You are here

Harvard Forest Data Archive


Detection Probability of Red Wood Ants in Friedenweiler, Germany 2015

Related Publications



  • Lead: Gabriele Berberich, Aaron Ellison
  • Investigators: Carsten Dormann, Dietrich Klimetzek, Martin Berberich, Nathan Sanders
  • Contact: Aaron Ellison
  • Start date: 2015
  • End date: 2015
  • Status: completed
  • Location: Friedenweiler, Germany
  • Latitude: +47.90
  • Longitude: +8.27
  • Elevation: 850 to 920 meter
  • Taxa: Formica rufa (red wood ant)
  • Release date: 2016
  • Revisions:
  • EML file: knb-lter-hfr.286.2
  • DOI: digital object identifier
  • Related links:
  • Study type: short-term measurement, modeling
  • Research topic: biodiversity studies; conservation and management; ecological informatics and modelling
  • LTER core area: populations
  • Keywords: abundance, ants, census, distribution, populations
  • Abstract:

    Estimation of population sizes and species ranges is central to population and conservation biology. It is widely appreciated that imperfect detection of mobile animals must be accounted for when estimating population size from presence-absence data. Sessile organisms also are imperfectly detected, but correction for detection probability in estimating their population sizes is rare. We illustrate challenges of detection probability and population estimation of sessile organisms using censuses of red wood ant (Formica rufa-group) nests as a case study. These ants, widespread in the northern hemisphere, can make large (up to 2m tall), highly visible nests. Using data from a two-day mapping campaign by eight individuals of 147 ant nests spread across sixteen 3600-m2 plots in the Black Forest region of southwest Germany, we developed a Bayesian model for quantifying detection probability of sessile organisms. Detection probabilities by individual observers of red wood ant nests ranged from 0.31 – 0.56, and depended on experience of the observers, size and density of nests, and habitat characteristics. Robust estimation of population density of sessile organisms—even highly apparent ones such as red wood ant nests—requires unbiased estimation of detection probability, just as it does when estimating population density of rare or cryptic species.

  • Methods:


    Field work was done during April 2015 in sixteen 60 × 60-m plots near Friedenweiler (47.90 °N, 8.27 °E, 850 – 920 m. a.s.l.) in the Black Forest region of southwest Germany. Eight observers (two experienced ones and six inexperienced ones) independently mapped RWA nests for one hour in each of the 16 plots. The inexperienced observers were trained beforehand to recognize RWA nests in the field and to map them by holding a GPS receiver (Garmin 60CSx/62S/64S; 10-m precision) directly above a RWA nest and registering its location. Each observer also took a photograph of every mapped nest to facilitate its subsequent identification and to avoid double-counting when nearby nests were within the precision of the GPS. Each GPS receiver was pre-loaded with 1:50.000 topographic maps onto which the boundaries of all 16 study plots had been transferred so that plot boundaries could be observed and maintained during each census. All cameras and GPS receivers were synchronized to local time and projection (WGS84 projection; Datum: Potsdam). Each observer mapped the plots in a specific sequence; the track of each observer in each plot was recorded continuously to quantify speed, total distance covered, and individual search strategy. Finally, to minimize errors in delimiting plot boundaries in the field, a buffer region of 10 m around each plot was included during field recording to allow for GPS imprecision. All GPS data were downloaded immediately after collection and transferred into a GIS database. Forest stand types were classified in the field, and nest heights and diameters were classified from nest photographs.

    Estimating and correcting for false positives

    Sampling may lead to two kinds of errors: false positives and false negatives. False positives for each observer i sampling in plot s were tabulated manually from the number of reported nests (Nreported [i,s]); we refer to the number of observed real nests as Nobs [i,s] (Nobs[.] ≤ Nreported[.]), and Nobs was determined by cross-matching all mapped entities identified as RWA nests with their GPS coordinates, photographs, and recorded census tracks. The number of false positives (per observer and site) = Nreported[i,s] – Nobs[i,s]. After eliminating false positives, we linked GPS coordinate positions for each actual RWA nest recorded by each observer and averaged them to obtain a unique GPS position for each nest, which was then assigned a unique identifier. In the final dataset, real RWA nests and non-RWA nests were flagged as such.

    Estimating the number of nests in each plot and accounting for detection probability

    We estimated the total number of nests in each plot, N*s, and the total number of nests among the 16 plots, N*, using standard bias-corrected species richness estimators (Chao’s S, jackknife 1 [Jack1] S, and Jack2 S) implemented in the specpool function of the vegan library in R, version 3.2. These estimators are based on the observed number of nests that were detected by only one (“singletons”) or two (“doubletons”) observers. However, none of these three estimates are corrected explicitly for detection probability. We thus compared these three standard estimates of total number of nests with maximum likelihood and Bayesian estimates that explicitly included estimates of detection probabilities by individual observers (Pi,s).

