Construction and evaluation of a semi-portable radiotelemetry tower system at Prairie Fork Conservation Area, Missouri.

Abstract: Triangulation techniques are often used to estimate the location of radio-marked animals, but many hand-held and mobile antennae systems have relatively low accuracy. Use of fixed-station radiotelemetry towers can increase location range and accuracy, yet the inability to move towers ultimately limits their effectiveness in the field. We designed and tested a unique semi-portable telemetry tower system at Prairie Fork Conservation Area in mid-Missouri during the fall and winter of 2000 and 2001. Each tower mast telescopes to a height of 9.1 meters, rotates 360 degrees, and can be easily moved by two people between permanent concrete base sites with complete set up taking approximately 20 to 30 minutes. Construction materials for each tower, not including antennae and cables, was $248 with the base materials adding an additional $18. After construction, we assessed error of the tower system and a three-element hand held Yagi antennae for comparison. Accuracy was assessed by placing 8 beacon transmitt ers of two different sizes (turtle and deer) at known coordinates throughout the study site. We compared known azimuths to the estimated azimuths to assess accuracy. Mean error for the tower system (e = 2.5[degrees], SD = 2.0, n = 122) was significantly lower (P < 0.001) than for the handheld antenna (e = 10.9[degrees] SD = 6.6, n = 121); no difference in mean error was noted between deer and turtle transmitters (P = 0.623). The convenience, efficiency, and accuracy of the tower system provides a unique option for radiotelemetry data collection while reducing costs by limiting the amount of equipment needed to adequately cover a study area with a traditional tower system.

Key words: Missouri, portable mast, radiotelemetry, radio tracking, telemetry, telemetry error, tower, tower construction, transmitter

Introduction

Radiotelemetry is commonly used to assess space use patterns and survival of wild vertebrates (Millspaugh and Marzluff 2001a). In fact, roughly one-third of all papers published in The Journal of Wildlife Management over the past 20 years were telemetry-based (Millspaugh and Marzluff 2001b). Triangulation and homing (i.e., direct observation) are two common techniques for estimating the location of radio-collared animals. Direct observation of radio-collared animals is generally considered a more accurate method of determining animal locations but may influence animal behavior (Kemohan et al. 2001). Furthermore, many free-ranging animals are relatively inaccessible, secretive, and sensitive to human influence, and thus are not good candidates for homing. Triangulation on the other hand is often more practical and efficient when monitoring large numbers of animals and animals that range over great distances.

Triangulation involves recording two or more azimuths from known locations to the radio-transmitter on the animal (White and Garrott 1990). Using azimuths recorded from known locations and trigonometric relationships, the location of a radio-tagged animal may be estimated with an associated measure of error (Heezen and Tester 1967, Leuth 1981). Azimuths may be obtained from either large, fixed receiving stations (i.e., towers) or smaller, mobile devices (e.g., hand-held antennas, vehicle mounted antennas). Often, azimuths obtained from towers are more accurate (Slade et al. 1965, Anderson and DeMoor 1971, Merson et al. 1982), varying by as little as [+ or -] 2 degrees. Use of fixed-station radiotelemetry towers can also increase location range, yet the inability to move towers ultimately limits their effectiveness in the field.

The goal of this study was to design and construct a semi-portable telemetry tower system at Prairie Fork Conservation Area (PFCA) that was easy to set up and maintained a high level of accuracy. After construction in the fall/winter of 2000, we tested the accuracy of the system in the fall/winter of 2001 and compared tower accuracy with a hand-held 3-element Yagi antenna.

Study Area

PFCA is a publicly owned tract of land of approximately 288 hectares located in Callaway County, Missouri, Township 48, Range 7 West. The site lies within the oak-hickory forest/prairie transition zone at the southern border of what was once the Grand Prairie, now the largely agricultural northern half of Missouri. Approximately 70% of the site is open, previously tilled prairie. Over 350,000 people live within an 80-km radius of PFCA, with the major metropolitan area of St. Louis being less than 130 km away. Agriculture is the primary land use within a 40-km radius of PFCA, with an increase recently in the breakup of larger farms into smaller acreages owned by absentee landowners.

The topography of PFCA is gentle, with a total elevational gradient of approximately 37 m; no steeply incised banks or cliffs are present. There is one permanent stream running through the property with several intermittent streams throughout the site. Twenty small ponds are scattered over the area, most of them having been created to hold water for livestock use.

