Philippa Kaur

Literature reviewWe conducted a comprehensive review of relevant literature published since 1995. Studies were extracted from the China National Knowledge Infrastructure and Web of Science using the following keywords: “N (nitrogen) loss OR NO (nitric oxide) emission OR N2O (nitrous oxide) emission OR NH3 (ammonia volatilization) emission OR NO3− (nitric leaching) OR N (nitrogen) runoff AND wheat AND China”. We excluded the following types of experiment: experiments not covering the entire wheat growing season, experiments conducted in greenhouses or laboratories, experiments without zero-N control, and experiments including manure, controlled release fertilizer, or inhibitors. In total, we extracted 941 observations from 138 articles, consisting of 121 observations of NO emission, 383 of N2O emission, 185 of NH3 emission, 188 of NO3− leaching, and 64 of Nr runoff. We also extracted data on N application rates, and climate and soil variables (Fig. 1). Missing climate data were obtained from China Meteorological Data Network (https://data.cma.cn/), miss values of soil organic carbon (SOC) and total N content were obtained from the National Scientific Fertilizer Network (http://kxsf.soilbd.com/), and missing soil silt, clay, sand content, bulk density, cation exchange capacity (CEC), and pH data were obtained from the Harmonized World Soil Database (HWSD) v. 1.2 (http://www.fao.org/soils-portal/soil-survey/soilmaps-and-databases/harmonized-world-soildatabase-v12/en). Based on this dataset, the EFs of Nr loss pathways were calculated by the following equation:$$E{F}_{i}=left({E}_{treatment}{rm{-}}{E}_{control}right){rm{/}}N;applied$$
(1)
where i = 1–5, represented NO, N2O, NH3, NO3− leaching and Nr runoff, respectively. Etreatment is the loss rate of experimental treatments with applied N fertilizer, Econtrol is the loss rate of experimental control without applied N fertilizer, and N applied is the N application rate corresponding to Etreatment. The resulting data was used to develop RF models to predict EFs of the five Nr loss pathways.Fig. 1The generate framework of the Nr loss from Chinese wheat system (Nr-Wheat) 1.0 database.Full size imageRF modelsRF models outperformed empirical models in previous studies15,18,19. We employed RF models to predict the EFs of NO, N2O, NH3, NO3− leaching, and Nr runoff. Environmental factors were selected via redundancy analysis20. Redundancy analysis, a basic ordination technique for gradients analysis, produces an ordination summarizing the variation in several response variables that can be best explained by a matrix of explanatory variables based on multiple linear regression. We conducted redundancy analysis using Canoco 5 to further analyze the effects of 10 environmental factors, including 4 soil physical factors (bulk density, silt, clay, and sand content), 4 soil chemical factors (pH, SOC, CEC and total N content), and 2 weather factors (total rainfall and mean temperature during the wheat growing period) of different EFs. Ultimately, the dataset of each pathway contained an ensemble of different environmental factors (Table 1).Table 1 Environmental factors were employed to build RF model for each pathway and total explanatory rates.Full size tableWhen establishing the RF model, the first step was to select k features from a total of m (k  More

Ambitious new targets are needed to conserve nature by protecting parks and species.Credit: Tang Dehong/VCG/Getty

