Quantifying the impact of habitat disturbance on ecosystem function is critical for understanding and predicting the future of tropical forests. Many studies have examined post-disturbance changes in animal traits related to mutualistic interactions with plants, but the effect of disturbance on plant traits in diverse forests has received much less attention. 2. Focusing on two study regions in the eastern Brazilian Amazon, we used a trait-based approach to examine how seed dispersal functionality within tropical plant communities changes across a landscape-scale gradient of human modification, including both regenerating secondary forests and primary forests disturbed by burning and selective logging. 3. Surveys of 230 forest plots recorded 26,533 live stems from 846 tree species. Using herbarium material and literature, we compiled trait information for each tree species, focusing on dispersal mode and seed size. 4. Disturbance reduced tree diversity and increased the proportion of lower wood-density and smaller-seeded tree species in study plots. Disturbance also increased the proportion of stems with seeds that are ingested by animals and reduced those dispersed by other mechanisms (e.g. wind). Older secondary forests had functionally similar plant communities to the most heavily disturbed primary forests. Mean seed size and wood density per plot were positively correlated for plant species with seeds ingested by animals. 5. Synthesis. Anthropogenic disturbance has major effects on the seed traits of tree communities, with implications for mutualistic interactions with animals. The important role of animalmediated seed dispersal in disturbed and recovering forests highlights the need to avoid defaunation or promote faunal recovery. The changes in mean seed width suggest larger vertebrates hold especially important functional roles in these human-modified forests. Monitoring fruit and seed traits can provide a valuable indicator of ecosystem condition, emphasising the importance of developing a comprehensive plant traits database for the Amazon and other biomes.

The present analysis builds on previous evidence for the California Vowel Shift in San Francisco English (Hall-Lew 2009, 2013) with data on the lowering and retraction of BET (Kennedy & Grama 2012) and the BAT nasal split (Eckert 2008). Based on interview speech from a socially stratified sample of 22 San Franciscans, women lead men in the retraction changes, and European Americans lead Chinese Americans in both BAT retraction and BAN raising. We also find the first evidence for gender-differentiated change in BAN raising when the nasal is velar. Furthermore, preliminary data suggest a pre-velar effect for BET and BAT, which is best described as inhibition of retraction and lowering, not the raising movement of Pacific Northwest varieties (Becker et al. current volume, Wassink current volume). Overall, San Francisco English exhibits precisely the Northern Californian vowel system expected, rather than being an exceptional dialect island (cf. Labov et al. 2006). 1 The automatic align ment and extraction of vowel data is only possible due to long hours of initial orthographic transcription. Our thanks go to research funds from the 2009-2010 Andrew W. Mellon postdoctoral fellowship which funded transcription by RAs Cla ire Drohan, Annabel Schwenk, and Amanda Wall, as well as to the 2010-2011 PPLS Pilot Scheme wh ich funded transcription by RAs Keelin Murray and the first author. A special note of thanks goes to RAs Julie Saigusa and Kieran Wilson, who, along with the authors of this paper, volunteered their time to pro ject transcription during 2014-2015. We also thank Josef Fruehwald and the editors and other contributors to this collection for the conversations that made this paper possible. Lastly, the biggest debt goes to those speakers whose voices are represented here. All shortcomings are our own.

.

No abstract