Fig 2 - uploaded by Joaquín Hortal
Content may be subject to copyright.
The Iberian Peninsula showing the most important rivers and mountain ranges. The study area is indicated in light grey
Source publication
Knowledge on the distribution and abundance of species is plagued by significant taxonomic and geographic biases that influence the analyses on biodiversity patterns. Due to this, standard, easy-to-use methods are needed to design efficient field campaigns that minimize data deficiencies. We evaluate the applicability, usefulness and effectiveness...
Citations
... Well-sampled sites are spatially clustered and in easily accessible locations. Asymmetries and inaccuracies in the geographical and environmental representation of species can strongly in uence the results of pattern analyses and also compromise conservation action planning (Medina et al. 2013). The results of sampling completeness indicate a need for more braconid collections and studies in large areas of the northern, northeastern, and midwestern regions of the country. ...
Understanding the distributional patterns of species is a challenge to science, mainly because knowledge of the occurrence of species is often scarce and biased. This study aimed to compile available information on the distribution of braconids in Brazil and evaluate the sampling gaps and biases using accessibility metrics. The full dataset includes 2,280 records of 1,015 species of these wasps in Brazil, distributed in 28 subfamilies and 229 genera. Opiinae is the subfamily containing the largest number of records, while Doryctinae stands out in terms of species richness. The genus with the greatest number of records is Doryctobracon Enderlein 1920, while Opius Wesmael 1835 attracts attention for its number of species. Doryctobracon areolatus (Szépligeti 1911) is the species with the largest number of records in all biomes. Most species records are from the Brazilian Southeast, especially from the Atlantic Forest. Data indicates accessibility bias towards roads, rivers, and urban centers.
Implication to the conservation of insects: Our results compile the information available on the distribution of braconids in Brazil and, through them, we examined the extension of deficiencies in the sampling coverage to subsidize future studies and the prioritization of sampling areas, as well as important conservation strategies that are efficient for conservation.
... To date, most efforts aimed at selection of areas for survey and inventory efforts have been based on proxies and approximations (Hill et al., 2005); only a few efforts have been made to optimize sampling in both geographical and environmental spaces (D'Antraccoli et al., 2020;Funk et al., 2005;Hortal & Lobo, 2005;Medina et al., 2013;Velásquez-Tibatá, 2019). As a result, most biodiversity patterns derived from inventories include different types of biases (Oliveira et al., 2016;Sastre & Lobo, 2009;Yang et al., 2013), which could be prevented if more comprehensive considerations are taken when planning systems for inventory. ...
... Hortal and Lobo (2005) proposed another approach using a rule-step siteallocation procedure, based partially on Faith and Walker's 'ED' criterion (a framework linking species data and environmental information to explore underlying environmental variation related to a biological pattern; Faith, 2003;Faith & Walker, 1996). Using similar considerations, Funk et al. (2005) employed a method to complement survey systems by selecting sampling localities based on a survey-gap analysis (see also Medina et al., 2013). These methods require certain knowledge of the biodiversity in the region such that application in areas where biodiversity data are scarce could be difficult. ...
... We also found that the use of preselected sites, corresponding to sites already well inventoried, and a mask to restrict analysis to sites that are interesting and/or accessible, has positive effects on the effectiveness of the set of selected sites. These ideas have been explored in previous studies (Funk et al., 2005;Hortal & Lobo, 2005;Medina et al., 2013), in which definition of areas suitable or unsuitable for surveys, and use of existing information from previous surveys play critical roles in implementation of methods for sampling site selection (see also Gillespie et al., 2017;Hoffmann et al., 2019;Tessarolo et al., 2021;Xu et al., 2017). In the example presented, our mask was used to focus analyses in areas with natural vegetation in Mexico. ...
Biodiversity inventory is among the major challenges for conservation biology in the face of global change. Species exist in two spaces that are linked in the so‐called Hutchinsonian Duality: distributions in geographical space and ecological niches in environmental space. We explore implications of using distinct methods to select locations for biodiversity inventories, based on this idea of two‐space distributions.
We combined empirical and statistical methods to facilitate selecting localities for biodiversity inventory based on either or both of geographical and environmental considerations. These approaches were applied to select sites for inventory in four example countries. For one of our examples, we tested how effective distinct methods were in sampling biodiversity.