    We used maximum likelihood methods to estimate the observed number of nests in each plot N*s and their detection probability P*i,s as Nobs[i,s] = Pi,s × Ns, where Ns is the “true” number of nests occurring in each plot. These calculations were facilitated by noting that Nobs[i,s] ~ Bin(Pi,s, Ns). These calculations yielded 1 - P*i,s as an estimate of the probability of false negatives for each observer i in plot s, and Bs = N*s - Nobs[i,s] as an estimate of the number of missed nests in each plot.

    We also already had some lower bound for the estimate of missed nests based on the attribution of records to unique nests. Therefore, we could use Nobs[s] as a prior for the minimum number of nests (min(Ns)) at each site in a Bayesian model (See R code in the Supplement). The Bayesian model followed the same logic as above, but was drawn from a Poisson distribution with parameter BsNobs. We truncated the Poisson distribution so that its minimum value was the minimum number of observed nests at each site (min[i](Nobs[s,i])).

    Covariates of detection probability

    The Bayesian approach also allowed us to incorporate four covariates into the analysis. Here, we used a quasi-binomial generalized linear model to test whether the height-class (1-10, 11-50, 51-100, and >100 cm) or diameter-class (1-50, 51-100, 101-150 and >150 cm) of each ant nest (classified from nest photographs); the forest type (dominated by spruces [Picea], pines [Pinus] or beech [Fagus]) in which it occurred (classified in the field); or its location (within the forest, along forest roads, or along forest edges, as classified in the field and from GIS layers) affected the number of nests detected by each observer. Because the number of small nests (Nobs = 70 and 87, for nests in the smallest height and diameter classes, respectively) far exceeded the number of larger nests (Nobs = 24 nests 51-100 cm in height and 1 nest > 100-cm high), and Nobs = 19 nests 101-150 cm in diameter and 3 nests > 150 cm in diameter), we pooled the largest size classes when regressing detection probability on nest size.

    Determining the number of observers needed to accurately estimate the number of nests

    The analyses described above assumed that detection probability and the probability of false positives were independent of each observer. However, our data suggested that many nests were recorded by all observers whereas others were found only by some. In other words, we could not assume independence of observations: adding more observers to the team led to records largely similar to what had already been reported. We computed the amount of effort required to accurately estimate numbers of nests assuming a constant detection probability among observers and serial correlation among observers.

  • Use:

    This dataset is released to the public under Creative Commons license CC BY (Attribution). Please keep the designated contact person informed of any plans to use the dataset. Consultation or collaboration with the original investigators is strongly encouraged. Publications and data products that make use of the dataset must include proper acknowledgement.

  • Citation:

    Berberich G, Ellison A. 2016. Detection Probability of Red Wood Ants in Friedenweiler, Germany 2015. Harvard Forest Data Archive: HF286.

Detailed Metadata

hf286-01: number of nests

  1. plot: plot number
  2. o1: number of nests observed by observer 1 (unit: number / missing value: NA)
  3. o2: number of nests observed by observer 2 (unit: number / missing value: NA)
  4. o3: number of nests observed by observer 3 (unit: number / missing value: NA)
  5. o4: number of nests observed by observer 4 (unit: number / missing value: NA)
  6. o5: number of nests observed by observer 5 (unit: number / missing value: NA)
  7. o6: number of nests observed by observer 6 (unit: number / missing value: NA)
  8. o7: number of nests observed by observer 7 (unit: number / missing value: NA)
  9. o8: number of nests observed by observer 8 (unit: number / missing value: NA)
  10. n.min: minimum number of nests per plot = total number of uniquely observed nests in plot (unit: number / missing value: NA)

hf286-02: nest records

  1. unique identifier for each observed nest; matches variable in hf286-03-nest-sizes.csv
  2. o1: whether or not observed by observer 1
    • 1: observed
    • 0: unobserved
  3. o2: whether or not observed by observer 2
    • 1: observed
    • 0: unobserved
  4. o3: whether or not observed by observer 3
    • 1: observed
    • 0: unobserved
  5. o4: whether or not observed by observer 4
    • 1: observed
    • 0: unobserved
  6. o5: whether or not observed by observer 5
    • 1: observed
    • 0: unobserved
  7. o6: whether or not observed by observer 6
    • 1: observed
    • 0: unobserved
  8. o7: whether or not observed by observer 7
    • 1: observed
    • 0: unobserved
  9. o8: whether or not observed by observer 8
    • 1: observed
    • 0: unobserved

hf286-03: nest sizes

  1. unique identifier for each observed nest, matches variable in hf286-02-nest-records.csv
  2. x: x-coordinate of nest location, UTM (unit: meter / missing value: NA)
  3. y: y-coordinate of nest location, UTM (unit: meter / missing value: NA)
  4. height: height category (cm) of nest (unit: centimeter / missing value: NA)
  5. diameter: diameter category (cm) of nest (unit: centimeter / missing value: NA)
  6. forest: forest type
  7. vegetation at nest
  8. location: location

hf286-04: R markdown file and output for all analyses

  • Compression: zip
  • Format: zip
  • Type: zip file