KEY-WORDS
autonomous recording units, point count surveys, avian diversity, biodiversity monitoring, elevational gradient, species richness, high mountain biodiversity, alpine
Species surveys are used to determine the presence, relative abundance, and diversity of taxa over space and time (Roberts, 2011; Sauer et al., 2017; Schramm et al., 2020). cornerstone of ecological studies used to identify biodiversity hotspots, infer the impact of natural or anthropogenic disturbance on communities, assess the effectiveness of management practices, and identify important habitats for species of conservation concern (e.g. Dorji et al., 2019; Friedlander et al., 2019; Ibarra & Martin, 2015; Rosenberg et al., 2017). For effective conservation decision-making to occur, biases associated with any given survey technique should be quantified and, where possible, corrected for. When abundance and diversity data are compared across broad regions and divergent communities, any interaction between detection bias due to survey method and the landscapes and/or communities being surveyed is a concern. The use of survey method can reveal such problems and may increase project coverage and efficiency.
For birds in terrestrial habitats, point counts have been the standard survey method for more than 80 years (Ralph et al., 1995). Point counts employ 1-2 trained observers to identify and count birds by sight and sound from a single location for a set period of time. Within the past 20 years, the use of autonomous recording units (ARUs) as an alternative to point count surveys has become increasingly popular (Darras et al., 2019). ARUs are installed at survey sites and record ambient sound that is then in the lab, with species identified by their vocalizations either manually or using identification software. Both methods have benefits and limitations as techniques for surveying avian diversity. Key among the benefits of point counts is the ability to visually species (Acevedo & Villanueva-Rivera, 2006; Hutto & Stutzman, 2009; Vold et al., 2017) and use distance to obtain better density estimates than can be assessed by audio alone (Shonfield & Bayne, 2017). Because point countobservers can assess call direction and they outperform ARUs when calls occur outside the ARU microphone(s) “line-of-sight” (Castro et al., 2019). ARUs, on the other hand, overcome logistical constraints experienced by point counts that can impact species detections. ARUs can collect data simultaneously from multiple sites, allowing projects to survey during peak diel activity for both diurnal and nocturnal species (Goyette et al., 2011) and eliminating potential time bias present in point counts (Darras et al., 2019). ARUs can be left in high latitude and high elevation habitats year-round and programed to start recording in spring, before observers can access these regions (e.g. Shonfield & Bayne, 2017). They can, therefore, better-sample peak seasonal activity for resident species and detect shifts in bird phenology (Klingbeil & Willig, 2015). Finally, as inanimate objects, ARUs are less likely to alter bird behaviour (Shonfield & Bayne, 2017, Darras et al., 2019,but see Hutto & Hutto 2020.
Effort is a consideration for research programs. Point counts and ARUs differ in their time costs. A single point count is completed in a single site visit. Establishing an ARU site and collecting data entails a minimum of two site visits; however, ARU recordings can subsequently be intensively sampled without increased field costs or increased site disturbance. ARUs can have notable drawbacks in terms of processing time in the lab: without automated data processing, the time costs of uploading and interpreting audio files, replaying sections of audio, and then transcribing observations is greater than for detections and transcriptions of equivalent length point counts (e.g. this study; Celis-Murillo et al., 2009; Alquezar & Machado, 2015). Even with automated processing, the need to proof output can eliminate any time advantages over manual scanning (Joshi et al., 2017 but seeKnight et al., 2020).
Despite the fact that mountains support important bird diversity, most high elevation systems in the Americas are poorly monitored (Boyle & Martin, 2015). Mountain habitats present challenging conditions in which to conduct avian surveys. Access is often limited by difficult terrain, late snowmelt, and poor infrastructure. Surveys may be disrupted by extreme weather. By necessity, mountain surveys are typically conducted in a linear fashion upslope or downslope, creating time bias in point counts across elevation. Given their field advantages, ARUs offer a compelling alternative to point counts at high elevation sites. Here we examine the performance of ARUs and point count surveys in detecting and quantifying avian diversity across a gradient of temperate mountain habitats in both North and South America. In both countries sampling encompassed three structurally similar habitats: densely forested upper montane, open subalpine, and highly exposed alpine. Using species detections at shared sites, we directly compare diversity index values and species accumulation curves produced by these two methods. We investigate the underlying causes of differences in diversity values obtained by each method by modeling method effects on the detection probabilities of bird families within the two regions. In order to make recommendations for future monitoring protocols, we examine the efficiency (time cost versus species detection return) of point counts and ARU sampling on their own, and for combined-method protocols.