Bunching up the background betters bias in species distribution models

Sets of presence records used to model species’ distributions typically consist of observations collected opportunistically rather than systematically. As a result, sampling probability is geographically uneven, which may confound the model's characterization of the species’ distribution. Modelers frequently address sampling bias by manipulating training data: either subsampling presence data or creating a similar spatial bias in non‐presence background data. We tested a new method, which we call ‘background thickening’, in the latter category. Background thickening entails concentrating background locations around presence locations in proportion to presence location density. We compared background thickening to two established sampling bias correction methods – target group background selection and presence thinning – using simulated data and data from a case study. In the case study, background thickening and presence thinning performed similarly well, both producing better model discrimination than target group background selection, and better model calibration than models without correction. In the simulation, background thickening performed better than presence thinning when the number of simulated presence locations was low, and vice versa. We discuss drawbacks to target group background selection, why background thickening and presence thinning are conservative but robust sampling bias correction methods, and why background thickening is better than presence thinning for small sample sizes. Particularly, background thickening is advantageous for treating sampling bias when data are scarce because it avoids discarding presence records.     

BioDeepTime: A database of biodiversity time series for modern and fossil assemblages

Motivation

We have little understanding of how communities respond to varying magnitudes and rates of environmental perturbations across temporal scales. BioDeepTime harmonizes assemblage time series of presence and abundance data to help facilitate investigations of community dynamics across timescales and the response of communities to natural and anthropogenic stressors. BioDeepTime includes time series of terrestrial and aquatic assemblages of varying spatial and temporal grain and extent from the present-day to millions of years ago.

Main Types of Variables Included

BioDeepTime currently contains 7,437,847 taxon records from 10,062 assemblage time series, each with a minimum of 10 time steps. Age constraints, sampling method, environment and taxonomic scope are provided for each time series.

Spatial Location and Grain

The database includes 8752 unique sampling locations from freshwater, marine and terrestrial ecosystems. Spatial grain represented by individual samples varies from quadrats on the order of several cm² to grid cells of ~100 km².

Time Period and Grain

BioDeepTime in aggregate currently spans the last 451 million years, with the 10,062 modern and fossil assemblage time series ranging in extent from years to millions of years. The median extent of modern time series is 18.7 years and for fossil series is 54,872 years. Temporal grain, the time encompassed by individual samples, ranges from days to tens of thousands of years.

Major Taxa and Level of Measurement

The database contains information on 28,777 unique taxa with 4,769,789 records at the species level and another 271,218 records known to the genus level, including time series of benthic and planktonic foraminifera, coccolithophores, diatoms, ostracods, plants (pollen), radiolarians and other invertebrates and vertebrates. There are to date 7012 modern and 3050 fossil time series in BioDeepTime.

Software Format

SQLite, Comma-separated values.     

Biased data reduce efficiency and effectiveness of conservation reserve networks

Complementarity-based reserve selection algorithms efficiently prioritize sites for biodiversity conservation, but they are data-intensive and most regions lack accurate distribution maps for the majority of species. We explored implications of basing conservation planning decisions on incomplete and biased data using occurrence records of the plant family Proteaceae in South Africa. Treating this high-quality database as 'complete', we introduced three realistic sampling biases characteristic of biodiversity databases: a detectability sampling bias and two forms of roads sampling bias. We then compared reserve networks constructed using complete, biased, and randomly sampled data. All forms of biased sampling performed worse than both the complete data set and equal-effort random sampling. Biased sampling failed to detect a median of 1-5% of species, and resulted in reserve networks that were 9-17% larger than those designed with complete data. Spatial congruence and the correlation of irreplaceability scores between reserve networks selected with biased and complete data were low. Thus, reserve networks based on biased data require more area to protect fewer species and identify different locations than those selected with randomly sampled or complete data.     

