Fig 5 - uploaded by Uri Roll
Content may be subject to copyright.
![Flowchart for classification tasks in automated extinction risk assessment method, using the XGBoost algorithm [29]. Green boxes represent outcomes of the binary task and red boxes represent the outcome of the specific tasks. Steps taken for each classification task (blue circle) are indicated after the asterisk. CR, Critically Endangered; EN, Endangered; LC, Least Concern; NT, Near Threatened; VU, Vulnerable.](profile/Uri-Roll/publication/360877506/figure/fig5/AS:1160075527897093@1653594967484/Flowchart-for-classification-tasks-in-automated-extinction-risk-assessment-method-using.png)
Flowchart for classification tasks in automated extinction risk assessment method, using the XGBoost algorithm [29]. Green boxes represent outcomes of the binary task and red boxes represent the outcome of the specific tasks. Steps taken for each classification task (blue circle) are indicated after the asterisk. CR, Critically Endangered; EN, Endangered; LC, Least Concern; NT, Near Threatened; VU, Vulnerable.
Source publication
The Red List of Threatened Species, published by the International Union for Conservation of Nature (IUCN), is a crucial tool for conservation decision-making. However, despite substantial effort, numerous species remain unassessed or have insufficient data available to be assigned a Red List extinction risk category. Moreover, the Red Listing proc...
Context in source publication
Context 1
... initial separation enabled different hyperparameter tuning, feature selection, and model fitting for each extent of occurrence class. Next, we used a decision tree (Fig 5) involving 4 hierarchical classification tasks for each extent of occurrence class: (1) separating threatened (CR, EN, and VU) from nonthreatened (NT and LC) species (binary classification); (2) separating CR species from other threatened species (EN and VU); (3) separating EN from VU in the remaining threatened species; and (4) separating NT from LC in the pool of nonthreatened species. We repeated this modeling approach after excluding threatened species not categorized under criterion B (360 species), to explore the amount of uncertainty introduced by the other Red List assessment criteria, which are less commonly used for reptiles. ...
Citations
... For instance, Meiri et al. (2018) showed that 927 lizards were found at a single location, with the last confirmed sightings of 171 species between 1830 and 1967. A recent automated assessment (Caetano et al., 2022) suggested that datadeficient species were more likely to be threatened than categorized species (27% vs. 21%). The underestimation of extinction risk mostly relates to incomplete distribution records, cryptic diversity and taxonomic uncertainties (Linnaean and Wallacean shortfalls ;Meiri et al., 2018). ...
... For instance, we found that newly described species were more likely to be narrow endemics and at higher risk of extinction. This reflects that, in many cases, the truly rare and small-ranged species lack information to be assigned an extinction risk category or to be included in conservation prioritization (Caetano et al., 2022). ...
... Therefore, we should be concerned that many species might go extinct in the next few decades if no immediate conservation action is taken (Caetano et al., 2022). ...
Aims: Identifying major reasons for species imperilment is a necessary step for conservation, yet the degree to which we can generalize is hard for species-rich yet less-studied taxa, such as lizards. Here, we aim to bridge the gap by providing comprehensive analyses of the correlates and processes of species extinction and threats for global lizards.
Location: Global
Time period: Current
Major taxa studied: Lizards
Methods: We compiled a dataset comprising extinction risk status, six intrinsic traits, and seven extrinsic factors for 5256 lizard species. We carried out binomial distribution tests for 43 families and seven realms to check the non-randomness in species’ extinction risk and then employed phylogenetic linear regressions to identify the key factors that relate to the extinction proneness of lizards and species subgroups. Based on the IUCN threat assessment, we identified major threats for global lizards and for major families and regions.
Results: We found strong evidence of taxonomic and geographical non-randomness in the extinction risk of lizards. Geographical range size, human footprint and density, insular endemism, temperature and precipitation seasonality, and body size were key predictors of extinction risk, and the first three factors were also important across families and realms. Moreover, newly described species were more likely to have a restricted range size and a higher extinction risk. Globally, the most detrimental threat was habitat destruction, while overexploitation, species invasion, and climate change varied widely in importance among species groups.
