| 1. |
- Gendre, Matthieu, et al.
(author)
-
Benchmarking imputation methods for categorical biological data
- 2024
-
In: METHODS IN ECOLOGY AND EVOLUTION. - 2041-210X .- 2041-2096.
-
Journal article (peer-reviewed)abstract
- Trait datasets are at the basis of a large share of ecology and evolutionary research, being used to infer ancestral morphologies, quantify species extinction risks, or evaluate the functional diversity of biological communities. These datasets, however, are often plagued by missing data, for instance, due to incomplete sampling, limited data and resource availability. Several imputation methods exist to predict missing values and recent studies have explored their performance for continuous traits in biological datasets. However, less is known about the accuracy of these methods for categorical traits. Here we explore the performance of different imputation methods on categorical biological traits combining phylogenetic comparative methods, machine learning and deep learning models. To this end, we develop an open-source R package, to impute trait data while integrating a simulation framework to evaluate their performance on synthetic datasets. We run a range of simulations under different missing rates, mechanisms, biases and evolutionary models. We propose an integration between phylogenetic comparative methods and machine learning imputation, and an ensemble approach, in which selected imputation methods are combined. Our simulations show that this approach provides the most robust and accurate predictions. We applied our imputation pipeline to an incomplete trait dataset of 1015 elasmobranch species (i.e. sharks, rays and skates) and found a high imputation accuracy of the predictions based on an expert-based assessment of the missing traits. Overall, our R package facilitates the comparison of multiple imputation methods and allows robust predictions of missing trait values. Our study highlights the benefits of coupling phylogenetic evolutionary models with machine learning inference to augment incomplete biological datasets.
|
|
| 2. |
|
|
| 3. |
- Bukkuri, Anuraag, et al.
(author)
-
Integrating eco-evolutionary dynamics into matrix population models for structured populations : Discrete and continuous frameworks
- 2023
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 14:6, s. 1475-1488
-
Journal article (peer-reviewed)abstract
- State-structured populations are ubiquitous in biology, from the age-structure of animal societies to the life cycles of parasitic species. Understanding how this structure contributes to eco-evolutionary dynamics is critical not only for fundamental understanding but also for conservation and treatment purposes. Although some methods have been developed in the literature for modelling eco-evolutionary dynamics in structured population, such methods are wholly lacking in the (Formula presented.) function evolutionary game theoretic framework. In this paper, we integrate standard matrix population modelling into the (Formula presented.) function framework to create a theoretical framework to probe eco-evolutionary dynamics in structured populations. This framework encompasses age- and stage-structured matrix models with basic density- and frequency-dependent transition rates and probabilities. For both discrete and continuous time models, we define and characterize asymptotic properties of the system such as eco-evolutionary equilibria (including ESSs) and the convergence stability of these equilibria. For multistate structured populations, we introduce an ergodic flow preserving folding method for analysing such models. The methods developed in this paper for state-structured populations and their extensions to multistate-structured populations provide a simple way to create, analyse and simulate eco-evolutionary dynamics in structured populations. Furthermore, their generality allows these techniques to be applied to a variety of problems in ecology and evolution.
|
|
| 4. |
- Dunkley, Katie, et al.
(author)
-
A low-cost, long-running, open-source stereo camera for tracking aquatic species and their behaviours
- 2023
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 14:10, s. 2549-2556
-
Journal article (peer-reviewed)abstract
- Ecologists are now widely utilising video data to quantify the behaviours and interactions of animals in the wild. This process can be facilitated by collecting videos in stereo, which can provide information about animals' positions, movements and behaviours in three-dimensions (3D). However, there are no published designs that can collect underwater 3D stereo data at high spatial and temporal resolutions for extended periods (days). Here, we present complete hardware and software solutions for a long-running, open-source, underwater stereo camera rig, costing £1337. This stereo camera can continuously record aquatic species and their behaviours/interactions in high resolution (1080 p and 30 fps) and in 3D, over multiple days. We provide full design guides for the cameras and a travel-friendly rig, and include guidance and open-source code for calibrating the cameras in space and time. We also show how these cameras could be used to track animals' body parts and positions, and how their size, posture and behaviour can be inferred. This stereo camera will facilitate the collection of high-resolution ecological and behavioural data, such as affiliative, agonistic or trophic interactions between species, which can inform us about the health and structure of ecosystems. These data will assist ecologists and conservationists in monitoring and understanding the impacts of current environmental pressures on ecosystem functioning.
|
|
| 5. |
- Gardner, Emma, et al.