It will take ample time and money to slow the world’s catastrophic loss of plant and animal species — and right now, both are running dangerously low. This year, nations are due to agree to an action plan to protect global biodiversity at the 15th Conference of the Parties (COP15) to the United Nations Convention on Biological Diversity. But the meeting is already two years late because of the pandemic, and China, which will host the conference in Kunming, has yet to set a new date.Now, conflicts over financing are adding to the tension. Conservation groups and advocates suggest that rich nations must donate at least US$60 billion annually to help less-affluent ones to fund projects such as protecting areas where wildlife can thrive and tackling the illegal wildlife trade that is driving hundreds of species to extinction. This is much more than the $4 billion to $10 billion that they are estimated to be spending today, and well below the amount they are giving low- and middle-income countries (LMICs) to fight climate change, which reached around $50 billion in 2019 according to one estimate. Yet limited overseas development funds are spread ever thinner as donors deal with the pandemic and now the fallout from Russia’s invasion of Ukraine. This is where COP15 is meant to deliver: as well as agreeing to the action plan, called the Global Biodiversity Framework, nations will be encouraged to pledge more money.A mix of public and private money has started to trickle in. Currently, biodiversity funding on the table ahead of COP15 amounts to roughly $5.2 billion per year, according to estimates by a group of five leading conservation organizations. Most comes from six governments, including France, the United Kingdom and Japan, and the European Union. In April, the Global Environment Facility (GEF) — a multilateral fund to support international environmental agreements — announced that, over the next four years, around $1.9 billion will go to projects dedicated to biodiversity. However, it’s unclear how much of this will come from the coffers that donor countries have already pledged.Some cash for conservation is coming from private philanthropic donors — such as $2 billion committed by entrepreneur Jeff Bezos last year. And starting in 2020, a group of financial institutions (now 89 of them) promised to annually report their financing activities and investments that affect biodiversity, and to move away from those that do harm — a form of ecological accounting that could help to shrink the budget needed to protect biodiversity. Donors will need to reach much deeper into their pockets to meet the demands of LMICs, the custodians of much of the world’s biodiversity. In March, a group of LMICs, led by Gabon, asked for $100 billion per year in new funding when officials met in Geneva, Switzerland, to discuss progress on the Global Biodiversity Framework. The LMICs want the money placed in a new multilateral fund for biodiversity, separate from, but complementary to, the GEF.Aside from cash, the fund will need to find a new home and structure — and there are a few options. A proposal from Brazil, circulated at the Geneva meeting, suggests the fund be governed by a board of 24 members, with an equal number from rich and lower-income nations. The board would be responsible for funding decisions and would prioritize projects that help to achieve the biodiversity convention’s goals. The pitch generated interest among some countries, but also concerns that it’s an attempt by Brazil to divert attention from its failure over the past few years to protect the Amazon rainforest and prevent other environmental harm.Another option is the Kunming Biodiversity Fund, which China announced in October last year to help LMICs to safeguard their ecosystems. It allocated 1.5 billion yuan (US$223 million) to seed the fund and invited other countries to contribute, but so far none has. Sources knowledgeable about the fund say that donor countries are reluctant to pitch in because China is holding on too tightly to the reins and is not involving others in its deliberations. Details of how the fund will operate are scarce, but Nature has learnt that China is floating the idea of housing it at the Asian Infrastructure Investment Bank (AIIB), based in Beijing. Set up in 2016, the AIIB has $100 billion in total capital and 105 members, including Germany, France and the United Kingdom. The AIIB has big green plans. By 2025, it wants half of all infrastructure projects it finances to focus on climate issues. With rigorous oversight and transparency, the AIIB would make a good home for the Kunming fund.As countries prepare to meet in Nairobi on 20–26 June in a last-ditch attempt to push the biodiversity framework forwards before COP15, China, as the host, must urgently provide stronger leadership on financing, including more transparency and engagement. Progress will require quick, generous contributions from donor nations — which should prioritize grants, not loans, for biodiversity projects.Holding the COP15 meeting must be a priority, too. As China tightens restrictions in the face of a COVID-19 surge, some researchers fear that delays will stretch on, stalling conservation work and leaving less time to meet biodiversity targets. China must either commit to holding the meeting this year or let it proceed elsewhere. One option being quietly discussed is moving the meeting to Canada — home of the United Nations biodiversity convention’s secretariat — and this deserves consideration. The world needs an ambitious biodiversity plan now — nature cannot wait. More

