No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Prediction of butterfly diversity hotspots in Belgium: A comparison of statistically focused and land use-focused models
Abstract
Aim We evaluate differences between and the applicability of three linear predictive models to determine butterfly hotspots in Belgium for nature conservation purposes.
Location The study is carried out in Belgium for records located to Universal Transverse Mercator (UTM) grid cells of 5 × 5 km.
Methods We first determine the relationship between factors correlated to butterfly diversity by means of modified t-tests and principal components analysis; subsequently, we predict hotspots using linear models based on land use, climate and topographical variables of well-surveyed UTM grid cells (n = 197). The well-surveyed squares are divided into a training set and an evaluation set to test the model predictions. We apply three different models: (1) a ‘statistically focused’ model where variables are entered in descending order of statistical significance, (2) a ‘land use-focused’ model where land use variables known to be related to butterfly diversity are forced into the model and (3) a ‘hybrid’ model where the variables of the ‘land use-focused model’ are entered first and subsequently complemented by the remaining variables entered in descending order of statistical significance.
Results A principal components analyses reveals that climate, and to a large extent, land use are locked into topography, and that topography and climate are the variables most strongly correlated with butterfly diversity in Belgium. In the statistically focused model, biogeographical region alone explains 65% of the variability; other variables entering the statistically focused model are the area of coniferous and deciduous woodland, elevation and the number of frost days; the statistically focused model explains 77% of the variability in the training set and 66% in the evaluation set. In the land use-focused model, biogeographical region, deciduous and mixed woodland, natural grassland, heathland and bog, woodland edge, urban and agricultural area and biotope diversity are forced into the model; the land use-focused model explains 68% of the variability in the training set and 57% in the evaluation set. In the hybrid model, all variables from the land use-focused model are entered first and the covariates elevation, number of frost days and natural grassland area are added on statistical grounds; the hybrid model explains 78% of the variability in the training set and 67% in the evaluation set. Applying the different models to determine butterfly diversity hotspots resulted in the delimitation of spatially different areas.
Main conclusions The best predictions of butterfly diversity in Belgium are obtained by the hybrid model in which land use variables relevant to butterfly richness are entered first after which climatic and topographic variables were added on strictly statistical grounds. The land use-focused model does not predict butterfly diversity in a satisfactory manner. When using predictive models to determine butterfly diversity, conservation biologists need to be aware of the consequences of applying such models. Although, in conservation biology, land use-focused models are preferable to statistically focused models, one should always check whether the applied model makes sense on the ground. Predictive models can target mapping efforts towards potentially species-rich sites and permits the incorporation of un-surveyed sites into nature conservancy policies. Species richness distribution maps produced by predictive modelling should therefore be used as pro-active conservation tools.
... Four types of environmental data were collected (Table 1 - Maes et al., 2003): (1) land use variables (Corine Land Cover -CEC, 1994), (2) topographic variables (digital elevation model for Belgium with a resolution of 20 m -National Geographical Institute), (3) climate variables for the period 1996-2001 (Royal Meteorological Institute of Belgium) and (4) soil variables (Soil Service of Belgium). The area of each land use and topographic classes in each grid cell were estimated using Arc-View3.2 ...
... The relatively short and recent time period for the climatic data (1996)(1997)(1998)(1999)(2000)(2001) was preferred because a much larger number of point data was available compared to the data for a longer time period (1970-2000 for example). This allowed a more accurate interpolating of the climatic variables to the whole of Belgium (see Maes et al., 2003 for more details). The range in elevation was calculated as the difference between the highest and lowest altitude in the grid cell. ...
... Furthermore, only woodland sites in grid cells for which all 10 model runs predicted the presence of the species, are, very conservatively, proposed for incorporation into the Natura 2000 network in Belgium. Many papers predict species distribution in different resolutions (5 · 5 km, 10 · 10 km or even 50 · 50 km; e.g., Maes et al., 2003;Araú jo, 2004). Relatively large grid cells are very useful to detect general patterns and changes in distribution under, for example, climate change scenario's on a larger scale (e.g., Bakkenes et al., 2002;Harrison et al., 2006). ...
Despite its size and attractiveness, many Lucanus cervus sites remain undetected in NW Europe because of its short flight period and its nocturnal activity. Therefore, present-day designated conservation areas for L. cervus are probably insufficient for a sustainable conservation of the species. We applied eight species distribution modelling techniques (artificial neural networks, classification tree analysis, generalised additive models, generalised boosting models, generalised linear models, mixture discriminant analysis, multiple adaptive regression splines and random forests) to predict the distribution of L cervus in Belgium using 10 randomly generated calibration and evaluation sets. We used AUC, sensitivity (% correctly predicted presences in the evaluation set) and specificity (% correctly predicted absences in the evaluation set) and Kappa statistics to compare model performances. To avoid the incorporation of only marginally suitable woodland sites into the Natura 2000 network, we, conservatively, considered the species as being present only in grid cells where all 10 randomly generated model sets predicted the species as such. Model performance was, on average, good allowing to predict the potential distribution of L. cervus accurately. According to the predicted distribution using the more robust prevalence threshold, only 5731 ha (11% of the potentially suitable area) is protected under the Natura 2000 scheme in Belgium. Subsequently, we categorised the potentially suitable woodlands into three conservation priority categories based on their surface area and the already designated Natura 2000 area. Including the most suitable L. cervus woodlands previously not included in the Natura 2000 sites within such network would require protecting an area of 15,260 ha. Finally, we discuss the implications of using species distribution modelling for nature policy decisions in designating conservation networks.
... iv) Quantification of habitat quality (Dennis & Eales, 1997). v) Prediction of species distributions (Dennis & Eales, 1999; Cowley et al., 2000; Fleishman et al., 2001) as well as species richness and biodiversity hotspots (Maes et al., 2003; Hortal et al., 2004) on different spatial and temporal scales. vi) Generating habitat suitability maps, i.e. maps of predicted occurrence probabilities (Osborne et al., 2001; Austin, 2002; Joy & Death, 2004; Luoto et al., 2005;). ...
... Similarly, Krauss et al.(2004) and Pywell et al. (2004) analysed the effects of some environmental variables on butterfly abundances in intensively managed arable farmland. On a larger scale, Maes et al. (2003; ) applied multiple linear regression models to predict butterfly diversity from atlas data in Belgium, whereas Konvicka et al. (2003) analysed Czech atlas data with respect to elevational shifts in butterfly distribution due to climate change. Stefanescu et al. (2004) assessed the effect of climate, topography, vegetation structure and human disturbance on butterfly species richness in Catalonia, and Kerr et al. (2001) as well as White and Kerr (2006) applied a similar analysis with variables derived from remote sensing to predict butterfly species richness in Canada. ...
... The simplest way to deal with multicollinearity is to delete one of two correlated predictors, if the correlation coefficients exceed a certain threshold (r S > 0.7 according to Fielding & Haworth, 1995 ). Other approaches to deal with multicollinearity are either residual and sequential regression (Graham, 2003) or principal component regression as applied by Maes et al. (2003; Stefanescu et al. (2004). By conducting a principal component analysis (PCA, Legendre & Legendre, 1998) prior to model building, statistically independent linear combinations of predictors are estimated. ...
... So far, most published models for predicting species distribution correlate patterns of incidence to combinations of relatively coarse-scaled environmental variables (e.g., climate, topography, and land-use patterns). These approaches are useful for predicting general distribution patterns of species that may help conservation policy making, including reserve-site selection (Maes et al. 2003; Araújo et al. 2004; Cabeza et al. 2004). But for other conservation applications, including management strategies of individual nature reserves, the output of models based on such general environmental variables can be too coarse scaled (Guisan & Thuiller 2005). ...
... We tested the accuracy and transferability of functional habitat models that predict distributions of two regionally threatened butterfly species within and among three nature reserves. We adopted a functional resource-based concept of habitat as discussed by Dennis et al. (2003) for butterflies (see also Shreeve et al. 2004; Dennis et al. 2006). In this approach a species requires a set of specific resources and conditions to survive and reproduce . ...
... The use of functional resources may greatly increase the accuracy of habitat definitions (Dennis et al. 2003 ). Therefore , we selected for both species those habitat variables for which a functional relationship is expected based on the best available ecological knowledge (Table 2). ...
Numerous models for predicting species distribution have been developed for conservation purposes. Most of them make use of environmental data (e.g., climate, topography, land use) at a coarse grid resolution (often kilometres). Such approaches are useful for conservation policy issues including reserve-network selection. The efficiency of predictive models for species distribution is usually tested on the area for which they were developed. Although highly interesting from the point of view of conservation efficiency, transferability of such models to independent areas is still under debate. We tested the transferability of habitat-based predictive distribution models for two regionally threatened butterflies, the green hairstreak (Callophrys rubi) and the grayling (Hipparchia semele), within and among three nature reserves in northeastern Belgium. We built predictive models based on spatially detailed maps of area-wide distribution and density of ecological resources. We used resources directly related to ecological functions (host plants, nectar sources, shelter, microclimate) rather than environmental surrogate variables. We obtained models that performed well with few resource variables. All models were transferable--although to different degrees--among the independent areas within the same broad geographical region. We argue that habitat models based on essential functional resources could transfer better in space than models that use indirect environmental variables. Because functional variables can easily be interpreted and even be directly affected by terrain managers, these models can be useful tools to guide species-adapted reserve management.
... Hotspots have been defined using species richness (e.g. Myers et al., 2000;Veech, 2000;Brummitt & Nic Lughadha, 2003;Maes et al., 2003;Ovadia, 2003;Fattorini, 2006a;Guilhaumon et al., 2008;Jenkins et al., 2013), species rarity or conservation status (Prendergast et al., 1993a;Dobson et al., 1997;Troumbis & Dimitrakopoulos, 1998;Griffin, 1999;Possingham & Wilson, 2005;Funk & Fa, 2010), a combination of richness and endemism (e.g. Kier & Barthlott, 2001;Hobohm, 2003), evolutionary distinctiveness (Cadotte & Davies, 2010;Jetz et al., 2014), functional diversity (Stuart- Smith et al., 2013) and a combination of species richness, phylogenetic diversity and functional diversity (Mazel et al., 2014). ...
... When species richness is used to identify hotspots, the areas with an exceptionally large number of species are considered as hotspots. If the compared areas are of the same size, as in cases where a grid system is used, it is possible to simply rank them according to the number of species they contain and to designate as hotspots those that rank highest, that is, that contain the highest species richness (see Williams et al., 1996;Maes et al., 2003;Balletto et al., 2010). However, if the compared areas are of different size, their species richness values cannot be compared directly, because larger areas tend to have more species, a pattern known as the species-area relationship (SAR). ...
The species–area relationship (SAR) describes a range of related phenomena that are fundamental to the study of biogeography, macroecology and community ecology. While the subject of ongoing debate for a century, surprisingly, no previous book has focused specifically on the SAR. This volume addresses this shortfall by providing a synthesis of the development of SAR typologies and theory, as well as empirical research and application to biodiversity conservation problems. It also includes a compilation of recent advances in SAR research, comprising novel SAR-related theories and findings from the leading authors in the field. The chapters feature specific knowledge relating to terrestrial, marine and freshwater realms, ensuring a comprehensive volume relevant to a wide range of fields, with a mix of review and novel material and with clear recommendations for further research and application.
... In a fi rst model (hereafter called 'bird model'), we applied generalized linear regression models (GLM) using butterfl y species' presence/absence as dependent variables and the number of kilometre squares (ranging from 1-8) per grid cell occupied by the heathland bird species as predictor variables. Here, we performed full models using all typical heathland bird species to make the model ecologically more meaningful since these birds share some of their environmental space-use and ecological resources with the heathland butterfl ies (Mac Nally, 2000 ; Mac Nally and Fleishman, 2002 ; Maes et al., 2003 ). Seven bird species are considered as typical for heathlands in the sandy regions of Flanders and the Netherlands (SOVON 2002): the Meadow Pipit ( Anthus pratensis ), the Tree Pipit ( Anthus trivialis ), the Nightjar ( Caprimulgus europaeus ), the Yellowhammer ( Emberiza citrinella ), the Woodlark ( Lullula arborea ), the Curlew ( Numenius arquata ) and the Stonechat ( Saxicola torquata ). ...
... variables per grid cell (hereafter called 'biotope model'). Th e 44 land use categories distinguished on the CORINE land cover map were lumped into 13 major land use types that are present in Flanders (Maes et al., 2003 ). CORINE land cover data are fairly broad scaled biotope classifi cations (e.g., urban area, deciduous woodland, heathland etc.) and are available on a European extent (CEC, 1994 ). ...
National or regional conservation strategies are usually based on available species distribution maps. However, very few taxonomic groups achieve a full coverage of the focal region. Distribution data of well-mapped taxonomic groups could help predict the distribution of less well-mapped groups and thus fill gaps in distribution maps. Here, we predict the distribution of five heathland butterflies in Flanders (north Belgium) using typical heathland bird distribution data as predictor variables. We compare predictions with those using only biotope or a combination of both biotope and bird data as variables. In addition, we test the transferability of 'bird', biotope and combined models to the Netherlands, an ecologically similar region. Transferability was tested in three separate sandy regions in the Netherlands at different distances from the region in which the models were built. For each of the five heathland butterflies, we applied logistic regressions on ten random model sets and tested the models on ten random evaluation sets within Flanders. We used the area under the curve (AUC) of the receiver-operating characteristics (ROC) plots to estimate model accuracy. Overall, bird models performed significantly better than biotope models but were not significantly different from the combined models in Flanders. In the Netherlands, the transferred biotope and the combined models performed better than the transferred 'bird models'. We conclude that on a local scale, birds can, to some extent, serve as proxies for biotope quality, but that biotope models are more robust when transferred to another region.
... To understand the impacts of future climate change, it is imperative to predict the current and future potential distributions of a species. Species distribution models (SDM) have a broad range of applications, and have been used to identify hot-spots of endangered species (Godown and Peterson 2000) or predict biodiversity (Maes et al. 2003), prioritize areas for conservation (Chen and Peterson 2002), establish suitable locations for species translocations or cultivation (Cunningham et al. 2002), and assess magnitude of changes in the distribution of multiple species in response to climate change (Williams et al. 2003;Meynecke 2004). ...
