Department of Ecology and Natural Resource ...

MARCH 2009

SHORT COMMUNICATIONS

69

J. Raptor Res. 43(1):69–74 E 2009 The Raptor Research Foundation, Inc.

A PORTABLE DIGITAL VIDEO SURVEILLANCE SYSTEM TO MONITOR PREY DELIVERIES AT RAPTOR NESTS RONNY STEEN1 Department of Ecology and Natural Resource Management, Norwegian University of Life Science, NO-1432 A˚s, Norway KEY WORDS: digital video surveillance system; diet; mini DVR; monitoring; nest boxes; prey delivery; raptor.

1

Email address: [email protected]

Video technology has improved rapidly, and wildlife video monitoring has successfully used modern VHS time lapse and hard disk recorders (e.g., McQuillen and Brewer 2000, Booms and Fuller 2003, Rogers et al. 2005, Reif and Tornberg 2006). The greater storage capacity enables extended continuous recording, and some systems use

70


VOL. 43, NO. 1

prey deliveries at raptor nests. In this study I evaluated whether the system was appropriate for monitoring prey deliveries at Eurasian Kestrel (Falco tinnunculus) nests located in nest boxes. Details on the food habits of these kestrels will be presented elsewhere. METHODS

Figure 1. Schematic drawing of the monitoring system: (A) color CCD camera, (B) nest box on a tree (approximately 5–7 m aboveground), (C) RCA video cable and power supply cable (both 10 m), (D) waterproof plastic box, containing the mini DVR, voltage converter and two fuses (one 1AT glass fuse each for the camera and the mini DVR), and (E) power supply, 12-volt battery. motion sensors to trigger recordings (Reif and Tornberg 2006). Although web-based camera systems have improved video monitoring (Locke et al. 2005), such systems require high power consumption and are constructed for a public electricity network. In recent years space-efficient and lowtension (i.e., operating under low voltage) video surveillance systems have been developed and used for wildlife surveys (Pierce and Pobprasert 2007). Continuous video recording, which requires substantial review time, has been used in numerous studies (McQuillen and Brewer 2000, Booms and Fuller 2003, Reif and Tornberg 2006). To reduce the amount of time necessary to review video footage, some surveys have used infrared detectors (e.g., Cutler and Swann 1999), where an individual’s approach triggered the sensor. Unfortunately, inherent in this system is a time lag from the time of detection to the onset of recording. Further, infrared detectors are sensitive to environmental influences, such as sunbeams and moving vegetation (Cutler and Swann 1999, Swann et al. 2004), potentially resulting in incomplete event recordings. Here I describe the use of a portable video surveillance system for monitoring prey deliveries at raptor nests. Although a comparable system has been used for monitoring nest predation (Bolton et al. 2007), this is the first study using mini DVR with a built-in VMD sensor for monitoring

For monitoring prey deliveries, I studied ten pairs of Eurasian Kestrels during June and July 2007 that nested in 10 nest boxes in an area of about 2000 km2 in Trysil municipality, Hedmark County in southeastern Norway (61uN, 12uE). For each pair, I replaced the original nest box, when the oldest nestlings were 8–12 d old, with a new box (40 cm 3 30 cm 3 50 cm) designed to accommodate video monitoring. At this age the nestlings were able to thermoregulate without parental assistance (Village 1990). The installation required 30–60 min, and parental activities were resumed almost immediately, as in my pilot study in 2005 and 2006 (R. Steen unpubl. data). Inside each nest box, a camera was placed in an upper back corner, and pointed towards the entrance of the nest box. To cover a large view inside the nest box the camera was equipped with a wide-angle lens. The camera was connected via cable to a mini digital video recorder (mini DVR), which stored data on SD cards. The mini DVR was placed in a waterproof plastic container on the ground (Fig. 1). The total weight of the equipment, battery excluded, was 1382 g, and the total cost was approximately $700. To operate the mini DVR, I used a portable LCD TV, Denver model DFT-709 (65 mm 3 90 mm 3 29 mm), powered by the same battery used for the mini DVR and camera. I used a CCD (charge-coupled device) camera, model MS-C850G (Misumi Electronic Corp., Taipei, Taiwan). The signal system consisted of PAL (phase alternating line) with 470 TV lines. The camera, fitted with a 2.1mm wide-angle lens and powered by a 12 V DC battery, had fine resolution (effective pixels at 512 3 582) and high light sensitivity (0.2 lux with a focal ratio of 1.4). The mini DVR (Cybereye model DV-100; Menix Co., Ltd., Daejeon, S. Korea) supports both NTSC (national television system committee) and PAL video systems. The video compression format was MPEG4-SP (moving picture expert group – simple profile), with ASF (advanced system format) media file format. A frame rate of 10 pictures per sec was chosen (resolution 704 3 560), which allowed ca. 150 min of recordings on a 2 GB Sandisk UltraH II SDTM card when the highest picture quality was selected. Date and time were automatically recorded. The mini DVRs were equipped with a voltage converter (input 12 V, output 5 V, with a maximum current drain of 300 mA). Both the camera and mini DVR system were powered by a sealed marine 12 V DC (80 Ah) lead battery (21 kg). For the whole system, power requirement was a 12 V DC battery, and average current drain was measured to 226 mA during recording and 221 mA in standby mode.

