Three Dimensional Tracking of Exploratory Behavior of Barnacle Cyprids Using Stereoscopy
© The Author(s) 2012
Received: 4 June 2012
Accepted: 30 July 2012
Published: 21 August 2012
Surface exploration is a key step in the colonization of surfaces by sessile marine biofoulers. As many biofouling organisms can delay settlement until a suitable surface is encountered, colonization can comprise surface exploration and intermittent swimming. As such, the process is best followed in three dimensions. Here we present a low-cost transportable stereoscopic system consisting of two consumer camcorders. We apply this novel apparatus to behavioral analysis of barnacle larvae (≈800 μm length) during surface exploration and extract and analyze the three-dimensional patterns of movement. The resolution of the system and the accuracy of position determination are characterized. As a first practical result, three-dimensional swimming trajectories of the cypris larva of the barnacle Semibalanus balanoides are recorded in the vicinity of a glass surface and close to PEG2000-OH and C11NMe3+Cl− terminated self-assembled monolayers. Although less frequently used in biofouling experiments due to its short reproductive season, the selected model species [Marechal and Hellio (2011), Int Biodeterior Biodegrad, 65(1):92–101] has been used following a number of recent investigations on the settlement behavior on chemically different surfaces [Aldred et al. (2011), ACS Appl Mater Interfaces, 3(6):2085–2091]. Experiments were scheduled to match the availability of cyprids off the north east coast of England so that natural material could be used. In order to demonstrate the biological applicability of the system, analysis of parameters such as swimming direction, swimming velocity and swimming angle are performed.
Since 2008, the most successful of the heavy-metal based antifoulants, tributyltin (TBT) , has been banned and research has since focused on alternative strategies to mitigate the undesired accumulation of biomass on vessels submerged in the marine environment. In order to develop environmentally benign coatings, a better understanding of the colonization mechanisms of the target organisms is required. Ideally, improved knowledge of surface cues that promote settlement of fouling species as well as investigation of surface cues that repel settlement will lead towards effective but environmentally inert antifouling materials. For most biofoulers it is not the macroscopically visible adult organism that is relevant in this context, but the colonization stage, which is responsible for initial surface attachment .
Barnacles, as one of the most prevalent marine fouling groups , contribute significantly to increased hydrodynamic drag and higher fuel consumption of vessels. Much of the research on settlement behavior of thoracican barnacles has been driven by the perceived need for a better understanding of larval settlement in order to interfere with the process.
In barnacles, the cyprid, or cypris larva is the colonization stage. The evolution of barnacles for a sessile mode of life is best epitomized at this stage. The cyprid, armed with a complex array of sensory setae , has a remarkable ability to search the substratum using reversible adhesion to locate a suitable place to settle for the remainder of its life . Two-dimensional tracking has already been used to derive statistically relevant data from the behavior of cyprids on surfaces, revealing differences in exploratory behavior depending on surface chemistry . However, the three dimensional nature of cyprid behavior makes gathering information of satisfactory quality a challenging task. Particularly, distinguishing between cyprids that are actively swimming, passively floating, or exploring surfaces is next to impossible on the basis of 2D data alone. Thus, 3D tracking techniques are a natural and necessary progression.
In the case of small microorganisms such as bacteria or algal spores, digital holographic microscopy has been used to extract 3D traces of moving objects [8–11]. It has also been successfully implemented to reveal predator–prey behavior in dinoflagellates . Holography, however, requires significant expertise and infrastructure if it is to be successfully applied. Alternatively, stereoscopy is a technique able to track objects as diverse as people [13, 14], small particles in particle tracking velocimetry  or dusty plasmas under microgravity . Compared to holography, a major advantage of stereoscopy is that no reconstruction of the recorded data is required.
We developed a system with two consumer camcorders which allows the determination of three dimensional trajectories of barnacle cyprids based on stereoscopy. The hardware setup, as well as the technical issues for obtaining the 3D swimming trajectories, are described in this article, including an empirical error analysis. We tested the system with cyprids of Semibalanus balanoides on chemically different surfaces (glass, PEG2000-OH and C11NMe3+Cl−) and first trajectories are visually presented. Several descriptive features are extracted from the trajectories in order to show the feasibility of the system in biological applications.