An aquatic cash crop, ‘makhana’ (Euryale ferox Salisb.) was studied for predicted habitat suitability through ‘BioClim’ model for the year 1950–2000 and projected climate data for 2050 and 2070 for the Indian region based on primary data from survey and exploration for germplasm collection by ICAR-National Bureau of Plant Genetic Resources (ICAR-NBPGR), New Delhi, India. A total of 362 diverse ‘makhana’ or foxnut germplasm accessions were augmented mainly from different districts of Indian state of Bihar and conserved at the National Genebank (NGB) at ICAR-NBPGR, New Delhi. Predicted habitat suitability map of climate for the period 1950–2000 showed that the current locations of the highly suitable areas of cultivation of ‘makhana’ in Bihar were extended to other states namely- Jharkhand, Uttar Pradesh and West Bengal in eastern India. Out of the total highly suitable estimated areas, maximum were located in Bihar (65.0 per cent) followed by Uttar Pradesh (12.3 per cent), Jharkhand (12.2 per cent) and West Bengal (10.5 per cent). Based on future predicted climate data for the year 2050 the habitat suitability map indicated that out of the total high suitable areas, Bihar had maximum area as compared to the other three states. Predicted climate data using ‘BioClim’ variables showed that high suitable areas shifted from major part in Jharkhand, Uttar Pradesh and West Bengal and negligible areas bordering Katihar district of Bihar. While predicted climate data for the year 2050 and 2070 showed high suitable areas in West Bengal, especially Malda and Dakschin Dinajpur districts completely shrank due to temperature rise; while Uttar Dinajpur (Goal Pokhar-I, Goal Pokhar-II and Islampur districts) appeared as new areas. In Uttar Pradesh, new districts viz. Balrampur and Shravasti were identified as high suitable areas for ‘makhana’ cultivation. As ‘makhana’ is a popular cash crop of eastern part of the country, and has not been cultivated widely in other regions, hence the knowledge on its botany, distribution, cultivation practices and uses are provided.
... Overall, 5.2% of hill forest, 11.2% of deciduous Sal forest, and 23.3% of mangrove forest is protected in the country, while the percentage is minimal for other habitat types (Mukul et al., 2008). Butterflies need diverse vegetation structures, and habitat requirements vary markedly between species from different groups e some species prefer grassland, while others prefer canopy (Dennis and Shreeve, 1997;Maes et al., 2003;Chowdhury et al., 2017Chowdhury et al., , 2020. For example, Euploea crameri, the single Critically Endangered butterfly, prefers mangroves, whereas Parantica melaneus, an Endangered butterfly, prefers hilly forest areas (IUCN Bangladesh, 2015). ...
Protected areas have been established around the world to preserve samples of biodiversity from pressing threats. Yet the adequacy of protected area systems in many tropical nations is poorly understood, and assessments are usually focused on vertebrates. Here, we model the occurrence of 246 butterfly species, and determine the extent to which they occur in protected areas in Bangladesh, a country that forms part of the Indo-Burma biodiversity hotspot. We develop ecological niche models, and measure overlap with protected areas using three methods to map species distributions (habitat suitability, area of occupancy, and extent of occurrence). Suitability maps identify the north-east and south-east regions as the main centres of butterfly diversity, yet there is marked variation among families, and between non-threatened and threatened species. Using the suitability map approach, a mean of 1.27% of the geographic range of species is covered by protected areas. Only two species (Euploea crameri and Danaus melanippus) have >15% coverage, 25% of species have no coverage and 70% of species have <1% coverage. Overall, protected area coverage is slightly higher for threatened species. Tracts of suitable, but unprotected habitat still exist in the north-east and south-east regions of Bangladesh, and designation of new protected areas in these regions will strengthen butterfly conservation in the country. Enhanced management of existing protected areas, and a strategy for conserving butterflies and other insects outside protected areas will also help secure the long-term future for biodiversity in Bangladesh.
... Species distribution models have been used to assess the potential threat of pests or invasive species (Ungerer et al., 1999), to obtain insights into the biology and biogeography of species (Steinbauer et al., 2002), to identify hotspots of endangered species (Godown and Peterson, 2000) or predict biodiversity (Maes et al., 2003), to prioritize areas for conservation (Chen and Peterson, 2002) and to establish suitable locations for species translocations or cultivation (Cunningham et al., 2002). Notably, species distribution models are currently the only means to assess the potential magnitude of changes in the distributions of multiple species in response to climate change (Meynecke, 2004). ...
... Species distribution models have been used to assess the potential threat of pests or invasive species (Ungerer et al., 1999), to obtain insights into the biology and biogeography of species (Steinbauer et al., 2002), to identify hotspots of endangered species (Godown and Peterson, 2000) or predict biodiversity (Maes et al., 2003), to prioritize areas for conservation (Chen and Peterson, 2002) and to establish suitable locations for species translocations or cultivation (Cunningham et al., 2002). Notably, species distribution models are currently the only means to assess the potential magnitude of changes in the distributions of multiple species in response to climate change (Meynecke, 2004). ...
Phytophagous Fruit flies (Diptera: Tephritidae) cause heavy losses on fruits and vegetable crops, and pose a threat to the commercialisation of the horticulture industry in Uganda. In order to develop an effective management strategy against fruit flies, it is important to understand the diversity, patterns of host utilization and ecological niche of the major fruit species, which were the objectives of this study. Major differences in species richness and community structure occurred among the three major mango growing regions. The alien Bactrocera invadens was noted to be displacing native fruit fly species. Similarly, fruit infestation was predominated by B.invadens, while damage by native fruit flies was negligible. Tropical almonds showed the highest fruit fly infestation incidence (87.9%), and were mainly infested by B. invadens (82.1%). Psidium guajava and Mangifera indica were also favorable hosts. There was significant difference in infestation among mango varieties (p < 0.0001). Among the host fruit species, female B. invadens fruit flies frequently oviposited most on fruits that gave better adapted offsprings (support for Preference Performance Hypothesis-PPH), with overall coefficient of determination (R2) for infestation averaging 75.4%. However, PPH was poorly evident among the various mango varieties, with the trends suggestive of an Optimal Foraging Theory (OFT)
(oviposition on readily available fruits). B. invadens from different agro ecological zones and fruit hosts were significantly different in morphology (p < 0.0001), which suggested that geographic and host-associated adaptations could produce phenotypic variations that can lead to ecotype and host populations. Precipitation (61.41%) and temperature (29.21%) were the most important determinants of fruit fly distribution in the country. On that basis, the most suitable habitats were central and mid north zones, while the western, north-eastern areas were marginal. Future potential fruit fly habitats were projected to decline by 25.4% on average. Dacus bivittatus, Bactrocera cucurbitae and Ceratitis anonae were projected to be the least climate change resilient species. D. cilliatus (249.3%), B. invadens (-1.9%) and C. cosyra (-2.2%) were projected to be the most climate change resilient species. Future fruit fly niches were predicted to shift northwards, mainly to the northern moist farmlands. This study has provided knowledge
on several aspects of the ecology of fruit and crucial information that can help in the
development of adaptative pest management strategies in Uganda.
... Since the principal components are orthogonal (uncorrelated), there is no multicollinearity in the regression. A number of studies have adopted this application of PCA in previous studies of species occurrence, distributions and vulnerability (WallisDeVries 2014; Shreeve et al. 2001;Maes et al. 2003;Kitahara and Fujii 2005;Summerville et al. 2006;Lütolf et al. 2009;Dapporto and Dennis 2013;Keddy et al. 2002;Oyarzabal et al. 2008). The relative contribution of trait components based on life-history vs. climatic niche traits in explaining the variance of each vulnerability indicator was calculated as their summed F values relative to total F value for the continuous variables Range size and Habitat specificity and as their summed χ2 values relative to total χ2 value for the binary variables Endemicity, Red List status and Affinity for natural habitats. ...
In drawing up Red Lists, the extinction risks of butterflies and other insects are currently assessed mainly by using information on trends in distribution and abundance. Incorporating information on species traits may increase our ability to predict species responses to environmental change and, hence, their vulnerability. We summarized ecologically relevant life-history and climatic niche traits in principal components, and used these to explain the variation in five vulnerability indicators (Red List status, Endemicity, Range size, Habitat specialisation index, Affinity for natural habitats) for 397 European butterfly species out of 482 species present in Europe. We also evaluated a selection of 238 species to test whether phylogenetic correction affected these relationships. For all but the affinity for natural habitats, climatic niche traits predicted more variation in vulnerability than life-history traits; phylogenetic correction had no relevant influence on the findings. The life-history trait component reflecting mobility, development rate, and overwintering stage, proved the major non-climatic determinant of species vulnerability. We propose that this trait component offers a preferable alternative to the frequently used, but ecologically confusing generalist-specialist continuum. Our analysis contributes to the development of trait-based approaches to prioritise vulnerable species for conservation at a European scale. Further regional scale analyses are recommended to improve our understanding of the biological basis of species vulnerability.
... Thus, these techniques compensate our lack of knowledge on distribution of species richness and their ecological requirements (Scott, 1998). Moreover, such techniques are not only able to predict the distribution of species richness and determine the influence of environmental predictors on species richness, APPLIED but can also indicate where the biodiversity conservation efforts should be concentrated (Myers et al., 2000;Maes et al., 2003). Thus, spatial models have become substantially important tools for biodiversity conservation Lobo and Martín-Piera, 2002), and may have been used to guide decision makers towards the impacts of environmental modifications (Fleishman et al., 2001b). ...
Spatial distribution pattern of butterfly species richness were explored using geographically weighted regression (GWR) and ordinary least square (OLS) regression. These models were compared to assess their abilities in modelling butterfly species richness and, further the spatial variation in the relationship between butterfly species richness and environmental predictors was questioned. Data on the occurrence of butterflies from "Die Tagfalter der Türkei unter besonderer Berücksichtigung der angrenzenden Länder" (The Butterflies of Turkey with special attention to the adjacent countries) and three groups of environmental predictors (climatology, topology, and physical features) were incorporated in the analyses after eliminating highly correlated, redundant predictors. Furthermore, Monte Carlo permutation test was applied simultaneously to assess non-stationarity in the relationship between butterfly species richness and environmental predictors. The results indicated that GWR model predicted butterfly species richness better than the OLS model and also, demonstrated spatial non-stationarity in the relationship between butterfly species richness and environmental predictors. In addition, it was found that most of the variation in butterfly species richness was associated with minimum temperature in January, maximum temperature in July, diurnal range, and solar radiation. This result indicated that the distribution of butterfly species richness is mostly governed by climatic environmental predictors, particularly temperature related predictors, indicating that many butterfly species may respond to projected climate changes rapidly.
... To determine the potential distribution of the Ilex Hairstreak in Flanders, we applied three different modelling techniques using the following variables: biotope type (arable land, agricultural grasslands, dunes, semi-natural grasslands, heathland, urban area, deciduous woodland, marshes, coniferous woodland, shrub and water- Vriens et al. 2011), soil texture (Titeux et al. 2009), moisture, nitrogen deposition and biotope diversity (calculated as the Shannon-Wiener index using only '(semi-)natural' habitats (dunes, semi-natural grasslands, heathland, deciduous woodland, marshes and shrub)-cf. Maes et al. 2003). The three modelling techniques used were: Generalised Additive Models (GAM- Hastie and Tibshirani 1987), Boosting Regression Trees (GBM- Friedman et al. 2000) and Maximum Entropy (MaxEnt-Phillips and Dudik 2008; Elith et al. 2011). ...
Ecotones are often species-rich and harbour specific resources and environmental conditions for invertebrates. Despite their functional significance for conservation, they are often not explicitly included in biotope typologies relevant to conservation policy and management (e.g., the European Habitats Directive). The Ilex Hairstreak (Satyrium ilicis) is a species of European conservation concern and a typical ecotone species inhabiting gradients from open (e.g., heathland, grassland) to closed vegetation (e.g., woodland). Here, we investigated its occurrence and habitat use at different spatial scales in Flanders (north Belgium). At a regional scale, species distribution modelling predicted 1,152 grid cells of 1 × 1 km2 to be suitable of which 190 were presumed to occur within colonization capacity (±2.5 km). At a local scale, adult butterflies were more abundant on sites sheltered by bushes and small trees and with nectar sources in the vicinity of tall oak trees (mate locating sites). For egg-laying, females preferred oaks of intermediate height (50–150 cm) with many low branches at some distance from the nearest woodland edge (12 m). Additionally, Alder Buckthorn (Frangula alnus) was abundant as well as a herb layer of 10–15 cm. 73 % of the eggs were parasitized and parasitism occurred more often within sites where small oaks were very abundant. Making use of our results, we suggest conservation measures at different scales for this endangered ecotone species: policy measures at a regional level to delineate functional conservation units and local management measures using a resource-based approach.
... Even though many studies have reported statistically significant relationships between land cover data and the regional or nationwide distribution of species (Eyre et al., 2004;Luoto et al., 2006;Heikkinen et al., 2004;Maes et al., 2003;Siriwardena et al., 2000), the causal relationship between classified land cover and species distributions is often indirect. This is in particular the case for plants, for which land cover is rather a generally limiting factor without having direct physiological impact (Thuiller et al., 2004). ...
... In spite of these problems with historical data, this kind of information represents an invaluable tool for describing ecological situations and for identifying potential losers. They represent the 'what we have' tool (New 2007) and are important for guiding policy makers and future conservation studies (see also Lobo et al. 1997; Dennis et al. 1999; Maes et al. 2003). ...
In depth studies of patterns of extinction are fundamental to understand species vulnerability, in particular when population
extinctions are not driven by habitat loss, but related to subtle changes in habitat quality and are due to ‘unknown causes’.
We used a dataset containing over 160,000 non-duplicate individual records of occurrence (referred to 280 butterflies and
43 zygenid moths), and their relative extinction data, to carry out a twofold analysis. We identified ecological preferences
that influence extinction probability, and we analysed if all species were equally vulnerable to the same factors. Our analyses
revealed that extinctions were non-randomly distributed in space and time, as well as across species. Most of the extinctions
were recorded in 1901–1950 and, as expected, populations at their range edges were more prone to become extinct for non-habitat-related
causes. Ecological traits were not only unequally distributed between extinction and non-extinction events, but also not all
ecological features had the same importance in driving population vulnerability. Hygrophilous and nemoral species were the
most likely to experience population losses and the most prone to disappear even when their habitat remained apparently unchanged.