Main conclusions: Overall, we highlight the detrimental influences of range restriction, climate variability, and anthropogenic threats to species persistence. We suggest that lizards are potentially at high risk of extinction due to widespread human disturbance and species with extinction-prone traits require conservation prioritization. Moreover, lizards of different families and regions require different management strategies because of variation in extinction-risk correlates and threats.
... Instead, when comparative extinction risk analyses have a predictive goal, species classified under criterion B are not excluded in order to generate the best possible predictions (see e.g. Zizka et al. 2021, Caetano et al. 2022. As the main goal of our models is to obtain the best predictions of extinction risk to assess possible mismatches with official RL assessments, rather than evaluating the actual importance of range size in comparison to other variables, we retained species assessed under criterion B. ...
... It should be considered, however, that when predictions are made for species in families not included in the training dataset, our family-block validation likely tends to under-estimate the models' true predictive ability. As expected, random cross-validation led to substantially better estimates, which are comparable to previous comparative extinction risk analysis of mammals (Bland et al. 2015), reptiles (Caetano et al. 2022), andplants (Zizka et al. 2021), and are higher than the latest analysis on amphibians (González-del-Pliego et al. 2019). ...
Assessing the extinction risk of species through the IUCN Red List is key to guiding conservation policies and reducing biodiversity loss. This process is resource-demanding, however, and requires a continuous update which becomes increasingly difficult as new species are added to the IUCN Red List. The use of automatic methods, such as comparative analyses to predict species extinction risk, can be an efficient alternative to maintaining up to date assessments. Using amphibians as a study group, we predict which species were more likely to change status, in order to suggest species that should be prioritized for reassessment. We used species traits, environmental variables, and proxies of climate and land-use change as predictors of the IUCN Red List category of species. We produced an ensemble prediction of IUCN Red List categories by combining four different model algorithms: Cumulative Link Models (CLM), phylogenetic Generalized Least Squares (PGLS), Random Forests (RF), Neural Networks (NN). By comparing IUCN Red List categories with the ensemble prediction, and accounting for uncertainty among model algorithms, we identified species that should be prioritized for future reassessments due to high prediction versus observation mismatch. We found that CLM and RF performed better than PGLS and NN, but there was not a clear best algorithm. The most important predicting variables across models were species range size, climate change, and land-use change. We propose ensemble modelling of extinction risk as a promising tool for prioritizing species for reassessment while accounting for inherent models' uncertainty.
... We suspect that many Data Deficient species would be assessed as threatened in the future because they often have similar attributes (e.g., small ranges, high levels of human pressure across their ranges) to species assessed as threatened (Gumbs et al., 2020). This assertion is supported by recent studies using an automated extinction risk assessment process through machine learning that has predicted that Data Deficient reptile species are substantially more likely to fall within a threat category than expected given the overall percentage of threatened reptiles (Borgelt et al., 2022;Caetano et al., 2022). A prediction of higher-than-average risk was also made for reptile species that have not yet been evaluated (i.e., ...
... A prediction of higher-than-average risk was also made for reptile species that have not yet been evaluated (i.e., Not Evaluated (NE) category; Caetano et al., 2022) which are usually small ranged (Meiri, 2016). Thirty-nine reptile species assessed as Data Deficient in 2009 have been reassessed as non-DD (IUCN, 2022). ...
... We view this as suggesting that newly described, and Data Deficient species, are more likely to be threatened than a naïve prediction based on the overall proportion of threatened species would suggest. Predictive modelling approaches (e.g., Borgelt et al., 2022;Caetano et al., 2022) could be used to identify priority Data Deficient, and unassessed (including newly described) species for targeted study, and subsequent assessment. Currently, such models are all we have to approximate the conservation status of Data Deficient and Not Evaluated species, and to potentially prioritize the development of research and conservation actions for them. ...
