Summary
Including predictors in species distribution models at inappropriate spatial scales can decrease the variance explained, add residual spatial autocorrelation (RSA) and lead to the wrong ...conclusions. Some studies have measured predictors within different buffer sizes (scales) around sample locations, regressed each predictor against the response at each scale and selected the scale with the best model fit as the appropriate scale for this predictor. However, a predictor can influence a species at several scales or show several scales with good model fit due to a bias caused by RSA. This makes the evaluation of all scales with good model fit necessary. With potentially several scales per predictor and multiple predictors to evaluate, the number of predictors can be large relative to the number of data points, potentially impeding variable selection with traditional statistical techniques, such as logistic regression.
We trialled a variable selection process using the random forest algorithm, which allows the simultaneous evaluation of several scales of multiple predictors. Using simulated responses, we compared the performance of models resulting from this approach with models using the known predictors at arbitrary and at the known spatial scales. We also apply the proposed approach to a real data set of curlew (Numenius arquata).
AIC, AUC and Naglekerke's pseudo R2 of the models resulting from the proposed variable selection were often very similar to the models with the known predictors at known spatial scales. Only two of nine models required the addition of spatial eigenvectors to account for RSA. Arbitrary scale models always required the addition of spatial eigenvectors. 75% (50–100%) of the known predictors were selected at scales similar to the known scale (within 3 km). In the curlew model, predictors at large, medium and small spatial scales were selected, suggesting that for appropriate landscape‐scale models multiple scales need to be evaluated.
The proposed approach selected several of the correct predictors at appropriate spatial scales out of 544 possible predictors. Thus, it facilitates the evaluation of multiple spatial scales of multiple predictors against each other in landscape‐scale models.
1. High-resolution vegetation maps are a valuable resource for conservation, land management and research. In Great Britain, the National Vegetation Classification (NVC) is widely used to describe ...vegetation communities. NVC maps are typically produced from ground surveys which are prohibitively expensive for large areas. An approach to produce NVC maps more cost-effectively for large areas would be valuable. 2. Creation of vegetation community maps from aerial or satellite images has often had limited success as the clusters separable by spectral reflectance frequently do not correspond well to vegetation community classes. Such maps have also been produced by exploring correlations between community occurrence and environmental variables. The latter approach can have limitations where anthropogenic activities have altered the distribution of vegetation communities. We combined these two approaches and classified 24 common NVC classes of the Yorkshire Dales and an additional class 'wood' consisting of trees and bushes at a resolution of 5 m from mostly remotely sensed variables with the algorithm random forest. 3. Classification accuracy was highest when environmental variables at low and high resolution (50 and 5—10 m, respectively) were added to aerial image information aggregated to a resolution of 5 m. Low-resolution environmental variables are likely to be correlated with the dominant vegetation surrounding a location and thus could represent critical area requirements or local species pool effects, while high-resolution environmental variables represent the environmental conditions at the location. 4. Overall classification accuracy was 87—92%. The median producer's and user's class accuracies were 95% (58—100%) and 92% (67—100%), respectively. 5. Synthesis and applications. The classification accuracies achieved in this study, the number of classes differentiated, their level of detail and the resolution were high compared with those of other studies. This approach could allow the production of good-quality NVC maps for large areas. In contrast to existing maps of broad land cover types, such maps would provide more detailed vegetation community data for applications like the monitoring of vegetation in a changing climate, the study of animal—habitat relationships, conservation management or land use planning.