Random and geographically uniform selections are generally biased towards the most common environments in the regions; selections aiming for uniform sampling of environments are concentrated spatially in areas of high heterogeneity in geographical context. Considering disparate geographical distributions of environments helped to cover geographical areas more broadly when selections were environmentally uniform. Generally, sets of sites selected considering environmental conditions perform better in sampling known biodiversity in regions of interest.
Our results underline the benefits of considering environmental and geographical conditions when selecting sites on the effectiveness of resulting inventories. Our tools, implemented in the r package biosurvey , will help researchers to design biodiversity survey systems taking into account the Hutchinsonian Duality and the crucial considerations that it suggests.
... The lack of information from these regions compromises any assessment of the processes behind species diversity patterns, as well as the implementation of conservation biogeography approaches (Reddy and Dávalos 2003, Lomolino 2004, Whittaker et al. 2005, Hortal et al. 2007, Hortal et al. 2008. Furthermore, the development of ecological, evolutionary and biogeographical research on Iberian mosses currently requires more surveys with an adequate spatial design and planning (see Hortal and Lobo 2005, Medina et al. 2013. This would maximise their effectiveness, as exemplified by the results of one performed on Iberian epiphytic mosses . ...
2020) Assessing spatial and temporal biases and gaps in the publicly available distributional information of Iberian mosses. Biodiversity Data Journal 8: e53474. https://doi. Abstract One of the most valuable initiatives on massive availability of biodiversity data is the Global Biodiversity Information Facility, which is creating new opportunities to develop and test macroecological knowledge. However, the potential uses of these data are limited by the gaps and biases associated to large-scale distributional databases (the so-called Wallacean shortfall). Describing and quantifying these limitations are essential to improve knowledge on biodiversity, especially in poorly-studied groups, such as mosses. Here we assess the coverage of the publicly-available distributional information on Iberian mosses, defining its eventual biases and gaps. For this purpose, we compiled IberBryo v1.0, a database that comprises 82,582 records after processing and checking the geospatial and taxonomical information. Our results show the limitations of data and metadata of the publicly-available information. Particularly, ca. 42% of the records lacked collecting date information, which limits data usefulness for time coverage analyses and enlarges the existing knowledge gaps. Then we evaluated the overall coverage of several aspects of the ‡ ‡ § | ¶ § ‡,#,¤ © Ronquillo C et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. spatial, temporal and environmental variability of the Iberian Peninsula. Through this assessment, we demonstrate that the publicly-available information on Iberian mosses presents significant biases. Inventory completeness is strongly conditioned by the recorders' survey bias, particularly in northern Portugal and eastern Spain and the spatial pattern of surveys is also biased towards mountains. Besides, the temporal pattern of survey effort intensifies from 1970 onwards, encompassing a progressive increase in the geographic coverage of the Iberian Peninsula. Although we just found 5% of well-surveyed cells of 30' of resolution over the 1970-2018 period, they cover about a fifth of the main climatic gradients of the Iberian Peninsula, which provides a fair-though limited-coverage. Yet, the well-surveyed cells are biased towards anthropised areas and some of them are located in areas under intense land-use changes, mainly due to the wood-fires of the last decade. Despite the overall increase, we found a noticeable gap of information in the southwest of Iberia, the Ebro river basin and the inner plateaus. All these gaps and biases call for a careful use of the available distributional data of Iberian mosses for biogeographical and ecological modelling analysis. Further, our results highlight the necessity of incorporating several good practices to increase the coverage of high-quality information. These good practices include digitalisation of specimens and metadata information, improvement on the protocols to get accurate data and metadata or revisions of the vouchers and recorders' field notebooks. These procedures are essential to improve the quality and coverage of the data. Finally, we also encourage Iberian bryologists to establish a series of re-surveys of classical localities that would allow updating the information on the group, as well as to design their future surveys considering the most important information gaps on IberBryo.