The influence of soil cover organization on the floristic and structural heterogeneity of a Guianan rain forest

The impact of soil cover organization on the forest community has been studied in a 19-ha tract at Piste de St Elie station in French Guiana. 195 species each represented by at least 10 individuals were chosen from records of the position, diameter at breast height (dbh) and precise identification by botanical sampling of 12104 ligneous plants (dbh 10 cm).Spatial variations in the soil were mapped using the method proposed by Boulet et al. (1982). The soil mapping units correspond to the successive stages of evolution of a currently unbalanced ferralitic cover. These stages describe firstly the thinning by erosion of the microaggregated upper horizon and secondly the mineralogical changes under more or less extended hydromorphic conditions. The degree of evolution of ferralitic cover is also related to the hydrodynamic functioning and chemical properties of the soil. Geological substrate, topography and slope have also been taken into account.Analysis of the influence of environmental variables on plant cover has been performed using the Ecological Profiles method and Correspondence Analysis (CA) of the table of ecological profiles.The forest community seems to be dependent on the soil and the topographical features that govern it. There are significant, exclusive soil-species links for each soil functioning mapping unit. However, the highest proportion of significant positive links is connected with a thick microaggregated horizon (25%). Several species are of real value as indicators and more particularly enable differentiation between the forest stands of typical ferralitic soil and the ones of thinned out, transformed and hydromorphic soils. The CA of the species by environmental variables matrices reveals two significant factorial axes. The first can be associated with the drainage mainly related to the thinning of the soil and the second with the hydromorphic conditions related to the topography. The vegetation ordination of the stands ( 0.25 ha) delimited in the various soil domains clearly shows that changes in ferralitic cover and in particular the transition from soil with deep vertical drainage to soil with superficial lateral drainage is accompanied by substantial changes in the forest community.     

Spatial capture–recapture with random thinning for unidentified encounters

1. Spatial capture–recapture (SCR) models have increasingly been used as a basis for combining capture–recapture data types with variable levels of individual identity information to estimate population density and other demographic parameters. Recent examples are the unmarked SCR (or spatial count model), where no individual identities are available and spatial mark–resight (SMR) where individual identities are available for only a marked subset of the population. Currently lacking, though, is a model that allows unidentified samples to be combined with identified samples when there are no separate classes of “marked” and “unmarked” individuals and when the two sample types cannot be considered as arising from two independent observation models. This is a common scenario when using noninvasive sampling methods, for example, when analyzing data on identified and unidentified photographs or scats from the same sites.
2. Here we describe a “random thinning” SCR model that utilizes encounters of both known and unknown identity samples using a natural mechanistic dependence between samples arising from a single observation model. Our model was fitted in a Bayesian framework using NIMBLE.
3. We investigate the improvement in parameter estimates by including the unknown identity samples, which was notable (up to 79% more precise) in low‐density populations with a low rate of identified encounters. We then applied the random thinning SCR model to a noninvasive genetic sampling study of brown bear (Ursus arctos) density in Oriental Cantabrian Mountains (North Spain).
4. Our model can improve density estimation for noninvasive sampling studies for low‐density populations with low rates of individual identification, by making use of available data that might otherwise be discarded.

     

Database records as a surrogate for sampling effort provide higher species richness estimations

The compilation of all the available taxonomic and distributional information on the species present in a territory frequently generates a biased picture of the distribution of biodiversity due to the uneven distribution of the sampling effort performed. Thus, quality protocol assessments such as those proposed by Hortal et al. (Conservation Biology 21:853–863, 2007) must be done before using this kind of information for basic and applied purposes. The discrimination of localities that can be considered relatively well-surveyed from those not surveyed enough is a key first step in this protocol and can be attained by the previous definition of a sampling effort surrogate and the calculation of survey completeness using different estimators. Recently it has been suggested that records from exhaustive databases can be used as a sampling-effort surrogate to recognize probable well-surveyed localities. In this paper, we use an Iberian dung beetle database to identify the 50 × 50 km UTM cells that appear to be reliably inventoried, using both data derived from standardized sampling protocols and database records as a surrogate for sampling effort. Observed and predicted species richness values in the shared cells defined as well-surveyed by both methods suggest that the use of database records provides higher species richness values, which are proportionally greater in the richest localities by the inclusion of rare species.     

