Surveying birds in tropical forests: comparing two methods in two widely-separated ecosystems

Flycatcher Indonesia

Male Rufous-throated Flycatcher (Ficedula rufigula), one of only a few species detected by mist-nets in Lambusango Forest Reserve, Buton, Indonesia

A comparison of two different survey techniques suggests that researchers should conduct preliminary studies before choosing the most appropriate survey method for that taxa or ecosystem, particularly when it is under-explored.

Effective survey techniques are essential for developing conservation strategies in all taxa. If it is not known which species inhabit an ecosystem, it is difficult to manage it effectively. This is very true for ornithologists. Birds are a well-known group, and as they are comparatively straightforward to sample and often act as reasonable bio-indicators for patterns in other taxa, information about their community composition and how this changes over time is often highly valuable.

However, while a range of methods exist for surveying tropical forest birds, little research has been conducted into how the effectiveness of these methods varies globally across different ecosystems. This is important, as forest habitats and their bird communities vary greatly throughout the tropics, and many of these habitats remain poorly explored.

There is a tendency in understudied areas to assume that tried-and-tested approaches will be as effective as they have been proven elsewhere, but this assumption can lead to inadequate sampling. Further research is therefore needed to determine the most appropriate methods for describing bird communities in different forest habitats.

Two survey techniques

This study examined how two of the most commonly used ornithological survey techniques – mist-netting and point counts – work in two poorly-explored tropical forests. Mist-netting involves sampling birds by capturing individuals in fine mesh nets. Point counts involve completing a series of timed counts at points spaced regularly along a set route, and recording all species seen and heard during these counts.

Local forest structure and bird community composition can have a strong influence on the effectiveness of common survey methods

A meta-analysis of previous research in lowland continental tropical forests suggests that point counts are generally the most effective method for describing bird communities, usually detecting between 70-80% of all species present. Mist-nets are typically capable of detecting between 20-30% of species present – fewer than point counts, although nets usually detect cryptic species missed by counts.

We investigated how well these two methods worked in cloud forest habitats in Cusuco National Park (CNP), Honduras, and insular lowland seasonal forest in the Lambusango Reserve (LR), Indonesia. As these two ecosystems possess unique habitat structures and bird community compositions, we hypothesise that the two target methodologies may produce results which differ from those of the meta-analysis results reported from lowland forest.

CNP’s dense cloud forest, for example, may limit visibility and so reduce the effectiveness of point-counts, although the lower-than-average canopy level may mean a greater percentage of birds occur in capture range for mist-nets, boosting their effectiveness. Bird communities in LR, on the other hand, are dominated by canopy-level species which may be untrappable for ground-level nets, although the low species richness and simple vocalisations of birds here may lead to an increased effectiveness of point counts.

Direct comparison

We ran 8-week bird surveys using both methods in both locations, and calculated the percentage of total bird species each method detected compared to the total number of birds known to exist in each forest. CNP fieldwork was completed in 2007. This survey was designed as a formal comparison, with point-counts and mist-netting conducted at the same points to ensure a direct comparison. The survey-effort totalled 63 person-hours of point-counting and 468 mist-net hours.

LR fieldwork was completed in 2008. Only point counts were completed as part of a formal survey; the mist-netting was conducted on an opportunistic basis with the nets being placed in sites predicted to produce the highest number of captures. Survey effort totalled 56 point count hours and 105 mist-net hours.

falcon, Honduras

A Barred Forest Falcon (Micrastur ruficollis) mist-netted in Cusuco National Park, Honduras

Results from CNP show patterns similar to those found in other studies: point counts detected around 60% of all known bird species in the park, slightly lower than average, while mist-netting detected 38%, slightly higher than average. This could result from the unusual vegetation structure and density of cloud forest ecosystems as previously hypothesised. Both methods proved effective, however, and complemented each other well, with mist nets detecting many species missed by the counts.

Results from LR were more polarised. Point counts proved very effective, detecting 72% of all known species. Mist-netting, however, was extremely poor, detecting just 5% of the known bird community. While netting survey effort was less than that employed in CNP, the low capture success rate still demonstrated the ineffectiveness of employing ground-level mist nets in forest ecosystems where population density is generally low and most species occur at mid-level or canopy level forest strata.

This research indicates that local forest structure and bird community composition can have a strong influence on the effectiveness of common survey methods. These results will be refined following a more detailed methodological study in LR in the near future.

However, preliminary results suggest that biologists undertaking surveys in relatively unknown ecosystems could benefit from conducting a series of preliminary methodological studies before implementing large-scale survey work, in order to ensure they are using the most efficient combination of methods available to them. This in likelihood applies to other taxonomical groups as well as birds.

Written by

Twitter Digg Delicious Stumbleupon Technorati Facebook Email

2 Responses to “Surveying birds in tropical forests: comparing two methods in two widely-separated ecosystems”

  1. Hey Tom – this is extremely helpful! My lab has worked different techniques like this in relatively un-sampled habitats. This article gives a nice foundation for thinking about what they have and approaching the next situation.

    I have been thinking about playback of predator mobbing events as another way to sample some cryptic species in high biodiversity areas – any thoughts?

    Katie Sieving

  2. Hi Katie – glad the article looks interesting! The results are still a little preliminary at the moment, but i hope we can expand the study into a full article after one more season of formal mist-net surveys in Indonesia.

    I think playback of predator mobbing could work very well for surveying certain cryptic species, but i think the results may vary again depending on the forest ecosystem you are surveying. Cryptic species assemblages in some of the tropical forest ecosystems Opwall has been working in have quite high ratios of small-bodied, furtive, undergrowth species that usually occur at low densities singularly or in pairs, and I would guess these might not respond as well to mobbing playback as other species groups. It might work better in temperate forests compared to tropical forests, perhaps. Still, even if it doesnt work for every single cryptic species, it could still work well for a good proportion, and any method that can boost the detectability of any cryptic species at all would definatly be extremely useful!

Leave a Reply