The IUCN recently coordinated the first assessment of extinction risk of the world's reptile species. This monumental undertaking allows, for the first time, an examination of threats and prioritization of conservation effort, not just for reptiles, but for land vertebrates as a whole. Reptiles are now the largest class of land vertebrates in terms of species numbers. The dynamic nature of reptile taxonomy, the 18 years it took for the Global Reptile Assessment to be completed, the poor state of knowledge for many species – especially of squamates – and the evolving nature of threats, however, all highlight the need for continued monitoring of reptile species and threats. Here we review the status of reptile conservation assessments, and identify the challenges facing the next reptile assessments. We then recommend potential avenues that could facilitate efficient, accurate and timely future assessments.
... This assessment gap is particularly evident in New Guinea and Eastern Wallacea (Fig. 3). Data Deficient species are likely to be at least as threatened (Bland and Böhm 2016), or even more so (Gumbs et al. 2020;Caetano et al. 2022), than the general threat level for other reptiles. Species that have not yet been assessed are likely to be even more threatened due to their smaller geographic range sizes (Meiri 2016). ...
Context. Skinks comprise the dominant component of the terrestrial vertebrate fauna in Oceania, New Guinea, and Eastern Wallacea (ONGEW). However, knowledge of their diversity is incomplete, and their conservation needs are poorly understood. Aims. To explore the diversity and threat status of the skinks of ONGEW and identify knowledge gaps and conservation needs. Methods. We compiled a list of all skink species occurring in the region and their threat categories designated by the International Union for Conservation of Nature. We used available genetic sequences deposited in the National Center for Biotechnology Information's GenBank to generate a phylogeny of the region’s skinks. We then assessed their diversity within geographical sub-divisions and compared to other reptile taxa in the region. Key results. Approximately 300 species of skinks occur in ONGEW, making it the second largest global hotspot of skink diversity following Australia. Many phylogenetic relationships remain unresolved, and many species and genera are in need of taxonomic revision. One in five species are threatened with extinction, a higher proportion than almost all reptile families in the region. Conclusions. ONGEW contain a large proportion of global skink diversity on <1% of the Earth’s landmass. Many are endemic and face risks such as habitat loss and invasive predators. Yet, little is known about them, and many species require taxonomic revision and threat level re-assessment. Implications. The skinks of ONGEW are a diverse yet underexplored group of terrestrial vertebrates, with many species likely facing extreme risks in the near future
... In a new PLOS Biology paper, de Oliveira Caetano and colleagues [6] presented an innovative machine learning analysis to estimate the extinction risk of 4,369 reptile species that were unassessed or data deficient in the Red List. Meanwhile, in a recent Nature paper, Cox and colleagues [7] presented the results of the Global Reptile assessment, including extinction risk categories for ca. ...
... The model presented in [6] was 84% accurate in predicting Red List categories during cross-validation and found unassessed species to face higher risk compared to assessed species (27% versus 21% species threatened with extinction). The model's performance was higher compared to previous similar exercises, albeit prediction accuracy for certain categories (e.g., near threatened) was substantially lower than others (e.g., least concern). ...
... The automated assessment model in [6] showed high accuracy both in the interpolation and extrapolation of least concern species: 92% of the species newly assessed as least concern were correctly predicted by the model. This reflects the ability of automated methods to separate least concern species from the rest, which is a promising implementation for facilitating periodic reassessments [10]. ...
In a new PLOS Biology paper, de Oliveira Caetano and colleagues presented an innovative method to estimate extinction risk in reptile species worldwide. The method shows a promising avenue to support Red List assessment, alongside some well-known challenges. This Primer explores the implications of a new PLOS Biology study that presents an innovative method for estimating extinction risk in reptile species worldwide; this method represents a promising avenue to support Red List assessment, alongside some well-known challenges.