2 Materials and Methods
2.1 Theoretical Framework for the Epipolar Geometry
With the projection matrix P it is possible to backproject any 2D point to a 3D projection ray. If this is done on both cameras, the intersection of the corresponding projection rays will represent the reconstructed 3D position.
If the values Δx and Δy are added to both vectors x and x′ and the 3D points are calculated for all combinations, the bounded 3D region defines the spatial resolution. However, this region is not spatially uniform for all values of x and x′ and the extreme values should be considered to determine the minimum, maximum and mean resolution.
2.2 Stereoscopy Setup
The video files recorded during the experiments were split into single frames, and their color format (24 bit RGB) was then converted to grayscale (8 bit). A “mean” background image was calculated by averaging five frames from the beginning, five frames from the middle and five frames towards the end of the video stream, in which the moving objects were present and thus filtered out. This frame, containing only the “static” parts of the image, was subtracted from every frame which was analyzed, thus eliminating most of the artifacts from container borders, edges and corners. To avoid a potential elimination of attached cyprids, the videos were manually inspected beforehand. In order to simplify and accelerate further calculations and processing, each image was transformed from grayscale to binary using the thresholding method described by Otsu . The centroid of the object of interest in each image was estimated applying an image moment calculation.
Since a low density of cyprids was used for these experiments we applied single object tracking to generate the swimming trajectories. An automatic algorithm was used where the user selected the starting position of the cyprid and its positions in the subsequent frames were determined evaluating the minimum Euclidean distance to all the candidate cyprids. The process was supervised by the user to avoid erroneous detection in case of crossing or overlapping trajectories. For the future, efficient multi object tracking approaches will be applied as described in .
2.3 System Calibration
In order to determine the projection and fundamental matrices for a given experimental situation, a calibration needs to be performed prior to the tracking experiment. During the calibration an object, with marks or points, the positions of which are known in real-world coordinates, is used to determine the corresponding positions in image coordinates for both camera images (left and right frame). The calibration object needs to possess at least 6 known points (in our case we used 8) in order to provide enough information to solve the system of Eq. (3). During the calibration, these points (X real ) are selected from different positions of the calibration object. The corresponding positions in image coordinates are detected in the left (x left ) and the right frames (x right ). Using the correspondences, the projection matrixes for the left (P left ) and the right (P right ) camera are calculated by solving Eq. (3).
2.4 Cyprids and Cyprid Collection
The tests described in this work were performed in the School of Marine Science and Technology of Newcastle University, UK. As target organisms, cyprids from the species Semibalanus balanoides were used. Cyprids were collected by plankton tow from the harbor wall at the Dove Marine Laboratory, Cullercoats, NE England and transported to the University where they were transferred to filter-sterilized (0.22 μm) seawater and stored at 6 °C in glass beakers. Cyprids sized ca. 800 μm, which do not feed, were used within 2 weeks of collection. The experiments were performed at room temperature of around 22 °C, and cyprids were left undisturbed for 10 min prior to each measurement in order for them to adapt to the change of water temperature.
2.5 Surface Preparation
Surfaces used for the experiment: PEG, Glass and TMA
Water contact angle (°)
Ellipsometric thickness (nm)
Acid washed glass
Nexterion® glass (Schott)
3 Results and Discussion
3.1 Characterization of the Accuracy in the Position Determination
During calibration it became obvious that the estimation of the exact coordinates of the calibration points at sub-pixel level is difficult and can lead to calibration errors. In addition, illumination and contrast between the swimming object and the background affect the exact estimation of the object’s coordinates. Depending on the reflectivity of the surface, mirror aberrations might also occur, which combined with the morphing of the non-spherical objects of interest (Fig. 2b) can induce inaccuracies in the detected position. Several tests (see Online Resource 1) were performed in order to characterize the system’s behavior when an error is introduced. By this analysis, the relationship between the error in the real position (in mm) and in the position of the object in the left and the right frames (in pixel) was empirically determined.
In Fig. 5a different motion patterns can be detected, which we labeled as A, B and C. Cyprids swimming in a region far away from the surface (2–5 mm distance) show relatively high velocities and rather directed motions. From the color coding it can be seen that the trajectory labeled “C” exhibits velocities of up to 15–30 mm/s. Traces indicated by “A” have slightly lower speeds in the range 5–15 mm/s. These values are below the mean velocity calculated (Fig. 5c).