The custom code used to clean occurrence records and construct SDMs is available at (github.com/RhettRautsaw/ VenomMaps). We used the following R16 packages for data cleaning, manipulation, species distribution modeling, and Shiny app creation: tidyverse17 readxl18, data.table19, sf20, sp21,22, rgdal23, raster24, smoothr25, ape26, phytools27, argparse28, parallel16, memuse29, dismo30, rJava31, concaveman32, spThin33, usdm34, ENMeval35, kuenm36, shiny37, leaflet38, leaflet.extras39, leaflet.extras240, RColorBrewer41, ggpubr42, ggtext43, and patchwork44.Updating occurrence record taxonomyOur goal was to update and reconstruct the distributions of New World pitvipers. We used the Reptile Database45 (May 2021) as our primary source for current taxonomy which included the following genera: Agkistrodon, Atropoides, Bothriechis, Bothrocophias, Bothrops, Cerrophidion, Crotalus, Lachesis, Metlapilcoatlus, Mixcoatlus, Ophryacus, Porthidium, and Sistrurus. However, to ensure we captured all New World pitvipers records, we incorporated all members of the family Viperidae (all vipers and pitvipers) into our pipeline for updating occurrence record taxonomy (i.e., to account for errors in the recorded latitude, longitude, or if subfamily was not recorded).First, we collected global occurrence records for “Viperidae” from GBIF (downloaded 2021-08-19)46, Bison (downloaded 2021-08-19)47, HerpMapper (only New World taxa; downloaded 2021-08-19)48, Brazilian Snake Atlas49, BioWeb (downloaded 2021-07-07)50, unpublished data/databases from RMR, GJV, EPH, LRVA, MM, and CLP, and georeferenced literature records totaling 373,673 species-level records, 292,425 of which are New World pitvipers. Given the fluidity of taxonomy, records were often associated with outdated names. For example, Crotalus mitchelli pyrrhus was elevated to Crotalus pyrrhus51, but may still be recorded as the former in a given repository (e.g., GBIF). To correct taxonomy in our database, we checked records against a list of synonyms found on the Reptile Database and compared them to current taxonomy. If species and subspecies columns matched the same taxon (or no subspecies was recorded), then species IDs were not altered. If species and subspecies IDs did not match the same taxon, we updated taxonomy by minimizing the number of changes required to a given character string. We then manually checked all changes.Constructing distribution mapsNext, we collected preliminary distribution maps from the International Union for Conservation of Nature (IUCN; downloaded 2018-11-27)52, Global Assessment of Reptile Distributions (GARD) v1.153, Heimes54, Campbell and Lamar55, and unpublished maps. We manually curated distribution maps for all New World pitvipers in QGIS using the occurrence records, previous distribution maps, and recent publications for each taxon (note that distributions for Old World Viperidae have not yet been updated). We used a digital relief map (maps-for-free.com) and The Nature Conservancy Terrestrial Ecoregions (TNG.org)56 to identify clear distribution boundaries (e.g., mountains). We then clipped the final distributions to a land boundary (GADM v3.6)57 and smoothed the distribution using the the “chaikin” method in the R package smoothr25.Occurrence-distribution overlapOur initial taxonomy check was only concerned with records for which a subspecies was recorded and had since been elevated to species status. Therefore, many records with no assigned subspecies likely remained associated with an incorrect or outdated generic and/or specific identification. Fortunately, taxonomic changes are typically associated with changes in the species’ expected distribution. For example, when Crotalus simus was resurrected from C. durissus, the distribution of C. durissus was split: the northern portion of its range in Central America now represented the resurrected species (C. simus) and the southern portion of its range remained C. durissus55. Yet, occurrence records in Central America often remain labelled as C. durissus in data repositories. Therefore, we spatially joined records with the newly reconstructed species distribution maps to determine if they overlapped with their expected distribution (Old World taxa were joined with the GARD 1.1 distributions53).Briefly, we developed a custom function (occ_cleaner.R) to perform the spatial join and update taxonomy. First, we calculated the distance for each record to the 20 nearest distributions within 50 km (full overlap resulted in a distance of 0 m). Next, we calculated the phylogenetic distance between the recorded species ID and each species with which that record overlapped using the tree from Zaher et al.58 and adding taxa based on recent clade-specific publications (bind.tip2.R; see github.com/RhettRautsaw/VenomMaps for full list of references and details). If records overlapped with their expected species, no changes were made. If records fell outside of their expected distribution, we filtered the potential overlapping and nearby species (within 50 km) to minimize phylogenetic distance. If multiple species were equally distant (i.e., share the same common ancestor), we attempted to minimize geographic distance. If multiple species remained equally distant in both phylogenetic and geographic distance, we flagged the record to be manually checked. We also flagged records if a species’ taxonomy had changed and records were additionally flagged as potentially dubious if the taxonomic change had a phylogenetic divergence greater than 5 million years. We manually checked all flagged records and returned records to their original species ID if species identity remained uncertain. We flagged these records as potentially dubious, along with records that fell outside of their expected distribution (within 50 km), and removed all flagged records for species distribution modeling. Our final cleaned database contained 344,998 global records, of which 275,087 were New World pitvipers.Species distribution modelingWe attempted to infer SDMs for the 158 species of New World pitvipers currently recognized by the Reptile Database (May 2021) and additionally modeled the three subspecies of Crotalus ravus separately based on recommendations for species status elevation by Blair et al.59 for a total of 160 species. We developed a unix-executable R script (autokuenm.R) designed to take occurrence records, distribution maps, and environmental data and prepare these data for species distribution