Search for: All records

Predicting insect responses to climate change is essential for preserving ecosystem services and biodiversity. Due to high daytime temperatures and low humidity levels, nocturnal insects are expected to have lower heat and desiccation tolerance compared to diurnal species. We estimated the lower (CT_Min) and upper (CT_Max) thermal limits ofMegalopta, a group of neotropical, forest-dwelling bees. We calculated warming tolerance (WT) as a metric to assess vulnerability to global warming and measured survival rates during simulated heatwaves and desiccation stress events. We also assessed the impact of body size and reproductive status (ovary area) on bees’ thermal limits.Megaloptadisplayed lower CT_Min, CT_Max, and WTs than diurnal bees (stingless bees, orchid bees, and carpenter bees), but exhibited similar mortality during simulated heatwave and higher desiccation tolerance. CT_Minincreased with increasing body size across all bees but decreased with increasing body size and ovary area inMegalopta, suggesting a reproductive cost or differences in thermal environments. CT_Maxdid not increase with increasing body size or ovary area. These results indicate a greater sensitivity ofMegaloptato temperature than humidity and reinforce the idea that nocturnal insects are thermally constrained, which might threaten pollination services in nocturnal contexts during global warming.

 

The dimensions of past ice sheets provide a record of palaeoclimate but depend on underlying topography, which evolves over geological timescales by tectonic uplift and erosional downcutting. Erosion during the Pleistocene epoch (2,580 to 11.650 thousand years ago) reduced glacier extent in some locations even as climate cooled, but whether other non-climatic influences impacted the glacial–geological record is poorly known. The Antarctic Peninsula provides an opportunity to examine this issue because of its long glacial history and preservation of remnants of a low-relief pre-glacial land surface. Here we reconstructed both palaeo-surface topography and long-wavelength variations of surface uplift for the Antarctic Peninsula by using inverse analysis that assimilates local topographic remnants with the branching structures of entire modern drainage networks. We found that the Antarctic Peninsula rose tectonically by up to 1.5 km due to dynamical support from the mantle. Glaciological models using the current climate and our palaeotopography show greatly reduced ice extent in the northern Antarctic Peninsula compared with modern, indicating that the onset of glaciation identified at offshore sites reflects tectonic uplift of the topography rather than climatic cooling. In the southern Antarctic Peninsula, however, we suggest the low-relief pre-glacial landscape supported a considerably greater ice volume than the modern mountainous topography, illustrating the influence of erosional sculpting on glaciation patterns.

 

We present a fast, differentially private algorithm for high-dimensional covariance-aware mean estimation with nearly optimal sample complexity. Only exponential-time estimators were previously known to achieve this guarantee. Given n samples from a (sub-)Gaussian distribution with unknown mean μ and covariance Σ, our (ε,δ)-differentially private estimator produces μ~ such that ∥μ−μ~∥Σ≤α as long as n≳dα2+dlog1/δ√αε+dlog1/δε. The Mahalanobis error metric ∥μ−μ^∥Σ measures the distance between μ^ and μ relative to Σ; it characterizes the error of the sample mean. Our algorithm runs in time O~(ndω−1+nd/ε), where ω<2.38 is the matrix multiplication exponent. We adapt an exponential-time approach of Brown, Gaboardi, Smith, Ullman, and Zakynthinou (2021), giving efficient variants of stable mean and covariance estimation subroutines that also improve the sample complexity to the nearly optimal bound above. Our stable covariance estimator can be turned to private covariance estimation for unrestricted subgaussian distributions. With n≳d3/2 samples, our estimate is accurate in spectral norm. This is the first such algorithm using n=o(d2) samples, answering an open question posed by Alabi et al. (2022). With n≳d2 samples, our estimate is accurate in Frobenius norm. This leads to a fast, nearly optimal algorithm for private learning of unrestricted Gaussian distributions in TV distance. Duchi, Haque, and Kuditipudi (2023) obtained similar results independently and concurrently.