Species vulnerability depends on both ecological requirements and threat type: in fact, each species showed a distinct pattern
of vulnerability, depending on threats. We concluded that the analysis may be an important step to prevent butterfly declines:
species that are strongly suffering due to ‘unknown changes’ are in clear and urgent need of more detailed auto-ecological
studies.
... Dennis & Hardy, 1999; Guisan et al., 1999; Luoto et al., 2002). In the context of multispecies analysis, atlases and GIS have already been used to provide environmental interpretation of avifaunal zonation (Pasinelli et al., 2001), to model species richness distribution ( Lobo & Martin-Piera, 2002; Maes et al., 2003) and to relate avian (Natuhara & Imai, 1996; Storch et al., 2003) or floristic (Guisan et al., 1999 ) assemblages to environmental conditions . In order to explain relationships between species assemblages and environmental variables, ordination methods are often used, especially direct gradient analyses in which species occurrence or abundance are directly related to environmental variables (ter Braak, 1986). ...
Aim To assess the relative roles of environment and space in driving bird species distribution and to identify relevant drivers of bird assemblage composition, in the case of a fine-scale bird atlas data set. Location The study was carried out in southern Belgium using grid cells of 1 · 1 km, based on the distribution maps of the Oiseaux nicheurs de Famenne: Atlas de Lesse et Lomme which contains abundance for 103 bird species. Methods Species found in < 10% or > 90% of the atlas cells were omitted from the bird data set for the analysis. Each cell was characterized by 59 landscape metrics, quantifying its composition and spatial patterns, using a Geographical Information System. Partial canonical correspondence analysis was used to partition the variance of bird species matrix into independent components: (a) 'pure' environmental variation, (b) spatially-structured environmental variation, (c) 'pure' spatial variation and (d) unexplained, non-spatial variation. Results The variance partitioning method shows that the selected landscape metrics explain 27.5% of the variation, whilst 'pure' spatial and spatially- structured environmental variables explain only a weak percentage of the variation in the bird species matrix (2.5% and 4%, respectively). Avian community composition is primarily related to the degree of urbanization and the amount and composition of forested and open areas. These variables explain more than half of the variation for three species and over one-third of the variation for 12 species.
... Using sets of environmental variables, a large number of studies at various resolutions have produced models that attempt to predict species-richness for different groups of organisms (e.g. Luoto, Toivonen & Heikkinen, 2002;Maes et al., 2003;Taplin & Lovett, 2003;Thomson et al., 2007), including bryophytes (e.g. Hill & Dominguez Lozano, 1994;Pharo et al., 2005). ...
Knowledge of the distribution of species is fundamental to their conservation. Hence, it is important to understand the completeness of distribution inventories and the location of the main gaps. In an attempt to do this within an area of north-west England, this study builds a model which predicts bryophyte species-richness in 1220 tetrads (2 × 2 km squares) based upon their environmental characteristics, then compares predictions with current knowledge. The number of raindays (?1 mm rain) each year was clearly the most important predictor of species-richness, followed by total nitrogen deposition and percentage cover of broad-leaved woodland. The main gaps in the current inventories are located and an assessment is made of their overall completeness. Suggestions for improvements in predicting bryophyte species-richness at tetrad resolution are provided.
... For instance, the Mediterranean Basin hotspot is widespread across 2 million km 2 and encompasses 20 countries around the Mediterranean Sea (M´edail & Myers, 2004). Consequently, diversity hotspots have also been defined at more local scales to take into account particular cases of endemism in focal groups (Olson & Dinerstein, 1998;Maes et al., 2003;Brehm et al., 2005). ...
Although the designation of biodiversity hotspots is a valuable tool to improve conservation efforts, this is a concept primarily based on species diversity. In consequence, another component of biodiversity, adaptive variation, is often ignored in conservation and to my knowledge no attempt has been made to identify hotspots of remarkable intraspecific patterns. My aim was to focus on the process of facultative paedomorphosis (i.e. the retention of larval traits such as gills in adult variants), a rare developmental pathway. One hundred and seventy-four ponds were inventoried in Larzac (France) to determine the distribution and abundance of paedomorphic palmate newts Triturus helveticus (Amphibia, Caudata) and to compare these results with the current distribution of paedomorphs in this and other species. During this study, paedomorphic newts were found in 46 ponds, 32 of which were described here for the first time. Seventy-nine per cent of known paedomorphic populations of this species were found there, whereas this area covers only 0.5% of the distribution area of the species. This represents the highest known density of facultatively paedomorphic populations, all species being considered. Because these populations face a high threat of disappearance, Larzac should be designated as an intraspecific biodiversity hotspot in order to protect adaptive intraspecific variation. Future conservation-oriented work should focus not only on species distributions but also on phenotypically diverse but spatially localized variation.
... Thus, species richness has become one of the most commonly used parameters in hotspot identification (e.g. Prendergast et al., 1993;Myers et al., 2000;Veech, 2000;Brummitt & Nic Lughadha, 2003;Hobohm, 2003;Maes et al., 2003;Ovadia, 2003). Both habitat loss (Myers et al., 2000) and human demography (Cincotta, Wisnewski & Engelman, 2000) have been proposed as measures of threat. ...
Selection of priority areas in conservation biology should incorporate an evaluation of the contribution of imperilled species to the total species richness. To rank areas according to their conservation value, a new index – termed biodiversity conservation concern (BCC) – is proposed. This index combines the conservation status of each species belonging to a given species assemblage with the total species richness. In the BCC index, species are classified into categories of endangerment and weighted by the respective risk of extinction. The new method is applied here to the butterflies of the Aegean Islands (Greece) using different (national and international) red lists. Results were consistent with both classifications based on multivariate analyses and findings from other researches. The BCC index has two important features: it assigns a biodiversity value to an area based on many species, and it facilitates multi-species assessments of ecological effects.
... As is well known, the eutrophication of habitats and thus the use of conventional fertilizers in general negatively influence biodiversity (Oostermeijer & van Swaay, 1998;). Therefore, the intensive agrarian land-use patterns of modern agriculture are believed to be one of the most important reasons for the decrease of butterfly biodiversity in many regions of Europe (van Swaay & Warren, 1999; Asher et al., 2001; Maes et al., 2003; Maes & van Dyck, 2004) and also could in the future have a severe impact on Romania's biodiversity. The other extreme represents the complete abandonment of little-productive areas. ...
Global biodiversity is decreasing as a result of human activities. In many parts of the world, this decrease is due to the destruction of natural habitats. The European perspective is different. Here, traditional agricultural landscapes developed into species-rich habitats. However, the European biodiversity heritage is strongly endangered. One of the countries where this biodiversity is best preserved is Romania. We analyse the possible changes in Romania's land-use patterns and their possible benefits and hazards with respect to biodiversity. As model group, we used butterflies, whose habitat requirements are well understood. We determined the ecological importance of different land-use types for the conservation of butterflies, underlining the special importance of Romania's semi-natural grasslands for nature conservation. We found that increasing modern agriculture and abandonment of less productive sites both affect biodiversity negatively — the former immediately and the latter after a lag phase of several years. These perspectives are discussed in the light of the integration of Romania into the European Union.
... Belgian climate point data for monthly rainfall (n = 186) and temperature (n = 114) were made available by the Royal Meteorological Institute of Belgium for the period 1996-2001. They were interpolated to the 5-km resolution squares (n = 1374) by universal kriging (Isaaks & Srivastava, 1989) with a linear drift (Maes et al., 2003). ...
Aim To analyse the effect of the inclusion of soil and land‐cover data on the performance of bioclimatic envelope models for the regional‐scale prediction of butterfly (Rhopalocera) and grasshopper (Orthoptera) distributions.
Location Temperate Europe (Belgium).
Methods Distributional data were extracted from butterfly and grasshopper atlases at a resolution of 5 km for the period 1991–2006 in Belgium. For each group separately, the well‐surveyed squares ( n = 366 for butterflies and n = 322 for grasshoppers) were identified using an environmental stratification design and were randomly divided into calibration (70%) and evaluation (30%) datasets. Generalized additive models were applied to the calibration dataset to estimate occurrence probabilities for 63 butterfly and 33 grasshopper species, as a function of: (1) climate, (2) climate and land‐cover, (3) climate and soil, and (4) climate, land‐cover and soil variables. Models were evaluated as: (1) the amount of explained deviance in the calibration dataset, (2) Akaike’s information criterion, and (3) the number of omission and commission errors in the evaluation dataset.
Results Information on broad land‐cover classes or predominant soil types led to similar improvements in the performance relative to the climate‐only models for both taxonomic groups. In addition, the joint inclusion of land‐cover and soil variables in the models provided predictions that fitted more closely to the species distributions than the predictions obtained from bioclimatic models incorporating only land‐cover or only soil variables. The combined models exhibited higher discrimination ability between the presence and absence of species in the evaluation dataset.
Main conclusions These results draw attention to the importance of soil data for species distribution models at regional scales of analysis. The combined inclusion of land‐cover and soil data in the models makes it possible to identify areas with suitable climatic conditions but unsuitable combinations of vegetation and soil types. While contingent on the species, the results indicate the need to consider soil information in regional‐scale species–climate impact models, particularly when predicting future range shifts of species under climate change.
... It is reassuring, however, that the predictor coefficients (see Supplementary material, Appendix SI) are realistic representations of the habitat requirements of this target species. Presumably this is because the variables selected for modelling are 'direct predictors' (i.e. are thought to correlate mechanistically with the response variables (Guisan & Zimmermann 2000; Augustin et al. 2001; Austin 2002; Maes et al. 2003; Randin et al. 2006). Habitat-based models were able to predict patterns of habitat availability in both training and testing landscapes. ...
Large‐scale conservation planning requires the identification of priority areas in which species have a high likelihood of long‐term persistence. This typically requires high spatial resolution data on species and their habitat. Such data are rarely available at a large geographical scale, so distribution modelling is often required to identify the locations of priority areas. However, distribution modelling may be difficult when a species is either not recorded, or not present, at many of the locations that are actually suitable for it. This is an inherent problem for species that exhibit metapopulation dynamics.
Rather than basing species distribution models on species locations, we investigated the consequences of predicting the distribution of suitable habitat, and thus inferring species presence/absence. We used habitat surveys to define a vegetation category which is suitable for a threatened species that has spatially dynamic populations (the butterfly Euphydryas aurinia ), and used this as the response variable in distribution models. Thus, we developed a practical strategy to obtain high resolution (1 ha) large scale conservation solutions for E. aurinia in Wales, UK.
Habitat‐based distribution models had high discriminatory power. They could generalize over a large spatial extent and on average predicted 86% of the current distribution of E. aurinia in Wales. Models based on species locations had lower discriminatory power and were poorer at generalizing throughout Wales.
Surfaces depicting the connectivity of each grid cell were calculated for the predicted distribution of E. aurinia habitat. Connectivity surfaces provided a distance‐weighted measure of the concentration of habitat in the surrounding landscape, and helped identify areas where the persistence of E. aurinia populations is expected to be highest. These identified successfully known areas of high conservation priority for E. aurinia . These connectivity surfaces allow conservation planning to take into account long‐term spatial population dynamics, which would be impossible without being able to predict the species’ distribution over a large spatial extent.
Synthesis and applications . Where species location data are unsuitable for building high resolution predictive habitat distribution models, habitat data of sufficient quality can be easier to collect. We show that they can perform as well as or better than species data as a response variable. When coupled with a technique to translate distribution model predictions into landscape priority (such as connectivity calculations), we believe this approach will be a powerful tool for large‐scale conservation planning.
... One of the potentially important factors is land cover, although its effects have only sporadically been considered in bioclimatic models (but seeDirnböck et al ., 2003;Pearson et al ., 2004;Thuiller et al ., 2004a). The scarcity of studies employing both climatic and land cover predictors in bioclimatic modelling is surprising, because many papers have reported statistically significant relationships between land cover variables and the regional or nationwide distribution and richness patterns of species (e.g.Fuller et al ., 1997;Siriwardena et al ., 2000;Maes et al ., 2003;Eyre et al ., 2004;Heikkinen et al ., 2004;Fuller et al ., 2005;Luoto et al ., 2006). The problems and impacts of scale have long been a central issue in ecological (Wiens et al ., 1987;Wiens, 1989;Levin, 1992;Palmer & White, 1994) and biogeographical studies (Rahbek & Graves, 2001;Blackburn & Gaston, 2002;Willis & Whittaker, 2002). ...
Aim We explored the importance of climate and land cover in bird species distribution models on multiple spatial scales. In particular, we tested whether the integration of land cover data improves the performance of pure bioclimatic models.
Location Finland, northern Europe.
Methods The data of the bird atlas survey carried out in 1986–89 using a 10 × 10 km uniform grid system in Finland were employed in the analyses. Land cover and climatic variables were compiled using the same grid system. The dependent and explanatory variables were resampled to 20‐km, 40‐km and 80‐km resolutions. Generalized additive models (GAM) were constructed for each of the 88 land bird species studied in order to estimate the probability of occurrence as a function of (1) climate and (2) climate and land cover variables. Model accuracy was measured by a cross‐validation approach using the area under the curve (AUC) of a receiver operating characteristic (ROC) plot.
Results In general, the accuracies of the 88 bird–climate models were good at all studied resolutions. However, the inclusion of land cover increased the performance of 79 and 78 of the 88 bioclimatic models at 10‐km and 20‐km resolutions, respectively. There was no significant improvement at the 40‐km resolution. In contrast to the finer resolutions, the inclusion of land cover variables decreased the modelling accuracy at 80km resolution.
Main conclusions Our results suggest that the determinants of bird species distributions are hierarchically structured: climatic variables are large‐scale determinants, followed by land cover at finer resolutions. The majority of the land bird species in Finland are rather clearly correlated with climate, and bioclimate envelope models can provide useful tools for identifying the relationships between these species and the environment at resolutions ranging from 10 km to 80 km. However, the notable contribution of land cover to the accuracy of bioclimatic models at 10–20‐km resolutions indicates that the integration of climate and land cover information can improve our understanding and model predictions of biogeographical patterns under global change.