MARCH 2009


71

Figure 2. Screenshots from recordings made in a kestrel nest box illustrating the video motion detection sensor. The upper left picture shows the camera view and the resting nestlings. The upper right picture shows area inside the nest box that was masked (cross-hatching). Only the unmasked entrance area is sensitive to movements that trigger recordings. In the lower left photograph, the approaching female triggered a recording. In this case (lower right), she delivered a common lizard (Lacerta vivipara).

Table 1. Time spent monitoring and reviewing files, relative to the number of prey deliveries and data storage use for ten Eurasian Kestrel nests in southeastern Norway, 2007. MONITORING TIME NEST

GROSS1

FAIL2

NET3

REVIEWING TIME (hr)

PREY DELIVERIES (N)

STORAGE USE (Gb)

1 2 3 4 5 6 7 8 9 10 TOTAL

400 379 405 454 426 358 334 334 432 456 3978

6 10 16 4 36

394 369 389 454 426 358 334 334 428 456 3942

14 13 10 14 15 12 12 12 14 14 130

209 360 338 512 472 355 249 357 508 507 3867

7.38 5.93 5.71 6.43 4.14 2.60 2.64 3.76 3.89 3.50 45.98

1 2 3

Total time monitoring. Time lost due to SD card or battery failure. Net monitoring (i.e., total time monitoring less time lost).

72


The mini DVR had a built-in VMD sensor, which detects movement in the monitoring area and automatically records these events on the SD card. Both the sensitivity and detection area, which may trigger recordings, were adjustable. Sensitivity was established by displaying the percent change in the monitoring area when a movement occurred, and selection of the detection area was set using a masking tool. During a simulated movement the detection area was set for the entrance of the nest box and the sensitivity was set to medium, since changes in the detection area were obvious when parents entered the nest box (Fig. 2). The recording time for each event was first set to 10 sec, but was later adjusted to 5 sec. In the present study, the ten nest boxes were monitored concurrently, with ten complete systems that were operated simultaneously. Each video system was checked daily in the beginning of the monitoring period, and subsequently on every second or third day. The disturbance caused by routine maintenance was minor, because these events lasted only ca. 1 min and were conducted on the ground below the nest box. I changed the SD card when 50–90% of the storage capacity had been used. I checked the capacity of the battery by measuring voltage with a digital voltmeter, and it was considered discharged at 12.05 V (and fully charged at 12.65 V). The system would operate at a lower voltage (i.e., just below 12.00-volt), but I chose a discharge level at 12.05 V to avoid power shortage. To maintain a continuous operation, rotation of batteries was required. The SD cards with stored material were transferred to a laptop with a SD card reader. The mini DVR recorded the events in ASF file format (see http://www.microsoft.com/ for details), and the files with prey deliveries were detected by viewing the files, using thumbnail function, where the start of the recording was displayed as a screenshot. In this manner, files were shown as small screenshots, and files with a prey delivery were easily separated from other files. In some instances the first frame was empty for files containing prey deliveries. This was a result of the parents approaching the nest after a recording was triggered by other environmental factors. This was easily detected because a single prey delivery triggered several recordings. Each file was automatically assigned a unique identification number with the date and time expressed. The recordings could be viewed in detail (frame by frame) with a DivX-player (see http://www.divx.com/ for detail) or with the mini DVR. I used Spearman’s rank correlation to test whether data space consumption of the SD cards (hereafter called storage use) was related to number of prey deliveries. The Spearman’s rank correlation was performed with the statistical software package JMP 4.0.0 (SAS Institute Inc. 2000). RESULTS A total of 164 d (3942 hr) of diurnal and nocturnal monitoring was recorded. No prey deliveries were recorded between 2336 H and 0312 H (average times of sunset