modeling with kuenm36. We chose to use kuenm – and MaxEnt v3.4.460 – because it has been shown to have good predictive power61 and fine-tuning of this algorithm has performance comparable to more computationally intensive ensembles62,63. Additionally, MaxEnt allows for flexibility in parameter selection64 and can function entirely with presence data14.Prior to autokuenm, to account for sampling/spatial bias during SDM, we created a bias file by using the pooled New World pitviper occurrence records as representative background data65,66,67,68. Specifically, we converted occurrence records to a raster and performed two-dimensional kernel density estimation (kde2d) with the MASS package with default settings69 and rescaled the kernel density by a factor of 1000 and rounded to three decimal places. This was then used as input to factor out sampling bias by MaxEnt. We then ran autokuenm, which is designed to subset/partition the cleaned occurrence records for a given species and prepare additional files for SDM. We first defined M-areas – or areas accessible to a given species – using the World Wildlife Fund Terrestrial Ecoregions70. Biogeographic regions represent distributional limits for many species and are reasonable hypotheses for the areas accessible to a given species71,72. To do this, we created alpha hulls from the subset of occurrence records for a given species using concaveman32 with default settings. We then identified regions with at least 20% of the region covered by the alpha hull and merged these regions together to form our final M-area. All environmental layers and the bias file were cropped to this M-area which was used as the geographic extent for modeling. We then randomly selected 5% of records to function as an independent test set for final model evaluation. Next, we generated 2000 random background points across the cropped environmental layers and used ENMeval to partition occurrence records into four sets using the checkerboard2 pattern35. Note that the background points here were not used in MaxEnt. One of the four partitions was selected at random to be used as the testing set; the remaining three partitions were used for training the MaxEnt models. If the number of occurrence records in the independent test set was less than five, then we used the training partition for final model creation and used the testing partition for final model evaluation.We tested the top-contributing variables from three sets of environmental layers: (1) bioclimatic variables, (2) EarthEnv topographic variables73, and (3) a combination of these variables. To select the top-contributing variables in each set, we wrote a custom function (SelectVariables) which used a combination of MaxEnt permutation importance and Variable Inflation Factors (VIF) to remove collinearity while keeping the variables that contributed the most to the model. Compared with variable selection via principal component analysis loadings, the permutation importance and VIF methodology demonstrated significant improvement in MaxEnt model fit. First, we designed SelectVariables to run MaxEnt using dismo::maxent with default settings and then extracted the permutation importance. We removed variables if they had 0% permutation importance. Next, we calculated VIF with usdm::vif and then iteratively removed variables by selecting the variables with two highest VIF values and removing whichever variable had the lowest permutation importance. We then recalculated VIF and repeated the process until the maximum VIF value was less than 10. Finally, we recalculated permutation importance with the remaining variables using dismo::maxent with default settings and removed variables with less than 1% permutation importance to create the final variable sets. This process was done for each species independently.With the final environmental variable, testing, and training sets, we generated SDMs using kuenm. First, we created candidate calibration models with multiple combinations of regularization multipliers (0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 2, 3, 4, 5, 6, 8, 10), feature classes (l, q, h, lq, lp, lt, lh, qp, qt, qh, pt, ph, th, lqp, lqt, lqh, lpt, lph, lth, qpt, qph, qth, pth, lqpt, lqph, lqth, lpth, qpth, lqpth), and sets of environmental predictors (bioclimatic, topographic, combination) totaling 2,958 candidate models per species. We then ran each model in parallel using GNU Parallel74. Next, we evaluated the candidate models and selected the best models using statistical significance (partial ROC), prediction ability (omission rates; OR), and model complexity (AICc) with the “kuenm_ceval” function with default settings. Specifically, models were only considered if they were statistically significant and had an OR less than 5%. If no models passed the OR criteria, the models with the minimal OR were considered. Finally, any remaining models were filtered to those within 2 AICc of the top model (Supplementary Table 1). In addition to evaluating and comparing all models together, we evaluated bioclimatic-only and combination-only models separately since these two sets of environmental variables were expected to be the best performing models given the ubiquity of bioclimatic variables in species distribution modeling (Supplementary Table 1).We generated 10 bootstrap replicates for each of the “best” calibration models using the “kuenm_mod” function. We also performed jackknifing to assess variable importance and models were output in raw format. We evaluated the final models using “kuenm_feval” with default settings. To select the best model for each comparative set (i.e., all, bioclimatic-only, and combination-only sets), we filtered the final evaluation results to minimize the OR and maximize the AUC ratio (Supplementary Table 2). If multiple models remained and were considered equally competitive, we averaged these models together (Supplementary Table 3). Because we performed three different set of comparisons, there were three “best” models per species, so we again aimed to minimize the OR and maximize the AUC ratio to select a final model for each species (Supplementary Table 4). We then converted our final models into cloglog format for visualization and threshold the models using a 10th percentile training presence cutoff (Fig. S2). Both conversion and thresholding functions are provided as R functions (raw2log, raw2clog, raster_threshold in functions.R; github.com/RhettRautsaw/VenomMaps). More