TetraDENSITY: A database of population density estimates in terrestrial vertebrates

Motivation

Population density is a key demographic parameter influencing many ecological processes, and macroecology has described both intra‐ and interspecific patterns of variation. Population density data are expensive to collect and contain many forms of noise and potential bias; these factors have impeded investigation of macroecological patterns, and many hypotheses remain largely unexplored. Population density also represents fundamental information for conservation, because it underlies population dynamics and, ultimately, extinction risk. Here we present TetraDENSITY, an extensive dataset with > 18,000 records of density estimates for terrestrial vertebrates, in order to facilitate new research on this topic.

Main types of variable contained

The dataset includes taxonomic information on species, population density estimate, year of data collection, season, coordinates of the locality, locality name, habitat, sampling method and sampling area.

Spatial location and grain

Global. Spatial accuracy varies across studies; conservatively, it can be considered at 1°, but for many data it is much finer.

Time period and grain

From 1926 to 2017. Temporal accuracy is yearly in most cases, but studies with higher temporal resolution (season, month) are also present.

Major taxa and level of measurement

Amphibians in terrestrial phase, reptiles, birds and mammals. Estimates derive from multiple methods, reflecting the study taxon, location and techniques available at the time of density estimation.     

Progression of ash canopy thinning and dieback outward from the initial infestation of emerald ash borer (Coleoptera: Buprestidae) in southeastern Michigan

Our objective was to characterize the rate at which ash (Fraxinus spp.) trees decline in areas adjacent to the leading edge of visible ash canopy thinning due to emerald ash borer, Agrilus planipennis Fairmaire (Coleoptera: Buprestidae). Trees in southeastern Michigan were surveyed from 2003 to 2006 for canopy thinning and dieback by comparing survey trees with a set of 11 standard photographs. Freeways stemming from Detroit in all directions were used as survey transects. Between 750 and 1,100 trees were surveyed each year. A rapid method of sampling populations of emerald ash borer was developed by counting emerald ash borer emergence holes with binoculars and then felling trees to validate binocular counts. Approximately 25% of the trees surveyed for canopy thinning in 2005 and 2006 also were sampled for emerald ash borer emergence holes using binoculars. Regression analysis indicates that 41-53% of the variation in ash canopy thinning can be explained by the number of emerald ash borer emergence holes per tree. Emerald ash borer emergence holes were found at every site where ash canopy thinning averaged > 40%. In 2003, ash canopy thinning averaged 40% at a distance of 19.3 km from the epicenter of the emerald ash borer infestation in Canton. By 2006, the point at which ash trees averaged 40% canopy thinning had increased to a distance of 51.2 km away from Canton. Therefore, the point at which ash trees averaged 40% canopy thinning, a state of decline clearly visible to the average person, moved outward at a rate of 10.6 km/yr during this period.     

A grid‐based method for sampling and analysing spatially ambiguous plants

Abstract. Spatial data can provide much information about the interrelations of plants and the relationship between individuals and the environment. Spatially ambiguous plants, i.e. plants without readily identifiable loci, and plants that are profusely abundant, present non‐trivial impediments to the collection and analysis of vegetation data derived from standard spatial sampling techniques. Sampling with grids of presence/absence quadrats can ameliorate much of this difficulty. Our analysis of 10 fully‐mapped grassland plots demonstrates the applicability of the grid‐based approach which revealed spatial dependence at a much lower sampling effort than mapping each plant. Ripley's K‐function, a test commonly used for point patterns, was effective for pattern analysis on the grids and the gridded quadrat technique was an effective tool for quantifying spatial patterns. The addition of spatial pattern measures should allow for better comparisons of vegetation structure between sites, instead of sole reliance on species composition data.     

