Skip to main content

Data from: Active restoration accelerates recovery of tropical forest bird assemblages over two decades

Cite this dataset

Joyce, Francis H. et al. (2024). Data from: Active restoration accelerates recovery of tropical forest bird assemblages over two decades [Dataset]. Dryad.


Choosing effective methods to restore habitat for the diverse faunal assemblages of tropical forests is hampered by lack of long-term data comparing multiple restoration treatments. We conducted area counts of bird assemblages over 12 years (~5-17 years since restoration) in a blocked experiment with two active planted treatments (tree plantations and applied nucleation) and a passive restoration treatment (natural regeneration) replicated at 11 sites in Costa Rica. We also surveyed six pastures and five remnant forest sites to assess recovery of avian species richness, composition, forest specialists, and range-restricted species in restoration plots relative to degraded and reference systems. Restoration treatments showed increased resemblance of avian assemblages to remnant forest over time. Applied nucleation proved equally effective as plantation, despite a reduced planted area, whereas natural regeneration recovered more slowly. Assemblage-level trends in avian species richness and compositional similarity to reference forest are underpinned by reductions in use by pasture birds and by gradual increases in richness of forest-affiliated species. Because forest-affiliated species tend to have narrower distributions than the open-country species they replace, forest restoration can reduce biotic homogenization at the local scale. Restoration practitioners should consider applied nucleation as an alternative to standard plantations if seeking rapid recovery of bird assemblages. However, the ecological return on investment from natural regeneration increases over a couple of decades. Managers should monitor trends in forest-affiliated and range-restricted species to track the recovery of the full avian assemblages, since coarse metrics like species richness and overall compositional similarity may plateau relatively quickly.

README: Data Documentation for: Active restoration accelerates recovery of tropical forest bird assemblages over two decades

General Information

Name of data set: Islas Bird Surveys 2010-2021

Data files:

(1) sites.csv

(2) bird_surveys_2010_2021.csv

Date the data set was last modified: February 6, 2023

Funder: U.S. National Science Foundation

How to cite data: Joyce, Francis H. et al. (2024). Active restoration accelerates recovery of tropical forest bird assemblages over two decades [Dataset]. Dryad.

Primary article: Joyce, F. H., Rosales, J. A., Holl, K. D., Zahawi, R. A., Bui, A., & Reid, J. L. (2024). Active restoration accelerates recovery of tropical forest bird communities over two decades. Biological Conservation.

Data Collection

Methodology for data collection:

This dataset contains surveys of bird communities from 2010 to 2021 at 11 restoration sites, 5 mature remnant forests, and 6 active cattle pastures in southern Costa Rica (Coto Brus canton).

We initially monitored birds in 14 restoration sites that had been established on degraded cattle pastures and coffee plantations in 2004-2006. Sites were chosen based on availability; some sites were established and subsequently converted to alternative land uses, and these were unavailable for study. We excluded from this version of the data set three restoration sites where we surveyed birds but that did not persist to 2021 (BR: data available 2009-2012; GN: 2009-2015; RS: 2011-2012). We also excluded data collected in 2009 (the year surveys began; see Reid et al. 2014), since plots were only surveyed once in 2009 (December), rather than for three times as in each of the remaining years (2010-2021).

Each restoration site included three 50 × 50 m plots, which were cleared of vegetation and randomly assigned to one of three treatments. Natural regeneration plots were abandoned to regenerate naturally; islands (i.e., applied nucleation treatments) were planted with patches of trees including two each of 4 × 4, 8 × 8, and 12 × 12 m patches; and plantations were planted with rows of seedlings across the entire plot. These local restoration treatments spanned a gradient of intervention intensity; the number of planted seedlings per plot was zero for natural regeneration, 86 for applied nucleation, and 313 for plantations. Seedlings planted in applied nucleation and plantations were a mix of two native timber species, Terminalia amazonia (J.F. Gmel.) Exell (Combretaceae) and Vochysia guatemalensis Donn. Sm. (Vochysiaceae), and two naturalized legumes, Erythrina poeppigiana Walp. Skeels and Inga edulis Mart. (Fabaceae). All plots were cleared with machetes for 2.5 years to allow planted seedlings to grow above existing grasses and forbs. Treatments had been in place for 3-5 y when bird counts began at which time there were large differences in vegetation structure and composition between treatments. Detailed descriptions of the restoration treatments and tree species selection are provided elsewhere (e.g., Cole et al. 2010, Holl et al. 2020).

In 2013-2021, we also monitored 5 mature forests as a reference condition and 6 active cattle pastures as a degraded condition. Reference forest locations were generally located near a restored block (4 of 5 were paired), whereas pastures were generally not paired with a restoration block (2 of 6 paired) but were spread across the same landscape and elevational range. For reference forests (50 x 50 m plots) that match restoration site codes, bird survey locations do not always exactly match the location of vegetation plots, given that (a) reference forest vegetation survey plots were smaller than 50 x 50 m, and (b) a suitable trail was necessary for accessibility (i.e. a 50 m long trail from which to survey 25 m on either side). Note that some reference forest and pasture locations originally received unique site codes but were subsequently re-coded given their proximity to restoration sites: the MP (Metaponto) pasture site was recoded as HB; the LC (Las Cruces) reference forest site was recoded as JG. The RS reference forest is located near by a restoration block, but the RS restoration plots were not monitored through 2021.