... Dennis & Hardy, 1999;Guisan et al., 1999;Luoto et al., 2002). In the context of multispecies analysis, atlases and GIS have already been used to provide environmental interpretation of avifa- unal zonation ( Pasinelli et al., 2001), to model species richness distribution ( Lobo & Martin-Piera, 2002;Maes et al., 2003) and to relate avian (Natuhara & Imai, 1996;Storch et al., 2003) or floristic ( Guisan et al., 1999) assemblages to environmental condi- tions. In order to explain relationships between species assemblages and environmental variables, ordination methods are often used, especially direct gradient analyses in which species occurrence or abundance are directly related to environmental variables (ter Braak, 1986). ...
Aim To assess the relative roles of environment and space in driving bird species distribution and to identify relevant drivers of bird assemblage composition, in the case of a fine‐scale bird atlas data set.
Location The study was carried out in southern Belgium using grid cells of 1 × 1 km, based on the distribution maps of the Oiseaux nicheurs de Famenne: Atlas de Lesse et Lomme which contains abundance for 103 bird species.
Methods Species found in < 10% or > 90% of the atlas cells were omitted from the bird data set for the analysis. Each cell was characterized by 59 landscape metrics, quantifying its composition and spatial patterns, using a Geographical Information System. Partial canonical correspondence analysis was used to partition the variance of bird species matrix into independent components: (a) ‘pure’ environmental variation, (b) spatially‐structured environmental variation, (c) ‘pure’ spatial variation and (d) unexplained, non‐spatial variation.
Results The variance partitioning method shows that the selected landscape metrics explain 27.5% of the variation, whilst ‘pure’ spatial and spatially‐structured environmental variables explain only a weak percentage of the variation in the bird species matrix (2.5% and 4%, respectively). Avian community composition is primarily related to the degree of urbanization and the amount and composition of forested and open areas. These variables explain more than half of the variation for three species and over one‐third of the variation for 12 species.
Main conclusions The results seem to indicate that the majority of explained variation in species assemblages is attributable to local environmental factors. At such a fine spatial resolution, however, the method does not seem to be appropriated for detecting and extracting the spatial variation of assemblages. Consequently, the large amount of unexplained variation is probably because of missing spatial structures and ‘noise’ in species abundance data. Furthermore, it is possible that other relevant environmental factors, that were not taken into account in this study and which may operate at different spatial scales, can drive bird assemblage structure. As a large proportion of ecological variation can be shared by environment and space, the applied partitioning method was found to be useful when analysing multispecific atlas data, but it needs improvement to factor out all‐scale spatial components of this variation (the source of ‘false correlation’) and to bring out the ‘pure’ environmental variation for ecological interpretation.
... Here, taxon-specific habitat-suitability models can provide a basis for optimizing the selection of potential hotspots for restoration (e.g. Maes et al. 2003). Although richness hotspots capture more species, our study suggests that focussing conservation efforts on rarity hotspots might be preferred as these sites seem to be currently the most critically threatened by land-use changes. ...
The success of the hotspot approach for biodiversity conservation depends on the spatial scale and the indicator species used.
In this study, we investigated grasshopper species richness in Switzerland at a 1ha resolution including a total of 111 species.
We compared the representativeness of common and of endangered grasshopper species for the overall grasshopper species richness
and we assessed the efficiency of the hotspot approach for grasshopper conservation. The pattern of overall grasshopper species
richness was well represented by both the number of common and the number of endangered grasshopper species. For evaluating
the efficiency of different hotspot approaches for conservation, we compared hotspots of common species, hotspots of endangered
species (rarity hotspots), and hotspots of all grasshopper species (richness hotspots). Among these hotspot types, richness
hotspots not only featured most common grasshopper species, but they even contained more endangered species than the rarity
hotspots. The combination of rarity hotspots and hotspots of common species featured more species than the other combinations
of hotspot types. However, the gain of combining two hotspot types compared to the single-hotspot approach was low (max. 3
species). About 24% of the species were not contained in any of the hotspots. These grasshopper species require species-specific
action plans. As rarity hotspots were located in areas that are rather strongly affected by landscape change, species richness
in rarity hotspots may decrease in the future. We conclude that, for grasshoppers, the hotspot approach on the 1ha scale
can be an effective way to conserve a high proportion of species richness.
... After appropriate validation, the model is used to obtain predictions of local species richness, which are less biased owing to differences in mapping intensity and incomplete survey coverage. This method was found to be successful in predicting species richness at different scales for a variety of taxonomic groups: terrestrial vertebrates in American national parks (Edwards et al. 1996); mammals in the North American continent (Badgley and Fox 2000); butterflies in countries like France (Dennis et al. 2002; Dennis and Shreeve 2003), Belgium (Maes et al. 2003) or in the Great Basin (Mac Nally et al. 2003). However, these studies were mostly focused on single taxonomic groups and analysing the degree of species richness coincidence among taxonomic groups has so far only been carried out with uncorrected and biased data (Maddock and Du Plessis 1999). ...
The present-day geographic distribution of individual species of five taxonomic groups (plants, dragonflies, butterflies, herpetofauna and breeding birds) is relatively well-known on a small scale (5 5 km squares) in Flanders (north Belgium). These data allow identification of areas with a high diversity within each of the species groups. However, differences in mapping intensity and coverage hamper straightforward comparisons of species-rich areas among the taxonomic groups. To overcome this problem, we modelled the species richness of each taxonomic group separately using various environmental characteristics as predictor variables (area of different land use types, biotope diversity, topographic and climatic features). We applied forward stepwise multiple regression to build the models, using a subset of well-surveyed squares. A separate set of equally well-surveyed squares was used to test the predictions of the models. The coincidence of geographic areas with high predicted species richness was remarkably high among the four faunal groups, but much lower between plants and each of the four faunal groups. Thus, the four investigated faunal groups can be used as relatively good indicator taxa for one another in Flanders, at least for their within-group species diversity. A mean predicted species diversity per mapping square was also estimated by averaging the standardised predicted species richness over the five taxonomic groups, to locate the regions that were predicted as being the most species-rich for all five investigated taxonomic groups together. Finally, the applicability of predictive modelling in nature conservation policy both in Flanders and in other regions is discussed.
... respectively). Although such datasets are a vital tool to assess species distributions and changes in population trends, which are pre-requisites to conservation planning (Sahlén et al. 2004), they generally suffer from coverage and recorder bias (Dennis and Shreeve 2003;Maes et al. 2003;Dennis et al. 2006). In this study, we show that this bias towards adult records in odonate datasets risks leading to misinterpretations of the environmental quality of sites. ...
Odonate populations and species numbers are declining globally. Successful conservation requires sound assessments of both
odonate distributions and habitat requirements. Odonates have aquatic (larval) and terrestrial (adult) stages, but most surveys
that are used to inform conservation managers are undertaken of the adult stage. This study investigates whether this bias
towards adult records in odonate recording is misinterpreting the environmental quality of sites. The habitat focus is farmland
ponds, a key feature of agricultural landscapes. We tested whether or not, adult, larval and exuvial surveys lead to similar
conclusions on species richness and hence on pond quality. Results showed that pond surveys based upon larvae and exuviae
are equally suitable for the reliable assessment of presence/absence of odonates, but that adult surveys are not interchangeable
with surveys of larvae/exuviae. Larvae were also found at ponds with no emerging individuals due to changes in habitat quality,
therefore presence of exuviae remains the only proof of life-cycle completion at a site. Ovipositing females were recorded
at all ponds where exuviae were totally absent hence adult surveys over-estimate pond quality and low-quality ponds are functioning
as ecological traps. Highly mobile and generalist species were recorded at more locations than other species. Adult surveys
also bias recording towards genera, species and populations with non-territorial mate-location strategies. Odonate biodiversity
monitoring would benefit from applying the best survey method (exuviae) to avoid wasting valuable financial resources while
providing unbiased data, necessary to achieve conservation objectives.
KeywordsOdonata-Survey methods-Exuviae-Ecological traps-Conservation
... Even though many studies have reported statistically significant relationships between land cover data and the regional or nationwide distribution of species (Eyre et al., 2004;Luoto et al., 2006;Heikkinen et al., 2004;Maes et al., 2003;Siriwardena et al., 2000), the causal relationship between classified land cover and species distributions is often indirect. This is in particular the case for plants, for which land cover is rather a generally limiting factor without having direct physiological impact (Thuiller et al., 2004). ...
In times of global change and rapid biodiversity loss, making spatially explicit assessments of species’ distributions is a core challenge in conservation planning. Thus, field surveys are essential to either collect target species or to conduct biological inventories in a given area. For sustainable management, large areas need to be covered - too costly in terms of time, effort and money to be done with field work alone. In the last two decades, correlative Species Distribution Models (SDMs) - that statistically link species records or abundances to environmental data - and remote sensing (RS) have become standard tools to interpolate between scattered field data using environmental information. Several SDM studies used land cover data, one of the standard RS products, to act as proxy for habitat availability. However, the combination of both techniques is still under-utilized – due to their different scientific background and remaining skepticism on both sides. This study summarizes several examples where multi-temporal RS imagery (representing differences in the vegetation seasonality) was directly integrated into SDMs (Maxent). As mentioned above, scale is a key factor in biodiversity assessments. When predicting species’ presence with remote sensing data as a mixed signal of the (plant) species itself and its environment we observe a continuous transition from “mapping” to “modeling”. What do we detect at which cell size? – This is the core question to be discussed.
... The selection of habitats for restoration could be optimized by having experts assess habitats or by applying taxon-specific habitat-suitability models (e.g. Maes et al., 2003). ...
Land-use change is a major driver for changes in biodiversity. In this study, we investigated how the objectives of two conservation strategies (large-scale conservation of species richness versus conservation of diversity hotspots) can be achieved for grasshopper diversity under different scenarios of environmental change (land-use and climate change).Based on surveys of 95 grasshopper species from 2001 to 2004 recorded by the Swiss Centre for Faunal Cartography, we modelled the actual richness pattern as a function of different environmental variables. The impact of potential future environmental change on species richness was evaluated by applying four land-use scenarios (‘liberalization’, ‘business as usual’, ‘lowered agricultural production’, and ‘complete conversion of intensive open land’) and one climate change scenario. The effects of the scenarios were assessed at the national scale, as well as on small-scale hotspots.Environmental change has considerable effect on grasshopper species richness. At the national scale, the number of grasshopper species decreased under the ‘liberalization’ scenario (−0.24 species per 1 ha pixel) and increased under a climate change scenario (+0.63 species per 1 ha pixel). For most environmental change scenarios, species richness in small-scale hotspots was more negatively affected than on average on the national scale. The response of species richness to the scenarios did not differ significantly between hotspots of endangered and the hotspots of common grasshopper species.We conclude that conservation efforts at the national scale and small-scale hotspot conservation programs should be combined to conserve species richness most successfully. For the long-term conservation of species richness, common species as well as the combined effects of land-use and climate change have to be considered.
... To understand the impacts of future climate change, it is imperative that we can confidently predict the current and future potential distributions of species. Species distribution models have a broad range of applications, and have been used to assess the potential threat of pests or invasive species (Ungerer et al., 1999;Sutherst et al., 2000), to obtain insights into the biology and biogeography of species Steinbauer et al., 2002), to identify hotspots of endangered species (Godown and Peterson, 2000) or predict biodiversity (Maes et al., 2003), to prioritise areas for conservation (Chen and Peterson, 2002), and to establish suitable locations for species translocations or cultivation (Jovanovic et al., 2000;Cunningham et al., 2002). Importantly, species distribution models are currently the only means by which we can assess the potential magnitude of changes in the distributions of multiple species in response to climate change (e.g. ...
Bioclimatic models are widely used tools for assessing potential responses of species to climate change. One commonly used model is BIOCLIM, which summarises up to 35 climatic parameters throughout a species’ known range, and assesses the climatic suitability of habitat under current and future climate scenarios. A criticism of BIOCLIM is that the use of all 35 parameters may lead to over-fitting of the model, which in turn may result in misrepresentations of species’ potential ranges and to the loss of biological reality. In this study, we investigated how different methods of combining climatic parameters in BIOCLIM influenced predictions of the current distributions of 25 Australian butterflies species. Distributions were modeled using three previously used methods of selecting climatic parameters: (i) the full set of 35 parameters, (ii) a customised selection of the most relevant parameters for individual species based on analysing histograms produced by BIOCLIM, which show the values for each parameter at all of the focal species known locations, and (iii) a subset of 8 parameters that may generally influence the distributions of butterflies. We also modeled distributions based on random selections of parameters. Further, we assessed the extent to which parameter choice influenced predictions of the magnitude and direction of range changes under two climate change scenarios for 2020. We found that the size of predicted distributions was negatively correlated with the number of parameters incorporated in the model, with progressive addition of parameters resulting in progressively narrower potential distributions. There was also redundancy amongst some parameters; distributions produced using all 35 parameters were on average half the size of distributions produced using only 6 parameters. The selection of parameters via histogram analysis was influenced, to an extent, by the number of location records for the focal species. Further, species inhabiting different biogeographical zones may have different sets of climatic parameters limiting their distributions; hence, the appropriateness of applying the same subset of parameters to all species may be reduced under these situations. Under future climates, most species were predicted to suffer range reductions regardless of the scenario used and the method of parameter selection. Although the size of predicted distributions varied considerably depending on the method of selecting parameters, there were no significant differences in the proportional change in range size between the three methods: under the worst-case scenario, species’ distributions decrease by an average of 12.6, 11.4, and 15.7%, using all parameters, the ‘customised set’, and the ‘general set’ of parameters, respectively. However, depending on which method of selecting parameters was used, the direction of change was reversed for two species under the worst-case climate change scenario, and for six species under the best-case scenario (out of a total of 25 species). These results suggest that when averaged over multiple species, the proportional loss or gain of climatically suitable habitat is relatively insensitive to the number of parameters used to predict distributions with BIOCLIM. However, when measuring the response of specific species or the actual size of distributions, the number of parameters is likely to be critical.