(author)
-
Reliably predicting pollinator abundance : Challenges of calibrating process-based ecological models
- 2020
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 11:12, s. 1673-1689
-
Journal article (peer-reviewed)abstract
- Pollination is a key ecosystem service for global agriculture but evidence of pollinator population declines is growing. Reliable spatial modelling of pollinator abundance is essential if we are to identify areas at risk of pollination service deficit and effectively target resources to support pollinator populations. Many models exist which predict pollinator abundance but few have been calibrated against observational data from multiple habitats to ensure their predictions are accurate. We selected the most advanced process-based pollinator abundance model available and calibrated it for bumblebees and solitary bees using survey data collected at 239 sites across Great Britain. We compared three versions of the model: one parameterised using estimates based on expert opinion, one where the parameters are calibrated using a purely data-driven approach and one where we allow the expert opinion estimates to inform the calibration process. All three model versions showed significant agreement with the survey data, demonstrating this model's potential to reliably map pollinator abundance. However, there were significant differences between the nesting/floral attractiveness scores obtained by the two calibration methods and from the original expert opinion scores. Our results highlight a key universal challenge of calibrating spatially explicit, process-based ecological models. Notably, the desire to reliably represent complex ecological processes in finely mapped landscapes necessarily generates a large number of parameters, which are challenging to calibrate with ecological and geographical data that are often noisy, biased, asynchronous and sometimes inaccurate. Purely data-driven calibration can therefore result in unrealistic parameter values, despite appearing to improve model-data agreement over initial expert opinion estimates. We therefore advocate a combined approach where data-driven calibration and expert opinion are integrated into an iterative Delphi-like process, which simultaneously combines model calibration and credibility assessment. This may provide the best opportunity to obtain realistic parameter estimates and reliable model predictions for ecological systems with expert knowledge gaps and patchy ecological data.
|
|
| 6. |
- Jonsson, Mattias, et al.
(author)
-
Ecological production functions for biological control services in agricultural landscapes
- 2014
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 5:3, s. 243-252
-
Journal article (peer-reviewed)abstract
- Research relating to ecosystem services has increased, partly because of drastic declines in biodiversity in agricultural landscapes. However, the mechanistic linkages between land use, biodiversity and service provision are poorly understood and synthesized. This is particularly true for many ecosystem services provided by mobile organisms such as natural enemies to crop pests. These species are not only influenced by local land use but also by landscape composition at larger spatial scales. We present a conceptual ecological production function framework for predicting land-use impact on biological control of pests by natural enemies. We develop a novel, mechanistic landscape model for biological control of cereal aphids, explicitly accounting for the influence of landscape composition on natural enemies varying in mobility, feeding rates and other life history traits. Finally, we use the model to map biological control services across cereal fields in a Swedish agricultural region with varying landscape complexity. The model predicted that biological control would reduce crop damage by 45-70% and that the biological control effect would be higher in complex landscapes. In a validation with independent data, the model performed well and predicted a significant proportion of biological control variation in cereal fields. However, much variability remains to be explained, and we propose that the model could be improved by refining the mechanistic understanding of predator dynamics and accounting for variation in aphid colonization. We encourage scientists working with biological control to adopt the conceptual framework presented here and to develop production functions for other crop-pest systems. If this kind of ecological production function is combined with production functions for other services, the joint model will be a powerful tool for managing ecosystem services and planning for sustainable agriculture at the landscape scale.
|
|
| 7. |
- Lürig, Moritz D.
(author)
-
phenopype : A phenotyping pipeline for Python
- 2022
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 13:3, s. 569-576
-
Journal article (peer-reviewed)abstract
- Digital images are an intuitive way to capture, store and analyse organismal phenotypes. Many biologists are taking images to collect high-dimensional phenotypic information from specimens to investigate complex ecological, evolutionary and developmental phenomena, such as relationships between trait diversity and ecosystem function, multivariate natural selection or developmental plasticity. As a consequence, images are being collected at ever-increasing rates, but extraction of the contained phenotypic information poses a veritable analytical bottleneck. phenopype is a high-throughput phenotyping pipeline for the programming language Python that aims at alleviating this bottleneck. The package facilitates immediate extraction of high-dimensional phenotypic data from digital images with low levels of background noise and complexity. At the core, phenopype provides functions for rapid signal processing-based image preprocessing and segmentation, data extraction, as well as visualization and data export. This functionality is provided by wrapping low-level computer vision libraries (such as OpenCV) into accessible functions to facilitate scientific image analysis. In addition, phenopype provides a project management ecosystem to streamline data collection and to increase reproducibility. phenopype offers two different workflows that support users during different stages of scientific image analysis. The low-throughput workflow uses regular Python syntax and has greater flexibility at the cost of reproducibility, which is suitable for prototyping during the initial stages of a research project. The high-throughput workflow allows users to specify and store image-specific settings for analysis in human-readable YAML format, and then execute all functions in one step by means of an interactive parser. This approach facilitates rapid program-user interactions during batch processing, and greatly increases scientific reproducibility. Overall, phenopype intends to make the features of powerful but technically involved low-level CV libraries available to biologists with little or no Python coding experience. Therefore, phenopype is aiming to augment, rather than replace the utility of existing Python CV libraries, allowing biologists to focus on rapid and reproducible data collection. Furthermore, image annotations produced by phenopype can be used as training data, thus presenting a stepping stone towards the application of deep learning architectures.
|
|
| 8. |
- Löfgren, Oskar, et al.