... KnowBR has the potential to become a standard tool not only to assess survey completeness, but also to provide fair estimates of the distribution of biodiversity, and to study the survey process in detail. To increase the current functionality of KnowBR, future R applications should incorporate tools: i) to identify the spatial units that would be most appropriate for surveying to maximize the spatial and environmental coverage provided by the set of well-surveyed territorial units (such as Medina et al., 2013); ii) to identify reliable absences for any focal species and model species distributions using techniques that take advantage of both occurrence and probabilistic absence data (Lobo et al., 2010); iii) to calculate the degree of uncertainty associated to the results of these models in poorly surveyed areas, based on their degree of completeness and the distance to well-surveyed areas (i.e. maps of ignorance sensu Rocchini et al., 2011); and iv) to describe the biases in the spatial distribution of sampling effort, and explore the factors behind these biases. ...
Biodiversity databases are typically incomplete and biased. We identify their three main limitations for characterizing the geographic distributions of species: unknown levels of survey effort, unknown absences of a species from a region, and unknown level of repeated occurrence of a species in different samples collected at the same location. These limitations hinder our ability to distinguish between the actual absence of a species at a given location and its (erroneous) apparent absence as consequence of inadequate surveys. Good practice in biodiversity research requires knowledge of the number, location and degree of completeness of relatively well-surveyed inventories within territorial units. We herein present KnowBR, an application designed to simultaneously estimate the completeness of species inventories across an unlimited number of spatial units and different geographical extents, resolutions and unit expanses from any biodiversity database. We use the number of database records gathered in a territorial unit as a surrogate of survey effort, assuming that such number correlates positively with the probability of recording a species within such area. Consequently, KnowBR uses a "record-by-species" matrix to estimate the relationship between the accumulated number of species and the number of database records to characterize the degree of completeness of the surveys. The final slope of the species accumulation curves and completeness percentages are used to discriminate and map well-surveyed territorial units according to user criteria. The capacity and possibilities of KnowBR are demonstrated through two examples derived from data of varying geographic extent and numbers of records. Further, we identify the main advances that would improve the current functionality of KnowBR.
... We sampled 43 forests that were selected to represent environmental and spatial variability within the studied territory. A complete description of the survey design and the protocol of locality selection can be found in Medina et al. (2015) and Medina et al. (2013). ...
Species richness is influenced by a nested set of environmental factors, but how do these factors interact across several scales? Our main aim is to disentangle the relative importance of environmental filters and the species pool on the richness of epiphytic bryophytes across spatial scales. To do so, we sampled epiphytic bryophytes in 43 oak forests across the northwest of the Iberian Peninsula. As predictors we used climate, descriptors of forest structure and micro-environment. We applied structural equation modeling to relate these variables with richness and cover at three scales: locality (forest), stand (three stands per forest), and sample (a quadrate in a tree). We assumed top–down relationships, so that large-scale variables influenced lower scale variables, and in which cover directly influenced richness. Richness at the next larger scale (locality to stand and stand to sample) is considered a surrogate of the species pool and included as a predictor of richness at the next smaller scale. Environmental variables explain locality richness, but as we decrease the spatial scale, its importance decreases and the dependence on species pool increases. In addition, we found unexpected bottom–up relationships (between micro-scale environment to locality richness). Our results point to the scale dependence of niche vs. neutral processes: niche processes are important at the locality (forest) scale, while neutral processes are significant at the small (sample) scale. We propose a modified conceptualization of the factors influencing biodiversity at different spatial scales by adding links across different scales (between micro-environment and locality-scale richness in our study).
... Hirzel & Guisan, 2002;Croft & Chow-Fraser, 2009). For instance, sampling designs based on the environmental diversity approach (Faith & Walker, 1996), such as stratified or heuristic methods, are suitable for maximizing the study of species diversity in a region by accounting for as much environmental heterogeneity as possible (Funk, Richardson & Ferrier, 2005;Medina et al., 2013). Ultimately, however, the degree of inventory completeness of our sampling design will depend on the available resources. ...
... The environmental diversity framework has shown contradictory results in capturing high species diversity (e.g. Araújo, Densham & Humphries, 2003;Medina et al., 2013). In our case, it performed relatively well because there was a seeming effect of both stratifying variables, giving support to our prior hypotheses for survey planning . ...