... The study area is located on a Mediterranean mountainous area, which also includes alluvial planes. Several studies have focused on the relationships between avian diversity and vegetation variables (Santos et al. 2002; Sandström et al. 2006) and many others have used geostatistical approaches to better understand bird distribution (Maes et al. 2003; Couteron and Ollier 2005). But only few studies have focused specifically on how priority bird species is spatially distributed over a protected area. ...
The effects of habitat structure and spatial variation on bird community composition were studied in the protected area “Antichasia-Meteora mountains” (GR1440003), 82.635 km2. The census of bird diversity was conducted from late April until mid June, and in October 2008. In 185 randomly selected sampling plots, the following parameters were recorded: (i) habitat type; (ii) habitat cover variables and (iii) spatial variables. Stepwise regression analysis was used to investigate the relationships between bird species richness and vegetation. Geostatistics were used to examine the spatial distribution of priority species by semivariography followed by kriging interpolation. The results showed that bird species richness is correlated significantly with fallow land (β=0.15, P<0.05), and also the presence of tall shrubs (>50cm tall) (P<0.001). Bird abundance is correlated significantly only with habitat type, specifically with fallow land (β=0.30, P<0.05) and farmland (β=0.24, P<0.05). Geographical location is associated with the presence of priority bird species.
... However, data on the composition and spatial distribution of biodiversity is largely insufficient, especially for the worldwide hotspots of biodiversity ( Myers et al., 2000). Given the present constraints in time and money for biodiversity assessments, an efficient tool is needed for identifying hotspot areas and high diversity loss areas (Luoto, Toivonen and Heikkinen, 2002b;Maes et al., 2003;Lobo, Jay-Robert and Lumaret, 2004). Predictive modelling combined with Geographical Information Systems (GIS) allows the development of more robust and reliable models, relating biological diversity with environmental factors ( Brito and Crespo, 2002;Soares and Brito, 2007;Martínez-Freiría et al., 2008). ...
The biogeographic patterns in species density of herptiles were analysed in the Iberian Peninsula. Geoclimatic regions were identified using a PCA. Individual habitat suitability (HS) models for 23 amphibians and 35 reptiles at 10 x 10 km scale were calculated with ENFA, using 12 environmental factors established with Remote Sensing (RS) techniques. The species presence proportion in each geoclimatic region was calculated through a cross-tabulation between each potential occurrence model and the geoclimatic regions. Species chorotypes were determined through Hierarchical Cluster Analysis using Jaccard's index as association measure and by the analysis of marginality and tolerance factors from individual HS models. Predicted species density maps were calculated for each geoclimatic region. Probable under-sampled areas were estimated through differences between the predicted species density maps and observed (Gap analysis). The selected PCA components divided the Iberian Peninsula in two major geoclimatic regions largely corresponding to the Atlantic and Mediterranean climates. The Jaccard's index clustered herptiles in two main taxonomic groups, with distribution similar to the Atlantic and Mediterranean geoclimatic regions (7 amphibian + 13 reptile species in three Atlantic subgroups and 16 amphibian + 22 reptile species in four Mediterranean subgroups). Marginality and tolerance factor scores identified species groups of herptile specialists and generalists. The highest observed and predicted species density areas were broadly located in identical regions. Predicted gaps are located in north-western, north-east and central Iberia. RS is a useful tool for biogeographical studies, as it provides consistent environmental data from large areas with high accuracy.
... Butterfly–environment studies have often relied on traditional regression methods: relationships between species richness or abundance of butterflies and the environment have commonly been examined with (stepwise ) regression techniques (e.g. Clausen et al. 2001; Krauss et al. 2003; Maes et al. 2003), and single species occupancy patterns using (stepwise) logistic-regression (e.g. Hill et al. 1996; Dennis & Eales 1997 Cowley et al. 2000; Thomas et al. 2001; Fleishman et al. 2002; WallisDeVries 2004). ...
Variation partitioning and hierarchical partitioning are novel statistical approaches that provide deeper understanding of the importance of different explanatory variables for biodiversity patterns than traditional regression methods. Using these methods, the variation in occupancy and abundance of the clouded apollo butterfly (Parnassius mnemosyne L.) was decomposed into independent and joint effects of larval and adult food resources, microclimate and habitat quantity. The independent effect of habitat quantity variables (habitat area and connectivity) captured the largest fraction of the variation in the clouded apollo patterns, but habitat connectivity had a major contribution only for occupancy data. The independent effects of resources and microclimate were higher on butterfly abundance than on occupancy. However, a considerable amount of variation in the butterfly patterns was accounted for by the joint effects of predictors and may thus be causally related to two or all three groups of variables. Abundance of the butterfly in the surroundings of the focal grid cell had a significant effect in all analyses, independently of the effects of other predictors. Our results encourage wider applications of partitioning methods in biodiversity studies.
We present and evaluate a quantitative method for delineation of ecophysiographic regions throughout the entire terrestrial landmass. The method uses the new pattern-based segmentation technique which attempts to emulate the qualitative, weight-of-evidence approach to a delineation of ecoregions in a computer code. An ecophysiographic region is characterized by homogeneous physiography defined by the cohesiveness of patterns of four variables: land cover, soils, landforms, and climatic patterns. Homogeneous physiography is a necessary but not sufficient condition for a region to be an ecoregion, thus machine delineation of ecophysiographic regions is the first, important step toward global ecoregionalization. In this paper, we focus on the first-order approximation of the proposed method – delineation on the basis of the patterns of the land cover alone. We justify this approximation by the existence of significant spatial associations between various physiographic variables. Resulting ecophysiographic regionalization (ECOR) is shown to be more physiographically homogeneous than existing global ecoregionalizations (Terrestrial Ecoregions of the World (TEW) and Bailey's Ecoregions of the Continents (BEC)). The presented quantitative method has an advantage of being transparent and objective. It can be verified, easily updated, modified and customized for specific applications. Each region in ECOR contains detailed, SQL-searchable information about physiographic patterns within it. It also has a computer-generated label. To give a sense of how ECOR compares to TEW and, in the U.S., to EPA Level III ecoregions, we contrast these different delineations using two specific sites as examples. We conclude that ECOR yields regionalization somewhat similar to EPA level III ecoregions, but for the entire world, and by automatic means.
Spatial distribution pattern of butterfly species richness were explored using geographically weighted regression (GWR) and ordinary least square (OLS) regression. These models were compared to assess their abilities in modelling butterfly species richness and, further the spatial variation in the relationship between butterfly species richness and environmental predictors was questioned. Data on the occurrence of butterflies from "Die Tagfalter der Turkei unter besonderer Berucksichtigung der angrenzenden Lander" (The Butterflies of Turkey with special attention to the adjacent countries) and three groups of environmental predictors (climatology, topology, and physical features) were incorporated in the analyses after eliminating highly correlated, redundant predictors. Furthermore, Monte Carlo permutation test was applied simultaneously to assess non-stationarity in the relationship between butterfly species richness and environmental predictors. The results indicated that GWR model predicted butterfly species richness better than the OLS model and also, demonstrated spatial non-stationarity in the relationship between butterfly species richness and environmental predictors. In addition, it was found that most of the variation in butterfly species richness was associated with minimum temperature in January, maximum temperature in July, diurnal range, and solar radiation. This result indicated that the distribution of butterfly species richness is mostly governed by climatic environmental predictors, particularly temperature related predictors, indicating that many butterfly species may respond to projected climate changes rapidly.
This is a Short Note without an abstract.
Since lichens are widely known for their high sensitivity towards environmental disturbances, both natural and human origin. Therefore, environmental changes result in alteration of habitats and ecosystems at local, regional as well as global scale resulting in loss of lichen biodiversity; extinction of sensitive species invasion of thermophilic species towards higher latitudes. Such changes can be best monitored by lichens as biomonitors. Lichen biomonitoring is not only suitable for monitoring levels of pollutants but also may be utilised as an effective ‘early alarms’ of climate change and spatio-temporal extent of pollutants along with its health impact. Lichens as indicators possess an undeniable appeal for conservationists and land managers as they provide a cost- and time-efficient means to assess the impact of environmental disturbances on an ecosystem. Information collected with different aims, such as air pollution, climate change, biodiversity and forest continuity studies, may be utilised for conservation purposes. This chapter discusses the need and utility of indicator species especially lichen biomonitoring data in sustainable forest management and conservation.
Van bij de start van de Z.W.V.V.K., begin 2000, ontstond het idee om een uitgebreide inventarisatie
van de dagvlinders te starten over de hele provincie. Ons eerste objectief was het verkrijgen van een gedetailleerd beeld over de huidige verspreiding. Het tweede en ultieme doel was het voortdurend stimuleren van vlindervriendelijk beheer. Zo beogen we onze resterende diversiteit aan dagvlinders te behouden en waar mogelijk te verstevigen. Dit boek is het resultaat van zeven jaar intensieve inventarisatie, studie en uitgebreide samenwerking met veel enthousiaste vrijwilligers. Zonder hen zou dit nooit mogelijk geweest zijn.
We hopen dat dit boek voor de lezer aan deze twee doelen van het Z.W.V.V.K.-project kan beantwoorden. Enerzijds geven we een actueel beeld over de verspreiding van de dagvlinders in West-Vlaanderen. Daarop aansluitend leveren we ook een bijdrage aan de toenemende interesse voor vlindervriendelijk beheer zodat alle dagvlinders in onze provincie zich hier voor altijd kunnen thuisvoelen.
Spatial environmental heterogeneity (EH) is an important driver of species diversity, and its influence on species richness has been analysed for numerous taxa, in diverse ecological settings, and over a large range of spatial scales. The variety and ambiguity of concepts and terminology, however, have hampered comparisons among studies. Based on a systematic literature survey of 192 studies including 1148 data points, we provide an overview of terms and measures related to EH, and the mechanisms that relate EH to species richness of plants and animals in terrestrial systems. We identify 165 different measures used to quantify EH, referred to by more than 350 measure names. We classify these measures according to their calculation method and subject area, finding that most studies have analysed heterogeneity in land cover, topography, and vegetation, whereas comparatively few studies have focused on climatic or soil EH. Overall, elevation range emerged as the most frequent measure in our dataset. We find that there is no consensus in the literature about terms (such as ‘habitat diversity’ or ‘habitat complexity’), their meanings and associated quantification methods. More than 100 different terms have been used to denote EH, with largely imprecise delimitations. We reveal trends in use of terms and quantification with respect to spatial scales, study taxa, and locations. Finally, we discuss mechanisms involved in EH–richness relationships, differentiating between effects on species coexistence, persistence, and diversification. This review aims at guiding researchers in their selection of heterogeneity measures. At the same time, it shows the need for precise terminology and avoidance of ambiguous synonyms to enhance understanding and foster among-study comparisons and synthesis.
The unambiguous recognition of a species’ habitat is a matter of debate. For terrestrial species, habitat is often defined as physical patches of a certain vegetation type in a matrix of non-habitat. Ecological resources that make up the habitat of a species may, however, only cover subsets of vegetation types or can be spatially dispersed in a complex way over different vegetation types. Here we present and test a procedure to recognize and delineate habitat according to a resource-based approach instead of a vegetation-based approach. We used the green hairstreak butterfly (Callophrys rubi) in a heathland landscape as a study case. Our resource-based habitat approach selects those zones that comprise essential resources and conditions within an appropriate spatial window. Variables that were retained in a logistic regression model were used to calculate larval, adult and combined habitat indices in a GIS, taking into account thermal constraints on resource-use, as this is a key habitat aspect for this heliothermous insect. To group different (and sometimes scattered) ecological resources into functional habitat zones, we derived a measure of space-use from mark-release-recapture data. By least-cost modelling this spatial window was adapted to the nature of the vegetation between sets of resources. The habitat zones that were delineated using this approach matched the observed distribution of butterflies significantly better than did a classic approach based on vegetation types with host plants only. Our approach provides concrete output for conservation purposes, like recognizing zones with the highest potential for habitat restoration.
The ecogeographical data used to calculate potential distribution models of species is obtained mainly from thematic maps. Although this data is usually available in developed areas of the world, this may not be the case for other areas, when a particular type of data does not exist and may be necessary to obtain this information with fieldwork. An alternative source of ecogeographical data is the satellite imagery. In the present paperwork, two sets of models, calculated exclusively with variables obtained from (1) Landsat 5 TM and SRTM, or (2) thematic maps, were compared to know if Remote Sensing is a reliable source of data to apply in remote areas. However, the comparison must be made in areas where the distribution of species is well known and where satellite imagery and thematic data are available. The two methodologies have a similar prediction capacity, although the models were spatially different. The remote sensing models had a larger and more fragmented surface. The satellite imagery is a useful data source, developing similar predictive models to the thematic map ones.
Distribution de Lucanus cervus (Coleoptera, Lucanidae) en Belgique : survivre dans un paysage changeant. — Le présent article décrit les résultats mettant à jour la distribution passée et actuelle du Lucane cerf-volant en Belgique. Sur la base de ces données une modélisation de la distribution a été effectuée en prenant en compte l’usage des terres et les paramètres climatiques et topographiques afin d’identifier les zones correspondant aux exigences de l’espèce. Les besoins écologiques et en habitat de l’espèce en Belgique sont décrits et discutés.
Aim To evaluate geostatistical approaches, namely kriging, co-kriging and geostatistical simulation, and to develop an optimal sampling design for mapping the spatial patterns of bird diversity, estimating their spatial autocorrelations and selecting additional samples of bird diversity in a 2450 km2 basin.
Location Taiwan.
Methods Kriging, co-kriging and simulated annealing are applied to estimate and simulate the spatial patterns of bird diversity. In addition, kriging and co-kriging with a genetic algorithm are used to optimally select further samples to improve the kriging and co-kriging estimations. The association between bird diversity and elevation, and bird diversity and land cover, is analysed with estimated and simulated maps.
Results The Simpson index correlates spatially with the normalized difference vegetation index (NDVI) within the micro-scale and the macro-scale in the study basin, but the Shannon diversity index only correlates spatially with NDVI within the micro-scale. Co-kriging and simulated annealing simulation accurately simulate the statistical and spatial patterns of bird diversity. The mean estimated diversity and the simulated diversity increase with elevation and decrease with increasing urbanization. The proposed optimal sampling approach selects 43 additional sampling sites with a high spatial estimation variance in bird diversity.