Globally, major efforts are being made to restore peatlands to maximise their resilience to anthropogenic climate change, which puts continuous pressure on peatland ecosystems and modifies the ...geography of the environmental envelope that underpins peatland functioning. A probable effect of climate change is reduction in the waterlogged conditions that are key to peatland formation and continued accumulation of carbon (C) in peat. C sequestration in peatlands arises from a delicate imbalance between primary production and decomposition, and microbial processes are potentially pivotal in regulating feedbacks between environmental change and the peatland C cycle. Increased soil temperature, caused by climate warming or disturbance of the natural vegetation cover and drainage, may result in reductions of long-term C storage via changes in microbial community composition and metabolic rates. Moreover, changes in water table depth alter the redox state and hence have broad consequences for microbial functions, including effects on fungal and bacterial communities especially methanogens and methanotrophs. This article is a perspective review of the effects of climate change and ecosystem restoration on peatland microbial communities and the implications for C sequestration and climate regulation. It is authored by peatland scientists, microbial ecologists, land managers and non-governmental organisations who were attendees at a series of three workshops held at The University of Manchester (UK) in 2019–2020. Our review suggests that the increase in methane flux sometimes observed when water tables are restored is predicated on the availability of labile carbon from vegetation and the absence of alternative terminal electron acceptors. Peatland microbial communities respond relatively rapidly to shifts in vegetation induced by climate change and subsequent changes in the quantity and quality of below-ground C substrate inputs. Other consequences of climate change that affect peatland microbial communities and C cycling include alterations in snow cover and permafrost thaw. In the face of rapid climate change, restoration of a resilient microbiome is essential to sustaining the climate regulation functions of peatland systems. Technological developments enabling faster characterisation of microbial communities and functions support progress towards this goal, which will require a strongly interdisciplinary approach.
When one or several classes are much less prevalent than another class (unbalanced data), class error rates and variable importances of the machine learning algorithm random forest can be biased, ...particularly when sample sizes are smaller, imbalance levels higher, and effect sizes of important variables smaller. Using simulated data varying in size, imbalance level, number of true variables, their effect sizes, and the strength of multicollinearity between covariates, we evaluated how eight versions of random forest ranked and selected true variables out of a large number of covariates despite class imbalance. The version that calculated variable importance based on the area under the curve (AUC) was least adversely affected by class imbalance. For the same number of true variables, effect sizes, and multicollinearity between covariates, the AUC variable importance ranked true variables still highly at the lower sample sizes and higher imbalance levels at which the other seven versions no longer achieved high ranks for true variables. Conversely, using the Hellinger distance to split trees or downsampling the majority class already ranked true variables lower and more variably at the larger sample sizes and lower imbalance levels at which the other algorithms still ranked true variables highly. In variable selection, a higher proportion of true variables were identified when covariates were ranked by AUC importances and the proportion increased further when the AUC was used as the criterion in forward variable selection. In three case studies, known species–habitat relationships and their spatial scales were identified despite unbalanced data.
A recent paper by Heinemeyer et al. (2018) in this journal has suggested that the use of prescribed fire may enhance carbon accumulation in UK upland blanket bogs. We challenge this finding based on ...a number of concerns with the original manuscript including the lack of an unburned control, insufficient replication, unrecognised potential confounding factors, and potentially large inaccuracies in the core dating approach used to calculate carbon accumulation rates. We argue that burn‐management of peatlands is more likely to lead to carbon loss than carbon gain.e00075
Large areas of upland mire and moorland in Northwest Europe are regarded as degraded, not actively peat-forming, and releasing carbon. Conservation agencies have short-term targets to restore such ...areas, but often have no clear knowledge of the timing and nature of degradation. It has been suggested that palaeoecology can be used to inform conservation management about past vegetation states, so as to help identify feasible restoration targets. Our research study in northern England, commissioned by the national statutory conservation agency, applied multiple palaeoecological techniques to establish the vegetation history of several mire and moorland sites, specifically to ascertain the nature and timing of degradation. Techniques applied included pollen analysis, plant macrofossil and charcoal analyses, determination of peat humification and mineral magnetic susceptibility, with ages ascertained using spheroidal carbonaceous particle analysis, 210Pb and 14C dating. Data are presented from case-study sites in the North York Moors, North- and South Pennines to illustrate how palaeoecology can extend long-term monitoring and guide conservation management. Palaeoecological data from a site within a National Nature Reserve, subject to exceptionally long-term (half-centennial) ecological monitoring, showed that this period does not include its pre-degradation state and that its current valued vegetation is novel and may have established after major fire. Overall, the studies suggest that the principal vegetation change at the sites took place after the start of the Industrial Revolution, and that the current landscape appearance not only has no long history, but that valued aspects, such as extensive heather moorland, feature only recently in the cultural landscape. These findings pose challenging questions for conservation management. We offer a non-specialist guide to the palaeoecological techniques that considers level of skill, cost, and comparability with ecological aspects of conservation and monitoring interest. We suggest palaeoecological data can provide valuable information and insights to aid practical conservation. While mires are particularly suitable, palaeoecological techniques could be applied in many other degraded landscapes internationally.
