Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Use of oviduct-inserted acoustic transmitters and positional telemetry to estimate timing and location of spawning: a feasibility study in lake trout, Salvelinus namaycush

Use of oviduct-inserted acoustic transmitters and positional telemetry to estimate timing and... Background: Oviduct-inserted transmitters have shown promise for determining precise location of spawning in fishes. Use of traditional manual tracking to locate expelled oviduct transmitters is laborious and accurate estimates of time of transmitter expulsion require frequent surveys. We tested the feasibility of using oviduct-inserted transmitters with positional telemetry to estimate time and location of spawning in lake trout (Salvelinus namaycush). Three assumptions were tested: (1) oviduct transmitters remain within fish until spawning, (2) oviduct transmitters are expelled with the eggs during spawning, and (3) time and location of oviduct transmitter expulsion can be accurately determined. Results: In the laboratory, 39 of 44 (89%) lake trout retained an oviduct transmitter until end of the spawning period and all premature transmitter expulsions occurred before maturation. Natural spawning in the laboratory was not feasible; however, of 35 ripe trout that retained transmitters, 31 (89%) expelled their transmitter with eggs when subjected to manual stripping. Ability to position transmitters with a telemetry array at known spawning sites in Lake Huron (North America) was poor when oviduct transmitters were placed in the substrate compared to transmitters suspended 1 m above substrate - 78% of transmitters in substrate could not be positioned. However, in simulations, time and location of spawning were determined with reasonable accuracy by double-tagging trout with one transmitter that is expelled with the eggs during spawning while another transmitter remains in the fish. Accuracy of estimated time and location of transmitter spatial separation varied with distance traveled from spawning site and swimming speed, and was dependent on transmission delay. Conclusions: Our results satisfied the three assumptions of oviduct tagging and suggested that oviduct transmitters can be used with positional telemetry to estimate time and location of spawning in lake trout and other species. In situations where oviduct transmitters may be difficult to position once expelled into substrate, pairing oviduct transmitters with a normal-sized fish transmitter that remains in the fish is recommended, with spawning inferred when the two tags separate in space. Optimal transmitter delay will depend on expected degree of spawning site residency and swim speed. Keywords: Acoustic telemetry, Changepoint analysis, Gymnovarian, Spawning behavior, Transmitter retention, Detection probability * Correspondence: tr.binder@gmail.com Department of Fisheries and Wildlife, Michigan State University, Hammond Bay Biological Station, 11188 Ray Rd, Millersburg, MI 49759, USA Full list of author information is available at the end of the article © 2014 Binder et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Binder et al. Animal Biotelemetry 2014, 2:14 Page 2 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Background The main disadvantage of using oviduct transmitters An understanding of the reproductive ecology of fishes in conjunction with mobile telemetry is the amount of is becoming increasingly important in light of wide- labor required to frequently manual track transmitters spread ecosystem alterations, such as introduction of [14,15]. The effort is compounded by the fact that ac- exotic species [1-3], critical habitat degradation [4-6], curacy of estimated time and location of transmitter and climate change [7-9]. Fishery managers need to have expulsion will be proportional to the amount of effort an understanding of spawning habitat, and the physio- expended in locating transmitters - for example, accur- chemical and biological variables that promote success- acy of the estimated time of spawning will be greater ful reproduction. Historically, spawning sites have been if manual tracking is done daily than if it is done once identified by surveying (for example, visually, by fishing, per week. Positional telemetry using fixed receivers is or hydroacoustics) for aggregations of ripe adult fish an attractive alternative to mobile telemetry when the [10], which have then allowed researchers to focus on general spawning area (scale of kilometers) is known specific areas and to deduce spawning locations at scales because positional telemetry provides nearly continuous of hundreds of meters to kilometers. More recently, ad- tracking data 24 h per day and requires little manual vances in biotelemetry have allowed researchers to track labor beyond deployment and retrieval of the receiver the fine-scale movements of individuals during spawning array [14,15,19]. In addition, depending on array design, season [11-13]. The current state-of-the-art in animal accuracy of position estimates from positional telemetry