Main conclusions Small-scale variations dominate the total spatial variation of the observed diversity due to a lack of spatial information and insufficient sampling. However, simulations of bird diversity consistently capture the sampling statistics and spatial patterns of the observed bird diversity. The data thus accumulated can be used to understand the spatial patterns of bird diversity associated with different types of land cover and elevation, and to optimize sample selection. Co-kriging combined with a genetic algorithm yields additional optimal sampling sites, which can be used to augment existing sampling points in future studies of bird diversity.
Being ectotherms, insects are predicted to suffer more severely from climate change than warm-blooded animals. We forecast possible changes in diversity and composition of butterflies, grasshoppers and dragonflies in Belgium under increasingly severe climate change scenarios for the year 2100. Two species distribution modelling techniques (Generalised Linear Models and Generalised Additive Models), were combined via a conservative version of the ensemble forecasting strategy to predict present-day and
future species distributions, considering the species as potentially present only if both modelling techniques made such a prediction. All models applied were fair to good, according to the AUC (area under the curve of the receiver operating characteristic plot), sensitivity and specificity model performance measures based on model evaluation data. Butterfly and grasshopper diversity were predicted to decrease significantly in all scenarios and species-rich locations were predicted to move towards higher altitudes. Dragonfly diversity was predicted to decrease significantly in all scenarios, but dragonfly-rich locations were predicted to move upwards only in the less severe scenarios. The largest turnover rates were predicted to occur at higher altitudes for butterflies and grasshoppers, but at intermediate altitudes for dragonflies. Our results highlight the challenge of building conservation strategies under climate change, because the changes in the sites important for different groups will not overlap, increasing the area needed for protection. We advocate that possible conservation and policy measures to mitigate the potentially strong impacts of climate change on insect diversity in Belgium should be much more pro-active and flexible than is the case presently.
Aim Taxocenes are regulated by different kinds of predictors, but can broad-scale patterns of prokaryotes (bacteria) and eukaryotes (like fungi and invertebrates) be ascribed to soil acidity? We sought to test for relationships between the numerical abundances of bacteria, microfungi, nematodes and arthropods along a pH gradient.
Location 284 agro-ecosystems on Pleistocene sand across the Netherlands.
Methods Generalized Linear Models (GLM) and stepwise regressions were applied, using soil and leaf-litter organisms from a land-covering network.
Results The major variation in the numerical abundance of the organisms belonging to the investigated taxocenes could be ascribed to soil acidity. Contrary to the expectations, the effects of temperature on numerical abundance were significant only for Fungi and Nematoda (P < 0.0001). Geographical co-ordinates always play a minor role. The often-suggested close correlation between the numerical abundance of eukaryotes and their local taxonomic diversity applied only to Arthropoda and Fungi (P < 0.00001). Only the number of bacterial DNA bands seemed to reflect the taxa-area relationship (F-value = 22.45, P < 0.0001).
Main conclusions There were strong relationships between the numerical abundances of all the investigated taxocenes and the field-measured soil acidity (P < 0.0001). The largest effects were detected in the Fungi, which tended to be much more acid-tolerant than Bacteria. These patterns imply ecological shifts in the detrital soil food web and deserve further investigation.
Mountain butterfly species are often restricted in their distribution and under threat from habitat destruction and climate change. Due to the inaccessibility of their habitats the distributions of many such species are unknown. We have investigated whether information on the habitat requirements of the Alpine endemic species Erebia calcaria could be used for modelling its potential distribution. We surveyed part of its range using transects and recorded habitat and environmental parameters. The most important parameters determining the presence of the species were average height of the vegetation, maximum height of the vegetation, percentage area of bare ground, number of food plants and slope. Furthermore, the abundance of E. calcaria is strongly affected by site exposure and grazing intensity. Using these results we modelled the potential distribution of the species in its known historical range in Slovenia. In the region covered by the model 70% of the records of E. calcaria were within the predicted distribution. It is reasonable to propose that such a high detection rate justifies the use of distribution models for predicting a species range and providing important additional information for their conservation. In the case of E. calcaria, we have shown that endemic mountain butterflies can be strongly threatened by fragmentation of their habitat, overgrazing and succession, which could be further amplified by changes in climate.
An analysis of factors affecting butterfly diversity over the British Isles. A response to a paper by JRG Turner. In Antenna the RES house journal
Taxonomic databases as tools in spatial biodiversity research. - The conservation of biodiversity constitutes an international goal today, but it needs a good knowledge of the ecology and distribution of fauna and flora. Electronic taxonomic database programs are being developed by many organizations around the world. Once taxonomic databases are established for the purpose of inventory and monitoring, it is necessary to enhance and correct the databases. Through baseline samplings, it is possible to record a very large proportion of the biodiversity present. But it is necessary to evaluate the representativity of the information in order to verify the relationship between the number of discovered species and the number of available samplings. The aim of the present study was to show how the French dung beetle database (Coleoptera, Scarabaeoidea Laparosticti) could provide data for conservation planners in order to encourage greater use of the work of systematists in conservation programmes. The data allow the presence and absence in France (continental part and Corsica) of species in each 0.8 × 0.4 grades squares (301 squares in total). France was also divided into five biogeographic subprovinces, and each square was allotted to one subprovince. The curve showing the accumulation of records in new squares was almost asymptotic. Well sampled squares were those where more than 70 % of the species potentially present in the square were found. The number of records to obtain well sampled squares depended on the subprovince : 311, 260, 193, 355 and 115 samples for the oceanic, continental, mediterranean, alpine and corsican subprovinces, respectively. Sixty four squares among 301 showed a good quality of information, and could be used later to detect the hotspots of species richness.
A long-standing objective of ecology has been to explain the basis for diversity patterns. Empirical evidence suggests that regional variation in richness of both animals and plants depends strongly on energy availability. The generality of the richness-energy hypothesis is limited by the paucity of analyses of invertebrates, which are much more diverse than the more thoroughly investigated vertebrate taxa. In this study, we consider two groups of North American Lepidoptera for which large-scale distribution data are available: the Papilionidae (swallowtail butterflies) and forest lepidopterans (moths based on the Canadian Forest Insect Survey). Energy, as measured by potential evapotranspiration (PET), statistically explains between 61 and 72% of the variability in the richness patterns of the Lepidoptera we have examined. It is the single best predictor of the richness of these groups, and the relationships have a very similar form to richness-PET relationships observed earlier in vertebrate taxa. After PET, Papilionidae richness is related to topographical heterogeneity. These patterns are true both within and among biomes. These results suggest that the richness-energy hypothesis applies generally to both vertebrates and insects in cold and temperate regions.
The distributions of most pollinator species are poorly documented despite their importance in providing ecosystem services. While these and other organisms are threatened by many aspects of the human enterprise, anthropogenic climate change is potentially the most severe threat to pollinator biodiversity. Mounting evidence demonstrates that there have already been biotic responses to the relatively small climate changes that have occurred this century. These include wholesale shifts of relatively well-documented butterfly and bird species in Europe and North America. Although studies of such phenomena are supported by circumstantial evidence, their findings are also consistent with predictions derived from current models of spatial patterns of species richness. Using new GIS methods that are highly precise and accurate, I document spatial patterns of Canadian butterfly diversity. These are strongly related to contemporary climate and particularly to potential evapotranspiration. An even more noteworthy finding is the fact that, for the first time, habitat heterogeneity, measured as the number of land cover types in each study unit, is proven to be an equally strong predictor of butterfly richness in a region where energy alone was thought to be the best predictor of diversity. Although previous studies reveal similar relationships between energy and diversity, they fail to detect the powerful link between richness and habitat heterogeneity. The butterflies of Canada provide a superb baseline for studying the effects of climate on contemporary patterns of species richness and comprise the only complete pollinator taxon for which this sort of analysis is currently possible.
The distributions of most pollinator species are poorly documented despite their importance in providing ecosystem services. While these and other organisms are threatened by many aspects of the human enterprise, anthropogenic climate change is potentially the most severe threat to pollinator biodiversity. Mounting evidence demonstrates that there have already been biotic responses to the relatively small climate changes that have occurred this century. These include wholesale shifts of relatively well-documented butterfly and bird species in Europe and North America. Although studies of such phenomena are supported by circumstantial evidence, their findings are also consistent with predictions derived from current models of spatial patterns of species richness. Using new GIS methods that are highly precise and accurate, I document spatial patterns of Canadian butterfly diversity. These are strongly related to contemporary climate and particularly to potential evapotranspiration. An even more noteworthy finding is the fact that, for the first time, habitat heterogeneity, measured as the number of land cover types in each study unit, is proven to be an equally strong predictor of butterfly richness in a region where energy alone was thought to be the best predictor of diversity. Although previous studies reveal similar relationships between energy and diversity, they fail to detect the powerful link between richness and habitat heterogeneity. The butterflies of Canada provide a superb baseline for studying the effects of climate on contemporary patterns of species richness and comprise the only complete pollinator taxon for which this sort of analysis is currently possible.
We use data from the Mersey Valley zone (3×2 km area; N = 600 I ha squares) of the Greater Manchester butterfly atlas to investigate whether recorder visits are biased by access (viz. distance from recorder's home base) and by the locations of potential hot spots. In a multiple regression analysis, visits were found to correlate significantly both with distance from home base of the recorder and with the mean and maximum number of species found in squares. Sites close to the home base of the recorder were visited more frequently than those further afield and squares with more species were visited more frequently than those squares with fewer species. Visits were also made significantly more frequently to squares with greater numbers of butterfly resources (e.g. hostplants, nectar). Furthermore, recording is biased to and away from distinct land uses, which vary significantly in species richness. Reasons are given why these biases are to be expected at all scales. The message is that future distribution mapping should be based on rigorous sampling approaches.
We illustrate the strong decrease in the number of butterfly species in Flanders (north Belgium) in the 20th century using data from a national butterfly mapping scheme. Nineteen of the 64 indigenous species went extinct and half of the remaining species are threatened at present. Flanders is shown to be the region with the highest number of extinct butterflies in Europe. More intensive agriculture practices and expansion of house and road building increased the extinction rate more than eightfold in the second half of the 20th century. The number of hot spots decreased considerably and the present-day hot spots are almost exclusively in the northeast of Flanders. Species with low dispersal capacities and species from oligotrophic habitats decreased significantly more than mobile species or species from eutrophic habitats. We discuss these results in a northwest European context and focus on concrete measures to preserve threatened butterfly populations in Flanders.
Data for the United Kingdom (UK) Manchester Butterfly Atlas produced conflicting species loss rates for increased urban development. In particular, a very low rate of loss was recorded (0.19 species for every 10% increase in urban cover) for the Mersey Valley mapped at a high resolution of 1 ha units. It was suggested that sampling artefacts (uneven recording) or failure to distinguish vagrant individuals from breeding populations cause this. Herein, results are reported for 30 sample squares, within the Mersey Valley, surveyed uniformly throughout 1999. It is shown that loss rates are as high as areas mapped at lower resolution over wider areas (0.67–0.68 species for every 10% increase in urban cover) and that increasingly stringent definitions of urban cover result in higher loss rates. Comparison with the data from the Atlas, but for the same 30 sample squares, indicate that the low rates at a fine scale for the complete Atlas data are more likely to be caused by uneven recording than from failure to record species status. However, progressive sampling of squares, despite uniform recording, will inevitably cause a reduction in loss rates of total species for increases in urban development.
The diversity of future landscapes might depend on our ability to predict their potential species richness. The predictability
of patterns of vascular plant species richness in a Finnish agricultural river landscape was studied using generalized linear
modeling, floristic records from fifty-three0.25-km grid squares in the “core” study area, and environmental variables derived
from Landsat TM images and a digital elevation model. We built multiple regression models for the total number of plant species
and the number of rarities, and validated the accuracy of the derived models with a test set of 52 grid squares. We tentatively
extrapolated the models from the core study area to the whole study area of 601 km2 and produced species richness probability maps using GIS techniques. The results suggest that the local ‘hotspots’ of total
flora (grid squares with > 200species) are mainly found in river valleys, where habitat diversity is high and a semi-open
agricultural-forest mosaic occurs. The ‘hotspots’ of rare species (grid squares with > 4 rare species) are also found in river
valleys, in sites where extensive semi-natural grasslands and herb-rich deciduous forests occur on steep slopes. We conclude
that environmental variables derived from satellite images and topographic data can be used as approximate surrogates of plant
species diversity in agricultural landscapes. Modeling of biological diversity based on satellite images and GIS can provide
useful information needed in land use planning. However, due to the potential pitfalls in processing satellite imagery and
model-building procedures, the results of predictive models should be carefully interpreted.
In many large-scale conservation or ecological problems where experiments are intractable or unethical, regression methods are used to attempt to gauge the impact of a set of nominally independent variables (X) upon a dependent variable (Y). Workers often want to assert that a given X has a major influence on Y, and so, by using this indirection to infer a probable causal relationship. There are two difficulties apart from the demonstrability issue itself: (1) multiple regression is plagued by collinear relationships in X; and (2) any regression is designed to produce a function that in some way minimizes the overall difference between the observed and predicted Ys, which does not necessarily equate to determining probable influence in a multivariate setting. Problem (1) may be explored by comparing two avenues, one in which a single best regression model is sought and the other where all possible regression models are considered contemporaneously. It is suggested that if the two approaches do not agree upon which of the independent variables are likely to be significant, then the deductions must be subject to doubt.
Ecologists and conservation biologists frequently use multipleregression (MR) to try to identify factors influencing response variables suchas species richness or occurrence. Many frequently used regression methods maygenerate spurious results due to multicollinearity. argued that there are actually two kinds of MR modelling: (1)seeking the best predictive model; and (2) isolating amounts of varianceattributable to each predictor variable. The former has attracted most attentionwith a plethora of criteria (measures of model fit penalized for modelcomplexity – number of parameters) and Bayes-factor-based methods havingbeen proposed, while the latter has been little considered, althoughhierarchical methods seem promising (e.g. hierarchical partitioning). If the twoapproaches agree on which predictor variables to retain, then it is more likelythat meaningful predictor variables (of those considered) have been found. Therehas been a problem in that, while hierarchical partitioning allowed the rankingof predictor variables by amounts of independent explanatory power, there was no(statistical) way to decide which variables to retain. A solution usingrandomization of the data matrix coupled with hierarchical partitioning ispresented, as is an ecological example.
