Analysis of Rogadinae Species
(Species Accumulation Curves)

Question A: How long must a trap be kept in one place before the site is sufficiently sampled?

    To determine the answer to this question, we analyzed the data for a number of traps for a 6-month period, with samples taken every two weeks.  To obtain a data set of a reasonable size, we only used the species of the subfamily Rogadinae (Hymenoptera: Braconidae) in our analysis.  We used the program EstimateS (Colwell, 1997) to generate species accumulation curves and species richness estimators.

    EstimateS uses a randomizing algorithm to obtain the approximate average rate at which species are collected.  This is represented in the graphs by the orange boxes, and illustrates how many species have been collected from each trap (y axis) for each two-week collection period (x axis).  A curve which shows little change near the end of the sampling period (horizontal or nearly horizontal line) shows that few or no new species are being collected - in other words, the site has been sufficiently sampled.  A curve which continues to raise sharply near the end shows that many new species are still being found, and the sampling is nowhere near finished.  Most curves will fall somewhere between these two extremes.  EstimateS also generates estimates of the total number of species using algorithms based on abundance (ACE, Chao and Lee 1992; Chao, Ma, and Yang 1993) and incidence (ICE, Lee and Chao 1994).  Using the shape of the curve and the calculated estimates, one can make an educated guess on how long to sample particular sites.

Tayrona - Canaveral

Tayrona - Pueblito

Tayrona - Zaino


    All of the sites in PNN Tayrona show that they have been fairly well sampled, but they still require more time before they have been sufficiently sampled.  Canaveral is the only curve which flattens out at the end suggesting that there are only a few more species to be found.  The estimates for Canaveral suggest that there are probably about 2 more species to be found at this site, while the Zaino data suggest that there are probably about 1.5 times as many species as have been currently found.  The Pueblito data indicate that less than half of the species at the site have been collected.  Note that a site refers to the exact location of a Malaise trap.  Moving the trap 100 meters may have a drastic effect on the species composition trapped.

La Planada - Parcela Permanente

Amacayacu - San Martin

    The data from La Planada, although not extremely diverse, show that there are probably a few more species yet to be found.  The most impressive results come from PNN Amacayacu.  The trap at the San Martin site averaged two new species of this single subfamily every time the trap was collected!  Although the curve does not look to be quickly increasing, in reality it is quite sharp, and is made to look flat by the magnitude of the estimates.  The incidence and abundance of the species found here suggest that we have only begun to sample this site.  

    Answer A: For most sites, 9-15 months should be adequate, however for rainforest sites it will probably take much longer.

Question B: How many times must we move traps within a park before the park as a whole is sufficiently sampled?

    To answer this question, species accumulation curves were generated for Tayrona and Amacayacu using all of the trap data from all sites.  We then compared these curves with those of the individual traps to understand the rate difference between the time it takes an individual trap to complete sampling and for all traps in the park to be complete sampling.

Tayrona - All traps

    Both the shape of the curve and the estimates are about an average of each of the individual traps, suggesting that there is little similarity between the traps.  However we can better answer this question by counting the shared species between each trap.

Traps compared (# of species)

Shared species
Canaveral (11) - Pueblito (11)


Canaveral (11) - Zaino (14) 5
Pueblito (11) - Zaino (14) 6

     Here we see that there is indeed a good bit of similarity between the traps.  However, each trap still has 5-8 species not found in any other trap.  This tells us that we must continue to move the traps until the number of unique species for each new trap is greatly decreased so that then trap samples more becomes redundant.  It must be noted however that this data is incomplete since the curves above indicate that there are still more species to be found from all three of these sites.  This question could be more accurately answered with more exhaustive sampling.

Amacayacu - All traps

    The data for Amacayacu closely resembles the graph for a single trap in the park, with the numbers slightly increased.  However, the expected total species has only increased by about 10, whereas the total species found has increased by over 20, showing that the original estimate for total species was fairly accurate.  Comparing individual traps within the park would be pointless, as the park technicians (aware of the vast diversity of the area) were frequently changing sites and setting up many traps, making comparison difficult.

    Answer B: Unfortunately we cannot determine any specific answer until we have traps that are sufficiently sampled, but we know that three traps is grossly inadequate.

    Question C: How many individual parks must be analyzed before Colombia is adequately sampled as a whole?

    This question may prove to be the most difficult of all to answer.  Colombia is remarkable for its biological diversity.  Out of the six parks analyzed for this study of Rogadinae, the diversity in the species speaks for itself.

Of the 87 species identified from these parks, only 2 were found in multiple parks.  This illustrates just how isolated each habitat is. With the exception of human-related species such as the house fly, the fauna of each region is very nearly unique.

    Answer: It appears as though all of Colombia's 48 national parks will have to be sampled and we still will have sampled only a fraction of Colombia's fauna.

References Cited

Chao, A. & S.-M Lee. 1992 Estimating the number of classes via sample coverage. Journal of the American Statistical Association 87, 210-217.

Chao, A., M.-C. Ma, and M. C. K. Yang. 1993. Stopping rules and estimation for recapture debugging with unequal failure rates. Biometrika 80, 193-201.

Colwell, R. K. 1997. EstimateS: Statistical estimation of species richness and shared species from samples. Version 5. User's Guide and application published at:

Lee, S.-M., and A. Chao. 1994. Estimating population size via sample coverage for closed capture-recapture models. Biometrics 50, 88-97.

Last updated January 15, 2001.
All graphs generated by EstimateS, except for shared species between parks, generated by Microsoft Excel.