tracking is positional telemetry - use of three or more should be better than those obtained from mobile stationary receivers to triangulate the precise location of telemetry [14]. a transmitter-implanted fish to within a few meters of its Regardless of tracking method, use of oviduct-implanted true location [14,15]. In many cases, use of positional transmitters to estimate time and location of egg depos- telemetry could allow researchers to infer spawning loca- ition requires three assumptions: (1) oviduct transmitters tions at scales of tens to hundreds of meters. Nonetheless, remain within fish until spawning (that is, no premature the above techniques have two distinct disadvantages. expulsion), (2) oviduct transmitters are expelled with the First, estimation of the precise time of spawning is not eggs during spawning, and (3) time and location of ovi- possible. Second, the above techniques rely on the as- duct transmitter expulsion can be accurately determined. sumption that presence of a fish at a given location during The biggest unknown with respect to use of oviduct trans- the spawning season indicates that an individual spawned mitters in conjunction with positional telemetry is the at that location. Therefore, additional techniques outside likelihood that the telemetry system will be able to calcu- of those described above (for example, sampling egg de- late positions of oviduct transmitters after they have been position) are required to determine time of and location expelled onto or into the substrate. Nonetheless, all three of spawning. A possible solution to these limitations is use assumptions should be addressed before oviduct transmit- of oviduct-inserted transmitters to estimate the precise ters can be deemed a useful tool for studying spawning time and location of egg deposition; the idea being that behaviors in a given species. an oviduct transmitter will be expelled from the fish Lake trout (Salvelinus namaycush) are an ecologically with eggs during spawning, and then located resting and economically important fish species in the Laurentian on the spawning substrate. The technique was first Great Lakes, where they have been the focus of intensive demonstrated by Pierce [16], who inserted miniaturized restoration efforts [20,21], and in parts of western United radio transmitters into the oviduct of northern pike States, where they are invasive [22-24]. Lake trout typically (Esox lucius) to identify critical shoreline spawning spawn during autumn, and unlike most salmonines, they habitat. Since then, the technique has been used suc- tend to spawn in open-lake environments [25,26]. The cessfully in muskellunge, Esox masquinongy [17], and reproductive biology of lake trout presents two unique challenges to use of oviduct transmitters in conjunction European perch, Perca fluviatilis [18]. In all three of those studies, mobile tracking was used to determine with positional telemetry. First, unlike the majority of the location of expelled transmitters - Pierce and col- teleost fishes (including northern pike, muskellunge, and European perch), lake trout are gymnovarian, which leagues [16,17] searched daily for expelled transmitters and Skovrind et al. [18] searched for expelled transmit- means that no direct connection exists between the ovary ters twice per week. Because expelled transmitters were and urogenital pore. Instead, eggs are first released loose into the body cavity and then expelled through the oviduct not recovered in these studies, it is not possible to assess the accuracy of transmitter expulsion locations deter- [27,28]. Second, lake trout often spawn on cobble mined by mobile tracking. However, Skovrind et al. [18] shoals with clean, deep interstitial spaces [25,26]. Expelled estimated that accuracy of estimated transmitter loca- oviduct transmitters would likely settle in the spaces tions in their acoustic telemetry study was between 10 between rocks, where their acoustic signals may be and 125 m dependent on the acoustic environment. blocked and, therefore, less likely to be positioned. In Binder et al. Animal Biotelemetry 2014, 2:14 Page 3 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Table 1 Summary of results from a laboratory study of this study, we used a series of field and laboratory retention and expulsion of oviduct transmitters in female experiments and random walk models to test the lake trout three assumptions of using oviduct transmitters (that Stage Inserted Implanted is, transmitter retention until spawning, transmitter expulsion during spawning, and ability to accurately Initial sample size (N) 22 22 determine time and location of transmitter expulsion) Retained oviduct transmitter until end 18 (82%) 21 (95%) of spawning period to study the spawning behavior of lake trout. Our goal was to assess the potential for using oviduct transmit- Became ovulatory (ripened) 17 (94%) 18 (86%) ters in conjunction with positional telemetry to study Percent of ripe trout that expelled 15 (88%) 16 (89%) reproductive ecology, not only in lake trout, but in other