Storey, M. et al. Timing of hot spot—Related volcanism and the breakup of Madagascar and India. Science (80-) 267, 852–855 (1995).CAS 
Article 
ADS 

Google Scholar 
Wilmé, L., Goodman, S. M. & Ganzhorn, J. U. Biogeographic evolution of Madagascar’s microendemic biota. Science (80-) 312, 1063–1065 (2006).Article 
ADS 
CAS 

Google Scholar 
Myers, N., Mittermeler, R. A., Mittermeler, C. G., Da Fonseca, G. A. B. & Kent, J. Biodiversity hotspots for conservation priorities. Nature 403, 853–858 (2000).CAS 
PubMed 
Article 
ADS 

Google Scholar 
Vences, M., Wollenberg, K. C., Vieites, D. R. & Lees, D. C. Madagascar as a model region of species diversification. Trends Ecol. Evol. 24, 456–465 (2009).PubMed 
Article 

Google Scholar 
Rakotomanana, H., Jenkins, R. K. B. & Ratsimbazafy, J. Conservation challenges for Madagascar in the next decade. In Conservation Biology: Voices from the Tropics (eds Raven, P. H., Sodhi, N. S. & Gibson, L.) 33–39 (Wiley-Blackwell, 2013). https://doi.org/10.1002/9781118679838.ch5.Jenkins, R. K. B. et al. Extinction risks and the conservation of Madagascar’s reptiles. PLoS ONE 9, 1 – 14 (2014). https://doi.org/10.1371/journal.pone.0100173Velosoa, J. et al. An integrated research, management, and community conservation program for the Rere (Madagascar Big-headed turtle), Erymnochelys madagascariensis. In Chelonian Research Monographs, Contributions in Turtle and Tortoise Research (eds Rhodin, A. G. J.) 171–177 (Chelonian Research Foundation, 2014). https://doi.org/10.3854/crm.6.a27p171.Leuteritz, T., Kuchling, G., Garcia, G. & Velosoa, J. Erymnochelys madagascariensis. In Chelonian Research Monographs, Contributions in Turtle and Tortoise Research (eds Rhodin, A. G. J.) 56–58 (Chelonian Research Foundation, 2014). https://doi.org/10.3854/crm.6.a11p56.Rafeliarisoa, T., Shore, G., Engberg, S., Louis, E. & Brenneman, R. Characterization of 11 microsatellite marker loci in the Malagasy big-headed turtle (Erymnochelys madagascariensis). Mol. Ecol. Notes 6, 1228–1230 (2006).CAS 
Article 

Google Scholar 
Roca, V., García, G. & Montesinos, A. Gastrointestinal helminths found in the three freshwater turtles (Erymnochelys madagascariensis, Pelomedusa subrufa and Pelusios castanoides) from Ankarafantsika National Park, Madagascar. Helminthologia 44, 177–182 (2007).Article 

Google Scholar 
Kuchling, G. & Garcia, G. Pelomedusidae, freshwater turtles. In The Natural History of Madagascar (eds Goodman, S. M. & Benstead, J. P.) 956–960 (University of Chicago Press, 2003).
Google Scholar 
Pedrono, M. & Smith, L. Overview of the natural history of Madagascar’s endemic tortoises and freshwater turtles: Essential components for effective conservation. In Chelonian Research Monographs, Contributions in Turtle and Tortoise Research (eds Rhodin, A. G. J.) 59–66 (Chelonian Research Foundation, 2014). https://doi.org/10.3854/crm.6.a12p59.Kuchling, G. Population structure, reproductive potential and increasing exploitation of the freshwater turtle Erymnochelys madagascariensis. Biol. Conserv. 43, 107–113 (1988).Article 

Google Scholar 
Allnutt, T. F. et al. A method for quantifying biodiversity loss and its application to a 50-year record of deforestation across Madagascar. Conserv. Lett. 1, 173–181 (2008).Article 

Google Scholar 
Leuteritz, T., Kuchling, G., Garcia, G. & Velosoa, J. Erymnochelys madagascariensis (errata version published in 2016). The IUCN Red List of Threatened Species. 2008, 1–3 (2008).Kuchling, G. Concept and design of the Madagascar side-necked turtle Erymnochelys madagascariensis breeding facility at Ampijoroa, Madagascar. Dodo 36, 62–74 (2000).
Google Scholar 
Witzenberger, K. A. & Hochkirch, A. Ex situ conservation genetics: A review of molecular studies on the genetic consequences of captive breeding programmes for endangered animal species. Biodivers. Conserv. 20, 1843–1861 (2011).Article 

Google Scholar 
Stanton, D. W. G. et al. Genetic structure of captive and free-ranging okapi (Okapia johnstoni) with implications for management. Conserv. Genet. 16, 1115–1126 (2015).Article 

Google Scholar 
Boumans, L., Vieites, D. R., Glaw, F. & Vences, M. Geographical patterns of deep mitochondrial differentiation in widespread Malagasy reptiles. Mol. Phylogenet. Evol. 45, 822–839 (2007).CAS 
PubMed 
Article 

Google Scholar 
Orozco-Terwengel, P., Andreone, F., Louis, E. & Vences, M. Mitochondrial introgressive hybridization following a demographic expansion in the tomato frogs of Madagascar, genus Dyscophus. Mol. Ecol. 22, 6074–6090 (2013).CAS 
PubMed 
Article 

