data analysis is usually limited to call a species present when its oligomer exceeds a certain intensity threshold. While this strategy works reasonably well for distantly related species, it does not work well for very closely related species: Cross-hybridization of nontarget DNA prevents a simple identification based on signal intensity. The majority of species of the same genus has a sequence similarity of over 90%. For biodiversity studies down to the species level, it is therefore important to increase the detection power of closely related species. We propose a simple, cost-effective and robust approach for biodiversity studies using DNA microarray technology and demonstrate it on scenedesmacean green algae. The internal transcribed spacer 2 (ITS2) rDNA sequence was chosen as marker because it is suitable to distinguish all eukaryotic species even though parts of it are virtually identical in closely related species. We show that by modelling hybridization behaviour with a matrix algebra approach, we are able to identify closely related species that cannot be distinguished with a threshold on signal intensity. Thus this proof-of-concept study shows that by adding a simple and robust data analysis step to the evaluation of DNA microarrays, species detection can be significantly improved for closely related species with a high sequence similarity.