When the method for mapping species matters: defining priority areas for conservation of African freshwater turtles

Aim Defining priority areas for conservation is essential to minimize biodiversity loss, but the adoption of different methods for describing species distributions influences the outcomes. In order to provide a robust basis for the conservation of freshwater turtles in Africa, we compared the effect that different species‐mapping approaches had on derived patterns of species richness, species vulnerability and protected‐area representativeness. Location Africa. Methods We adopted three different approaches with increasing complexity for generating species distribution maps. The first approach was based on the geographic intersection of species records and grid squares; the second on the union of local convex polygons; and the third on inductive distribution modelling techniques. We used distribution maps, generated using these three approaches, to determine conservation priorities based on geographic patterns of species richness and vulnerability, as well as for conducting gap and irreplaceability analyses. Results We obtained markedly different distribution maps using the three methods, which in turn caused differences in conservation priorities. The grid‐square approach underestimated range sizes and species richness, while the polygon approach overestimated these attributes. The distribution modelling approach provided the most realistic outcome in terms of diversity patterns, by minimizing both commission and omission errors. An integrated map of conservation priority – derived by combining individual measures of priority based on the distribution modelling approach – identified the Gulf of Guinea coast and the Albertine Rift as major priority areas. Main conclusions Each species‐mapping approach has both advantages and disadvantages. The choice of the most appropriate approach in any given situation depends on the availability of locality records and on the relative importance of mitigating omission and commission errors. Our findings suggest that in most circumstances, the use of distribution modelling has many advantages relative to the other approaches. The priority areas identified in this study should be considered for targeting efforts to conserve Africa freshwater turtles in the coming years.     

Geographic sampling bias in the South African Frog Atlas Project: implications for conservation planning

Quality conservation planning requires quality input data. However, the broad scale sampling strategies typically employed to obtain primary species distribution data are prone to geographic bias in the form of errors of omission. This study provides a quantitative measure of sampling bias to inform accuracy assessment of conservation plans based on the South African Frog Atlas Project. Significantly higher sampling intensity near to cities and roads is likely to result in overstated conservation priority and heightened conservation conflicts in urban areas. Particularly well sampled protected areas will also erroneously appear to contribute highly to amphibian biodiversity targets. Conversely, targeted sampling in the arid northwest and along mountain ranges is needed to ensure that these under-sampled regions are not excluded from conservation plans. The South African Frog Atlas Project offers a reasonably accurate picture of the broad scale west-to-east increase in amphibian richness and abundance, but geographic bias may limit its applicability for fine scale conservation planning. The Global Amphibian Assessment species distribution data offered a less biased alternative, but only at the cost of inflated commission error.     

How well do the existing and proposed reserve networks represent vertebrate species in Chile?

Increasingly, biogeographical knowledge and analysis are playing a fundamental role in assessing the representativeness of biodiversity in protected areas, and in identifying critical areas for conservation. With almost 20% of the country assigned to protected areas, Chile is well above the conservation target (i.e. 10–12%) proposed by many international conservation organizations. Moreover, the Chilean government has recently proposed new conservation priority sites to improve the current protected area network. Here, we used all 653 terrestrial vertebrate species present in continental Chile to assess the performance of the existing and proposed reserve networks. Using geographical information systems, we overlaid maps of species distribution, current protected areas, and proposed conservation priority sites to assess how well each species is represented within these networks. Additionally, we performed a systematic reserve selection procedure to identify alternative conservation areas for expanding the current reserve system. Our results show that over 13% of the species are not covered by any existing protected area, and that 73% of Chilean vertebrate species can be considered partial gaps, with only a small fraction of their geographical ranges currently under protection. The coverage is also deficient for endemic (species confined to Chile) and threatened species. While the proposed priority sites do increase coverage, we found that there are still several gaps and these are not the most efficient choices. Both the gap analysis and the reserve selection analysis identified important areas to be added to the existing reserve system, mostly in northern and central Chile. This study underscores the need for a systematic conservation planning approach to redefine the conservation priority sites in order to maximize the representation of species, particularly endemic and threatened species.     

Data Acquisition for Conservation Assessments: Is the Effort Worth It?

When identifying conservation priorities, the accuracy of conservation assessments is constrained by the quality of data available. Despite previous efforts exploring how to deal with imperfect datasets, little is known about how data uncertainty translates into errors in conservation planning outcomes. Here, we evaluate the magnitude of commission and omission error, effectiveness and efficiency of conservation planning outcomes derived from three datasets with increasing data quality. We demonstrate that investing in data acquisition might not always be the best strategy as the magnitude of errors introduced by new sites/species can exceed the benefits gained. There was a trade-off between effectiveness and efficiency due to poorly sampled rare species. Given that data acquisition is limited by the high cost and time required, we recommend focusing on improving the quality of data for those species with the highest level of uncertainty (rare species) when acquiring new data.     

