Knowledge of habitat selection derived from surveys is an important component of conservation planning. However, mismatches between survey timing and the behaviour of target species can result in observation bias, which can misconstrue the ability to determine species-habitat relationships. For birds, large-scale, long-term surveys are based on the songbird dawn chorus, suggesting that birds which exhibit non-dawn chorusing behaviour may either be only incidentally detected or overlooked. I quantified bias and improved standard morning surveys to evaluate coarse-scale models to predict fine-scale occupancy of non-dawn chorusing birds. I found that nocturnal vocalizations occur in at least 30% of 749 species across 18 of 22 orders, establishing the need for an investigation of bias in standard morning surveys that do not account for birds that exhibit non-dawn chorusing behaviour. Subsequently, I used automated acoustic recorders to collect repeated recordings throughout 24-h periods to compare with results from standard morning surveys, and found that the latter surveys underestimated total species richness (Chapter 2) and waterfowl and songbird occupancy (Chapter 3). Further, I developed a novel subsampling approach for extended acoustic recordings and compared statistical estimators, to efficiently estimate total species richness (Chapter 2). I also investigated the effect of revisitation schedules - same or different day, as well as increased sampling effort- to estimate occupancy for birds with different diel vocalization patterns (Chapter 3). In Chapter 4, I used improved estimates of waterfowl occupancy from extended acoustic recordings to evaluate the ability of previously published coarse-scale models to predict fine-scale distributions, as well as models augmented with additional fine-scale habitat data. Lack of significant increase in model performance with the inclusion of fine-scale habitat data suggested that waterfowl select habitat based more on coarser than finer cues. Nevertheless, no models predicted distribution well enough at fine scales for practical application, suggesting no available shortcuts to conducting fine-scale surveys for local conservation planning. Ultimately, my thesis comprehensively demonstrated that an understanding of vocal behaviour is required for developing effective surveys for birds and illustrated how improved sampling designs can be applied to address important questions in conservation and management.