VOL. 43, NO. 1

and sunrise during the study were 2250 H and 0335 H, respectively). A total of 3867 prey deliveries was recorded (Table 1), and 130 hr were spent extracting the prey deliveries and identifying the prey (see below). The capacity of the SD card varied between the nests and there was no correlation between the storage use and number of prey deliveries recorded (rs 5 0.07, P 5 0.85, Table 1). Thus, other factors seemed to determine storage use (i.e., female resting at the entrance and nestlings flapping their wings in the VMD area). To have a safe margin, I replaced the SD cards when 50–90% of the storage capacity was used, and this resulted in an average of 3.5 cards per nest for the monitoring period (i.e., 7 GB). However, only 4.59 GB on average was used per nest (i.e., 65% of the available storage capacity). The mini DVR was set for 5-sec recordings, but some of the stored files lasted up to 10 sec. In addition, some recordings were triggered by flying insects and lasted only 0.01–5 sec. Consequently, the actual recording time may in some cases vary despite the 5-sec setting. The batteries were changed after approximately 10 d. The system worked well, but a few hours were lost due to a full SD card, one case of SD card error, and shortage of battery power (Table 1). The video quality was satisfactory and enabled identification useful for prey studies. The prey items were preliminary classified, and more time will be spent on prey identification to species level. However, prey species like common lizard (Lacerta vivipara) and wood lemming (Myopus schisticolor) were easily classified to species. Shrews were classified only to genus (Sorex), and species of the genus Microtus and Clethrionomys were sometimes only classified to Cricetidae. Prey identification of small mammals was based upon a combination of size, shape, fur structure, and color. Most birds were only classified to group (e.g., small passerines and thrushes), since they were usually delivered decapitated and plucked, but sometimes plumage color, shape and size of bill and tarsus made identification possible. Prey identification was more difficult when the nestlings were older than 23–25 d, because the duration of prey deliveries was very short. By viewing the footage frame by frame it was usually possible to classify or identify prey items. Of the delivered prey items, 95% were identified at least to class level, 92% at least to family level, 70% at least to genus level, and 5% were impossible to identify or classify. I tested the reliability of the system prior to the monitoring period described above to determine whether all prey deliveries were detected by the VMD. In 2006, I videorecorded six nests continuously with approximately the same apparatus and settings described above, using a hard disk drive (HDD; Sony RDR-HX910). I selected the recordings from one of these nests, with 29 prey deliveries distributed throughout ca. 12 hr of continuously recorded video. In the laboratory, these recordings were later transferred via a video cable to the mini DVR by using the HDD recorder as a player. The mini DVR interpreted these recordings identically to video signals that were ob-

MARCH 2009


tained from the monitoring cameras at the nest box. In fact, the mini DVR detected all recorded prey deliveries. In 2008, I monitored one nest for 24 hr both continuously and with VMD. For this test, I used two mini DVRs simultaneously, one with the VMD activated (as in 2007), and one with continuous recording at the lowest quality setting. A RCA split cable was used to provide both mini DVRs with the same video signal from the camera in the nest box. Both the continuous recordings and VMD documented the same 20 prey deliveries. DISCUSSION To sort out and classify the 3867 unique prey deliveries from the 3942 hr of video monitoring data, I spent 130 hr reviewing the recordings, which translated to about 2 min of review per prey delivery and about 2 min of review per monitoring hour. However, more time will be needed to classify prey to the species level. Compared to other studies, this method required less effort because sorting files by small screenshots was effective. Previous studies using continuous video recordings required considerable time to review video footage (e.g., McQuillen and Brewer 2000, Booms and Fuller 2003, Margalida et al. 2005, Rogers et al. 2005, Reif and Tornberg 2006). For example, Booms and Fuller (2003) used about 20 min of review time per prey delivery, and about 7 min of review time per monitoring hr of nesting Gyrfalcons (Falco rusticolus). The monitoring period consisted of 164 days/nights, distributed among the ten nest boxes. Because of the sensitivity of the camera, motion was recorded even at night, although the picture quality was reduced. This did not bias my results because the kestrels were resting at night. The quality of the recordings was satisfactory and enabled prey identification. However, at times, especially late in the nestling period, the prey was visible for only a few seconds and prey identification was more difficult. This underscored the importance of using a high-resolution camera and recording at least 10 frames per sec to enable adequate frame-by-frame viewing. The capacity of the SD card varied between the nests and was affected by factors other than number of prey deliveries. Some females approached the VMD area more frequently than others, and some rested at the edge of the entrance and triggered recordings. Also, some single prey deliveries triggered several recordings because the parents remained at the entrance for .5 sec. Further, parents triggered the sensor when feeding the nestlings and when leaving the nest. The incidence of such recordings peaked early in the nesting period, and decreased as the nestlings grew older. As the nestlings became more active with age, their wing flapping triggered the sensor, resulting in stored files containing unwanted events. For raptors breeding in nest boxes with a small entrance hole, this problem can be minimized by limiting the VMD area to the entrance hole. To minimize the time spent at the nest location when adjusting the mini DVR, I recommend using long signal