The effect of urban development on butterfly species' richness and species' incidence is tested for the Greater Manchester conurbation and two sample areas, mapped at finer scales, within the southern part of the conurbation. The tests include measures of bias for recording effort (number of visits). Species' richness is found to increase with percentage urban cover for Greater Manchester (tetrad scale) and decrease with urban cover for the two sample areas in South West Manchester (1 km scale) and the Mersey Valley (100 m scale). For Greater Manchester, the increase in species' richness with increased urban cover is largely explained by lower species' richness at higher altitude in the Pennines bounding the conurbation. For the two sample areas, decreasing species' richness associated with increasing urban cover corresponds with reductions in the areas of a number of semi-natural habitats, hostplants and nectar sources. Despite these statistically significant correlations, the impact of urban cover on species' richness is weak. The maximum loss rate identified anywhere within the region is 0.81 species per 10% change in urban cover for South West Manchester. This finding may reflect on the generally low species diversity of the region. However, these results could be influenced by recording and sampling artefacts, particularly the failure of mapping programmes to distinguish vagrant individuals from breeding populations and a bias of records to vagrants. This is supported by the higher correlations between species' incidence and nectar sources than between species' incidence and their hostplants. Adult butterflies are opportunistic nectar users and nectar sources are more widely spread and thus less influenced by urban development than are specific butterfly hostplants. The finding may also reflect on the capacity of most of the butterfly species to breed successfully on tiny areas of hostplant existing within extensively built-up areas. Moreover, the capacity of butterfly species to persist by using small fragments of hostplants would be enhanced by vagrancy. If this is indeed the case, it is a finding that would support the value of small patches in butterfly metapopulations, albeit ones comprising incomplete complements of resources required during the life cycle. The incidence of most species decreases with increase in urban cover. Multivariate analyses indicate that this is owing to corresponding declines in hostplant-habitats and nectar sources. Five species increase with urban cover, but none attain formal significance. Associations among hostplants and habitat variables in a principal components analysis suggest that, in three cases (Pieris brassicae, P. rapae, Celastrina argiolus), this is owing to increasing areas of their hostplants within urban environments.
In ecological modelling, limitations in data and their applicability for predictive modelling are more rule than exception. Often modelling has to be performed on sub-optimal data, as explicit and controlled collection of (more) appropriate data would not be feasible. An example of predictive ecological modelling is given with application of generalized additive and generalized linear models fitted to presence–absence records of plant species and site condition data from four nutrient-poor Flemish lowland valleys. Standard regression procedures are used for modelling, although explanatory and response data do not meet all the assumptions implicit in these procedures. Data were non-randomly collected and are spatially autocorrelated; model residuals retain part of that correlation. The scale of most site-condition records does not match the scale of the response variable (species distribution). Hence, interpolated and up-scaled explanatory variables are used. Data are aggregated from distinct phytogeographical regions to allow for generalized models, applicable to a wider population of river valleys in the same region. Nevertheless, ecologically sound models are obtained, which predict well the distribution of most plant species for the Flemish river valleys considered.
Clifford, Richardson, and Hemon presented modified tests of association between two spatially autocorrelated processes, for lattice and non-lattice data. These tests are built on the sample covariance and on the sample correlation coefficient, they require the estimation of an effective sample size that takes into account the spatial structure of both processes. Clifford et al. developed their method on the basis of an approximation of the variance of the sample correlation coefficient and assessed it by Monte Carlo simulations for lattice and non-lattice networks of moderate to large size. In the present paper, the variance of the sample covariance is computed for a finite number of locations, under the multinormality assumption, and the mathematical derivation of the definition of effective sample size is given. The theoretically expected number of degrees of freedom for the modified t test with renewed modifications is compared with that computed on the basis of equation (2.9) of Clifford et al. The largest differences are observed for small numbers of locations and high autocorrelation, in particular when the latter is present with opposite sign in the two processes Basic references that were missing in Clifford et al. are given and inherent ambiguities are discussed.
A quick dip into the literature on diversity reveals a bewildering range of indices. Each of these indices seeks to characterize the diversity of a sample or community by a single number. To add yet more confusion an index may be known by more than one name and written in a variety of notations using a range of log bases. This diversity of diversity indices has arisen because, for a number of years, it was standard practice for an author to review existing indices, denounce them as useless, and promptly invent a new index. Southwood (1978) notes an interesting parallel in the proliferation of new designs of light traps and new permutations of diversity measures.
Clifford, Richardson, and Hemon (1989, Biometrics 45, 123-134) presented modified tests of association between two spatially autocorrelated processes, for lattice and non-lattice data. These tests are built on the sample covariance and on the sample correlation coefficient; they require the estimation of an effective sample size that takes into account the spatial structure of both processes. Clifford et al. developed their method on the basis of an approximation of the variance of the sample correlation coefficient and assessed it by Monte Carlo simulations for lattice and non-lattice networks of moderate to large size. In the present paper, the variance of the sample covariance is computed for a finite number of locations, under the multinormality assumption, and the mathematical derivation of the definition of effective sample size is given. The theoretically expected number of degrees of freedom for the modified t test with renewed modifications is compared with that computed on the basis of equation (2.9) of Clifford et al. (1989). The largest differences are observed for small numbers of locations and high autocorrelation, in particular when the latter is present with opposite sign in the two processes. Basic references that were missing in Clifford et al. (1989) are given and inherent ambiguities are discussed.
The available potential ecological factors have been scored in the form of presence/absence in U.T.M. squares in Beltium. A correspondence analysis shows a strong underlying gradient in the data set which induces an extraordinary horseshoe effect. This gradient follows closely the altitude component. Applying the k-means clustering method on U.T.M. squares produces geographically compact groups which are largely hierarchically nested. This indicates strong regional trends in the ecological data set. As homogeneous groups may also be artefacts created by the clustering algorithms on a continuous gradient, the relevance of the borders between homogeneous areas is tested. In general, k-means borders correspond to the main breaking lines between adjacent U.T.M. squares. They can be referred to as natural borders.
We used comprehensive data on butterfly distributions from six mountain ranges in the Great Basin to explore three connected biogeographic issues. First, we examined species richness and occurrence patterns both within and among mountain ranges. Only one range had a significant relationship between species richness and area. Relationships between species richness and elevation varied among mountain ranges. Species richness decreased as elevation increased in one range, increased as elevation increased in three ranges, and was not correlated in two ranges. In each range, distributional patterns were nested, but less vagile species did not always exhibit greater nestedness. Second, we compared our work with similar studies of montane mammals. Results from both taxonomic groups suggest that it may be appropriate to modify existing general paradigms of the biogeography of montane faunas in the Great Basin. Third, we revisited and refined previous predictions of how butterfly assemblages in the Great Basin may respond to climate change. The effects of climate change on species richness of montane butterflies may vary considerably among mountain ranges. In several ranges, few if any species apparently would be lost. Neither local species composition nor the potential order of species extirpations appears to be generalizable among ranges.
If occurrence of individual species can be modeled as a function of easily quantified environmental variables (e.g., derived from a geographic information system [GIS]) and the predictions of these models are demonstrably successful, then the scientific foundation for management planning will be strengthened. We used Bayesian logistic regression to develop predictive models for resident butterflies in the central Great Basin of western North America. Species inventory data and values for 14 environmental variables from 49 locations (segments of canyons) in the Toquima Range ( Nevada, U.S.A.) were used to build the models. Squares of the environmental variables were also used to accommodate possibly nonmonotonic responses. We obtained statistically significant models for 36 of 56 (64%) resident species of butterflies. The models explained 8–72% of the deviance in occurrence of those species. Each of the independent variables was significant in at least one model, and squared versions of five variables contributed to models. Elevation was included in more than half of the models. Models included one to four variables; only one variable was significant in about half the models. We conducted preliminary tests of two of our models by using an existing set of data on the occurrence of butterflies in the neighboring Toiyabe Range. We compared conventional logistic classification with posterior probability distributions derived from Bayesian modeling. For the latter, we restricted our predictions to locations with a high ( 70%) probability of predicted presence or absence. We will perform further tests after conducting inventories at new locations in the Toquima Range and nearby Shoshone Mountains, for which we have computed environmental variables by using remotely acquired topographic data, digital-terrain and microclimatic models, and GIS computation.
Interpretation of large-scale faunal and floral survey data is often frustrated by the bias caused by variation in recording intensity. Using distribution data for Odonata and Hepaticae from the Biological Records Centre, a technique for correcting this bias is described The method is used to locate species-rich hotspots for the two taxa and comparisons are made with uncorrected data.
With the rise of new powerful statistical techniques and GIS tools, the development of predictive habitat distribution models has rapidly increased in ecology. Such models are static and probabilistic in nature, since they statistically relate the geographical distribution of species or communities to their present environment. A wide array of models has been developed to cover aspects as diverse as biogeography, conservation biology, climate change research, and habitat or species management. In this paper, we present a review of predictive habitat distribution modeling. The variety of statistical techniques used is growing. Ordinary multiple regression and its generalized form (GLM) are very popular and are often used for modeling species distributions. Other methods include neural networks, ordination and classification methods, Bayesian models, locally weighted approaches (e.g. GAM), environmental envelopes or even combinations of these models. The selection of an appropriate method should not depend solely on statistical considerations. Some models are better suited to reflect theoretical findings on the shape and nature of the species’ response (or realized niche). Conceptual considerations include e.g. the trade-off between optimizing accuracy versus optimizing generality. In the field of static distribution modeling, the latter is mostly related to selecting appropriate predictor variables and to designing an appropriate procedure for model selection. New methods, including threshold-independent measures (e.g. receiver operating characteristic (ROC)-plots) and resampling techniques (e.g. bootstrap, cross-validation) have been introduced in ecology for testing the accuracy of predictive models. The choice of an evaluation measure should be driven primarily by the goals of the study. This may possibly lead to the attribution of different weights to the various types of prediction errors (e.g. omission, commission or confusion). Testing the model in a wider range of situations (in space and time) will permit one to define the range of applications for which the model predictions are suitable. In turn, the qualification of the model depends primarily on the goals of the study that define the qualification criteria and on the usability of the model, rather than on statistics alone.
To assess conservation priorities, a means of measuring the distribution of a much larger part of overall biodiversity is needed that will at the same time reduce the colossal sampling problems of exhaustive surveys. One possibility is a ‘top-down’ taxonomic approach, in which the biodiversity of different areas may be compared using measures based on the number of higher taxa present in each. The advantage of this approach is that survey costs should be greatly reduced because identification to species level, particularly within the few hyper-rich higher taxa, would be unnecessary. We report that family richness is a good predictor of species richness for a variety of groups and regions, including both British ferns and British butterflies among 100 km × 100 km (10,000 km2) grid squares, Australian passerine birds among 5° × 5° grid squares (c. 220,000–310,000 km2) and 10° × 10° grid squares (c. 970,000–1,190,000 km2), and North and Central American bats among grid squares of c. 611,000 km2. With careful choice of higher-taxon rank, it may be possible to re-deploy effort from taxonomically intensive to taxonomically extensive surveys, in order to estimate the global distribution of a much larger proportion of overall biodiversity at the same cost.
New data appearing in the second of two French atlases within one and a half years confirm that there was substantial under-recording of butterfly species in France for the production of the first atlas, particularly in the south and west of the country. Under-recording is still a prominent feature of the southwest region and eastern border. The new data also reveal contractions in the ranges of 60 species suggesting real losses as a result of regional extinction especially in the north of the country. This finding links adjacent areas of ongoing high regional extinction in continental European Lepidoptera extending from the Netherlands through Belgium into northern France. The new data also demonstrate that predictions of species numbers and species incidences based on records in the first atlas, using regression techniques on geographical and neighbourhood variables, have been largely successful (76% correct prediction of new records for départements). This supports the application of such techniques to targeting surveys for mapping spatial units and species to improve atlas databases; the recent rapid changes in distributions underlines the importance of having a suitable framework for continuing recording after atlas publication.
To examine the importance of management practices and landscape structure on diversity of butterflies 16 farms with organic or conventional management were censused during 1997 and 1998. On each farm a transect route was walked during July and the beginning of August, six times in 1997 and five times in 1998. The farms were located in the central part of Sweden in two adjacent regions with the same pool of species. The organic and conventional farms were paired with help of the Bray-Curtis dissimilarity index according to land use to control for landscape structure on the farm level.
On each farm calculations were made of large- and small-scale landscape heterogeneity with the help of GIS. A grid with a mesh size of 400 m was placed over each farm and the small-scale heterogeneity was calculated as the mean habitat diversity of four squares. The large-scale landscape heterogeneity described the landscape in which the farms were imbedded, and covered an area of 5x5 km.
No differences in butterfly diversity, number of species or number of observations were noted between organic and conventional farms. Butterfly diversity was positively correlated with small-scale landscape heterogeneity while butterfly abundance was positively correlated with large-scale heterogeneity. Both large-scale and small-scale heterogeneity were important for the composition of species. The landscape structure seemed to be more important for butterfly diversity and species composition than the farming system in itself.
Summary A comparison of species richness patterns of butterflies and birds was made using data from two grids of squares (small squares 137.5 km on a side and large squares 275 km on a side) covering western North America. Using geostatistical procedures, we found that the spatial patterns of species richness of these two taxa were related. The influence of grain size on the strength of this relationship was investigated by analysing the two data sets. For both data sets, the number of butterfly species in a square was a statistically significant predictor of the corresponding number of bird species. However, cross-validation techniques showed that the marginal improvement in prediction accuracy due to including butterflies as a predictor was greater in the large-square data. We explored the effect of areal extent on cross-taxon congruencies by investigating species richness patterns in four subsets of the small-square data. In regions with smaller areal extent, the cross-taxon congruence patterns were not substantially different from the pattern found in the full data set. Finally, using data-splitting techniques, we explored the relationships between prediction accuracy of species richness, sample size, areal extent of the sample, and grain size.
