Evidence regarding the ability of agroforests to conserve biological diversity has been mixed; they tend to maintain avian communities with species richness similar to that of undisturbed forest ecosystems but generally do not completely preserve community composition. Using a combination of occupancy modeling and non-metric multidimensional scaling on point-count data, we assessed changes in avian community diversity and composition along a successional gradient in traditional Lacandon Maya agroforests and compared them to protected areas in the region. Bird species richness and diversity in Lacandon agroforests peaked in early secondary forest stages. These agroforests’ mean Shannon-Weiner diversity was 5% higher than that of nearby protected areas, but their species richness was similar. Community composition in Lacandon agroforests changed throughout succession, with earlier stages supporting communities distinctly characterized by generalist species, while subsequent, less-intensively managed stages tended to support more forest-dwellers. The bird community observed in even the most mature secondary forest stages in Lacandon agroforests differed from that of undisturbed rainforest ecosystems. These results demonstrate the potential of traditional Lacandon agroforestry management to conserve avian biodiversity while ensuring food sovereignty for farmers. However, because the community composition of early successional stages was different than later stages, shortening fallow cycles and reducing forest cover to increase agricultural production will limit the species this system can support. This study illustrates the value of incorporating traditional agroecosystems into conservation planning as well as maintaining protected areas, because the latter serve as refugia for species that require undisturbed forest habitat in an agroecological matrix.

We observed Lacandon agroforest avian communities in Lacanjá Chansayab, Chiapas, Mexico (16°45'43.7"N, 91°07'44.3"W).  As a reference, we also studied avian communities in protected areas in Lacanjá Chansayab and the Ejido Reforma Agraria, Chiapas, Mexico (16°15'18.0"N, 90°51'33.7"W), both of which are located within or directly adjacent to the Montes Azules Biosphere Reserve (MABR).  We conducted unlimited-radius point counts in all seven Lacandon agroforest successional stages and protected areas.  Sampling occurred from May-August in 2013, 2015, and 2017; we pooled data across all years.  We performed counts during the first three hours after sunrise, which was roughly 0630 h to 0930 h in the MABR region.  In Lacandon agroforests, we randomly selected patches for each successional stage and established observation sites (i.e., points) at the approximate center of each patch.  We ensured at least 70 m of separation between sites to safeguard against including observations of birds in surrounding stands.  In large, uniform, contiguous forest tracts, we performed point counts at approximately 200 m intervals. Counts lasted for 20 minutes immediately upon arrival at a site.  We made aural observations and recordings of bird vocalizations during point counts using a Sennheiser ME66/K6 Super-Cardioid Mic Capsule with a wind shield and an Olympus LS-10 Linear PCM Recorder.  Recordings were analyzed using Adobe Audition software to include bird vocalizations not identified in the field.  In both cases, we noted the time to detection (in minutes) for each observed species, as well as the spatial coordinates and successional stage of each site. We also classified bird species into feeding guilds and habitat preference groups as per Van der Wal et al. (2012). 

• Recording: Unique identifier for each point count
• Location: Categorical variable specifying the location of each observation
• Management: Categorical variable specifying the land management (either Lacandon agroforest or protected area) in which the point count was carried out
• Stage: Lacandon agroforestry successional stage (in Lacandon Maya) in which species was observed, if applicable
• Year: Year observed
• Sci.Name: Observed bird Latin binomial
• Com.Name: Common name in English
• Time: Time observed during point count (mm:ss)
• Lat: Latitude of point coordinate
• Longitude: Longitude of point coordinate
• Zone: UTM zone of point coordinate
• Northing: Northing of point coordinate
• Feeding: Feeding guild of observed species as per Van der Wal et al. (2012)
• Habitat: Habitat preference of observed species as per Van der Wal et al. (2012)
• Vegetation: Dominant vegetation of stand where point was located