These were chosen by Juan Abel and Leighton, and Juan Abel defined their boundaries as 50 × 50 m plots comparable to the Islas restoration plots. Most of the cattle pastures have at least one remnant tree, whereas the restoration plots were carefully located to avoid remnant trees.

Birds in restoration sites were surveyed three times per year, in April-May (mostly April), July-August, and November-December. Each plot was actively searched for 20 minutes per observation, and all birds seen or heard within the plot were recorded. Birds flying over or through the plot without using it were also recorded and marked as such so that they can be easily excluded from analysis if desired. Note that because a single row of data can include an abundance >1 as well as an "X" for flyover, there is some uncertainty about how many birds were observed flying over versus interacting with the plot. Nineteen plot surveys (1.3%) were missing; pasture and reference forest were missing eight and six surveys, respectively, whereas each restoration treatment was missing just one or two surveys. The 2021 Nov-Dec survey for one pasture site (LA) is missing because the land use changed from pasture to row crops in late 2021.

Data collector: Juan Abel Rosales

Dates of data collection: 2010-2021

Contacts for project: Francis H. Joyce (; and J. Leighton Reid (

Data entry

Software used to prepare data set: Microsoft Excel (16.83), R v3.0.1-4.2.2

Data processing that was performed: Bird species names were last updated in 2023 to match the American Ornithological Society's Check-list of North American Birds (Chesser et al. 2022). This version of the data set excludes species detections that were recorded as flyovers (birds flying over or through the plot without interacting with it).

People who entered data: J. Leighton Reid, Francis H. Joyce. J. Leighton Reid was responsible for data entry and proofing for surveys through 2018. Francis Joyce took over responsibility for the data set curation on 2019-06-07.

People who proofed data: J. Leighton Reid, Francis H. Joyce

People to contact with questions: F. H. Joyce (;

File: sites.csv


-site_code: two-letter code used in combination with

-site_name: full site name

-Type: type of location.

-Elevation: meters above sea level, to the nearest 10 m. For restoration sites, values are those reported in Table 1 from Holl et al. (2020). Elevations at pastures and reference forests were obtained from GPS waypoints collected in the field by Francis Joyce and from NASA's SRTM1 data set.

-Latitude: latitude in decimal degrees (CRS: WGS84)

-Longitude: longitude in decimal degrees (CRS: WGS84)

-year: year restoration plots were established. Value is NA (not applicable) for pasture and reference forest locations.

File: bird_surveys_2010_2021.csv


-season: factor denoting the time of year of sampling. apr_may = April-May; jul_aug = July-August; nov_dec = November-December

-year: year of sampling

-month: month of sampling

-day: day of sampling

-site: two-letter code denoting the study site. Does not necessarily uniquely identify each survey plot unless combined with the treatment variable. Restoration site codes follow other Islas Project data sets, as do 4 of 5 reference forest site codes. See sites.csv for more site details.

-observer: character string denoting the person who surveyed birds (Juan Abel Rosales)

-treatment: Factor (character string) denoting experimental treatment (habitat type): B = mature remnant forest (bosque); C = natural regeneration (control), I = applied nucleation (island), P = plantation, Z = active cattle pasture (zacate)

-start: time of day when the 20-min survey started. First digit is hours, second and third digits are minutes.

-species: species of bird detected (factor). Matched to the Sixty-third supplement to the AOS's Check-list of North American Birds (Chesser et al. 2022).

-abundance1_3min: number of individuals detected in the first three minutes of observation (0:00-3:00).

-abundance3_10min: number of individuals detected in minutes 3-10 (3:01-10:00)

-abundance10_20min: number of individuals detected in minutes 10-20 (10:01-20:00)

-abundance: total number of individuals detected in the 20-min survey (i.e., sum of abundance1_3, abundance_3_10, abundance10_20)

Missing code: NA


A note about abundances. We recorded bird abundances in three periods, minutes 1-3, 3-10, and 10-20. This was done to allow for estimating detection probabilities. Only new individuals were recorded in minutes 3-10 and 10-20 - i.e., these are not independent counts of the full community.


Chesser, R. T., Billerman, S. M., Burns, K. J., Cicero, C., Dunn, J. L., Hernández-Baños, B. E., Jimenez, R. A., Kratter, A. W., Mason, N. A., Rasmussen, P. C., Remsen, J. V., Jr., Stotz, D. F., & Winker, K. (2022). Sixty-third supplement to the American Ornithological Society's Check-list of North American Birds. Ornithology, 139(3), ukac020.

Cole, R.J., Holl, K.D., Zahawi, R.A., 2010. Seed rain under tree islands planted to restore degraded lands in a tropical agricultural landscape. Ecological Applications 20, 1255-1269.

Holl, K. D., Reid, J. L., Cole, R. J., Oviedo-Brenes, F., Rosales, J. A., & Zahawi, R. A. (2020). Applied nucleation facilitates tropical forest recovery: Lessons learned from a 15-year study. Journal of Applied Ecology, 57(12).

Reid, J. L., Mendenhall, C. D., Rosales, J. A., Zahawi, R. A., & Holl, K. D. (2014). Landscape context mediates avian habitat choice in tropical forest restoration. PLOS ONE, 9(3), e90573.


National Science Foundation, Award: 05-15577, DEB

National Science Foundation, Award: 09-18112, DEB

National Science Foundation, Award: 14-56520, DEB

National Science Foundation, Award: 20-16623, DEB