Tower & Base Placement and Design

Tower base placement was determined using a modification of White's (1985) recommendations for a square, flat study site and by taking into account topographical variation and interference from power lines. First, White's (1985) ideal geometric tower placement was overlaid on a map of PFCA. Next, site coordinates were moved to the nearest high point away from power-lines (White and Garrott 1990), resulting in a configuration with the best possible use of the site topography and geometric design.

A semi-portable tower design was created by modifying the towers and materials reported by Banks et al. (1975), Medina and Smith (1986), and Merson et al. (1982) (Figure 1). Each tower was constructed using a stacked, horizontally-oriented RA-4A 5-element Yagi array (Telonics, Inc., Mesa, AZ, USA) attached to a 9.1 m telescoping mast (Channel Master, Smithville, NC, Model #1830) (Figure 1). A compass rosette attached to the tower base and a pointer attached to the telescoping mast were used to estimate direction to the nearest degree (Figure 2). Dual coaxial RW-5 cables were routed inside the hollow mast to a TAC-5 null precision combiner box (Telonics, Inc., Mesa, AZ, USA). Tower masts were guyed with nylon rope (Figure 1); to prevent interference, no steel cables were used. Also, no guys were attached to the mast within 3 meters of the antenna array. Eight guy wires were affixed to the mast; 4 were attached to the mast at 3 meters and 4 at 6 meters (Figure 1). Guy wires were anchored to four steel posts bur ied horizontally 1/2 meter deep and 4.3 meters from the base center in the four cardinal directions. A length of steel rope that exited the ground was attached to each post to allow for the connection of the tower guy wires. This design has withstood wind speeds up to 80 km/hr. With 2 people, complete tower setup at a pre-constructed base site takes approximately 20-30 minutes. The towers rotate 360 degrees and are easily moved by two people between permanent concrete bases.

Sixteen concrete bases, 0.6 m x 0.6 m x 10 cm, were established throughout PFCA (Figure 2). Each base was reinforced with steel rebar, and all bases were given at least two days to set before use. Three lengths of 0.9525 cm all-thread rod were set vertically into each base with 10.5 cm exposed above the concrete. This exposed rod was then used to attach a two-piece steel thrust bearing (J & L Technologies, Columbia, MO, USA) and a two-piece wooden compass rosette (Figure 2). Complete construction of one base, including the buried steel posts, took approximately 2 hours. The base stations were established low to the ground, permitting mowing and prescribed burning with no negative effects on the bases. Coordinates of the base stations were determined with a global positioning system (GPS) unit corrected with Differential Code OPS (DGPS) to an accuracy of [+ or -] 1 meter. A pre-drilled wooden block was placed over the exposed rod at each unused base to protect the rods from damage that might occur from tires or fee t unexpectedly finding the base. Construction materials for each tower, not including antennae and cables, was $248 with the base materials adding an additional $18 (Table 1).

Error Assessment

Bearing accuracy was evaluated for the tower system in fall 2001; a hand-held antenna error assessment was conducted concurrently for comparison. Data were collected from all 16 permanent receiving stations. A folding 3-element hand-held Yagi (AF Antronics, Inc., Urbana, IL, USA) held 1.8 meters high was used during the hand-held assessment. Only the null signal (Banks et al. 1975) was used to determine directionality for the towers while the loudest signal method (Springer 1979) was used with the hand-held antenna. ATS (Advanced Telemetry Systems, Isanti, Minnesota) radio receivers (Model #R2000) with headphones were used by the three field personnel.

Transmitter test locations were determined by placing a 400m x 400m grid over the study site (White and Garrott 1990) to include diverse topographic and vegetation conditions. Transmitters were attached to wooden stakes at heights of 0.66 meters for deer-size transmitters and 0.33 meters for turtle-size transmitters. Four deer (Lotek Engineering Inc., Newmarket, Ontario, Canada) and four turtle (Advanced Telemetry Systems, Isanti, Minnesota) transmitters (164-165 MHz) were placed at the nearest 4(x = 0.703 kin, Range = 0.2 - 1.7 km, SE = 0.047 km) of 18 grid locations to the base location being tested for bearing accuracy. Azimuths were randomly taken from the receiving station being evaluated to each of the 4 beacon locations to avoid observer bias (Lee et al. 1985); the sequence was repeated five times for each of the 8 transmitters. A second observer recorded azimuths so a previous reading would not influence the investigator. Thus, there were five bearing estimates for each of the eight transmitters from each of the 16 tower locations.