Google Scholar 
Pearson, R. G. & Raxworthy, C. J. The evolution of local endemism in Madagascar: Watershed versus climatic gradient hypotheses evaluated by null biogeographic models. Evolution (New York) 63, 959–967 (2009).
Google Scholar 
Sunde, J., Yıldırım, Y., Tibblin, P. & Forsman, A. Comparing the performance of microsatellites and RADseq in population genetic studies: Analysis of data for pike (Esox lucius) and a synthesis of previous studies. Front. Genet. 11, 218 (2020).PubMed 
PubMed Central 
Article 

Google Scholar 
Hulce, D., Li, X., Snyder-Leiby, T. & Liu, J. GeneMarker® genotyping software: Tools to increase the statistical power of DNA fragment analysis. J. Biomol. Tech. https://doi.org/10.1002/wps.20394 (2011).Article 
PubMed Central 

Google Scholar 
Van Oosterhout, C., Hutchinson, W. F., Wills, D. P. M. & Shipley, P. MICRO-CHECKER: Software for identifying and correcting genotyping errors in microsatellite data. Mol. Ecol. Notes 4, 535–538 (2004).Article 
CAS 

Google Scholar 
Carlsson, J. Effects of microsatellite null alleles on assignment testing. J. Hered. 99, 616–623 (2008).CAS 
PubMed 
Article 

Google Scholar 
Bossuyt, F. & Milinkovitch, M. C. Convergent adaptive radiations in Madagascan and Asian ranid frogs reveal covariation between larval and adult traits. Proc. Natl. Acad. Sci. U. S. A. 97, 6585–6590 (2000).CAS 
PubMed 
PubMed Central 
Article 
ADS 

Google Scholar 
Rousset, F. GENEPOP’007: A complete re-implementation of the GENEPOP software for Windows and Linux. Mol. Ecol. Resour. 8, 103–106 (2008).PubMed 
Article 

Google Scholar 
Beaumont, M. A. Detecting population expansion and decline using microsatellites. Genetics 153, 2013–2029 (1999).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Bulut, Z. et al. Microsatellite mutation rates in the eastern tiger salamander (Ambystoma tigrinum tigrinum) differ 10-fold across loci. Genetica 136, 501–504 (2009).CAS 
PubMed 
Article 

Google Scholar 
Brooks, S. P. & Gelman, A. General methods for monitoring convergence of iterative simulations. J. Comput. Graph. Stat. 7, 434–455 (1998).MathSciNet 

Google Scholar 
Plummer, M. & Murrell, P. CODA: Convergence Diagnosis and Output Analysis for MCMC. R News. 6, 7–11 (2006).R Development Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2008). https://doi.org/10.1017/CBO9781107415324.004.Book 

Google Scholar 
Pritchard, J. K., Stephens, M. & Donnelly, P. Inference of Population Structure Using Multilocus Genotype Data. Genetics 155, 945–959 (2000). https://doi.org/10.1111/j.1471-8286.2007.01758.x.Puechmaille, S. J. The program structure does not reliably recover the correct population structure when sampling is uneven: Subsampling and new estimators alleviate the problem. Mol. Ecol. Resour. 16, 608–627 (2016).PubMed 
Article 

Google Scholar 
Hale, M. L., Burg, T. M. & Steeves, T. E. Sampling for microsatellite-based population genetic studies: 25 to 30 individuals per population is enough to accurately estimate allele frequencies. PLoS ONE 7, e45170 (2012).CAS 
PubMed 
PubMed Central 
Article 
ADS 

Google Scholar 
Francis, R. M. Pophelper: An R package and web app to analyse and visualize population structure. Mol. Ecol. Res. 17, 27–32 (2017).Evanno, G., Regnaut, S. & Goudet, J. Detecting the number of clusters of individuals using the software STRUCTURE: A simulation study. Mol. Ecol. 14, 2611–2620 (2005).CAS 
PubMed 
Article 

Google Scholar 
Kearse, M. et al. Geneious basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28, 1647–1649 (2012).PubMed 
PubMed Central 
Article 

Google Scholar 
Dieringer, D. & Schlötterer, C. Microsatellite analyser (MSA): A platform independent analysis tool for large microsatellite data sets. Mol. Ecol. Notes 3, 167–169 (2003).CAS 
Article 

Google Scholar 
Narum, S. R. Beyond Bonferroni: Less conservative analyses for conservation genetics. Conserv. Genet. 7, 783–787 (2006).CAS 
Article 

Google Scholar 
Goudet, J. FSTAT (version 1.2): A computer program to calculate F-statistics. J. Hered. 86, 485–486 (1995).Article 

Google Scholar 
Excoffier, L. & Lischer, H. E. L. Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Mol. Ecol. Resour. 10, 564–567 (2010).PubMed 
Article 

Google Scholar 
Librado, P. & Rozas, J. DnaSP v5: A software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451–1452 (2009).CAS 
PubMed 
Article 

Google Scholar 
Prost, S. & Anderson, C. N. K. TempNet: A method to display statistical parsimony networks for heterochronous DNA sequence data. Methods Ecol. Evol. 2, 663–667 (2011).Article 