A Model-Based Approach for Making Ecological Inference from Distance Sampling Data

Summary .  We consider a fully model-based approach for the analysis of distance sampling data. Distance sampling has been widely used to estimate abundance (or density) of animals or plants in a spatially explicit study area. There is, however, no readily available method of making statistical inference on the relationships between abundance and environmental covariates. Spatial Poisson process likelihoods can be used to simultaneously estimate detection and intensity parameters by modeling distance sampling data as a thinned spatial point process. A model-based spatial approach to distance sampling data has three main benefits: it allows complex and opportunistic transect designs to be employed, it allows estimation of abundance in small subregions, and it provides a framework to assess the effects of habitat or experimental manipulation on density. We demonstrate the model-based methodology with a small simulation study and analysis of the Dubbo weed data set. In addition, a simple ad hoc method for handling overdispersion is also proposed. The simulation study showed that the model-based approach compared favorably to conventional distance sampling methods for abundance estimation. In addition, the overdispersion correction performed adequately when the number of transects was high. Analysis of the Dubbo data set indicated a transect effect on abundance via Akaike's information criterion model selection. Further goodness-of-fit analysis, however, indicated some potential confounding of intensity with the detection function.     

A journey through time: exploring temporal patterns amongst digitized plant specimens from Australia

Online access to species occurrence records has opened new windows into investigating biodiversity patterns across multiple scales. The value of these records for research depends on their spatial, temporal, and taxonomic quality. We assessed temporal patterns in records from the Australasian Virtual Herbarium, asking: (1) How temporally consistent has collecting been across Australia? (2) Which areas of Australia have the most reliable records, in terms of temporal consistency and inventory completeness? (3) Are there temporal trends in the completeness of attribute information associated with records? We undertook a multi-step filtering procedure, then estimated temporal consistency and inventory completeness for sampling units (SUs) of 50?km ×?50?km. We found temporal bias in collecting, with 80% of records collected over the period 1970–1999. South-eastern Australia, the Wet Tropics in north-east Queensland, and parts of Western Australia have received the most consistent sampling effort over time, whereas much of central Australia has had low temporal consistency. Of the SUs, 18% have relatively complete inventories with high temporal consistency in sampling. We also determined that 25% of digitized records had missing attribute information. By identifying areas with low reliability, we can limit erroneous inferences about distribution patterns and identify priority areas for future sampling.     

A Fixed-Precision Sequential Sampling Plan for the Potato Tuberworm Moth, <Emphasis Type="Italic">Phthorimaea operculella</Emphasis> Zeller (Lepidoptera: Gelechidae), on Potato Cultivars

Phthorimaea operculella Zeller is an important pest of potato in Iran. Spatial distribution and fixed-precision sequential sampling for population estimation of the pest on two potato cultivars, Arinda^® and Sante^®, were studied in two separate potato fields during two growing seasons (2013–2014 and 2014–2015). Spatial distribution was investigated by Taylor’s power law and Iwao’s patchiness. Results showed that the spatial distribution of eggs and larvae was random. In contrast to Iwao’s patchiness, Taylor’s power law provided a highly significant relationship between variance and mean density. Therefore, fixed-precision sequential sampling plan was developed by Green’s model at two precision levels of 0.25 and 0.1. The optimum sample size on Arinda^® and Sante^® cultivars at precision level of 0.25 ranged from 151 to 813 and 149 to 802 leaves, respectively. At 0.1 precision level, the sample sizes varied from 5083 to 1054 and 5100 to 1050 leaves for Arinda^® and Sante^® cultivars, respectively. Therefore, the optimum sample sizes for the cultivars, with different resistance levels, were not significantly different. According to the calculated stop lines, the sampling must be continued until cumulative number of eggs + larvae reached to 15–16 or 96–101 individuals at precision levels of 0.25 or 0.1, respectively. The performance of the sampling plan was validated by resampling analysis using resampling for validation of sampling plans software. The sampling plant provided in this study can be used to obtain a rapid estimate of the pest density with minimal effort.     

A Pragmatic Approach for Determining Otter Distribution from Disparate Occurrence Records