We require representative data of species occurrence to explain plant diversity patterns, but most of the available information is incomplete and biased. To improve our knowledge, we suggest that species inventorying should be an iterative process encompassing the following: (1) the detection of taxonomic and geographical gaps; (2) the planning of a survey design to reduce such gaps; and (3) the evaluation of field sampling results. Here, we focus on the latter phase for the bryophytes of Terceira Island (Azores) for which we have previously estimated < 1% of the area as well surveyed based on historical collections. To examine the performance of our stratified survey based on two factors (land use and environmental regions), we used rarefaction curves, ANOVA tests and bootstrap sampling. We recorded 40% of all the species known for the island and presented eight new citations. The species assemblages remained similar between historical and current inventories. Most localities had completeness values > 85%, but we always exceeded the optimal sampling effort. Land uses and environmental regions affected species diversity, but, unexpectedly, to a different degree. Our study illustrates the difficulties of planning field surveys to obtain reliable biodiversity patterns, even when prior information and standardized sampling protocols are explicitly considered. © 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, ●●, ●●–●●.
... Sampling effort, in terms of specimens collected or area surveyed, can heavily influence species richness estimates and thus the analyses of geographic patterns of species diversity and their environmental drivers (Gotelli and Colwell 2010;Yang et al. 2013;Zhang et al. 2014). This is particularly true for bryophytes, since many regions have not been adequately surveyed due to their small size, simplicity in structure and subsequent difficulties in field identification (Aranda et al. 2010;Mutke and Geffert 2010;Medina et al. 2013). To be able to compare richness patterns between bryophyte studies and relate these patterns to environmental variables, the effect of sampling effort has to be taken into account. ...
... To be able to compare richness patterns between bryophyte studies and relate these patterns to environmental variables, the effect of sampling effort has to be taken into account. However, it has long been recognized that most bryophyte biogeographical studies suffer from incomplete sampling (Aranda et al. 2010;Mutke and Geffert 2010;Medina et al. 2013). Studies that did include measures of sampling completeness generally only considered the effect of sampling area, but not specimen number (e.g., von Konrat et al. 2008;Mutke and Geffert 2010), even though the latter is a more direct measure of sampling effort (Willott 2001). ...
Our knowledge of the spatial distribution of bryophyte diversity still suffers from low sampling efforts. Here we try to determine the spatial diversity patterns of liverworts and mosses and their environmental drivers more accurately by correcting for this sampling bias. We compiled bryoflora from 49 localities in eastern China, including data on sampling effort. Both sampling bias uncorrected (raw) species richness and bias corrected (estimated) species richness, as derived from species-sampling curves, were used as response variables. Model selection based on Akaike’s information criterion was used to evaluate the impact of collection bias on the selection of environmental and spatial variables in the regression models. Variation partitioning was used to assess the independent and joint effects of environmental, spatial and sampling variables on raw and estimated species richness. Liverwort richness increased significantly with decreasing latitude, while moss richness showed no latitudinal pattern, whether for raw or estimated species richness. However, estimated species richness showed stronger correlation with environmental variables than raw species richness. Importantly, selected environmental variables in the raw species richness models changed after correcting for collection bias. Despite their ability to produce copious amounts of spores, our sampling bias corrected models indicated that bryophyte richness showed strong spatial structuring, indicating dispersal limitation. Environmentally, liverwort richness was primarily controlled by water availability, while the richness of mosses was mainly determined by available energy. Our results highlight that biological features, dispersal ability and environmental sorting may account for the
discrepancies between species richness of liverworts and mosses. Given the considerable impact that controlling for sampling effort had on analysis outcome, we like to stress the importance of controlling for sampling bias when studying spatial richness patterns in bryophytes.
... It follows that an adequate spatial design of the surveys increases the value of biodiversity data collections to answer different questions in ecology, evolution and biogeography (Hortal & Lobo, 2005;Albert et al., 2010). Many methods to design surveys have been proposed with the objective of maximizing the amount of biodiversity captured, while incorporating time and cost limitations (Austin & Heyligers, 1989;Pereira & Itami, 1991;Hirzel & Guisan, 2002;Funk et al., 2005;Hortal & Lobo, 2005;Medina et al., 2013). However, even planned sample designs can vary in the efficiency with which they detect biodiversity patterns . ...