transmitter with eggs fishes as well. Percent of total trout that expelled 68% 73% transmitter with eggs Results Oviduct transmitters were either ‘inserted’ into the body cavity via the oviduct or ‘implanted’ surgically just anterior to the genital pore. Values are the Given the methodological focus of this study, we recom- number of lake trout that reached each stage and percentages in parentheses mend that readers review the Methods section at the are based on the number of lake trout that made it to the previous stage. end of the paper prior to reading Results. expel the oviduct transmitter. It is possible that these Retention and expulsion of oviduct-inserted transmitters trout would have become ovulatory after a few more In a laboratory study, 48 transmitters were inserted via days in the holding tank, but any results obtained would the oviduct (hereafter ‘inserted’) or surgically implanted have been confounded by previous handling. Therefore (hereafter ‘implanted’) into adult female lake trout. In- lake trout that were not ovulatory were excluded from sertion and implantation procedures had no observable analysis of transmitter expulsion. effect on the behavior or condition of lake trout during Transmitter expulsion rates during stripping were high the 10-week lab study; however, four trout died during for both the inserted and implanted groups (88% and the study (two from the inserted group: both V7 trans- 89%, respectively) and did not differ from one another mitters; two from the implanted group: one V6 transmit- (Fisher’s exact test: P >0.999). The number of stripping ter and one V7 transmitter). Mortalities occurred at 13, motions required to expel the transmitters ranged from 15, 36, and 55 days post tagging, but cause of death was 1 to 17 (mean (±SD) = 7.0 ± 3.8, median = 6) and was not apparent. Between the two different sized oviduct not different between the two treatment groups (t-test: transmitters tested (V6 and V7, VEMCO, Nova Scotia, t = 1.03, P = 0.31). In lake trout that did not expel the Canada), no transmitter-size effect was observed on re- transmitter during stripping, postmortem autopsies re- tention (time-to-event analysis: χ = 1.6, P = 0.21) or ex- vealed that the oviduct transmitter had moved anteri- pulsion (Fisher’s exact test: P >0.999), so data from the orly, away from the oviduct. two tag types were pooled for subsequent analyses to in- crease statistical power. Data from the four lake trout Estimating time and location of spawning using oviduct that died during the study were not included in analyses. transmitters Most lake trout in the inserted and implanted groups Inferences from a single oviduct transmitter in each fish retained their oviduct transmitters until the end of The ability of acoustic receivers to detect stationary spawning season (Table 1). Five (11.3%) trout expelled transmitters at known lake trout spawning reefs in the their transmitter prematurely. More trout in the inserted Drummond Island Refuge (northern Lake Huron, North group (4 of 22 fish) than the implanted group (1 of America; Figure 1) was reduced when transmitters were 22 fish) prematurely expelled their transmitter. The dif- manually placed in the substrate to simulate a transmit- ference was not significant (survival analysis: χ = 2.0, ter expelled during spawning versus a transmitter that P = 0.16), but because our sample size was relatively was suspended 1 m above the substrate, simulating a small, statistical power was marginal (for example, transmitter still within the fish (Figure 2A; linear mixed power to identify a 25% difference in premature expul- model: t = -13.91, P <0.001). Mean (±SE) detection prob- sion rate was 0.68). All five premature expulsions oc- ability (that is, number of transmissions detected by re- curred during the first 2 weeks after tagging, well before ceivers as a proportion of total transmissions multiplied maturity and the start of spawning season and ovulation. by the number of receivers) of control transmitters sus- Not all lake trout were ovulatory at the end of the pended 1 m above substrate was 0.74 ± 0.27 and 0.70 ± spawning season when eggs were manually stripped to 0.16 for V6 and V7 transmitters, respectively. For trans- determine if oviduct transmitters would be expelled with mitters located in the substrate, probability of detection the eggs. Three trout from the implanted group and one dropped to 0.25 ± 0.17 and 0.23 ± 0.14 for V6 and V7 from the inserted group did not shed eggs, nor did they transmitters, respectively. Detection probabilities did Binder et al. Animal Biotelemetry 2014, 2:14 Page 4 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 1 Map of Drummond Island Lake Trout Refuge field study site used for testing the ability of the VEMCO Positioning System (VPS) to position lake trout oviduct transmitters after they are expelled into substrate. Left: Location of the refuge in Lake Huron (North America; shown in orange). The blue star indicates the actual study site (45°55.840ˈ N, 83°39.572ˈ W). Right: Location of three small VPS arrays (at known lake trout spawning sites) overlaid on high-resolution multibeam sonar bathymetry. Blue squares indicate location of receivers and white circles represent test transmitter locations. One transmitter at each test transmitter location was suspended 1 m above the substrate to act as a control and two more were placed in the substrate. A synchronization transmitter was placed at the central test transmitter site to synchronize receiver clocks. not differ between V6 and V7 transmitters (linear mixed 0.03 ± 0.10, respectively. No positions were obtained for model: t = -15.56, P = 0.38). 