... However, a paucity of data for many taxonomic groups hinders such census efforts. Considering that, even for some well-studied taxonomic groups, the current rates of extinction are unknown or seem to have been substantially underestimated [3,4], global biodiversity is in jeopardy. ...
... A cursory survey of biodiversity-related papers that have been published in the journal in the past year reveals several principal topics. In this issue alone, we feature work addressing the biodiversity of marine communities half a billion years apart [5, 6] and the use of artificial intelligence to automate the surveillance of threatened species [3]. ...
... One study in this area reveals global and national inequities in monitoring species distributions [10], whereas a study of data in Australia-a nation with clear vested interest in its biological resources-shows that even there, many endangered species remain taxonomically undocumented, compromising their conservation [11]. Another paper proposes the use of machine learning to avoid the time-consuming and potentially biased reliance on human experts for judging extinction risk [3]. A fourth study intriguingly reveals the Newfoundland, Canada, subjects of a metacommunity analysis of shifts in biodiversity more than 540 million years ago [5]. ...
As the UN International Day for Biological Diversity enters its twentieth year, we take stock of recent developments and trends in biodiversity research and renew the call to build a better shared future for all life.
The eastern clades of the Spur-thighed Tortoise, Testudo graeca (Family Testudinidae),include small to medium-sized (maximum straightline carapace lengths [SCL] typically < 35.8 cm, but possibly up to 46.0 cm) tortoise subspecies clades native to the Balkans (southeastern Europe) and southwestern Asia. Within this vast geographic distribution, the species occupies Mediterranean,
Mediterranean-continental, and steppe climates, as well as the margins of hot deserts. In southeastern Turkey and Iran, it reaches elevations of 2500 m, the record for the species. Testudo graeca is characterized by the presence of a small conical keratinous spur on each posterior thigh, present in all eastern clade subspecies. The eastern clades show a very high degree of phenotypic variability,
ranging from almost entirely black individuals (typically, but not exclusively in the north of its range) to yellowish (flavistic) individuals (primarily in southern Israel). They also show variability in the structure and shape of the carapace, ranging from flatter steppe forms to domed, highershelled forms. Adult females are generally slightly larger and heavier than males in most eastern clade populations; however, in approximately a quarter of surveyed populations across the range, males reach a similar or larger average size. Female mean SCLs range from 14.5–21.9 cm per
population, with mean body mass of 700 g; male mean SCLs range from 13.6–24.1 cm with mean body mass of 420 g. Other male dimorphic characters are similar to the western clade, including a proportionally longer and thicker tail and plastral concavity. Females produce 1–3 clutches of 2–19 eggs annually between May and July. The species is still relatively common and widespread in a large part of its range. Some populations have declined drastically since the second half of the
20th century (as in Azerbaijan) and in others this decline has accelerated more recently, mainly in southwestern Russia, Armenia, and in the coastal regions of Israel. The main threat to the eastern subspecies clades of T. graeca is the degradation and loss of habitats by urbanization, bush fires, and agricultural mechanization, while the legal collection for the international pet trade, especially
to Europe, is currently much more regulated than in previous decades.
