Friday, 24 January 2014

Tumbes, plantcutters and people

Posted by Christian Devenish

I’ve just finished four months of intensive ecological fieldwork in the dry forests of Northern Peru. This is the major fieldwork for my PhD, which aims to develop tools, combining modelling software, field data and GIS, to provide improved estimates of extinction risk in threatened Neotropical birds. As part of this, I have surveyed key birds at a range of sites and I will use these data to assess abundance and population sizes, and to gauge reactions to habitat change. Most importantly, I will investigate the relationship between probability of occurrence for Neotropical birds derived from presence only niche modelling methods (MAXENT) and bird abundance (densities derived using DISTANCE software). If there is a workable relationship then the former can be used as a surrogate for the latter in cases where empirical abundance data are missing.

Photo: Christian Devenish (more at

My study area is part of the Tumbesian Region, which spans the coasts of Ecuador and northern Peru. It is part of an Endemic Bird Area, a biodiversity Hotspot, and covers critical habitat included in WWF’s global 200 ecoregions. The area is important for conservation due to the high number of endemics, but also due to the threats faced by the dry forest habitat and its inhabitants - not just birds. Peru has a burgeoning agricultural export industry, based mainly on the arid coastal plains, and fed with water from the Andes. This has placed pressure not only on the dry forest and scrub environments, but also on the traditional goatherding livelihoods in the region. Many communities are facing difficult choices as to whether to sell community land for large-scale agriculture, or continue with the harsh farming lifestyles practiced over the last few hundred years.

Male Peruvian Plantcutter looks like he has his seat belt on for a rough ride (Photo: Jorge Montejo)
Along with Elio Nuñez, an undergraduate student at the Universidad Nacional de Piura, I have been surveying endemic dry forest birds at 26 sites along Peru´s north coast, from the province of Tumbes, south through Piura and Lambayeque, where the majority of my field sites are concentrated, south to Ancash. Between June and September 2013, we walked 10 km of transects at each site, once for birds, and once to measure basic habitat variables. Birds under study include the Endangered Peruvian Plantcutter Phytotoma raimondii and Rufous Flycatcher Myiarchus semirufus, plus 18 other species, seven of which are restricted to Peru.

Sechuran Fox Lycalopex sechurae. Photo: Christian Devenish

I had two basic requirements in my study design, I wanted to get a spread of sites across the ranges of my species (all Tumbesian endemics with relatively restricted ranges), but also ensure that I would have a fair chance of obtaining data on them, especially the rarer ones. Given the large size of my whole study area, a random assignment of sites would not have been feasible. Therefore, I made distribution models using MAXENT for four species (two rare and two common sharing very similar ranges), summed the results and sampled sites from quantiles above a presence/absence threshold, with a heavier weighting for the top quantile (Figure 1). To ensure good geographic coverage, I created a 25 km grid over the study area, and repeated the sampling until at least one point fell within each of 10 cells with the highest average model scores (below in in blue).

Preliminary results are interesting. For example, there is a distinct distribution pattern for both the plantcutter and the Rufous Flycatcher, echoing two of Rabinowitz’s seven forms of rarity, on the one hand the plantcutter can be very abundant locally, but is very patchily distributed (it was only recorded at half the study sites, but reached encounter rates up to a staggering 12 individuals per km). On the other hand, the Rufous Flycatcher was widely distributed, but at very low abundances (recorded at all but three sites, but only reaching an encounter rate higher than 1 ind/km at one site).

A very preliminary look at encounter rates against predicted probability of occurrence (above) from the MAXENT models  shows a weak, but significant relationship for two of the more common species (Grey and White Tyrannulet Pseudelaenia leucospodia and Cinereous Finch Piezorhina cinerea), but not for both the Peruvian Plantcutter and Rufous Flycatcher. It may well be that there are other factors behind the rarity of these two species that determine their distribution, such as much more specific habitat requirements than are currently represented in the model?  With habitat and land use data from each site, I hope to answer questions like these in my PhD.

Photo: Christian Devenish
Christian’s PhD is being supervised by Stu, along with Graham Smith at MMU, and Graeme Buchanan of RSPB. The Conservation and Research Fund, The Birdfair/RSPB Research Fund for Endangered Birds and IdeaWild provided funding for the field project and equipment, and CONDESAN kindly allowed a flexible working arrangement during the field work.

1 comment:

  1. Why not a geralized linearl model analizing by confident intervals instead p value an R2?