What spatial data do we need to develop global mammal conservation strategies?

Spatial data on species distributions are available in two main forms, point locations and distribution maps (polygon ranges and grids). The first are often temporally and spatially biased, and too discontinuous, to be useful (untransformed) in spatial analyses. A variety of modelling approaches are used to transform point locations into maps. We discuss the attributes that point location data and distribution maps must satisfy in order to be useful in conservation planning. We recommend that before point location data are used to produce and/or evaluate distribution models, the dataset should be assessed under a set of criteria, including sample size, age of data, environmental/geographical coverage, independence, accuracy, time relevance and (often forgotten) representation of areas of permanent and natural presence of the species. Distribution maps must satisfy additional attributes if used for conservation analyses and strategies, including minimizing commission and omission errors, credibility of the source/assessors and availability for public screening. We review currently available databases for mammals globally and show that they are highly variable in complying with these attributes. The heterogeneity and weakness of spatial data seriously constrain their utility to global and also sub-global scale conservation analyses.     

Biased data reduce efficiency and effectiveness of conservation reserve networks

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

Patterns of species richness hotspots and estimates of their protection are sensitive to spatial resolution

Aim

Species richness is a measure of biodiversity often used in spatial conservation assessments and mapped by summing species distribution maps. Commission errors inherent those maps influence richness patterns and conservation assessments. We sought to further the understanding of the sensitivity of hotspot delineation methods and conservation assessments to commission errors, and choice of threshold for hotspot delineation.

Location

United States.

Methods

We created range maps and 30‐m and 1‐km resolution habitat maps for terrestrial vertebrates in the United States and generated species richness maps with each dataset. With the richness maps and the GAP Protected Areas Dataset, we created species richness hotspot maps and calculated the proportion of hotspots within protected areas; calculating protection under a range of thresholds for defining hotspots. Our method allowed us to identify the influence of commission errors by comparing hotspot maps.

Results

Commission errors from coarse spatial grain data and lack of porosity in the range data inflated richness estimates and altered their spatial patterns. Coincidence of hotspots from different data types was low. The 30‐m hotspots were spatially dispersed, and some were very long distances from the hotspots mapped with coarser data. Estimates of protection were low for each of the taxa. The relationship between estimates of hotspot protection and threshold choice was nonlinear and inconsistent among data types (habitat and range) and grain size (30‐m and 1‐km).

Main conclusions

Coarse mapping methods and grain sizes can introduce commission errors into species distribution data that could result in misidentifications of the regions where hotspots occur and affect estimates of hotspot protection. Hotspot conservation assessments are also sensitive to choice of threshold for hotspot delineation. There is value in developing species distribution maps with high resolution and low rates of commission error for conservation assessments.     

Nature reserve selection in the Transvaal, South Africa: what data should we be using?

Iterative reserve selection algorithms were applied to two mammal databases, generalized to sixteenth degree grid squares, for the Transvaal region of South Africa. Based on primary point data, 24 grid squares are required to represent all species at least once, while only 13 grid squares are required when based on distribution map data; only two of these grid squares are common to both analyses. As the number of representations per species is increased from one to five, the number of selected grid squares increased to 86 and 71 or 72 respectively, with only 17 of these common to both analyses. These differences in the selection of sites are further reflected in the degree of congruence between selected grid squares and existing conservation areas which is on average 63.3% for grid squares selected from the primary database and only 42.5% for those selected from the distribution map database. These results emphasize the importance of quality data input when evaluating regional reserve networks. Highly generalized distribution map data sets, on the one hand, are extrapolations of limited data sets and contain non-quantifiable levels of false-positives which could have significant implications if used for establishing regional reserve networks. On the other hand, although there are problems associated with the establishment of primary diversity databases, namely data currency and uneven and non-random sampling (leading to false negatives), they remain our most reliable option for assigning conservation value.     