Google Scholar 
Paquette, S. R. et al. Riverbeds demarcate distinct conservation units of the radiated tortoise (Geochelone radiata) in southern Madagascar. Conserv. Genet. 8, 797–807 (2007).CAS 
Article 

Google Scholar 
Bouchard, C., Tessier, N. & Lapointe, F. J. Watersheds influence the wood turtle’s (Glyptemys insculpta) genetic structure. Conserv. Genet. 20, 653–664 (2019).Article 

Google Scholar 
Perlman, S. J., Hodson, C. N., Hamilton, P. T., Opit, G. P. & Gowen, B. E. Maternal transmission, sex ratio distortion, and mitochondria. Proc. Natl. Acad. Sci. U. S. A. 112, 10162–10168 (2015).CAS 
PubMed 
PubMed Central 
Article 
ADS 

Google Scholar 
Pearse, D. E. et al. Estimating population structure under nonequilibrium conditions in a conservation context: Continent-wide population genetics of the giant Amazon river turtle, Podocnemis expansa (Chelonia; Podocnemididae). Mol. Ecol. 15, 985–1006 (2006).CAS 
PubMed 
Article 

Google Scholar 
Pearse, D. E. & Avise, J. C. Turtle mating systems: Behavior, sperm storage, and genetic paternity. J. Hered. 92, 206–211 (2001).CAS 
PubMed 
Article 

Google Scholar 
Claussen, M. et al. Simulation of an abrupt change in Saharan vegetation in the mid-Holocene. Geophys. Res. Lett. 26, 2037–2040 (1999).Article 
ADS 

Google Scholar 
Virah-Sawmy, M., Willis, K. J. & Gillson, L. Threshold response of Madagascar’s littoral forest to sea-level rise. Glob. Ecol. Biogeogr. 18, 98–110 (2009).Article 

Google Scholar 
Wahlund, S. Zusammensetzung von populationen und korrelationserscheinungen vom standpunkt der vererbungslehre aus betrachtet. Hereditas 11, 65–106 (1928).Article 

Google Scholar 
Hurst, G. D. D. & Jiggins, F. M. Problems with mitochondrial DNA as a marker in population, phylogeographic and phylogenetic studies: The effects of inherited symbionts. Proc. R. Soc. B Biol. Sci. 272, 1525–1534 (2005).CAS 
Article 

Google Scholar 
Hill, W. G. & Robertson, A. The effect of linkage on limits to artificial selection. Genet. Res. (Camb.) 89, 311–336 (2008).Article 

Google Scholar 
Valenzuela, N. Multiple paternity in side-neck turtles Podocnemis expansa: Evidence from microsatellite DNA data. Mol. Ecol. 9, 99–105 (2000).CAS 
PubMed 
Article 

Google Scholar 
Moritz, C. Defining ‘Evolutionarily Significant Units’ for conservation. Trends Ecol. Evol. 9, 373–375 (1994).CAS 
PubMed 
Article 

Google Scholar 
Volkmann, L., Martyn, I., Moulton, V., Spillner, A. & Mooers, A. O. Prioritizing populations for conservation using phylogenetic networks. PLoS ONE 9, 1–10 (2014). https://doi.org/10.1371/journal.pone.0088945Article 
CAS 

Google Scholar 
García-Dorado, A. & Caballero, A. Neutral genetic diversity as a useful tool for conservation biology. Conserv. Genet. 22, 541–545 (2021).Article 

Google Scholar 
Frankham, R. Genetic rescue of small inbred populations: Meta-analysis reveals large and consistent benefits of gene flow. Mol. Ecol. 24, 2610–2618 (2015).PubMed 
Article 

Google Scholar 
Teixeira, J. C. & Huber, C. D. The inflated significance of neutral genetic diversity in conservation genetics. Proc. Natl. Acad. Sci. U. S. A. 118, 1–10 (2021). https://doi.org/10.1073/pnas.2015096118Araki, H., Cooper, B. & Blouin, M. S. Genetic effects of captive breeding cause a rapid, cumulative fitness decline in the wild. Science (80-) 318, 100–103 (2007).CAS 
Article 
ADS 

Google Scholar  More

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript. More

In this International Year of Sustainable Mountain Development, we call for transdisciplinary research by Chilean scientists and for concerted action among all stakeholders to address the complex factors responsible for the degradation of Chile’s mountains. Mountains cover 64% of Chile’s surface and are a crucial source of water, food, energy, minerals and biodiversity.
Competing Interests
The authors declare no competing interests. More

None of the 21 targets of the United Nations’ Sustainable Development Goals (SDGs) set for 2020 was achieved. But, by our calculations, the target to protect 10% of the global ocean area (SDG14, target 5) could become a reality this year.
Competing Interests
The authors declare no competing interests. More