(author)
-
Landscape history confounds the ability of the NDVI to detect fine-scale variation in grassland communities
- 2018
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 9:9, s. 2009-2018
-
Journal article (peer-reviewed)abstract
- The NDVI is a remotely sensed vegetation index that is frequently used in ecological studies. There is, however, a lack of studies that evaluate the ability of the NDVI to detect fine-scale variation in grassland plant community composition and species richness. Ellenberg indicators characterize the environmental preferences of plant species—and community-mean Ellenberg values have been used to explore the environmental drivers of community assembly. We used variation partitioning to test the ability of satellite-based NDVI to explain community-mean Ellenberg nutrient (mN) and moisture (mF) indices, and the richness of habitat-specialist species in dry grasslands of different ages. The grasslands represent a gradient of decreasing soil nutrient status. If community composition is determined by the responses of individual species to the underlying environmental conditions and if, at the same time, community composition determines the optical characteristics of the vegetation canopy, then positive relationships between the NDVI and mN and mF are expected. Many grassland specialists are intolerant of nutrient-rich soils. If specialist richness is negatively related to soil-nutrient levels, then a negative association between the NDVI and specialist richness is expected. However, because grassland community composition is not only influenced by abiotic variables but also by other spatial and temporal drivers, we included spatial variables and grassland age in the statistical analyses. The NDVI explained the majority of the variation in mF, and also contributed to a substantial proportion of the variation in mN. However, variation in specialist richness and the lowest values of mN were explained by grassland age and spatial variables—but were poorly explained by the NDVI. Synthesis and applications. The NDVI showed a good ability to detect variation in plant community composition, and should provide a valuable tool for assessing fine-scale environmental variation in grasslands or for monitoring changes in grassland habitat properties. However, because the concentration of grassland specialists not only depends on environmental variables but also on the age and spatial context of the grasslands, the NDVI is unlikely to allow the identification of grasslands with high numbers of specialist species.
|
|
| 9. |
- Moller, Anders Pape, et al.
(author)
-
Clutch-size variation in Western Palaearctic secondary hole-nesting passerine birds in relation to nest box design
- 2014
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 5:4, s. 353-362
-
Journal article (peer-reviewed)abstract
- Secondary hole-nesting birds that do not construct nest holes themselves and hence regularly breed in nest boxes constitute important model systems for field studies in many biological disciplines with hundreds of scientists and amateurs involved. Those research groups are spread over wide geographic areas that experience considerable variation in environmental conditions, and researchers provide nest boxes of varying designs that may inadvertently introduce spatial and temporal variation in reproductive parameters. We quantified the relationship between mean clutch size and nest box size and material after controlling for a range of environmental variables in four of the most widely used model species in the Western Palaearctic: great tit Parus major, blue tit Cyanistes caeruleus, pied flycatcher Ficedula hypoleuca and collared flycatcher F.albicollis from 365 populations and 79610 clutches. Nest floor area and nest box material varied non-randomly across latitudes and longitudes, showing that scientists did not adopt a random box design. Clutch size increased with nest floor area in great tits, but not in blue tits and flycatchers. Clutch size of blue tits was larger in wooden than in concrete nest boxes. These findings demonstrate that the size of nest boxes and material used to construct nest boxes can differentially affect clutch size in different species. The findings also suggest that the nest box design may affect not only focal species, but also indirectly other species through the effects of nest box design on productivity and therefore potentially population density and hence interspecific competition.
|
|
| 10. |
- Olsson, Ola, et al.
(author)
-
Efficient, automated and robust pollen analysis using deep learning
- 2021
-
In: Methods in Ecology and Evolution. - 2041-210X. ; 12:5, s. 850-862
-
Journal article (peer-reviewed)abstract
- Pollen analysis is an important tool in many fields, including pollination ecology, paleoclimatology, paleoecology, honey quality control, and even medicine and forensics. However, labour‐intensive manual pollen analysis often constrains the number of samples processed or the number of pollen analysed per sample. Thus, there is a desire to develop reliable, high‐throughput, automated systems. We present an automated method for pollen analysis, based on deep learning convolutional neural networks (CNN). We scanned microscope slides with fuchsine stained, fresh pollen and automatically extracted images of all individual pollen grains. CNN models were trained on reference samples (122,000 pollen grains, from 347 flowers of 83 species of 17 families). The models were used to classify images of different pollen grains in a series of experiments. We also propose an adjustment to reduce overestimation of sample diversity in cases where samples are likely to contain few species. Accuracy of a model for 83 species was 0.98 when all samples of each species were first pooled, and then split into a training and a validation set (splitting experiment). However, accuracy was much lower (0.41) when individual reference samples from different flowers were kept separate, and one such sample was used for validation of models trained on remaining samples of the species (leave‐one‐out experiment). We therefore combined species into 28 pollen types where a new leave‐one‐out experiment revealed an overall accuracy of 0.68, and recall rates >0.90 in most pollen types. When validating against 63,650 manually identified pollen grains from 370 bumblebee samples, we obtained an accuracy of 0.79, but our adjustment procedure increased this to 0.85. Validation through splitting experiments may overestimate robustness of CNN pollen analysis in new contexts (samples). Nevertheless, our method has the potential to allow large quantities of real pollen data to be analysed with reasonable accuracy. Although compiling pollen reference libraries is time‐consuming, this is simplified by our method, and can lead to widely accessible and shareable resources for pollen analysis.
|
|