When the cyprids are active in the vicinity (<1 mm distance) of the surface, the swimming velocities are also low (5–7 mm/s). The “B” traces have very limited spatial displacements, which can be interpreted as close inspection/inactivity within a narrow region of the test surface rather than swimming.
In the region far away from the surface (2–10 mm), the traces for PEG (Fig. 7-II.) are qualitatively similar compared to the traces for glass. In agreement with the above discussion, trajectories extending into solution reveal in general relatively high swimming velocities (up to 10–25 mm/s) combined with a wide exploration region. On TMA, no volume trajectories were found as cyprids mostly stayed close to the surface (Fig. 7-III.). The analysis of motility close to the surface reveals that in the case of PEG the swimming velocities are still in the order of 10–20 mm/s and do not change significantly compared to motility further away from the surface. Also, the perimeter of exploration remains relatively wide. On the TMA surfaces, all the movements are highly localized at very low velocities (<5 mm/s). The cyprids stay on the surface during the whole observation period and seem to explore it slowly and thoroughly, while turning around in circles in a relatively narrow exploration region.
As mentioned above, the influence of surface charge on the resistant properties of surfaces has been studied extensively in the past. However, the influence of the presence of tethered, positively charged ammonium groups on barnacle cyprids is still equivocal. Petrone et al.  found higher settlement (50 %) of cyprids of the barnacle B. amphitrite on negatively charged (carboxyl) and only low settlement (<5 %) for positively charged surfaces (quaternary amines). Using cyprids of Semibalanus balanoides—the same species as in this study—deposition of the proteinaceous temporary adhesive (‘footprints’) has been investigated by Aldred et al.  using iSPR. It was found that the number of touchdowns made by the cyprids on chemically different surfaces did not differ significantly and was independent of surface charge. Also, the probability of leaving adhesive ‘footprints’ during exploration was ≈20 % on both, positively charged ammonium terminated surfaces (–NH3+) and on ethylene glycol terminated surfaces (mPEG). However the amount of deposited footprint material was found to be significantly higher on the positively charged surfaces compared to the mPEG coating. The fact that in both cases (–NH3+ and mPEG) only ≈20 % of the touchdowns resulted in footprints, but the amount of deposited material was quite different, suggests that footprint deposition could be connected with the resistant properties of the surfaces. Our observation that extensive exploration is observed on TMA is a further evidence for attractive cues present on these surfaces.
It is important to note that in this work we only attempted quantification of a limited number of swimming trajectories and described observations that were made to demonstrate the applicability of our stereoscopic technique to record surface exploration behavior. For future work a more sophisticated full statistical analysis will allow a quantitative correlation of surface properties and behavior. However, the results shown here already demonstrate that stereoscopy allows investigating how barnacle cyprids select the suitability of a surface for settlement.
4 Summary and Conclusions
A stereoscopic tracking system for sub-millimeter sized microorganisms has been developed and the theoretical framework for its calibration and use has been described. The physical resolution of the cameras, the 3D resolution in position determination, and the frame rate are suited to track the exploration behavior of the target larvae. As shown in the analysis section, the errors originate mainly from position uncertainties in both the calibration and the tracking steps, and accumulate to less than 1 % of the size of the larvae. Thus, the precision is sufficient for very accurate 3D tracking.
The future key application will be to track larvae of biofouling organisms . Barnacle cyprids are of primary interest as they are widely distributed and have been identified as the most common fouling marine invertebrates in the world [31, 32]. The first results obtained with our stereoscopic setup and cyprids of the barnacle Semibalanus balanoides demonstrated successful tracking and showed that differences in the swimming trajectories in solution and close to surfaces could be detected. First indications were obtained which suggested that the surface properties affected the exploration behavior at the surface. Even though all results presented need to be statistically verified with more comprehensive data sets, they clearly show the applicability of the experimental approach. As the setup is very compact and highly modular, a future implementation and tests in situ in the ocean will allow investigating the colonization of surfaces by biofouling larvae in their natural habitat. The developed device can easily be applied for the 3D analysis not only of organisms at sub-millimeter level but also for larger animals, and thus has numerous potential applications in biology, biochemistry, and biophysics.