Species distributions represent intraspecific genetic diversity of freshwater fish in conservation assessments

相似文献   

Logic for designing nature reserves for multiple species

We examine the logic of designing nature reserves to understand better how to integrate the concepts of representativeness and persistence. Simple models of viability are used to evaluate how the expected number of species in the reserve system changes with variation in the risk of extinction among species, their rate of occurrence, and the distribution of species. The optimal size of individual reserves increased with the mean and variance of the probability of extinction among species and with the rate at which the risk of extinction declines with the cost of each reserve. In contrast, the rate of occurrence of species within reserves and their rate of accumulation with increasing reserve area had a relatively minor influence on the optimal size of reserves. Patterns of endemism were most important for the location of reserves. Including differences among species in the analysis reduced the optimal number of individual reserves (and increased the size of each) when operating under a fixed budget compared with reserve designs based on single species. A case study in the city of Melbourne, Australia, demonstrates the conservation value of small (approximately 1 ha) grassland reserves and the underrepresentation of Melbourne's volcanic plains in the region's conservation network.     

How well do Important Bird Areas represent species and minimize conservation conflict in the tropical Andes?

Where high species richness and high human population density coincide, potential exists for conflict between the imperatives of species conservation and human development. We examine the coincidence of at‐risk bird species richness and human population in the countries of the tropical Andes. We then compare the performance of the expert‐driven Important Bird Areas (IBA) scheme against a hypothetical protected‐areas network identified with a systematic reserve selection algorithm seeking to maximize at‐risk bird species representation. Our aim is to assess the degree to which: IBAs contain a higher richness of at‐risk species than would be expected by chance, IBAs contain more people than would be expected by chance, and IBAs are congruent with complementary areas that maximize species representation with an equivalent number of sites. While the correlation of richness and population was low for the region as a whole, representation of all at‐risk bird species required many sites to be located in areas of high human population density. IBA sites contained higher human population densities than expected by chance (P < 0.05) and were markedly less efficient in representing at‐risk bird species of the region than sites selected using the reserve selection algorithm. Moreover, overlap between IBAs and these latter sites was very limited. Expert‐driven selection procedures may better reflect existing sociopolitical forces, including land ownership and management regimes, but are limited in their ability to develop an efficient, integrated network of sites to represent priority species. Reserve selection algorithms may serve this end by optimizing complementarity in species representation among selected sites, whether these sites are adopted independently or as a supplement to the existing reserve network. As tools of site selection, they may be particularly useful in areas such as the tropical Andes where complex patterns of species disjunction and co‐occurrence make the development of representative reserve networks particularly difficult. Furthermore, they facilitate making spatially explicit choices about how reserve sites are located in relation to human populations. We advocate their use not in replacement of approaches such as the IBA initiative but as an additional, complementary tool in ensuring that such reserve networks are developed as efficiently as practically possible.     

Threatened and endemic species: are they good indicators of patterns of biodiversity on a national scale?

Endemic and/or threatened species are often targeted to set conservation priorities. It is tempting to assume that a reserve network focusing on these species will be an effective umbrella for overall species richness of a country. For South Africa and Lesotho we tested whether complementary networks selected for threatened and/or endemic bird species satisfactorily represent all bird species, both in terms of capturing areas where other species are present or areas where they are more abundant (and, presumably, more viable). We found that areas selected for threatened and endemic species perform considerably better than areas selected at random. However, they do not guarantee the representation of overall bird species diversity, particularly not in peak abundance locations. Although nationally threatened and endemic species are important conservation targets, our results indicate that reserve networks focusing solely on these species may not be sufficient to preserve overall species diversity in a country.     

Reserve selection and persistence: complementing the existing Atlantic Forest reserve system

This study is an exercise to check the efficiency of the existing reserve system, and to show how systematic conservation planning—using information available and the complementarity concept—can improve the basis for decisions and minimize costs. We verified the performance, in number of cells and primate species representation, of the existing Atlantic Forest (Brazil) reserve network with a quarter-degree resolution grid, with 1,884 cells. We used occurrence data of 20 endemic primate species, and the maps of 237 existing reserves. Reserve networks were selected to represent primate species first considering no pre-existing reserves in Atlantic Forest, and then, considering the existing reserve system, taking into account the minimum area for viable population of the larger species (Northern muriqui Brachyteles hypoxanthus). Reserve selection was carried out using the complementarity concept implemented by a simulated annealing algorithm. Primate species representation (at least one occurrence in the network) could be achieved with 8% of the existing reserve system (nine cells in relation to the 120 in the existing reserve system). We found that today’s reserve system represents 89% of endemic primate species, excluding the species Coimbra Filho’s titi monkey (Callicebus coimbrai) and Marcgraf’s capuchin (Cebus flavius). The networks selected without considering existing reserves contained nine cells. The networks selected considering existing reserves (120 cells), had two new cells necessary to represent all the primates. This does not mean that a viable alternative is to start from zero (i.e., nonexistent reserves). Identifying critical supplementary areas using biodiversity information to fill the gaps and then starting “conservation in practice” in these areas should be priorities.     

