Bird Survey Around Mida Creek

2021-06-09

Raphaël Nussbaumer1,, Kirao Lennox1,, Colin Jackson 1,

1A Rocha Kenya, Watamu, Kenya

Corresponding author: Raphaël Nussbaumer (raphael.nussbaumer@arocha.org)

DOI:10.15468/jwnu2e

licensebuttons by-nc

Description

The initial goal of these surveys was to collect information about the location of Mangrove Kingfishers (Halcyon senegaloides) for a geolocator study. However, as surveys provided richer information, it was decided to publish this dataset on its own.

Each survey refers to a complete presence-only species list (called “event” in the dataset), each recorded over ~1km and during ~1hr. Four surveys were performed successively over a morning from 6am to 10am covering a transect. Each of the 23 transects was repeated once a month (grouping called “session”) from May 2020 to May 2021. The study area covers the habitat between Arabuko-Sokoke Forest and Mida Creek. Surveys were collected by bird guides from the area (Daniel Kazungu, Juma Badi, Kibwana Ali, Saddam Kailo and Mohammed Ali) and organized by Kirao Lennox, Raphaël Nussbaumer and Colin Jackson from A Rocha Kenya.

Summary of count statistics.

  • Number of transects : 23
  • Number of sessions: 11
  • Number of events (surveys): 873
  • Number of occurrences (sightings): 29195
  • Number of species: 217

Purpose:

  • Collect standardized data on common species between the forest habitat and the creek/coastal habitat.
  • Collect baseline data with a reproducible protocol in order to assess the change of avifauna over time.
  • Assess the impact of deforestation on the ‘shambas’ (fields/farms).

Type of data

This dataset is published as a Darwin Core Archive using a sampling event dataset type.

Data structure

Table 1: Events (surveys) table structure

Table 2: Occurrences (counts) table structure

Keywords

bird, ornithology, count, monitoring, survey, Kenya, Africa, East Africa, coast, Mida Creek, Arabuko-Sokoke Forest, habitat change, population, trend, Birdlasser, African Bird Atlas, Samplingevent

Geographic coverage

The surveys were performed between Arabuko-Sokoke Forest and Mida Creek, on the coast of Kenya. See for more information about the spatial extent of the dataset.

## NOTE: keeping only 23 wkbLineString of 23 features

Bounding box

The bounding box is -3.46669° to -3.271515° latitude and 39.891823° to 40.044845° longitude.

Taxonomic coverage

We record all birds seen (Aves sp.) at the species level (with the exception of the Hybrid Lovebird Agapornis fischeri x personatus). We use the taxonomy from the Checklist of the Birds of Kenya (5th Edition) (ISBN: 9966-761-37-3) for the scientificName and vernacularName. The taxonID and taxonRank are taken from the eBird/Clements Checklist of Birds of the World: v2019. See section Sampling Description for explanations on how the taxons were initially recorded in the field and later filtered and matched to these taxonomy lists.

Our dataset contains 217 unique species belonging to 64 families.

Taxonomic ranks

Kingdom: Animalia (animals)

Phylum: Chordata

Class: Aves (birds)

Temporal coverage

The temporal coverage is from 2020-05-19 to 2021-05-25.

Content providers

  • The surveys were designed by Raphaël Nussbaumer, Kirao Lennox and Colin Jackson
  • The logistics were organized by Kirao Lennox
  • The surveys were performed by Daniel Kazungu, Juma Badi, Kibwana Ali, Saddam Kailo and Mohammed Ali
  • The data was collected, registered, cleaned, formated for GBIF and uploaded by Raphaël Nussbaumer

Sampling Methods

Study extent

The Kenyan coast was once a large continuous forest called the Northern Zanzibar-Inhambane coastal forest mosaic, of which Arabuko-Sokoke Forest is the largest remainder of today. Extensively studied, this forest is an area of high endemism and was therefore gazetted as a National Park in the late 1980s. Since then, it has benefited from a hard protection from deforestation and human influence. However, the surrounding habitat has been strongly affected by human and agricultural extensions on what used to be forest land.

On the other side of our study area, Mida Creek is a 580 ha tidal marine multi-habitat ecosystem fringed with a diverse assemblage of mangrove species (Ceriops tagal, Rhizophora rnucronata, Bruguiera gyrnnorrhiza, Avicennia marina, Sonneratia alba and Xylocarpus benadirensis).

Surveys in this area allow us to study the distribution and abundance of common species not necessarily restricted to a specific habitat (such as forest or mangrove), as well as their association with certain habitats.

The transects were defined to cover all habitat types between Arabuko-Sokoke Forest, Mida Creek and the coast. They were designed such that the habitat remains relatively homogeneous during each 1hr survey (i.e. along the forest edge or mangrove edge). In addition, each survey starts and finishes near a main road to facilitate access and early start (6am). See for a map of the transects.

Sampling Description

  • Each survey is performed by 2 persons.
  • Each survey is designed to last 1 hour, covering a transect of 1km.
  • During a survey, each bird species seen and/or heard is recorded once upon the first encounter (i.e., no count data, just presence/absence)
  • The data is collected with the Birdlasser app, allowing to record the GPS location of the observer at the moment of the observation (i.e., not the bird), the exact time of observation (date and hour) as well as any additional comment.
  • Surveys can be paused if interrupted due to heavy rain or interaction with land owners, for instance. This information is recorded in the dynamicProperties and eventRemarks columns in events.
  • Surveys are performed in clusters of 4, starting at 6am up to 10am. Some remote transects (e.g. Kiperwe Island) required more time, thus surveys were performed until 12pm.
  • The Birdlasser data is then exported in CSV format, collected on A Rocha infrastructure and available at /data/birdlasser_files.

Quality control

Quality control is performed in the Rmarkdown script Processe2GBIF.rmd.

Protocol Ok Surveys were discarded if the protocol was not properly followed. This happened mainly at the beginning, before the guides were fully trained. We also deleted surveys when the rain lasted longer than 30 minutes.

Spatial error Phone GPS errors or sightings entered later at wrong locations were manually identified by plotting all sightings on a map. These locations were interpolated linearly between the surrounding sightings of the same survey.
The column georeferenceVerificationStatus indicates if such interpolation was performed and georeferenceRemarks provides the original coordinate.

Species list All species recorded were manually verified by Raphaël Nussbaumer, Colin Jackson and checked with the guides by Kirao Lennox. The spreadsheet species_lists_validation.xlsx was used to record the status (see table below) of each species. In general, each specie was either validated or flagged with an error code. Some errors could be corrected by providing the Equivalent specieID (e.g., taxonomic issue).

Status code Description
Unknown Unsure if it is an error or not.
Error–certain Erroneous species with unknown origin (typo, identification, taxonomy).
Error–typo Typing error on the phone. Corrected when possible with Equivalent specieID.
Error–taxonomy Issues with taxonomy (old/new name, split/merge, duplicate specieID etc…). ALWAYS enter Equivalent specieID.
Error–probable Possible species in the area but not sufficiently confident in the identification. Entry will be removed. See individual comments.
Valid–known Species known to occur here.
Valid–confirmed Uncommon/rare species confirmed by Mida guides. See individual comments.
Valid–likely Uncommon species but probable enough to keep the data.

The events data was cleaned accordingly: kept if valid, changed to the Equivalent specieID or deleted if erroneous. The status code and original specieID were recorded in events under the columns identificationVerificationStatus and identificationRemarks.

Step Description

All the processing steps from the original Birdlasser datafile to the GBIF formatted file are described in the Rmarkdown script Processe2GBIF.rmd.