73

cables (approx. 50–100 m) between the camera mounted at the nest and the mini DVR on the ground. To make accurate adjustments of the VMD, one may first use an HDD recorder for continuous recording and then transfer these recordings to the mini DVR. The mini DVR handles these recordings in the same way it does video signals obtained from the monitoring camera in the field. In this manner, any adjustment of the mini DVR could be conducted away from the nest, before coupling the mini DVR and camera at the nest location. SISTEMA DE VIGILANCIA CON VIDEO DIGITAL POR´ TIL PARA SUPERVISAR LAS ENTREGAS DE PRESAS TA A LOS NIDOS DE RAPACES RESUMEN.—Los sistemas de vigilancia por video desarrollados recientemente permiten registrar variables como la deteccioń del movimiento. En este estudio, describo el uso de un sistema de vigilancia con video porta´til para analizar las entregas de presas en cajas nido usadas por Falco tinnunculus. El sistema consistio´ en un mini DVR con deteccioń de movimiento por video, que detectaba movimiento y activaba las grabaciones inmediatamente. Las grabaciones se guardaron en una tarjeta digital de seguridad (Tarjeta-SD). La duracioń de la grabacioń activada de video fue ajustable, de 5 a 30 seg. Cuando se selecciono´ la resolucio´ n ma´ s alta, una tarjeta-SD (2 GB) almaceno´ aproximadamente 150 minutos de grabacio´ n de video (aproximadamente 1800 secuencias con intervalos de 5 seg.). Con una baterıá 12 volt DC (80 Ah) como fuente de suministro de electricidad, cada unidad podrıá durar aproximadamente 14 dıás. El costo total por unidad fue aproximadamente de US $700. Este estudio se realizo´ a lo largo de 164 dıás, se emplearon 10 cajas nido y fueron registradas 3867 entregas de presas. Comparado con los sistemas de vigilancia por video empleados recientemente para analizar las entregas de presas en rapaces anidantes, el mini DVR resulto´ ma´s liviano, presento´ un consumo de electricidad bajo y demostro´ tener un censor de movimiento confiable. [Traduccioń del equipo editorial] ACKNOWLEDGMENTS I thank Bjørn Foyn and Ole Petter Blestad for allowing me to film their kestrel nest boxes, Geir Homme for help with fieldwork, Tom Ringstad for technical advice and support, Vidar Sela˚s, Tore Slagsvold, Geir A. Sonerud, Brett Smithers, and an anonymous reviewer for valuable comments on the manuscript, Anne-May Ilestad for improvements of the English, and Luis Tapia and Andre´s Ordiz for Spanish translation of the abstract. The study was supported by the Directorate for Nature Management and the Hedmark County Governor. LITERATURE CITED BOLTON, M., N. BUTCHER, F. SHARPE, D. STEVENS, AND G. FISHER. 2007. Remote monitoring of nests using digital camera technology. J. Field Ornithol. 78:213–220.

74


BOOMS, T.L. AND M.R. FULLER. 2003. Time-lapse video system used to study nesting Gyrfalcons. J. Field Ornithol. 74:416–422. CUTLER, T.L. AND D.E. SWANN. 1999. Using remote photography in wildlife ecology: a review. Wildl. Soc. Bull. 27:571–581. LOCKE, S.L., M.D. CLINE, D.L. WETZEL, M.T. PITTMAN, C.E. BREWER, AND L.A. HARVESON. 2005. From the field: a web-based digital camera for monitoring remote wildlife. Wildl. Soc. Bull. 33:761–765. MARGALIDA, A., J. BERTRAN, AND J. BOUDET. 2005. Assessing the diet of nestling Bearded Vultures: a comparison between direct observation methods. J. Field Ornithol. 76:40–45. MCQUILLEN, H.L. AND L.W. BREWER. 2000. Methodological considerations for monitoring wild bird nests using video technology. J. Field Ornithol. 71:167–172.

VOL. 43, NO. 1

PIERCE, A.J. AND K. POBPRASERT. 2007. A portable system for continuous monitoring of bird nests using digital video recorders. J. Field Ornithol. 78:322–328. REIF, V. AND R. TORNBERG. 2006. Using time-lapse digital video recording for a nesting study of birds of prey. Eur. J. Wildl. Res. 52:251–258. ROGERS, A.S., S DESTEFANO, AND M.F. INGRALDI. 2005. Quantifying Northern Goshawk diets using remote cameras and observations from blinds. J. Raptor Res. 39:303–309. SWANN, D.E., C.C. HASS, D.C. DALTON, AND S.A. WOLF. 2004. Infrared-triggered cameras for detecting wildlife: an evaluation and review. Wildl. Soc. Bull. 32:357–365. VILLAGE, A. 1990. The kestrel. T. and A.D. Poyser, Berkhamsted, U.K. Received 26 March 2008; accepted 22 October 2008 Associate Editor: James W. Watson