One question arising with increasing quantities of data is the automatic analysis which can be realized as demonstrated for smaller organisms by Leal-Taixé et al.  or even using more powerful 3D tracking techniques specifically designed for stereoscopic systems . The same holds true for the subsequent automated classification of motion patterns, which is required to show that changes in the occurrence of patterns as result of surface cues are reproducible and have statistical relevance . We see great potential to even combine 3D tracking with further emerging techniques which are sensitive towards interaction of the cyprid temporary adhesive with surfaces, such as imaging SPR as recently reported by Andersson et al. .
The developed 3D tracking system will allow to gain a deeper understanding of the different stages of attachment of larvae of barnacles and other biofouling organisms to interfaces. Besides the relevance of such information for understanding behavior, responses to physical and chemical cues and sensory capabilities of the larvae, comparisons of different surfaces will support the development of environmentally friendly antifouling concepts, interfering at crucial stages of the surface selection process.
The authors would like to acknowledge the financial support by the following projects: DFG Ro 2524/2-2, Ro 2497/7-2, ONR N00014-08-1-1116, ONR N00014-12-1-0498, and the European Community’s Seventh Framework Programme FP7/2007-2013 under Grant Agreement number 237997 (SEACOAT). We thank S. Conlan (Newcastle University) for her support and suggestions in the biological experiments.
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.
- Marechal JP, Hellio C (2011) Int Biodeterior Biodegrad 65(1):92–101. doi:https://doi.org/10.1016/j.ibiod.2010.10.002View ArticleGoogle Scholar
- Aldred N, Ekblad T, Andersson O, Liedberg B, Clare AS (2011) ACS Appl Mater Interfaces 3(6):2085–2091. doi:https://doi.org/10.1021/am2003075View ArticleGoogle Scholar
- Townsin RL (2003) Biofouling 19:9–15. doi:https://doi.org/10.1080/0892701031000088535View ArticleGoogle Scholar
- Callow JA, Callow ME (2011) Nat Commun 2:244. doi:https://doi.org/10.1038/ncomms1251View ArticleGoogle Scholar
- Aldred N, Clare AS (2008) Biofouling 24(5):351–363. doi:https://doi.org/10.1080/08927010802256117View ArticleGoogle Scholar
- Maruzzo D, Conlan S, Aldred N, Clare AS, Høeg JT (2011) Biofouling 27(2):225–239. doi:https://doi.org/10.1080/08927014.2011.555534View ArticleGoogle Scholar
- Marechal JP, Hellio C, Sebire M, Clare AS (2004) Biofouling 20 (4–5):211–217. doi:https://doi.org/10.1080/08927010400011674
- Leal-Taixé L, Heydt M, Rosenhahn A, Rosenhahn B Automatic tracking of swimming microorganisms in 4D digital in-line holography data. In: Motion and Video Computing, 2009. WMVC ‘09. Workshop on, 8–9 Dec. 2009, pp 1–8Google Scholar
- Heydt M, Rosenhahn A, Grunze M, Pettitt M, Callow ME, Callow JA (2007) J Adhes 83(5):417–430. doi:https://doi.org/10.1080/00218460701377388View ArticleGoogle Scholar
- Xu W, Jericho MH, Kreuzer HJ, Meinertzhagen IA (2003) Opt Lett 28(3):164–166. doi:https://doi.org/10.1364/ol.28.000164View ArticleGoogle Scholar
- Heydt M, Pettitt M, Cao X, Callow M, Callow J, Grunze M, Rosenhahn A (2012) Biointerphases 7(1):1–7. doi:https://doi.org/10.1007/s13758-012-0033-yView ArticleGoogle Scholar
- Sheng J, Malkiel E, Katz J, Adolf J, Belas R, Place AR (2007) Proc Natl Acad Sci USA 104(44):17512–17517. doi:https://doi.org/10.1073/pnas.0704658104View ArticleGoogle Scholar
