BioGD: Bio-inspired robust gradient descent.

Recent research in machine learning pointed to the core problem of state-of-the-art models which impedes their widespread adoption in different domains. The models' inability to differentiate between noise and subtle, yet significant variation in data leads to their vulnerability to adversarial perturbations that cause wrong predictions with high confidence. The study is aimed at identifying whether the algorithms inspired by biological evolution may achieve better results in cases where brittle robustness properties are highly sensitive to the slight noise. To answer this question, we introduce the new robust gradient descent inspired by the stability and adaptability of biological systems to unknown and changing environments. The proposed optimization technique involves an open-ended adaptation process with regard to two hyperparameters inherited from the generalized Verhulst population growth equation. The hyperparameters increase robustness to adversarial noise by penalizing the degree to which hardly visible changes in gradients impact prediction. The empirical evidence on synthetic and experimental datasets confirmed the viability of the bio-inspired gradient descent and suggested promising directions for future research. The code used for computational experiments is provided in a repository at https://github.com/yukinoi/bio_gradient_descent.

Harvesting induced stability and instability in a tri-trophic food chain.

Non-equilibrium dynamics in the form of oscillations or chaos is often found to be a natural phenomenon in complex ecological systems. In this paper, we first analyze a tri-trophic food chain, which is an extension of the Rosenzweig-MacArthur di-trophic food chain. We then explore the impact of harvesting individual trophic levels to answer the following questions : a) when a non-equilibrium dynamics persists, b) whether it can locally be stabilized to a steady state, c) when the system switches from a stable steady state to a non-equilibrium dynamics and d) whether the Maximum Sustainable Yield (MSY) always exists when the top predator is harvested. It is shown that searching for a general theory to unify the harvesting induced stability must take into account the number of trophic levels and the degree of species enrichment, the outcomes that cannot be obtained from the earlier reports on prey-predator models. We also identify the situation where harvesting induces instability switching: the non-equilibrium state enters into a stable steady-state and then, upon more intensive harvesting, the steady-state again loses its stability. One of the new and important results is also that the MSY may not exist for harvesting the top predator. In general, our results contribute to biological conservation theory, fishery and ecosystem biodiversity management.

Predicting perceived level of disturbance of visitors due to crowding in protected areas.

Managing the disturbance of visitors due to crowding is an important management task in protected areas with high use levels. To achieve this, managers need to know how the use level affects the perceived disturbance due to crowding. Here we present a method to predict the level of disturbance as a function of use level measured by number of visitors. In contrast to the visual approach where subjects are asked to evaluate acceptability of use levels from manipulated images of scenery, our approach uses data gathered from actual experiences: actual (measured) use levels and concurrent on-site data on levels of disturbance experienced by visitors. Using the example of Nature Park Telascica, we show how these data can be acquired with limited resources (a smart-phone and short, time-stamped questionnaires), and demonstrate the subsequent analysis and model fitting. The resulting model estimates the probability that a visitor experiencing a given use level will report certain level of disturbance. We suggest a way of using the probability density functions to define an inherent limit of acceptable disturbance (LAD) due to crowding; the LAD can also be set to a desired value by management. Regardless of the definition, LAD can be used to determine the maximum acceptable use level as dictated by crowding considerations. The method gives predictions consistent with previous literature and can be used even when data are collected at low use levels.

Inferring physiological energetics of loggerhead turtle (Caretta caretta) from existing data using a general metabolic theory.

Loggerhead turtle is an endangered sea turtle species with a migratory lifestyle and worldwide distribution, experiencing markedly different habitats throughout its lifetime. Environmental conditions, especially food availability and temperature, constrain the acquisition and the use of available energy, thus affecting physiological processes such as growth, maturation, and reproduction. These physiological processes at the population level determine survival, fecundity, and ultimately the population growth rate-a key indicator of the success of conservation efforts. As a first step towards the comprehensive understanding of how environment shapes the physiology and the life cycle of a loggerhead turtle, we constructed a full life cycle model based on the principles of energy acquisition and utilization embedded in the Dynamic Energy Budget (DEB) theory. We adapted the standard DEB model using data from published and unpublished sources to obtain parameter estimates and model predictions that could be compared with data. The outcome was a successful mathematical description of ontogeny and life history traits of the loggerhead turtle. Some deviations between the model and the data existed (such as an earlier age at sexual maturity and faster growth of the post-hatchlings), yet probable causes for these deviations were found informative and discussed in great detail. Physiological traits such as the capacity to withstand starvation, trade-offs between reproduction and growth, and changes in the energy budget throughout the ontogeny were inferred from the model. The results offer new insights into physiology and ecology of loggerhead turtle with the potential to lead to novel approaches in conservation of this endangered species.

Nutrient quotas and carbon content variability of Prorocentrum minimum (Pavillard) Schiller, 1933.

Frequency, severity, and geographic range of harmful blooms caused by a dinoflagellate Prorocentrum minimum have been increasing significantly over the past few decades. The ability to adapt nutrient quotas and carbon content to a wide range of environmental conditions is one of the key factors for the proliferation of P. minimum. Understanding the limits of stoichiometric variability in terms of nutrient quotas and carbon content would help explain the observed trends and assist in P. minimum growth model creation. This manuscript aggregates information from 15 studies to investigate variability in nutrient quotas and carbon content for a broad range of P. minimum isolates and clonal lines. Nitrogen quota, phosphorus quota, and carbon content in the studies varied between 11-107.5pgNcell-1, 1.45-17.58pgPcell-1, and 70-656.36pgCcell-1, respectively. Regression analysis was used to estimate average nitrogen and phosphorus quotas as functions of carbon, and to show that carbon content variability explains 55% of nitrogen and 23% of phosphorus quota variability. Confidence intervals for data (CID) found during the analysis were used to define maximal and minimal nutrient quotas as functions of carbon content. The ratios of the upper and lower CID ranges can, therefore, be used to estimate nutrient storage capacity as a function of carbon content. The new results and comparison with other species show that, at least for P. minimum, carbon-based quotas are more suitable for modelling than cell-based quotas. Finally, results indicate that environmental nutrient availability affects quotas more than light does: while quota variability due to light remains within 80% CID, nutrient variability covers the 95% CID.

Impact of maximum sustainable yield on competitive community.

A system of n competing logistical species of the Volterra type under proportional harvesting strategy is analyzed. In case of selective harvesting, when the effort is adjusted to each species, the optimum effort may result in the total maximum sustainable yield (TMSY(1)). When it exists, reaching TMSY(1) does not affect the system stability character, but it does affect the state, and hence some populations may reach too small a value to persist in nature. If competition is strong, species with smaller biotic potential may be driven to extinction. In case the system is harvested with a common harvesting effort, such as in trawler fishery, the total maximum sustainable yield (TMSY(2)) is smaller than TMSY(1), and all the species with lower or equal biotic potential to the optimum harvesting effort will be driven to extinction. In this case a call for implementation of the MSY is equivalent to a call for the extermination of some species and it runs directly against the Convention on Biological Diversity (CBD, 1992). Therefore, all legal documents advocating MSY in ecosystems starting with the Johannesburg Plan of Implementation (JPI, 2002) must be urgently retracted and replaced with adaptive management which will respect CBD.