Opportunistic records of animal occurrence may be problematic for inferring species distribution and habitat requirements because of unknown and uncontrolled sources of sampling variance. In this study, we used occurrence records for river otters (Lontra canadensis) derived from sign surveys, road kills, trapper bycatch, and opportunistic sightings (n = 185 records collected 2001–2012) to assess the potential distribution and habitat relationships of otters across central and western New York, USA. To mitigate for obvious observation biases, we standardized observation intensity across regions a priori and restricted inference to readily accessible areas (i.e., ≤700 m from the nearest road). Model selection, and the direction of covariate effects, proved robust to these sampling biases although effect sizes varied −7.1% to +48.0% after bias correction, with the coefficient for the proportion of available shoreline being the most unstable. Ultimately, the top bias-corrected model proved a reliable index for otter probability of occurrence given a strong, positive, and linear relationship with a withheld set of standardized survey records for otters collected in winter 2016–2017 (n = 57; R² = 0.90). This model indicated that approximately 20% of the study area represented high probability of otter occurrence. We demonstrated that reliable inference on wildlife habitat requirements can be obtained from disparate records of animal occurrence provided that data biases are known and effectively mitigated. © 2020 The Wildlife Society.     

Bayesian prediction of spatial count data using generalized linear mixed models

Spatial weed count data are modeled and predicted using a generalized linear mixed model combined with a Bayesian approach and Markov chain Monte Carlo. Informative priors for a data set with sparse sampling are elicited using a previously collected data set with extensive sampling. Furthermore, we demonstrate that so-called Langevin-Hastings updates are useful for efficient simulation of the posterior distributions, and we discuss computational issues concerning prediction.     

Developing an integrated cloud-based spatial-temporal system for monitoring phenology

Geospatial cloud computing offers computing infrastructure, software and data services that enable rapid integration of ecological data from various resources. The objectives of this study were to utilize readily-available and low-cost technology (e.g., GPS–enabled cameras, Cloud photo storage, Google Drive) to create a cloud-based spatial-temporal inventory of plant (including flowering phenology) and other relevant information. An interactive ArcGIS Online Map of Lake Issaqueena, SC with sampling locations of flowering plants allows users to obtain additional information (plant, soil, weather data) by selecting sampling locations or soil polygons. The contents of the map can be filtered using any of the attributes (e.g., growth form) in the data tables by selecting specific information. Plant information can be viewed at custom time intervals using the settings in ArcGIS Online. Spatial patterns (e.g., clustering) in the plant data can be viewed using the ArcGIS Online heat map view. The map can be easily queried and viewed on both computers and hand-held devices. Services from multiple cloud infrastructures can be integrated for use by various species monitoring programs, improving workflow and assessment capabilities.     

Movement and vertical stratification of fruit-feeding butterflies in a managed West African rainforest

We analysed movement parameters and vertical stratification in fruit-feeding butterflies between a control, a thinned and a plantation site within a West African rainforest. Overall, distances moved between traps were largest in the plantation. Movement parameters were generally largest in species feeding on early successional hostplants of gap and margin habitats. In these species, distances between recaptures were significantly shorter in the thinned compared to the control forest. Conversely, forest floor species feeding on climbers and understorey shrubs showed significantly larger movement in the thinned forest. Higher strata species also flew larger distances in the thinned understorey. Including higher strata samples, they were significantly less abundant in the thinned plot, although understorey sampling alone indicated the contrary. Comparing vertical distribution patterns between thinned and control sites indicated a disruption of vertical stratification after thinning. Canopy species seem to fly in the upper strata of the more closed-canopy control forest, whereas they descend more frequently in the forest opened by the thinning management. Understorey sampling might therefore lead to biased conclusions due to differences in vertical distribution between forest plots. This study showed that thinning can affect the restricted-range forest floor butterflies as well as the more widespread canopy butterfly fauna.     

Temporal and spatial variation in an index of biological integrity for the middle Wabash River, Indiana

Few studies have addressed within-year temporal variation of IBI scores. We compared index of biotic integrity (IBI) scores for two summer sampling events from a large river during 25 annual periods. The results indicated that IBI scores calculated from June samples were not significantly different from July samples. Spatial autocorrelation was present, such that sites that were closer together produced similar IBI scores and could not be considered independent. Temporal autocorrelation was present, but was not strong. Lower quality sites (low IBI score) did not have higher variation than higher quality sites. Our results show that a single sample of fishes by boat electrofisher during the summer in a large river such as the Wabash River can produce a repeatable estimate of IBI score. Thus, repeated or additional sampling within the summer season to improve the quality of the evaluation is not warranted. Handling editor: J. Trexler     