The Little Ice Age (LIA) is a well-recognised palaeoclimatic phenomenon, although its causes, duration and severity have been matters of debate and dispute. Data from a wide range of archives have ...been used to infer climate variability before, during and after the LIA. Some published proxy-climate data from peatlands imply that two particularly severe episodes within the LIA may be contemporaneous between hemispheres; these echo a previous climatic downturn ca. 2800 cal BP of similar severity but lesser duration. Here, we present palaeoclimate data from the mid- to late-Holocene, reconstructed from three blanket peats in Yorkshire: Mossdale Moor, Oxenhope Moor and West Arkengarthdale. Multiproxy techniques used for palaeoclimatic reconstruction were plant macrofossil, pollen and humification analyses. Dating was provided by a radiocarbon-based chronology, aided by spheroidal carbonaceous particles (SCPs) for all sites, and 210Pb dates for one. The LIA presents as a distinct climatic event within each palaeoenvironmental record at the three sites. These indications are compared with terrestrial datasets from northwest Europe and elsewhere. A broad degree of synchronicity is evident, signifying that the LIA is one of the most pronounced downturns in global climate in the last ca. 6000 years, and arguably the most routinely recorded within the Holocene.
Here, we present results from the most comprehensive compilation of Holocene peat soil properties with associated carbon and nitrogen accumulation rates for northern peatlands. Our database consists ...of 268 peat cores from 215 sites located north of 45°N. It encompasses regions within which peat carbon data have only recently become available, such as the West Siberia Lowlands, the Hudson Bay Lowlands, Kamchatka in Far East Russia, and the Tibetan Plateau. For all northern peatlands, carbon content in organic matter was estimated at 42 ± 3% (standard deviation) for Sphagnum peat, 51 ± 2% for non-Sphagnum peat, and at 49 ± 2% overall. Dry bulk density averaged 0.12 ± 0.07 g/cm3, organic matter bulk density averaged 0.11 ± 0.05 g/cm3, and total carbon content in peat averaged 47 ± 6%. In general, large differences were found between Sphagnum and non-Sphagnum peat types in terms of peat properties. Time-weighted peat carbon accumulation rates averaged 23 ± 2 (standard error of mean) g C/m2/yr during the Holocene on the basis of 151 peat cores from 127 sites, with the highest rates of carbon accumulation (25–28 g C/m2/yr) recorded during the early Holocene when the climate was warmer than the present. Furthermore, we estimate the northern peatland carbon and nitrogen pools at 436 and 10 gigatons, respectively. The database is publicly available at https://peatlands.lehigh.edu.
Peatlands are wetland ecosystems with great significance as natural habitats and as major global carbon stores. They have been subject to widespread exploitation and degradation with resulting losses ...in characteristic biota and ecosystem functions such as climate regulation. More recently, large-scale programmes have been established to restore peatland ecosystems and the various services they provide to society. Despite significant progress in peatland science and restoration practice, we lack a process-based understanding of how soil microbiota influence peatland functioning and mediate the resilience and recovery of ecosystem services, to perturbations associated with land use and climate change.
We argue that there is a need to: in the short-term, characterise peatland microbial communities across a range of spatial and temporal scales and develop an improved understanding of the links between peatland habitat, ecological functions and microbial processes; in the medium term, define what a successfully restored ‘target’ peatland microbiome looks like for key carbon cycle related ecosystem services and develop microbial-based monitoring tools for assessing restoration needs; and in the longer term, to use this knowledge to influence restoration practices and assess progress on the trajectory towards ‘intact’ peatland status.
Rapid advances in genetic characterisation of the structure and functions of microbial communities offer the potential for transformative progress in these areas, but the scale and speed of methodological and conceptual advances in studying ecosystem functions is a challenge for peatland scientists. Advances in this area require multidisciplinary collaborations between peatland scientists, data scientists and microbiologists and ultimately, collaboration with the modelling community.
Developing a process-based understanding of the resilience and recovery of peatlands to perturbations, such as climate extremes, fires, and drainage, will be key to meeting climate targets and delivering ecosystem services cost effectively.
•Although microbes are key to peatland function the underpinning processes are unclear.•Microbial characterisation is needed across a range of sites, depths and conditions.•Temporal and spatial changes in microbial communities need to be linked to functions.•Potential to use microbiome as a monitoring tool for peatland restoration progress•Enhancing microbial communities could improve peatland resilience.