Ethics statementThe data collection and experiments were conducted in accordance with the collecting permits (SGPA/DGVS/07946/08, 03369/12, 00228/13, 07587/13, 01629/16, 01205/17, 02490/17, 06768/17, 000998/18, 002463/18, 002490/18, 002491/18, 003209/18, and 02523/19) approved by Dirección General de Vida Silvestre, México.Phylogeny and divergence time estimationTo estimate the phylogeny and divergence time among phrynosomatid species we used sequences of five mitochondrial and eight nuclear genes available in GenBank for 149 taxa (Supplementary Data 2). Accession numbers were the same as those used in Martínez-Méndez et al.58 for the Sceloporus torquatus, S. poinsettii and S. megalepidurus groups and the same as those in Wiens et al.59 for other phrynosomatid species. For taxa not included in the previous references, we searched GenBank for available sequences. We then performed alignments for each gene using MAFFT (ver. 7)60 and concatenation and manual refinement using Mesquite (ver. 3.6);61 obtaining a concatenated matrix of 9837 bp for 149 taxa (Supplementary Data 3). For the relaxed clock analyses, three nodes were calibrated using lognormal distributions based on two previous studies59,62. The first calibration was set for the Sceloporus clade (offset 15.97 million years ago (MYA)) based on a fossil Sceloporus specimen63). The second calibration point was set for the Phrynosoma clade (offset 33.3 MYA) based on the fossil Paraphrynosoma greeni64, and the last calibration point was for the Holbrookia-Cophosaurus stem group (offset 15.97 MYA) given the fossil Holbrookia antiqua63. We conducted dating analysis with the concatenated sequences matrix, partitioned the mitochondrial and nuclear information, each gene under GTR + I + Γ model, and allowed independent parameter estimation. We performed Bayesian age estimation with the uncorrelated lognormal relaxed clock (UCLN) model in BEAST (ver. 2.5.2)65,66 and run on CIPRES67. Tree prior (evolutionary model) was under the Birth-Death model, and we ran two MCMC analyses for 100 million generations each and stored every 20,000 generations. We assessed the convergence and stationarity of chains from the posterior distribution using Tracer (ver. 1.7)68. We combined independent runs using LogCombiner (ver. 2.5.2; BEAST distribution)69 and discarded 30% of samples as burn-in, obtaining values of effective sample size (ESS) greater than 200. We estimated the maximum clade credibility tree from all post-burnin trees using TreeAnnotator (ver. 1.8.4)69. The ultrametric tree is available as Supplementary Data 4. As we describe below, we accounted for phylogenetic uncertainty in our models by reperforming analyses using 500 trees that we randomly sampled from our posterior distribution. The 500 sampled trees are available as Supplementary Data 5.Data collectionParity modeWe categorized each species as either oviparous or viviparous based on previously published databases21,37,51,70, published references, and unpublished data (Supplementary Data 1). Our assignations align with other studies, except for one species, Sceloporus goldmani, which has been previously considered a viviparous species21,71,72,73. The only available sequence in GenBank (U88290) for that species is from a male (MZFC-05458) collected in Coahuila, Mexico72. However, in that same locality, one of us (F. R. Méndez-de la Cruz; unpubl. data) collected two females of the same species, and both laid eggs. Thus, the population of S. goldmani herein included is considered oviparous. Considering S. goldmani viviparous increases the number of originations of viviparity to 6 (from 5) in this lineage (Supplementary Fig. 4), but does not alter the outcome of our model-fitting analyses of trait evolution (Supplementary Table 7).Thermal physiologyWe compiled a database of four thermal physiological traits that influence the performance and fitness of ectotherms74 for 104 phrynosomatid species. These data were gathered from both published sources and from our own field and laboratory work (Supplementary Data 1). The thermal physiological traits we examined were the field body temperature (Tb) of active lizards, the preferred body temperature (Tpref) in a laboratory thermal gradient75, cold tolerance (critical thermal minimum, CTmin), and heat tolerance (critical thermal maximum, CTmax). These latter two traits (CTmin and CTmax) describe the thermal limits of locomotion; specifically, they describe the lower and upper temperatures, respectively, at which lizards fail to right themselves when flipped onto their backs55,76. To minimize the confounding effects of experimental design, we limited our data selection to species that were measured with similar methods. Correspondingly, our new data collection approach mirrored that of the published studies from which we extracted data. To obtain mean values for each thermal physiological trait (CTmin, Tb, Tpref, and CTmax) we did not mix data measured from different locations (instead, we used data from the population with the highest sample size).For species that we newly measured thermal physiological traits, we obtained the data as we describe below, and we based our methodology on the previous work55,56,75,76. We captured active (perching) adult lizards by lasso or by hand, and immediately ( More