An automatic method for removing empty camera trap images using ensemble learning

Camera traps often produce massive images, and empty images that do not contain animals are usually overwhelming. Deep learning is a machine‐learning algorithm and widely used to identify empty camera trap images automatically. Existing methods with high accuracy are based on millions of training samples (images) and require a lot of time and personnel costs to label the training samples manually. Reducing the number of training samples can save the cost of manually labeling images. However, the deep learning models based on a small dataset produce a large omission error of animal images that many animal images tend to be identified as empty images, which may lead to loss of the opportunities of discovering and observing species. Therefore, it is still a challenge to build the DCNN model with small errors on a small dataset. Using deep convolutional neural networks and a small‐size dataset, we proposed an ensemble learning approach based on conservative strategies to identify and remove empty images automatically. Furthermore, we proposed three automatic identifying schemes of empty images for users who accept different omission errors of animal images. Our experimental results showed that these three schemes automatically identified and removed 50.78%, 58.48%, and 77.51% of the empty images in the dataset when the omission errors were 0.70%, 1.13%, and 2.54%, respectively. The analysis showed that using our scheme to automatically identify empty images did not omit species information. It only slightly changed the frequency of species occurrence. When only a small dataset was available, our approach provided an alternative to users to automatically identify and remove empty images, which can significantly reduce the time and personnel costs required to manually remove empty images. The cost savings were comparable to the percentage of empty images removed by models.     

Insights into the area under the receiver operating characteristic curve (AUC) as a discrimination measure in species distribution modelling

Aim The area under the receiver operating characteristic (ROC) curve (AUC) is a widely used statistic for assessing the discriminatory capacity of species distribution models. Here, I used simulated data to examine the interdependence of the AUC and classical discrimination measures (sensitivity and specificity) derived for the application of a threshold. I shall further exemplify with simulated data the implications of using the AUC to evaluate potential versus realized distribution models. Innovation After applying the threshold that makes sensitivity and specificity equal, a strong relationship between the AUC and these two measures was found. This result is corroborated with real data. On the other hand, the AUC penalizes the models that estimate potential distributions (the regions where the species could survive and reproduce due to the existence of suitable environmental conditions), and favours those that estimate realized distributions (the regions where the species actually lives). Main conclusions Firstly, the independence of the AUC from the threshold selection may be irrelevant in practice. This result also emphasizes the fact that the AUC assumes nothing about the relative costs of errors of omission and commission. However, in most real situations this premise may not be optimal. Measures derived from a contingency table for different cost ratio scenarios, together with the ROC curve, may be more informative than reporting just a single AUC value. Secondly, the AUC is only truly informative when there are true instances of absence available and the objective is the estimation of the realized distribution. When the potential distribution is the goal of the research, the AUC is not an appropriate performance measure because the weight of commission errors is much lower than that of omission errors.     

Spatial and temporal variations in species occurrence rate affect the accuracy of occurrence models

Aim Predictive models of species occurrence have potential for prioritizing areas for competing land uses. Before widespread application, however, it is necessary to evaluate performance using independent data and effective accuracy measures. The objectives of this study were to (1) compare the effects of species occurrence rate on model accuracy, (2) assess the effects of spatial and temporal variation in occurrence rate on model accuracy, and (3) determine if the number of predictor variables affected model accuracy. Location We predicted the distributions of breeding birds in three adjacent mountain ranges in the Great Basin (Nevada, USA). Methods For each of 18 species, we developed separate models using five different data sets — one set for each of 2 years (to address the effects of temporal variation), and one set for each of three possible pairs of mountain ranges (to address the effects of spatial variation). We evaluated each model with an independent data set using four accuracy measures: discrimination ability [area under a receiver operating characteristic curve (AUC)], correct classification rate (CCR), proportion of presences correctly classified (sensitivity), and proportion of absences correctly classified (specificity). Results Discrimination ability was not affected by occurrence rate, whereas the other three accuracy measures were significantly affected. CCR, sensitivity and specificity were affected by species occurrence rate in the evaluation data sets to a greater extent than in the model‐building data sets. Discrimination ability was the only accuracy measure affected by the number of variables in a model. Main conclusions Temporal variation in species occurrence appeared to have a greater impact than did spatial variation. When temporal variation in species distributions is great, the relative costs of omission and commission errors should be assessed and long‐term census data should be examined before using predictive models of occurrence in a management setting.     

