Search for: All records

Montane ecosystems are crucial for maintaining global biodiversity and function that sustain life on our planet. Yet, these ecosystems are highly vulnerable to changing temperatures and may undergo critical transitions under ongoing climate change. What we do not know is to what extent montane biodiversity and ecosystem services will respond to local temperature variations in a gradual versus abrupt manner across global environments. To fill this knowledge gap, we conducted a global synthesis, including 4,462 observations from 290 elevation gradients, to investigate how biodiversity (spanning animals and plants) and ecosystem services (including plant production, soil carbon, and fertility) respond to local temperature variations along elevation gradients. We found that nearly one-third of these gradients exhibited abrupt shifts in multiple biodiversity and ecosystem services in response to local variations in temperature along elevation gradients. More specifically, we showed that once a particular local temperature level (~10 °C for mean annual temperature) was reached, even small increases in temperature resulted in dramatic variations in biodiversity and ecosystem services. We further showed that those abrupt shifts in response to local temperature increases were commonly positive for plant and animal diversity, as well as plant production, while soil carbon and fertility more commonly exhibit negative abrupt trends. Our work, based on the most comprehensive empirical evidence available so far, reveals the pervasive abrupt responses of biodiversity and ecosystem services to local temperature variations in montane ecosystems worldwide, highlighting the highly sensitive nature of montane ecosystems in the context of climate change. 

Existing Neural Architecture Search (NAS) methods either encode neural architectures using discrete encodings that do not scale well, or adopt supervised learning-based methods to jointly learn architecture representations and optimize architecture search on such representations which incurs search bias. Despite the widespread use, architecture representations learned in NAS are still poorly understood. We observe that the structural properties of neural architectures are hard to preserve in the latent space if architecture representation learning and search are coupled, resulting in less effective search performance. In this work, we find empirically that pre-training architecture representations using only neural architectures without their accuracies as labels improves the downstream architecture search efficiency. To explain this finding, we visualize how unsupervised architecture representation learning better encourages neural architectures with similar connections and operators to cluster together. This helps map neural architectures with similar performance to the same regions in the latent space and makes the transition of architectures in the latent space relatively smooth, which considerably benefits diverse downstream search strategies.