COBRA - Conservation Oriented Biodiversity Rapid Assessment
(username and password are "ennor" for access to papers in pdf format)

The COBRA is a protocol that intends to be a future reference for biodiversity assessment of arthropods (download pdf).
In this first stage it was developed for Iberian spiders and was already applied in a number of sites of Portugal (Table 1).
We are now starting to expande its use to the Macaronesian Islands.


Table 1. Comparison of all semi-quantitative sampling studies in Portugal that followed a similar strategy and that were used to delineate the COBRA protocol. Each sample corresponds to one person-hour of fieldwork. Sampling intensity is the number of individuals divided by the number of species. Inventory completeness is the proportion between observed and estimated richness.

SiteGerêsArrábidaGuadiana
Samples320320192
Individuals751655482675
Species richness185150110
Sampling intensity413724
Singletons21%17%29%
Estimated richness (Chao 1)  213162155
Inventory completeness87%92%71%



A COBRA COOKBOOK

At each site to be studied, a sampling plot with one hectare should be delimited. All the sampling methods will be made inside this plot and collectors are allowed to freely roam inside it. This can be a square (100x100m) area or an adapted shape for linear habitats (e.g. river margins).

A semi-quantitative sampling design, with a sampling unit defined as one person-hour of effective fieldwork, is to be followed. For the purpose, collectors should use a stopwatch to control the time that is spent collecting, and to be able to stop the counter when occupied with other activities (e.g. photography, maintenance of tools, eating, etc.; one hour of collecting can take up to 10 or 20 minutes more).

Previous work (see Cardoso et al., 2008a, 2008b, 2008c) has demonstrated that each combination of method and time of the day can be considered as a different method itself. For example, sweeping low vegetation during the day samples a different fraction of the spider community than sweeping that same vegetation during the night. Therefore, these should be regarded as different methods.

The proposed methods were chosen due to their proved efficiency to sample spiders, and also because they target different fractions of the sampled community, even if with some overlap in a few cases:
  • Aerial – this method consists in collecting all spiders found above knee-level by hand, forceps, pooter or brush and immediately transferring them into alcohol. All the time spent searching is to be accounted for.
  • Ground – Similar to the aerial method, but directed towards spiders seen below knee-level, including species in hidden sites such as below stones or inside hollow trunks.
  • Beating – A one by one meter square sheet with a frame should be used as a drop-cloth and a wooden pole used to beat tree branches, as high as possible. The effective time includes all the time spent in the activity, like beating, searching for fallen spiders on the sheet and transferring them to alcohol.
  • Sweeping – A round sweep net with an opening diameter of 40 cm is used to sweep bushes and tall herbs. All time spent sweeping or searching for dislodged spiders is accounted for.
  • Pitfall – Pitfall traps are placed next to the delimited plot (not inside to avoid interference with collectors). Each trap is placed 5 meters apart from the nearest traps. A sample is a group of 4 contiguous pooled traps. The clumping of traps makes individual sampling effort reasonably comparable with time-based samples, as the effort applied to rig and collect four traps is calculated to be roughly equivalent to one person-hour of work. Traps should be left in the field for two weeks. Preference should be given to standard 33 cl plastic cups, 8 cm wide at the top and 12 cm high. Two-thirds of each cup is filled with a preservative liquid containing 50% of ethylene glycol and a drop of detergent to break surface tension, covered with a square wooden plate placed about 2 cm above the ground.
Other factors like collector experience have been tested and only occasionally found to influence the results. It is therefore unnecessary to strictly control the experience of collectors, although I recommend that at least one of the collectors is experienced in this type of sampling and proficient with all the methods, so that consistency between different teams is maximized.

With these methods, and using an iterative optimization procedure explained elsewhere (Cardoso, subm.), three nested (sub-)protocols were reached (Table 2). The low-effort protocol is intended to capture around 50% of the existing species, regardless of the habitat type, and is part of the medium-effort protocol, which requires four times more effort to capture 70% of the richness. Concurrently, this medium-effort protocol is part of the high-effort protocol. This way, the flexibility is guaranteed, as different sampling teams may have different objectives and resources, but at the same time all sites are comparable, even if sampled with different effort.


Table 2. Proposed nested (sub-)protocols. Ad/An – Aerial searching day/night; Bd/Bn – Beating day/night; Gd/Gn – Ground searching day/night; Sd/Sn – Sweep day/night; Pf – Pitfall trapping (each sample being comprised by four pitfall traps as explained in text). Numbers of samples in parenthesis refer to alternative protocols when sites do not have arboreal cover. %S - expected % of species captured.

Samples%SAdAnBdBnGdGnSdSnPf
2450 0 4 (0) 2 (0)2 (0) 0 0 (4) 2 (4) 2 (4) 12
96 70 0 16 (0) 8 (0) 8 (0) 0 0 (16) 8 (16) 8 (16) 48
320 90 32 (0) 32 (0) 32 (0) 32 (0) 32 (64) 32 (64) 32 (64) 32 (64) 64



It has been suggested that it is more efficient to concentrate most effort during a single sampling season than to spread the effort during an entire year (Cardoso et al., 2007). Therefore, if a single season is to be chosen for sampling, it should be during May and June, when richness is highest (Table 3; Cardoso et al., 2007). If the objectives require that the sampling is done along the year, I propose a “96 + 24 + 24” or a “320 + 96 + 96” strategy, with the lower effort sampling seasons being done during January/February and September/October (Cardoso, subm.).


Table 3. Adequacy of periods of the year to sample according to geographical area in Portugal and habitat (tree cover density) characteristics (see Cardoso et al., 2007). Notation as follows: (-) Avoid; (o) Good; (+) Optimum.

 AprilMayJuneJuly
 1st half 2nd half 1st half 2nd half 1st half 2nd half 1st half 2nd half
North Dense - - o + + + o -
Sparse - - - o + o - -
Open - - - o + o - -
Centre Dense - o + + + + o -
Sparse - - o + + o - -
Open - - o + + o - -
South Dense - o + + + o - -
Sparse - - o + o - - -
Open - - o + o - - -



The protocol now proposed only optimizes the effort spent in the field. This is probably the most critical stage, given the remote location of many sampling sites and the logistics involved. Moreover, fieldwork is the only part of any project that is often impossible to repeat. In contrast, it is possible to (re-)identify some species if collections are available or to rerun statistic analyses afterwards. However, several ways for optimizing the identification process have been proposed for Iberian spiders. Among these are the identification of individuals only to genus level which will act as higher taxa surrogates (Cardoso et al., 2004a) or the identification of gnaphosids and theridiids which were proved to be good indicator taxa of the overall spider diversity (Cardoso et al., 2004b). With these options, the total number of species may be estimated with reasonable confidence and results remain comparable between areas, even if at different taxonomic resolution.