Design of optimal ecosystem monitoring networks: hotspot detection and biodiversity patterns

Abstract

Effective monitoring of ecosystems is crucial for assessing and possibly anticipating shifts, quantifying ecosystem services, and optimal decision making based on shifts and services. The selection of monitoring sites is typically suboptimal following local stakeholder or research interests that do not allow to capture ecosystem patterns and dynamics as a whole. The design of optimal monitoring network is crucial for the accurate determination of biodiversity patterns of ecosystems. A novel model for the design of optimal monitoring networks for biodiversity based on the concept of the value of information (VoI) is proposed. The VoI is assigned to species richness that is the economically and ecologically valuable metric. As a case study the trinational frontier ecosystem among Brazil, Peru, and Bolivia is considered for the model. A multiresolution texture-based model estimates species richness and turnover on satellite imagery calibrated on different sets of information coming from forest plot data organized in network topologies. The optimal monitoring network is the network that minimizes the integrated VoI defined as the variation of the VoI in the 28 years considered. This is equivalent to minimize the sum of the species turnover of the ecosystem. The small world network is identified as the optimal and most resilient monitoring network whose nodes are the hotspots of species richness. The hotspots are identified as the sites whose VoI is the highest for the whole period considered. Hence, the hotspots are the most valuable communities for inferring biodiversity patterns and the most ecologically/economically valuable according to the richness-resilience hypothesis. Most hotspots are honored by the small world network that can be thought as the ''backbone'' ecological network of the ecosystem. The small world monitoring network has an accuracy *50 % higher than other network topologies in predicting biodiversity patterns. This network Electronic supplementary material The online version of this article (has the highest VoI at any time step and scale considered; thus, it guarantees to track changes of ecosystems in space and time. The network that results from the optimal tradeoff between data value with their uncertainty and relevance, has deep implications for understanding ecosystem function and for management decisions. The model allows to include preferences for ecosystem communities by using differential weights on the VoI of these communities, and economic constraints that limit the extension of the network. Because of the optimal integration of environmental, social, and economical factors the model allows a sustainable monitoring and planning of biodiversity for the future.

The effects of marine protected areas over time and species' dispersal potential: a quantitative conservation conflict attempt

“…Models are used when experiments are costly, require significant labour effort, ethics, and effects of spatial or temporal scales associated. Cellular automata and agent-based models are useful tools for addressing such issues (Bastardie et al, 2013;Convertino et al, 2015;DeAngelis and Yurek, 2015;Eide, 2012Eide, , 2014Moustakas and Silvert, 2011;Moustakas et al, 2006).…”

Integrating modelling and smart sensors for environmental and human health

“…This is also true considering the advancement of a socially and computationally driven global system science (Convertino et al , 2014, 2015; Helbing et al , 2015). This has paved the way for some of the coupling and integration activities described in section 3.2 (for details on conceptual approaches for model integration see Argent, 2004a, b; Kelly et al , 2013).…”

“…The central role of information (factors, interpretation, values, uncertainty, transformation and context) is highlighted. Here, information is also depicted as input to the modelling stage, e.g., to reduce the size of ’Big Data’ by extracting only data with high information value for the question being asked (Shannon and Weaver 1949, Lazer et al ., 2014, Galelli et al , 2014; Convertino et al , 2014, 2015). Information in general and the policy questions to be assessed in particular include value judgements (Voinov et al , 2014).…”

“…for determination of regional greenhouse gas (GHG) emissions or vertical profile retrieval of GHGs) and recently applied for human exposure assessment in urban regions (Schlink and Fischer, 2014). While instruments must have well quantified error characteristics, the discussion is now centred on information content , and how this could be optimised by suitable deployments of instruments and networks (Convertino et al , 2014, 2015). The issue is no longer just ‘how good is an instrument’ , but ‘have we placed it where it provides maximum information’ ?…”

“…Such models rely on data for estimating non-physical transition rates among compartments, lack a spatial component, and very rarely are coupled to environmental and agent-causing-disease compartments whose information can be derived from integrated models (Convertino et al , 2014, 2015; Rinaldo et al , 2012; Helbing et al , 2014).…”