Analytical Methods

Deviate observations resulting from signal bounce and other factors may significantly affect error assessments (Garrott et al. 1986). Without the use of a technique that is relatively insensitive to gross outliers (Lenth 1981), these aberrant bearings must be identified and removed from bias and precision calculations (Lee et al. 1985). Consequently, prior to data analysis, we identified outliers from the data set based on frequency distributions. To estimate azimuth accuracy, we used the replicated measurements of error from each base station to calculate a mean error (e) and a standard deviation (SD). Error (e) was defined as the difference between the true azimuth ([theta]), determined using a hand held GPS unit, and the estimated azimuth ([theta]) of the transmitter for each azimuth i and replicate j as:

[e.sub.ij] = [[theta].sub.i] - [[theta].sub.ij] (1)

Mean error, an estimate of bias, was obtained by summing all errors for all stations including replicates from individual stations, and then dividing this number by the product of the number of locations and the number of replicates (Lee et al. 1985) as

e = [summation over (n/i=1)][summation over (r/j=1)] [e.sub.ij]/nr (2)

where n is the number of reference transmitter locations and r is the number of azimuth replicates. Finally, for a measure of the repeatability or amount of variation of estimated bearings, we estimated standard deviation as (Lee et al. 1985, White and Garrott 1990):

SD = [[[summation over (n/i=1)][summation over (r/j=1)] [([e.sub.ij]-e).sup.1/2]/(nr-1)].sup.1/2] (3)

To determine if there was a difference in the mean error between the tower and hand-held antenna we used one-way ANOVA (Zar 1996). We tested for differences in mean error and precision using transmitter type (deer or turtle), observers, and wind speed as factors in a one-way ANOVA model. Last, we used a backward selection procedure in a generalized linear model (McCullagh and Nelder 1989) to determine which of these factors were most correlated with mean error. All analyses were considered significant at P < 0.05.

Results

Based on a frequency distribution of tower error estimates, we noted that most tower azimuths (94.1%) were within 0 and 10 degrees of the true azimuth (Figure 3). Consequently, for our tower assessment, we defined absolute errors of [greater than or equal to]10 degrees as signal bounce and eliminated them from further consideration (Slade et al. 1965, Lee et al. 1985, Zimmerman and Powell 1995). There was no clear cut off point for handheld outliers, however, and to facilitate a valid comparison, we eliminated the same percentage (5.9%) of outliers from the handheld data set (Figure 4).

Mean error for the tower system (e = 2.5[degrees], SD = 2.0, n = 122) was significantly lower (P = 0.0001) than for the hand-held antenna (e 10.8[degrees], SD = 6.6, n = 121). We noted no difference in mean error between deer (x = 7.35, SD = 8.49, n = 126) and turtle transmitters (x = 8.11, SD = 9.41, n = 128) for the tower system (P = 0.623). Mild observer bias was noted for the tower (x 3.00, SD = 2.31, n = 48 observer 1; x = 2.52, SD 2.36, n = 23 observer 2; x = 2.08, SD = 1.42, n = 51 observer 3) (P = 0.077) but not hand-held (x 13.26, SD = 12.47, n =48 observer 1; x 11.88, SD = 10.01, n = 24 observer 2; x = 12.53, SD = 7.81, n = 56 observer 3) (P = 0.103). We observed no effect of wind speed on error (n = 47, 0-8 km/hr; n = 39, 0-16 km/br; n = 23, 8-24 km/br; n = 15, 16-32 km/hr)(x 2.06, SD = 1.01, n = 122); however, data were not collected when wind speeds exceeded 32 km/hr. Last, the backward selection procedure selected observer (P = 0.023) as the most influential factor in the generalized linear mod el; other factors were not significant.

Discussion

The semi-portable radiotelemetry tower system at PFCA was more accurate than traditional hand-held antennae for triangulation under the situations we evaluated. We believe the accuracy observed for the tower arrays at PFCA would adequately meet many research needs on similar sites. Hand-held systems will continue to be useful in studies where homing is necessary (e.g., behavioral studies) and feasible, and when strong winds preclude use of a tower system.

We offer the following suggestions to those interested in developing a semi-portable tower system. 1. All observers using towers should be trained to avoid observer bias. This should involve repeated practice and continued evaluation of their accuracy during the course of normal research activities. 2. In planning future tower sites, we recommend that areas under and near power lines and buildings be avoided because these structures cause signal bounce and difficulties in distinguishing a null signal. 3. Great care should be taken when collecting bearing estimates with the tower system when wind speeds exceed 16 km/br, and no estimates should be recorded when winds exceed 24 km/hr as the null signal can be difficult if not impossible to distinguish.