- Rosenhahn B, Kersting U, Powell K, Klette R, Klette G, Seidel H-P (2007) Mach Vis Appl 18(1):25–40. doi:https://doi.org/10.1007/s00138-006-0046-yView ArticleGoogle Scholar
- Leal-Taixé L, Pons-Moll G, Rosenhahn B Branch-and-price global optimization for multi-view multi-object tracking. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2012Google Scholar
- Maas HG, Gruen A, Papantoniou D (1993) Exp Fluids 15(2):133–146View ArticleGoogle Scholar
- Himpel M, Buttenschon B, Melzer A (2011) Rev Sci Instrum 82(5):053706View ArticleGoogle Scholar
- R. I. Hartley AZ (2004) Multiple View Geometry in Computer Vision. 2nd edn. Cambridge University PressGoogle Scholar
- Brauner H (1962) W. Blaschke, Kinematik und Quaternionen. (Mathematische Monographien) VIII + 84 S. Berlin 1960. Deutscher Verlag der Wissenschaften. Preis geb. DM 20,40. ZAMM. J Appl Math Mech/Zeitschrift für Angewandte Mathematik und Mechanik 42 (7–8):366–367. doi:https://doi.org/10.1002/zamm.19620420724
- Otsu N (1979) IEEE Trans Syst Man Cybern 9(1):62–66View ArticleGoogle Scholar
- Leal-Taixé L, Heydt M, Weisse S, Rosenhahn A, Rosenhahn B Classification of Swimming Microorganisms Motion Patterns in 4D Digital In-Line Holography Data. In: 32nd Annual Symposium of the German Association for Pattern Recognition (DAGM), 2010. Lecture Notes in Computer Science. Springer, pp 283–292Google Scholar
- Aldred N, Li G, Gao Y, Clare AS, Jiang S (2010) Biofouling 26(6):673–683. doi:https://doi.org/10.1080/08927014.2010.506677View ArticleGoogle Scholar
- Hills JM, Thomason JC, Davis H, Kohler J, Millett E (2000) Biofouling 16(2–4):171–179View ArticleGoogle Scholar
- DiBacco C, Fuchs HL, Pineda J, Helfrich K (2011) Marine Ecol-Prog Ser 433:131–148. doi:https://doi.org/10.3354/meps09186View ArticleGoogle Scholar
- Crisp DJ (1955) J Exp Biol 32(3):569–590Google Scholar
- Walker G (2004) J Marine Biol Assoc UK 84(4):737–742. doi:https://doi.org/10.1017/S002531540400983XhView ArticleGoogle Scholar
- Holmlin RE, Chen X, Chapman RG, Takayama S, Whitesides GM (2001) Langmuir 17(9):2841–2850View ArticleGoogle Scholar
- Ekblad T, Andersson O, Tai F-I, Ederth T, Liedberg B (2009) Langmuir 25(6):3755–3762. doi:https://doi.org/10.1021/la803443dView ArticleGoogle Scholar
- Rosenhahn A, Schilp S, Kreuzer HJ, Grunze M (2010) Phys Chem Chem Phys 12(17):4275–4286. doi:https://doi.org/10.1039/c001968mView ArticleGoogle Scholar
- Petrone L, Di Fino A, Aldred N, Sukkaew P, Ederth T, Clare AS, Liedberg B (2011) Biofouling 27(9):1043–1055. doi:https://doi.org/10.1080/08927014.2011.625474View ArticleGoogle Scholar
- Abarzua S, Jakubowski S (1995) Mar Ecol Prog Ser 123:301–312. doi:https://doi.org/10.3354/meps123301View ArticleGoogle Scholar
- Hills JM, Thomason JC, Muhl J (1999) Funct Ecol 13(6):868–875. doi:https://doi.org/10.1046/j.1365-2435.1999.00377.xView ArticleGoogle Scholar
- Yebra DM, Kiil S, Dam-Johansen K (2004) Prog Org Coat 50(2):75–104. doi:https://doi.org/10.1016/j.porgcoat.2003.06.001View ArticleGoogle Scholar
- Andersson O, Ekblad T, Aldred N, Clare AS, Liedberg B (2009) Biointerphases 4(4):65–68. doi:https://doi.org/10.1116/1.3274060View ArticleGoogle Scholar
- Bjørndal JO (2011) Marine aquarium blog. http://www.jonolavsakvarium.com/blog/200805/barnacle01_HTML.jpg. Accessed 22 June 2011
- Maleschlijski S, Leal-Taixé L, Weisse S, Di Fino A, Clare AS, Sendra GH, Rosenhahn B, Rosenhahn A (2011) A stereoscopic approach for three dimensional tracking of marine biofouling microorganisms. In: Microscopic Image Analysis with Applications in Biology, Heidelberg, 02 September 2011Google Scholar