Validating Landsat-based landscape metrics with fine-grained land cover data

ContextModerate-grained data may not always represent landscape structure in adequate detail which could cause misleading results. Certain metrics have been shown to be predictable with changes in scale; however, no studies have verified such predictions using independent fine-grained data.ObjectivesOur objective was to use independently derived land cover datasets to assess relationships between metrics based on fine- and moderate-grained data for a range of analysis extents. We focus on metrics that previous literature has shown to have predictable relationships across scales.MethodsThe study area was located in eastern Connecticut. We compared a 1 m land cover dataset to a 30 m resampled dataset, derived from the 1 m data, as well as two Landsat-based datasets. We examined 11 metrics which included cover areas and patch metrics. Metrics were analyzed using analysis extents ranging from 100 to 1400 m in radius.ResultsThe resampled data had very strong linear relationships to the 1 m data, from which it was derived, for all metrics regardless of the analysis extent size. Landsat-based data had strong correlations for most cover area metrics but had little or no correlation for patch metrics. Increasing analysis areas improved correlations.ConclusionsRelationships between coarse- and fine-grained data tend to be much weaker when comparing independent land cover datasets. Thus, trends across scales that are found by resampling land cover are likely to be unsuitable for predicting the effects of finer-scale elements in the landscape. Nevertheless, coarser data shows promise in predicting fine-grained for cover area metrics provided the analysis area used is sufficiently large.     

A social learning model of conflict and cooperation in human societies

Patterns of conflict and cooperation both within and between societies may be related to the degree of cultural similarity within and between the same societies. A simple model of social learning is used to predict patterns of conflict and cooperation in hypothetical societies that differ in the roles of relatives and nonrelatives in the enculturation of children. The model is illustrated by comparing its predictions to known differences in the patterns of conflict between males inpatrilocal and matrilocal societies.     

The role of context, colour and location cues in socially learned novel food source preferences in starlings, Sternus vulgaris

Although the opportunity for errors in social learning is widely recognised, as yet little research has been directed towards understanding specific inaccuracies, biases and limitations in social learning and the mechanisms that give rise to them. In two experiments I ask how starlings, Sternus vulgaris, identify exemplars of novel feeders previously learned about socially. I find that starlings have a stronger response to feeders in the same context as that in which social learning took place, compared to identical and nonidentical feeders in a different context. Within a context that matches where social learning took place, starlings prefer feeders that show the same location and colour as the feeder demonstrated by the demonstrator starling, and show no preference when colour and location cues are dissociated. This suggests that starlings are relatively accurate social learners, since they show strong responses to novel foraging options only if they match the context, colour and location of options learned about socially, and they do so after very few trials. Furthermore, the responses of the subjects were compatible with conditioned learning-like mechanisms, which provide a useful basis for the further investigation of the origins and implications of errors in social learning.     

Structure-based protein design with deep learning

Since the first revelation of proteins functioning as macromolecular machines through their three dimensional structures, researchers have been intrigued by the marvelous ways the biochemical processes are carried out by proteins. The aspiration to understand protein structures has fueled extensive efforts across different scientific disciplines. In recent years, it has been demonstrated that proteins with new functionality or shapes can be designed via structure-based modeling methods, and the design strategies have combined all available information — but largely piece-by-piece — from sequence derived statistics to the detailed atomic-level modeling of chemical interactions. Despite the significant progress, incorporating data-derived approaches through the use of deep learning methods can be a game changer. In this review, we summarize current progress, compare the arc of developing the deep learning approaches with the conventional methods, and describe the motivation and concepts behind current strategies that may lead to potential future opportunities.     

Ability of predatory native Australian fishes to learn to avoid toxic larvae of the introduced toad Bufo marinus

Of two native Australian fishes naïve to the introduced toad Bufo marinus most barramundi Lates calcarifer rapidly learned to avoid B. marinus tadpoles, while sooty grunter Hephaestus fuliginosus exhibited considerable intraspecific variation in their learning ability. Some sooty grunter learned to avoid tadpoles after only a few attacks, while other individuals continued to attack and reject tadpoles throughout the entire laboratory trials. Individuals of both species recognized and avoided tadpoles 1 day after their previous encounter. None of the fishes died during the trials. The observed variation in behavioural responses of fishes to B. marinus may be due to differences in (1) learning ability, (2) fish hunger levels, and or (3) tadpole palatability and toxicity. The results demonstrate that most barramundi and sooty grunter learn to avoid B. marinus tadpoles with minimal trauma. Consequently, it is anticipated that the toads are unlikely to have a significant negative impact on wild populations of these fishes through direct toxic effects.     

Foraging Efficacy of a Larval Parasitoid in a Cotton Patch: Influence of Chemical Cues and Learning

Plant-herbivore chemical signals and behavioral plasticity may enhance parasitoid host-foraging efficacy in the field; however, no studies have quantified the potential benefits from these factors under field-type conditions. The effect of plant-herbivore signals and learning on the foraging efficacy of Microplitis croceipes was quantified by directly observing and recording total and sequential duration of various foraging behaviors relative to 5 randomly placed herbivore-damaged and host-infested cotton plants and 20 undamaged and non-host-infested plants. Microplitis croceipes spent significantly more time searching (flying and antennation) on host infested versus uninfested plants. Antennation time was significantly and negatively correlated with successive host stings. Contrary to expectations of increased duration, flight time remained constant throughout the foraging bout, which may indicate that there was some learning associated with flight. These results suggest that plant-herbivore chemical signals and learning enhances the foraging efficacy of M. croceipes.