The convenience, efficiency, and accuracy of the tower system provides future investigators with a unique way to collect telemetry data. The ability to move towers easily allows the best use of a limited number of antennas, enables investigators a greater choice of tower locations, and reduces costs by limiting the amount of equipment needed to adequately cover a study area.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

Table 1

Materials and cost for tower and base construction of a semi-portable
radiotelemetry tower built at Prarie Fork Consevation Area, Missouri

Materials                                           Cost

RA-NX-5 Precision Direction Finding Antenna    $406 each
Arrays (Telonics, Inc., Mesa, AZ, USA)

9.1 meter telescoping mast model #1830          $50 each
(Channel Master, Smithville, NC, USA)

Rotating steel thrust bearings                 $170 each
(J&L Technologies, Columbia, MO, USA)

Null combiner box                              $285 each
(Telonics, Inc., Mesa, AZ, USA)

Cable and adapters, 1 tower                    $131 each
(Telonics, Inc., Mesa, AZ, USA)

27.2 kg sacks of concrete, 2 per base         $6.00 each

Various bolts, nuts, washers, wood, nylon,      $40 each
rope, steel cable, steel posts, etc.

Total Material Costs per tower               $1,070 each

Total Material Costs per base                   $18 each

Total Cost One Working Tower System          $1,088 each

Acknowledgments

We thank all of those individuals who helped with the construction and implementation of the tower design and assessment; C. Rittenhouse, J. Sumners, B. Washbum, and B. Woeck. Funding for this study was provided by the Prairie Fork Conservation Fund administered by the Missouri Department of Conservation and The School of Natural Resources, University of Missouri.

Literature Cited

Anderson, F. and P. P. DeMoor. 1971. A system for radio tracking monkeys in dense bush and forest. Journal of Wildlife Management 35:636-643.

Banks, E. M., R. J. Brooks, and J. Schnell. 1975. A radiotracking study of home range and acitivity of the brown lemming (Lemmus trimucronatus). Journal of Mammalogy 56:888-901.

Garrott, R.A., G.C. White, R.M. Bartmann, and D.L. Weybright. 1986. Reflected signal bias in biotelemetry triangulation systems. Journal of Wildlife Management 50:747-752.

Heezen, K. L. and J. R. Tester. 1967. Evaluation of radiotracking by triangulation with special reference to deer movements. Journal of Wildlife Management 31:124-141

Kernohan, B. J., R. A. Gitzen, and J. J. Millspaugh, 2001. Analysis of animal space use and movements. Pages 125-166 in J. J. Millspaugh and J. M. Marzluff, editors. Radio Tracking and Animal Populations. Academic Press, San Diego, California, USA. 467 pages.

Lee, J. E., G. C. White, R. A. Garrott, R. M. Bartmann, and A.W. Alldrege. 1985. Accessing accuracy of a radiotelemetry system for estimating animal locations. Journal of Wildlife Management 49:658-663.

Lenth, R. V. 1981. On finding the source of a signal. Technometrics 23:149-154.

McCullagh, P., and J. A. Nelder. 1989. Generalized linear models. Second edition. Monographs on statistics and applied probability No. 37. Chapman & Hall, London, England.

Medina, A. L. and H. D. Smith. 1986. Designs for an antenna boom and masts for telemetry applications. Wildlife Society Bulletin 14:291-297.

Merson, M. H., R. E. Byers, and L. D. Leta. 1982. A portable antenna base for fixed-station radiotracking. Wildlife Society Bulletin 10:44-45.

Millspaugh, J. J., and J. M. Marzluff, editors. 2001a. Radio Tracking and Animal Populations. Academic Press, San Diego, California, USA. 467 pages.

Millspaugh, J. J., and J. M. Marzluff. 2001b. Past trends and future needs. Pages 383-396 in I. J. Millspaugh and J. M. Marzluff, editors. Radio Tracking and Animal Populations. Academic Press, San Diego, California, USA. 467 pages.

Slade, N. A., J. J. Cebula, and R. J. Robel. 1965. Accuracy and reliability of biotelemetric instruments used in animal movement studies in Prairie grasslands of Kansas. Transactions of the Kansas Academy of Science 68:173-179.

Springer, J. T. 1979. Some sources of bias and sampling error in radio triangulation. Journal of Wildlife Management 43:926-935.

White, G. C. and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, San Diego, California. 383 pages.

White, G. C. 1985. Optimal locations of towers for triangulation studies using biotelemetry. Journal of Wildlife Management 49:190-196.

Zar, J. H. 1996. Biostatistical analysis. Third edition. Prentice-Hall, Englewood Cliffs, New Jersey, USA.

Zimmerman, J. W. and R. A. Powell. 1995. Radiotelemetry error: location error method compared with error polygons and confidence ellipses. Canadian Journal of Zoology 73: 1123-1133.