We use data from the French national butterfly atlas to compare the potential of direct geographical and neighbourhood models to account for numbers of species and incidence of species in French départements. Direct geographical models use data on latitude, longitude and altitude, whereas neighbourhood models use information from adjacent areas. Both geographical models and neighbourhood models account for a large proportion of the variance in species richness (68–78%). However, neighbourhood models are more successful than models based solely on simple geographical variables. A large number of individual species distributions are accounted for by logistic and autologistic regression models (222 of 246 species, 90.2%). The autologistic models incorporate information on neighbouring areas. The exceptions are rare species, five of six of which occur in a single administrative unit only (2.4%), or virtually ubiquitous species found in >90% of units (7.3%). Autologistic models dominate logistic models in accounting for species incidences using stepwise logit regressions, neighbourhood variables appearing in 64.5% of successful species models (absent in 22.8%) and then always entering first. A simple neighbourhood (distance-unweighted) measure (C2) dominates more models (89 of 246 species, 36.2%) than a distance-weighted neighbourhood measure (C1; 77 of 246 species, 31.3%). The models are here demonstrated to be potentially valuable for identifying under-recording and losses from regional extinction and for filling gaps in recording. The findings reveal substantial, apparent, losses of species in western and northern France as well as substantial discrepancies (differences) in numbers of species, for some administrative units (départements) and for both post-1970 and total records, compared with numbers predicted to occur. We use two distinct approaches on total species and individual species to provide comparative estimates of the numbers of species expected within spatial units and we present the number of additional units in which species are expected to occur. The probabilities for these species in French départements are available on Web site: http://www.brookes.ac.uk/schools/bms/research/data/ecology/butterfly.html.
SPECIES conservation in situ requires networks of protected areas selected for high conservation interest13. Throughout most of the world, however, there are neither the resources nor the time to carry out detailed inventories for most taxa2,4 before designating protected areas. Site selection (on grounds other than availability) would be easier and more effective if two things were true: (1) habitats that are species-rich for one taxon are also species-rich for others5; and (2) rare1 species occur in, and therefore benefit from the conservation of, species-rich habitats. Diversity (usually, species richness) and the presence of rare species are the most frequently cited criteria for site selection by conservationists68. Here, we use data on British plants and animals held by the Biological Records Centre (BRC)9 and the British Trust for Ornithology (BTO), mapped on a grid of 10 km × 10 km ('10 km squares') to examine the extent to which species-rich areas for different taxa coincide, and whether species-rich areas contain substantial numbers of rare species. The fine scale and high intensity of recording in Britain produces distributional datasets at least as good as and, in most cases, better than those available elsewhere. For Britain at least, we do not find strong support for either proposition. Species-rich areas ('hotspots'10) frequently do not coincide for different taxa, and many rare species do not occur in the most species-rich squares.
Summary By using a simple example a minimax type optimality of the minimum AIC procedure for the selection of models is demonstrated.
In Mediterranean countries, inventories of many animal groups, particularly insects, are incomplete or nonexistent. Hence, a feasible spatial picture of unequally surveyed areas is required to ascertain which faunistic surveys are good enough to produce reliable estimates of species richness. We used generalized linear models to build a multiple-regression function through which we predicted the distribution of Iberian dung beetle species richness. Given the scarcity and unevenness of the species-richness spatial distribution, the number of records of a dung beetle database ( BANDASCA ), falling within each of the 50 × 50 km grid squares, was chosen as a measure of the sampling effort for that square. Examining the asymptotic relationship between the number of dung-beetle species and database records for each physioclimatic Iberian subregion, we found that 82 grid squares (32% of the total) were adequately sampled. Dung-beetle species richness was related in each of these 82 cells to 24 explanatory variables. Curvilinear functions, interaction terms, and the significant third-degree polynomial terms of latitude and longitude were included to model species-richness distribution. The final model accounted for 62.4% of the total deviance after we eliminated seven outlier squares, with maximum elevation, grassland area, land-use diversity, forest area, geological diversity, interaction of terrestrial area and maximum elevation, and interaction between calcareous rock and geological diversity and latitude being the most significant independent variables. The residuals of the function were not spatially autocorrelated, and we validated the final model by a jackknife procedure. Large and environmentally complex hotspots in the Iberian Central, Baetic, and Subbaetic mountain ranges stand out from the emerging map of species richness. Further detailed research is required to determine the complementarity of the faunas of these two main hotspots, the key question in conservation planning for a dung-feeding beetle.
Resumen: Los inventarios de muchos grupos de animales, particularmente insectos, son incompletos o totalmente ausentes en los países Mediterráneos. Por ello, se requiere de un panorama espacial viable de áreas poco estudiadas para determinar que estudios faunísticos son suficientemente buenos para producir estimaciones confiables de la riqueza de especies. Utilizamos modelos lineales generalizados para construir una función de regresión múltiple con la cual predijimos la distribución de la riqueza de especies de escarabajos coprófilos ibéricos. Dada la escasez y desigualdad de la distribución espacial de la riqueza de especies, el número de registros de una base de datos de escarabajos coprófilos ( BANDASCA ) en cada uno de los cuadrantes de 50 × 50 km fue seleccionado como una medida del esfuerzo de muestreo para ese cuadro. Al examinar la relación asintótica entre el número de escarabajos coprófilos y los registros en la base de datos para cada región fisioclimática, encontramos que 82 cuadros (32 % del total) fueron adecuadamente muestreados. La riqueza de especies de escarabajos coprófilos se relacionó con 24 variables explicativas en cada uno de estas 82 celdas. Se incluyeron funciones curvilíneas, términos de interacción y los significativos términos polinomiales de tercer grado de latitud y longitud en el modelo de distribución espacial de la riqueza de especies. El modelo final consideró el 62.4 % de la desviación total después de eliminar siete cuadros externos, elevación máxima, superficie de pastizal, diversidad de uso de suelo, superficie forestal, diversidad geológica, la interacción entre superficie terrestre y elevación máxima, la interacción entre roca calcárea y diversidad geológica y latitud, les variables independiente más significativa. Los residuos de la correlación no fueron autocorrelacionados espacialmente, y validamos el modelo final con un procedimiento de navaja. En el mapa emergente de la riqueza de especies sobresalen sitios conflictivos extensos y ambientalmente complejos en las cadenas montañosas de Iberia Central, Baética y Subaética. Se requiere de más investigación detallada para determinar la complementariedad de la fauna de estos sitios conflictivos principales como el punto central en la planeación de la conservación de escarabajos coprófagos.
Data from a national butterfly monitoring scheme were analysed to test for relationships between temperature and three phenological measures, duration of flight period and timing of both first and peak appearance. First appearances of most British butterflies has advanced in the last two decades and is strongly related to earlier peak appearance and, for multibrooded species, longer flight period. Mean dates of first and peak appearance are examined in relation to Manley's central England temperatures, using regression techniques. We predict that, in the absence of confounding factors, such as interactions with other organisms and land-use change, climate warming of the order of 1 °C could advance first and peak appearance of most butterflies by 2–10 days.
Over the last three decades a great deal of research, money, and effort have been put into the development of theory and techniques designed to make conservation more efficient. Much of the recent emphasis has been on methods to identify areas of high conservation interest and to design efficient networks of nature reserves. Reserve selection algorithms, gap analysis, and other computerized approaches have much potential to transform conservation planning, yet these methods are used only infrequently by those charged with managing landscapes. We briefly describe different approaches to identifying potentially valuable areas and methods for reserve selection and then discuss the reasons they remain largely unused by conservationists and land-use planners. Our informal discussions with ecologists, conservationists, and land managers from Europe and the United States suggested that the main reason for the low level of adoption of these sophisticated tools is simply that land managers have been unaware of them. Where this has been the case, low levels of funding, lack of understanding about the purpose of these tools, and general antipathy toward what is seen as a prescriptive approach to conservation all play a part. We recognize there is no simple solution but call for a closer dialogue between theoreticians and practitioners in conservation biology. The two communities night be brought into closer contact in numerous ways, including carefully targeted publication of research and Internet communication. However it is done, we feel that the needs of land managers need to be catered to by those engaged in conservation research and that managers need to be more aware of what science can contribute to practical conservation.
Aim We compare the influence of contemporary geography and historical influences on butterfly diversity for islands in the Aegean archipelago.
Location The Aegean archipelago (Greece) and two islands (Cyprus and Megisti) in the Levantine Sea.
Methods Thirty-one islands were examined. Data are taken from own surveys (Coutsis and Olivier) and from the literature. Stepwise multiple regression is used to determine relationships between species richness, frequency, rarity and endemicity against potential geographical predictors. Stepwise logit regression is used to determine geographical predictors of species incidence on islands. Inter-island and inter-species associations have been examined using multivariate ordination and clustering techniques.
Results The Aegean butterfly fauna is characterized by decreasing diversity and rarity, and increasing homogeneity, from the periphery to the present geographical centre of the archipelago (Cyclades). Diversity and rarity are shown to relate closely to species richness, and species richness, in turn, is largely explained by contemporary geography, particularly the degree of isolation from the nearest mainland sources of Greece or Turkey, and island dimensions. Islands towards the centre of the archipelago are characterized by a group of mobile species (n ≥ 20 species) with extensive ranges across Europe; species that would have recolonized Santorini (Thira) following the VI6 eruption there c. 1630 bc. Endemic components, indicative of autochthonous evolutionary events, are few (5% of species are endemic) compared to known sedentary organisms (molluscs and isopods), but exceed those for more mobile animals (i.e. birds); their distribution is mainly confined to large isolated islands along the Aegean arc (i.e. Kriti) and in the Dodecanese group.
Main conclusions Contemporary geography, i.e. processes currently operating in ecological time, dominates butterfly diversity gradients (species richness, frequency, rarity and incidence) in the archipelago. Two reasons are suggested to account for the lack of endemism and the pattern of decreasing diversity into the Cyclades. First, relict butterfly elements may have become extinct on all but a few larger islands, particularly from environmental changes since the Neolithic (fire and overgrazing). Second, colonization from the continental landmasses is ongoing with more mobile species transferring even to the most isolated islands.
1. Predictions on species richness and incidence of species are made using data for three scales of mapping from the Greater Manchester butterfly atlas; for the whole of the conurbation (2 km×2 km scale) and for two sample areas centred on the Mersey Valley (1 km×1 km scale; 1 ha scale). Predictions are based on data for recording effort, altitude, biotopes, host‐plants and nectar resources.
2. Data for Greater Manchester indicate that substantial shortfalls may occur in recording butterfly species for atlases despite the fact that butterflies are generally easily identified and well supported with recorders. Shortfalls tend to be larger for species with fewer records, indicating that some species may be more easily overlooked than others.
3. The results demonstrate that targeting squares for re‐survey is necessary and feasible. The predictions have other valuable research applications, the most important of which is being able to assess the accuracy of distribution maps, to correct them, and to make projections of distribution changes.
4. Predictions may be enhanced by improvements to mapping in three ways: (i) Collecting data on recording effort. Variation in recording effort typically accounts for differences in species richness and incidence of species more than any other variable; (ii) Collecting data on biotopes and specific resources. The present results are promising and demonstrate that the collection of environmental data linked to a suitable sampling frame could facilitate knowledge of the distribution of species over extensive areas that remain under‐recorded; and (iii) Distinguishing between breeding individuals and vagrants. Vagrancy is a problem associated both with species and scale. Although species vary substantially in their capacity to migrate beyond their habitats, the effect of vagrancy on distribution maps becomes an increasingly large problem as the grain of mapping (size of recording units) decreases. It is suggested that over‐recording can be a problem, particularly when mapping is fine‐grained.
I draw attention to the need for ecologists to take spatial structure into account more seriously in hypothesis testing. If spatial autocorrelation is ignored, as it usually is, then analyses of ecological patterns in terms of environmental factors can produce very misleading results. This is demonstrated using synthetic but realistic spatial patterns with known spatial properties which are subjected to classical correlation and multiple regression analyses. Correlation between an autocorrelated response variable and each of a set of explanatory variables is strongly biased in favour of those explanatory variables that are highly autocorrelated - the expected magnitude of the correlation coefficient increases with autocorrelation even if the spatial patterns are completely independent. Similarly, multiple regression analysis finds highly autocorrelated explanatory variables “significant” much more frequently than it should. The chances of mistakenly identifying a “significant” slope across an autocorrelated pattern is very high if classical regression is used. Consequently, under these circumstances strongly autocorrelated environmental factors reported in the literature as associated with ecological patterns may not actually be significant. It is likely that these factors wrongly described as important constitute a red-shifted subset of the set of potential explanations, and that more spatially discontinuous factors (those with bluer spectra) are actually relatively more important than their present status suggests. There is much that ecologists can do to improve on this situation. I discuss various approaches to the problem of spatial autocorrelation from the literature and present a randomisation test for the association of two spatial patterns which has advantages over currently available methods.
Practical conservation activity is increasing globally and is being undertaken by many different government and nongovernmental organizations. In the majority of cases, justification for proposed actions is experience-based rather than evidence-based, action is often taken without monitoring or evaluation of effectiveness, and results are rarely widely disseminated. Conservation has been compared with medicine as a crisis discipline in which action is often required urgently in the absence of good information. The practice of medicine has recently gone through an effectiveness revolution that has improved the criteria upon which treatment strategies are based by progressing from reliance on personal experience to reliance on scientific' evidence. We draw parallels between medicine and conservation and present a practical framework to encourage evidence-based conservation action. Our rationale is that conservation actions for which scarce resources are sought should be justified by good scientific evidence. In our view this will also encourage more research addressing practical issues in conservation management.
Data from the Greater Manchester Butterfly Atlas (UK) reveal a highly significant and substantial impact of visits on both species' richness and species' incidence in squares. This effect has been demonstrated for three different zones mapped at different scales. The significant impact of number of visits persists when data are amalgamated for coarser scales. The findings demonstrate that it is essential for distribution mapping projects to record data on recording effort as well as on the target organisms. Suggestions are made as to how distribution mapping may be improved, including a geographically and environmentally representative structure of permanently monitored squares and closer links between distribution mapping and the Butterfly Monitoring Scheme (BMS), which primarily monitors changes in butterfly populations. The benefit to conservation will be data that can be better used to analyse the reasons for changes in ranges and distributions, fundamental for determining priorities and policy decisions.