Accessibility maps as a tool to predict sampling bias in historical biodiversity occurrence records

Historical biodiversity occurrence records are often discarded in spatial modeling analyses because of a lack of a method to quantify their sampling bias. Here we propose a new approach for predicting sampling bias in historical written records of occurrence, using a South African example as proof of concept. We modelled and mapped accessibility of the study area as the mean of proximity to freshwater and European settlements. We tested the model's ability to predict the location of historical biodiversity records from a dataset of 2612 large mammal occurrence records collected from historical written sources in South Africa in the period 1497–1920. We investigated temporal, spatial and environmental biases in these historical records and examined if the model prediction and occurrence dataset share similar environmental bias. We find a good agreement between the accessibility map and the distribution of sampling effort in the early historical period in South Africa. Environmental biases in the empirical data are identified, showing a preference for lower maximum temperature of the warmest month, higher mean monthly precipitation, higher net primary productivity and less arid biomes than expected by a uniform use of the study area. We find that the model prediction shares similar environmental bias as the empirical data. Accessibility maps, built with very simple statistical rules and in the absence of empirical data, can thus predict the spatial and environmental biases observed in historical biodiversity occurrence records. We recommend that this approach be used as a tool to estimate sampling bias in small datasets of occurrence and to improve the use of these data in spatial analyses in ecological and conservation studies.     

不同间伐强度对辽东栎林群落稳定性的影响

对黄土高原地区近自然经营间伐强度为13.4%(弱度间伐)和30.0%(强度间伐)的辽东栎林进行样地调查,并以未间伐林为对照,研究间伐后第5年建群种的更新潜力、林地生产力、土壤肥力和物种多样性;应用模糊数学中的隶属函数方法对辽东栎林群落稳定性进行评价.结果表明：与对照相比,弱度和强度间伐使辽东栎林地的更新潜力分别提高了14.2%和20.2%;立木蓄积量分别减少了9.0%和23.8%,但灌木生物量分别增加7.3%和12.2%,草本生物量分别增加10.5%和31.6%.间伐样地的土壤肥力和物种多样性指数均高于未间伐样地.辽东栎林群落稳定性表现为：强度间伐＞弱度间伐＞未间伐.近自然经营采伐强度30.0%更适合黄 土高原地区辽东栎次生林的抚育管理.     

Trapping intensities for sampling ants in Australian rangelands

Abstract I investigated the relationship between species richness and composition of ant faunas, and sampling intensity in two regions with different long‐term histories of grazing intensity in mulga (Acacia aneura) woodlands in northern New South Wales. There were two aims: (i) to examine the relationship between sampling intensity and species richness and composition; and (ii) to explore the differences in ant assemblages from two regions of markedly different grazing intensity when sampled at different intensities (i.e. when a higher proportion of the local ant fauna were collected). Ants were sampled in pit traps (120‐mm diameter) at densities of two, four, six and nine pits per 100 m². Each sampling‐intensity treatment was replicated three times within each region. Pit traps filled with preservative were opened for 3 days. Species richness was higher with each successive increase in sampling intensity but was not different between regions for a given trapping intensity. There was no obvious asymptote of the curve relating trapping intensity to cumulative species richness suggesting that even greater trap densities than those used in the present study would be needed to collect most of the species of ants using a patch of ground over a few days. Spatial replication of a low‐intensity sampling design did not capture as many species as one higher‐intensity sampling array with the same total number of pit traps. This result can be explained by aggressive numerically dominant species of ants monopolizing access to a greater proportion of the traps in low‐density arrays. Ordination reveals that regions and sampling‐intensity treatments could be discriminated and that differences between regions with different grazing histories were less apparent with high‐intensity sampling arrays than they were with low‐intensity sampling arrays. This suggests that differences between locations in space (or potentially samples in time) could be exaggerated by incomplete sampling of the patch‐scale fauna. Comparison of the present study with other studies suggests that most studies to date have used sampling intensities that would not give a thorough assessment of the patch‐scale ground‐dwelling fauna if sampled only by pit traps. The implications of the results for programmes of ant monitoring in rangelands are discussed.