AimSpecies distribution models (SDM) can be used to predict the location of unknown populations from known species occurrences. It follows that how the data used to calibrate the models are collected can have a great impact on prediction success. We evaluated the influence of different survey designs and their interaction with the modelling technique on SDM performance.LocationIberian Peninsula.Methods
We examine how data recorded using seven alternative survey designs (random, systematic, environmentally stratified by class and environmentally stratified using P-median, biased due to accessibility, biased by human density aggregation and biased towards protected areas) could affect SDM predictions generated with nine modelling techniques (BIOCLIM, Gower distance, Mahalanobis distance, Euclidean distance, GLM, MaxEnt, ENFA and Random Forest). We also study how sample size, species’ characteristics and modelling technique affected SDM predictive ability, using six evaluation metrics.ResultsSurvey design has a small effect on prediction success. Characteristics of species’ ranges rank highest among the factors affecting SDM results: the species with lower relative occurrence area (ROA) are predicted better. Model predictions are also improved when sample size is large.Main conclusionsThe species modelled – particularly the extent of its distribution – are the largest source of influence over SDM results. The environmental coverage of the surveys is more important than the spatial structure of the calibration data. Therefore, climatic biases in the data should be identified to avoid erroneous conclusions about the geographic patterns of species distributions.
... We sampled 43 forests that were selected to represent environmental and spatial variability within the studied territory. A complete description of the survey design and the protocol of locality selection can be found in Medina et al. (2015) and Medina et al. (2013). ...
Spatial variation in species richness is one of the most frequently studied topics on macroecology. However, the relative importance of the factors affecting richness across scales and their influence on some groups of small-sized organisms, such as bryophytes, remain unclear. We evaluate the relative importance of biogeographic region, climate, topography, forest structure and abundance in shaping epiphytic bryophyte richness at both local (forest) and sample (trunk) scale on the boundary between the Atlantic and Mediterranean regions in NW Spain. For that purpose we used simple, multiple and partial regressions, hierarchical partitioning and partial least squares path analyses. Although climatic variables related to water availability during spring and summer were the most important predictors of bryophyte richness, their effects were moderated by winter temperature. Abundance, in contrast, was mostly related to forest structure. Biogeographic region was not significantly related to richness. Interestingly, forest richness was the best predictor of trunk richness. Our results highlight the importance of seasonal distribution of rainfall and temperatures and support that the richness of bryophyte communities is constrained by mesoscale climatic factors, in particular the interplay between water and energy availability. In contrast, abundance seems to be controlled by habitat characteristics. We also detected a strong top-down structure between both scales of measurement evidencing a scaling down of the climatic effect: richness at the sample scale is controlled mainly by local richness and local richness is in turn controlled by climate, so mesoscale climatic gradients are indirectly limiting richness at the smallest scale.
Biodiversity sampling with sufficient geographical and environmental representativeness is essential for understanding biodiversity patterns and processes and designing effective conservation and management strategies. Yet, environmental representativeness is generally overlooked in existing sampling designs, and there are no user-friendly tools available for grassroots investigators that account for transportation accessibility and phased funding. We developed a novel sampling site selection approach using Marxan, by simply modifying default input datasets to include geographical and environmental data and transportation accessibility. Using the Nu-Salween River as a case study, we compiled an occurrence database of freshwater fish and then designed phased sampling strategies aiming at optimizing the coverage of species through geographical and environmental gradients, improving cost-effectiveness, and ensuring sampling feasibility compared to a random approach. Our approach was highly flexible in determining the optimal number of sampling units based on the number of surveying features (i.e., geographical units and environmental clusters). The output and visualization of the geographical locations of newly added sampling sites were adaptable to sampling strategies with phased funding. Compared to the random approach, the Marxan approach showed a two-fold increase in the mean species richness for selected sampling units and improved sampling effectiveness by 81.18 % (± 0.83 %) and cost efficiency by 81.75 % (± 0.02 %). We provide an efficient and customized sampling design that yields high geographical and environmental representativeness. It is accessible to grassroots investigators and applicable to various taxa, ecosystems, and regions, giving it the potential to significantly contribute to biodiversity inventory and conservation efforts.
