Search

ABSTRACTNature provides numerous functions and services that can contribute to societal climate change adaptation. These ‘adaptation services’ can be sustained, latent, or novel, depend on persistence or transformation of ecosystems, and require varying co-production by people. Adaptation services also include climate mitigation. We present an approach to explore an extensive set of land use and climate scenarios that outline possible futures for a landscape, and quantify the contributions of adaptation services. We quantified adaptation success across six criteria relevant to the region, characterised the contributions of different types of adaptation services, and mapped spatial variation in contributions across the landscape. We built an integrated model of the Mackenzie District (an upland landscape of Aotearoa New Zealand), and analysed 1200 hypothetical scenarios for the period 2060–2070. We found many adaptation options, with 46% of scenarios meeting the criteria for successful adaptation. Four sustained, two latent and five novel services co-produced through financial input made diverse contributions to economic profit, profit resilience, climate change mitigation, climate risk adaptation, landscape cultural heritage and biodiversity. Successful adaptation scenarios were multifunctional, relying on alternative combinations of services allowed by spatial heterogeneity. By accounting for the numerous relationships between people and natural components within complex landscape systems, our advanced simulation approach can inform participatory pathway development by quantifying the potential for nature to contribute to future climate change adaptation.

The gap in resolution between existing global climate model output and that sought by decision-makers drives an ongoing need for climate downscaling. Here we test the extent to which developments in deep learning can out-perform existing statistical approaches for downscaling historical rainfall in the highly complex terrain setting of New Zealand. While deep learning removes the need for manual feature selection when extracting spatial-temporal information from predictor fields, several key considerations need to be addressed. These include: the chosen complexity of the network architecture, suitable loss functions tailored to the problem, as well as input data considerations of domain size and amount of training data required to provide adequate out-of-sample generalization. Sensitivity testing to these considerations reveals that a relatively simple convolutional neural network (CNN) architecture with carefully selected loss functions can considerably outperform existing statistical downscaling models based on multiple linear regression with manual feature selection. When aggregated across the entire region, the fraction of explained variance on wet days increased from 0.35 to 0.52, the root-mean-squared error reduced by over 20% and percentage biases for the 90th percentile of rainfall improved by over 25%. Using interpretable machine learning methods, we demonstrate that the CNN has been capable of self-learning physically plausible relationships between the large-scale atmospheric environment and extreme localized rainfall events. The historical performance and physical interpretability documented here lends support for wider development and application of deep learning in climate downscaling.

Calcineurin (CaN) is a calcium/calmodulin-dependent serine/threonine phosphatase with a crucial role in cellular homeostasis. It is also the target of the Food and Drug Administration (FDA) approved immunosuppressant drugs FK506 and cyclosporine A. Recent work from our group and others indicated that an uncontrolled increase in CaN activity causes synaptic dysfunction and neuronal death in various models of neurodegenerative diseases associated with calcium dysregulation. Furthermore, pharmacological normalization of CaN activity can prevent disease progression in animal models. However, none of the FDA-approved CaN inhibitors bind CaN directly, leading to adverse side effects. The development of direct CaN inhibitors is required to reduce off-target effects, but its highly conserved active site and similar mechanism of action with other protein serine/threonine phosphatases impose a significant challenge. In this work, we developed a novel pharmacophore model to screen for CaN-specific inhibitors. Then, we performed a virtual screen for molecules having the pharmacophore model. We also show that the molecules identified in this screen can inhibit CaN with a low micromolar IC50. Interestingly, the inhibitors identified from the screen do not inhibit phosphoprotein phosphatase 2A, a member of the serine/threonine phosphatase family that shares 43% sequence identity with the CaN active site. The pharmacophore model that we developed and validated in this work may help to accelerate the development of specific CaN inhibitors.