Evaluating the costs and benefits of systematic data acquisition for conservation assessments

Effective decision‐making in conservation often is constrained by data quality. Uncertainties associated with poor quality or sparse data can lead to the misuse of limited resources and potentially the failure of conservation practice. Data acquisition, which can help improve decision‐making, is constrained by limited budgets and time. This is especially concerning for rare species, the most in need of conservation, but the most difficult to accurately represent in conservation plans. Here we test the suitability of three different sampling design strategies (two systematic vs random) designed to improve the quality of information available for conservation planning involving rare species. We modelled the spatial distribution of freshwater fish species in a data rich area in northern Australia using a large dataset (representing the best attainable data or true distribution) and simulate increasing subsets of data acquired through the three alternative sampling designs. We then evaluated omission and commission errors in conservation planning outcomes, efficiency and return on investment of data acquisition for conservation planning outcomes obtained from the different data availability × sampling design strategies. Even though we were able to find new species more effectively through systematic sampling designs, this did not 1) translate into reduced errors in conservation planning outcomes for rare species and 2) meet our goal of enhancing cost‐effectiveness of conservation planning. Our results suggest that collecting more biodiversity data, irrespective of the sampling design used, does not necessarily reduce data uncertainty issues and could lead to the misuse of the limited resources and ultimately the failure of conservation practice.     

Complementary representation and zones of ecological transition

Minimum complementary sets of sites that represent each species at least once have been argued to provide a nominal core reserve network and the starting point for regional conservation programs. However, this approach may be inadequate if there is a tendency to represent several species at marginal areas within their ranges, which may occur if high efficiency results from preferential selection of sites in areas of ecological transition. Here we use data on the distributions of birds in South Africa and Lesotho to explore this idea. We found that for five measures that are expected to reflect the location of areas of ecological transition, complementary sets tend to select higher values of these measures than expected by chance. We recommend that methods for the identification of priority areas for conservation that incorporate viability concerns be preferred to minimum representation sets, even if this results in more costly reserve networks.     

The value of coarse species range maps to inform local biodiversity conservation in a global context

Range maps of thousands of species, compiled and made freely available by the International Union for Conservation of Nature, are being increasingly applied to support spatial conservation planning. However, their coarse nature makes them prone to commission and omission errors, and they lack information on the variations in abundance within species’ distributions, calling into question their value to inform decisions at the fine scales at which conservation often takes place. Here, we tested if species ranges can reliably be used to estimate the responsibility of sites for the global conservation of species. We defined ‘specific responsibility’ as the fraction of a species’ population within a given site, considering it useful for prioritising species within sites; and defined ‘overall responsibility’ as the sum of specific responsibility across species within a site, assuming it informative of priorities among sites. Taking advantage of an exceptionally detailed dataset on the distribution and abundance of bird species at a near‐continental scale – a level of information rarely available to local decision‐makers – we created a benchmark against which we tested estimates of responsibility derived from range maps. We investigated approaches for improving these estimates by complementing range maps with plausibly available local data. We found that despite their coarse nature, range maps provided good estimates of sites’ overall responsibility, but relatively poor estimates of specific responsibility. Estimates were improved by combining range maps with local species lists or local abundance data, easily available through local surveys on the sites of interest, or simulated expert knowledge. Our results suggest that combining range maps with local data is a promising route for improving the effectiveness of local conservation decisions at contributing to reducing global biodiversity losses. This is all the more urgent in hyper‐diverse poorly‐known regions where conservation‐relevant decisions must proceed despite a paucity of biodiversity data.     

Misuse of bird digital distribution maps creates reversed spatial diversity patterns in the Amazon

It is well known that bird richness in the Amazon is greater in upland forests and that seasonally flooded forest is particularly species poor. However, the misleading pattern of greater bird richness in seasonally flooded forest has emerged seemingly unnoticed numerous times in richness maps in the literature. We hypothesize that commission errors in digital distribution maps (DDMs) are the cause behind the misleading richness pattern. In the Amazon, commission errors are a consequence of the different methodological treatment given to large‐ranged versus small‐ranged habitat specialists when mapping distributions. DDMs of 1007 Amazonian birds were examined, and maps that had commission errors were corrected. We generated two richness maps, one from the overlay of original DDMs and another from the overlay of the corrected ones. We identified 291 species whose distribution maps had errors. In the original data, seasonally flooded forests showed higher species richness than upland forest, but this pattern was reverted in the corrected richness map. Commission errors were 35 times more likely in the seasonally flooded forest. We conclude that DDMs accurately portray the distribution of single species in the Amazon. Commission errors in individual maps, however, accumulate when they are overlaid, explaining the misleading pattern for birds in the Amazon. DDMs can continue to be used mapping richness, as long as, at a regional scale: (1) basic map refinements are carried, or (2) only small‐range species are used for mapping species richness.