Skip to main content

Avian Response to Hurricane Maria in Coffee Plantations

Cite this dataset

Collazo, Jaime; Irizarry, Amarilys; Perfecto, Ivette; Vandermeer, John (2021). Avian Response to Hurricane Maria in Coffee Plantations [Dataset]. Dryad.


Insights on impacts and resiliency of avian species with respect to hurricanes in the Caribbean have largely focused on responses measured in protected habitats.  We assessed avian responses in non-protected habitat, specifically shade-restored coffee plantations, because their structural complexity retains many attributes of secondary forests, and may contribute to landscape scale species resiliency.  We tallied species richness and estimated occupancy probability of 12 resident avian species, after adjusting for imperfect detection, to assess the impact of hurricane Maria (20 September 2017) in shade-restored coffee plantations.  For five of those species, we also estimated local colonization and extinction probabilities to assess their prospect of rebounding (resiliency) in the context of two stages of shade restoration.  We used survey data collected March-June 2015-2017 (pre-hurricane) and 2018 (post-hurricane) in 58 coffee farms and satellite imagery to assess vegetation structure.  Restored farms were grouped into two categories based on time-since-restoration: newly-restored and fully-restored. We predicted that mean percent forest cover in fully-restored farms (~30-40%) would revert to levels in newly-restored farms (<15%), with concomitant changes in occupancy by avian species.  As predicted, mean percent forest cover (16.17 ± 4.27%) in fully-restored farms post-hurricane reverted to pre-hurricane levels in newly-restored farms (15.00 ± 5.61%).  The loss represented 30-38% relative to the pre-hurricane cover levels.  Detections of focal species dropped an average of 41% post-hurricane, with associated reductions in occupancy for 9/11 species.  Occupancy of the Puerto Rican Bullfinch and Puerto Rican Spindalis reverted to levels detected in newly-restored plantations prior to the hurricane as predicted.  Prospects of rebounding were more likely for species with invariant or increases in colonization probability (e.g., Yellow-faced Grassquit, Northern Mockingbird, Puerto Rican Spindalis).  Rebounding for frugivores like the Puerto Rican Bullfinch would be protracted given that colonization rates dropped from 0.56 ± 0.12 (pre-hurricane) to 0.04 ± 0.2 (post-hurricane), regardless of restoration stage.  Our work showed that the avian community associated with restored coffee farms exhibited a high degree of ecological resistance as the similarity in species composition before and after the hurricane was 81%, and all 12 focal species continued to occupy farms under both restoration stages.  The prospect of the focal species to rebound (resiliency) was species-specific, and in some cases, mediated by their affinity to a particular farm restoration stage.  The strength of hurricanes is projected to intensify with global warming.  Pockets of undamaged or partially damaged shade-grown or fully-restored coffee plantations may contribute to species resiliency by increasing landscape level habitat redundancy, and facilitate habitat shifts to secure food resources or harbor source populations to colonize recovering, hurricane-damaged habitat tracts.


Data were collected by means of surveys at each farm (58 total)  three consecutive years (2015-17) prior to the hurricane and one year after the hurricane (2018).  Of the 58 farms, 22 were newly-restored (0-4 years since shade layer restoration) and 36 were fully-restored (≥6 years since shade layer restoration) farms.  We conducted three (3) surveys at each farm between March and June each year.  We used fixed-radius point counts in which two observers recorded all bird species detected visually or by sound, within 50 meters of the center of the station for 10 minutes.  All surveys were conducted between 0600 and 1000 hrs under favorable weather conditions.  We converted survey data into a binary encounter history (1=species detected, 0=species not detected).  We used these data to tally the number of unique species before and after the hurricane, and calculated the Jaccard Similarity Index as the number of species detected in both periods/the number of species in either period* 100.  We also calculated the cumulative number of detections per focal species (12 of them) for the years with equal number of surveys per year (i.e., constant effort; 174 surveys/year). Lastly, we conducted two types of occupancy analyses.  The first type was a single-season, single-species modeling approach to contrast occupancy estimates between 2017 (pre-hurricane) and 2018 (post-hurricane).  The second type of analysis was a single species-multi season occupancy model. 

Usage notes

There were three sampling occasions (surveys) per year (season) which are denoted as "v1, v2, v3", which is followed by the year the surveys was conducted.  Missing values are indicated by "-" or ".", depending on the program used to open files.   Lastly NR - newly restored, and FR = fully restored.  "Open" means percent area with no forest cover, and "forest" means percent area with forest cover.  Focal species codes are: ANEU= Antillean Euphonia, SNPI= Scaly-naped pigeon, PUEB= Puerto Rican Bullfinch, YFGQ=Yellow-faced grassquit, PRFL= Puerto Rican Flycatcher, PRTA= Puerto Rican Tanager, LAPE= Lesser Antillean Pewee, PRVI= Puerto Rican Vireo, NOMO= Northern Mockingbird, LOKI= Loggerhead Kingbird, PRSP= Puerto Rican Spindalis, and GRAK= Gray Kingbird.


Puerto Rico Department of Natural Resources, Award: W-35 Federal Aid Program

United States Fish and Wildlife Service, Award: F14AC01125

National Institute of Food and Agriculture, Award: 2017-67019-26292

National Institute of Food and Agriculture, Award: 2018-67030-28239

Puerto Rico Department of Natural Resources, Award: W-35 Federal Aid Program