The Cameroon Volcanic Line, which divides the Congo Basin fauna from the West African fauna, is a known area of high endemism for various taxa, but the region’s bat fauna has received little attention. We review variation in morphological and molecular (mitochondrial Cytochrome b) characters in the Tropical African vespertilionid bat genus Pseudoromicia. Assessment of this variation indicates the existence of a new species of Pseudoromicia, from the Mbam Minkom Massif in the Centre Region of Cameroon. The new species is diagnosable by sequence data and is morphologically similar to its putative sister taxon, P. kityoi, from Uganda. Although we suggest that it be assigned to the IUCN category of Data Deficient, there should be some concern as to the conservation status of this species: the Mbam Minkom Massif ecosystem is threatened due to lack of legal conservation frameworks and exposure to increasing human pressure. The new species is in a clade with P. roseveari and P. kityoi. These three species may be relicts of a single widespread species originating in the West African “white-winged” group of Pseudoromicia that then spread east across the tropical moist broadleaf forest into East Africa, and are now restricted to a few upland rainforest patches in West Africa (P. roseveari), in outliers of the Cameroon Volcanic Line region (Pseudoromicia sp.), and the Lake Victoria area (P. kityoi). The smaller, white-winged species are hypothesized to be ancestral, with one extant putative species (P. rendalli) also extending into East and southern Africa (Van Cakenberghe & Happold, 2013 Van Cakenberghe, V., & Happold, M. (2013). Pipistrellus rendalli, Rendall’s Pipistrelle. In M. Happold & D. Happold, (Eds.), Mammals of Africa. Volume IV. Hedgehogs, shrews and bats (pp. 645–647). Bloomsbury Publishing. [Google Scholar]). The larger, dark-winged taxa likely dispersed to East Africa and subsequently back to West Africa (e.g., Pseudoromicia sp., P. roseveari). Our data illustrate the potential importance of the Dahomey Gap and climatic changes in the evolution of this group of species.
http://zoobank.org/urn:lsid:zoobank.org:pub:30929BAA-4043-422D-B209-16CAB826043E
Chameleons represent one of the most endangered groups of lizards, with most being characterized as threatened or near-threatened. Anthropogenic impacts on their native habitat are responsible for their declines. To protect these animals, we must develop functional assisted reproductive programs. However, a full understanding of their basic reproductive biology is required to develop such a program. This study represents the first attempt at characterizing the reproductive cycles of male chameleons. Sixteen veiled chameleons (Chamaeleo calyptratus) and 16 panther chameleons (Furcifer pardalis) being housed in captivity under static conditions were used for this study. Chameleons were anesthetized twice monthly for semen collection via electroejaculation, blood sampling to measure plasma testosterone concentrations, and ultrasonographic testicular measurements. Plasma testosterone concentrations were found to be significantly different between species (P = 0.02) and the interaction of species and month (P = 0.04). In veiled chameleons, plasma testosterone concentrations were significantly elevated during October through December, while for panther chameleons, plasma testosterone concentrations were elevated in July through September, and December. Electroejaculation success was found to be seasonal in both species. In veiled chameleons, semen collection was highest in April, May, and June, while in panther chameleons semen collection success was highest in March, April, May, and June. Median spermatozoa concentrations for veiled and panther chameleons were 6.66 and 7.52 x 107 million sperm/mL, respectively. Motility was lower than expected for both veiled chameleon (median 23%) and panther chameleons (median 25%). Spermatozoa clumping was a common finding. The majority of the spermatozoa (mean 51%) were...
The continuous decline in biodiversity despite global efforts to create new protected areas calls into question the effectiveness of these areas in conserving biodiversity. Numerous habitats are absent from the global protected area network, and certain taxonomic groups are not being included in conservation planning. Here, we analyzed the level of protection that the current protected area system provides to viper species in the Neotropical region through a conservation gap analysis. We used distribution size and degree of threat to set species-specific conservation goals for 123 viper species in the form of minimum percentage of their distribution that should be covered by protected areas, and assessed the level of protection provided for each species by overlapping their distribution with protected areas of strict protection. Furthermore, using species richness and evolutionary distinctiveness as priority indicators, we conducted a spatial association analysis to detect areas of special concern. We found that most viper species have <1/4 of their distribution covered by protected areas, including 22 threatened species. Also, the large majority of cells containing high levels of species richness were significantly absent from protected areas, while evolutionary distinctiveness was particularly unprotected in regions with relatively low species richness, like northern Mexico and the Argentinian dry Chaco. Our results provide further evidence that vipers are largely being excluded from conservation planning, leaving them exposed to serious threats that can lead to population decline and ultimately extinction.
