The majority of published articles were focused on a single taxonomic group
(95%). Mammals and birds were the most modeled taxa (40.87% and 33.83% of
studies), followed by amphibians (10.93%), invertebrates (6.29%),
reptiles (5.69%), fish (5.09%) and plants (2.25%). When more than one
taxonomic group was modeled, the selected taxa included combinations such as mammals and
birds, amphibians and reptiles, reptiles and plants, or the entire
vertebrate community.
Almost half of the studies used data obtained from visual surveys
(49.55%), followed by acoustic surveys (28.59%), camera traps
(25.60%) and traps (16.47%). Less frequently used data were obtained by
recording tracks (3.89%), online databases (3.29%), eDNA (1.80%), and
interviews (1.65%). Nearly 25% of articles used more than one method
for data collection. Such studies combined visual and acoustic
surveys (141 out of 166 articles combining data collection methods),
while the remaining studies combined camera traps and track or traps,
camera traps with interviews, or visual survey and eDNA.
Most data collection designs were identified as grids (40.57%),
followed by transects (24.25%), feature-based (23.80%), opportunistic
collection (11.53%), and territorial unit (4.64%). Only 5% of articles
combined study designs, with the most common combination being transects inside a grid or
feature-based area. Few articles (6 articles) supplemented species data collected from a grid or transect with online databases or interviews.
Hierarchical modeling studies spanned 89 countries and territories
(e.g., Antarctica). Notably, 46.26% of the published papers were
conducted in the USA. The other countries in the top 10 list are Brazil, Canada, Australia, Argentina, India, South Africa,
Spain, China, and Italy, each accounting for only 5% to 2% of the
studies. The majority of studies focused on specific regions within
a single country, with only 4% of papers involving research
conducted in more than one country.
With respect to the type of hierarchical modeling approach, over 30% of studies employed single-season occupancy models,
followed by single-season multi-species (community) occupancy (18.26%),
single-season N-mixture (14.97%), multi-season (dynamic) occupancy
(14.82%), multi-season multi-species (community) occupancy (9.13%),
multi-season (dynamic) N-mixture (6.89%), multi-state (single or
multi-season)(3.89%), multi-scale (single or multi-season)(3.44%), and
co-occupancy (3.29%). Nearly all studies used only one method, with
only 5.23% involving a second method (e.g., occupancy and N-mixture
models for abundant species).
Data collection methods were principally taxa-specific; most bird data were
obtained from acoustic and visual surveys and analyzed using a wide range of models. Data for mammals were mainly based on camera traps
and tracks and were largely modeled using single-season occupancy and
multi-species (community) occupancy. Amphibian data were mostly
collected via acoustic monitoring and visual survey and were modeled using
the entire spectrum of models (Figure \ref{695701}).