78% (47 of 60) of transmitters in the substrate compared The ability of the VEMCO Positioning System (VPS) to only 10% (3 of 30) of suspended transmitters that to calculate positions for transmitters was reduced for returned no positions. V6 and V7 transmitters did transmitters placed in the substrate (Figure 2B; linear not differ in positioning probability (linear mixed model: mixed model: t = -15.56, P <0.001) in comparison to t = -0.44, P =0.66). suspended transmitters. The decrease in probability of The poor ability of the VPS to position transmitters in positioning (that is, number of positions obtained as a the substrate, though many were still being detected by proportion of total number of transmissions) was greater the array (albeit at a lower rate than for suspended than the decrease in probability of detection described transmitters), occurred because transmissions from above (Figure 2A and B). Mean (±SE) positioning prob- transmitters planted in the substrate were less likely to ability for suspended transmitters was 0.69 ± 0.35 and be detected simultaneously by three or more receivers. 0.64 ± 0.27 for V6 and V7 transmitters, respectively. The detection failure by multiple receivers was due, in For transmitters placed in the substrate, mean (±SE) part, to lower elevation of transmitters in substrate com- probability of positioning decreased to 0.04 ± 0.10 and pared to suspended tags and complex rough bathymetry Figure 2 Probability of detecting (A) and positioning (B) V6 and V7 oviduct transmitters (VEMCO, Halifax, Canada) following expulsion into substrate at known spawning sites in the Drummond Island Lake Trout Refuge, Lake Huron. V6 transmitters operated at 180 kHz and V7 transmitters operated at 69 kHz. Control transmitters were suspended 1 m above the substrate. Transmitter type had no significant effect on probability of detection or positioning (linear mixed model: P <0.001); however, oviduct transmitters in the substrate were significantly less likely to be detected and positioned than control transmitters (linear mixed model; P <0.001 for both detection and positioning). Seventy-eight percent of oviduct transmitters in the substrate were not positionable. Binder et al. Animal Biotelemetry 2014, 2:14 Page 5 of 14 http://www.animalbiotelemetry.com/content/2/1/14 of the cobble substrate in the spawning area, both of (low residency group) (Figure 4; general linear model: which restricted line of sight between transmitters and t = -5.512, P <0.001). For example, mean (±SE) location receivers. Probability of detection was negatively associ- error was 10 times greater for simulated trout in the low ated with the area obstructing line of sight (cross-sec- residency group than that for simulated trout in the high tional area of substrate located above line of sight residency group (186.1 ± 3.7 m location error vs. 15.8 ± between transmitter and receiver) between transmitters 0.3 m location error). Location error also increased with and receivers (linear mixed model: t = 2.094, P = 0.037). increasing transmission delay, but size of the effect Probability of detection for both suspended transmitters was also greater for fish that moved farther from the and transmitters in substrate was also negatively associ- spawning site (Figure 4; general linear model: t = 45.940, ated with distance between transmitter and receiver P = <0.001). In general, location error increased with (linear mixed model: t = -8.051, P <0.001). swim speed, but size of this effect varied with both dis- tance travelled from spawning site and transmission Pairing oviduct transmitters with larger transmitters to delay (Figure 4; general linear model: t = 33.6, P <0.001 track fish after spawning and t = 10.3, P <0.001 for interaction with distance trav- In light of the poor performance of the VPS array with elled from spawning site and transmission delay, respect- respect to positioning transmitters expelled into the sub- ively). Time errors ranged from 0.0 min to 569.2 min, strate, an alternative paired-transmitter approach was but in contrast to location error, decreased with distance tested via simulation of fish double-tagged with a small travelled from spawning site (Figure 4; general linear oviduct transmitter (to be expelled during spawning) model: t = -5.51, P <0.001). Mean (±SE) time error was and a larger transmitter that is retained after spawning. 6.4 ± 0.1 min for simulated trout in the low residency Spawning was inferred when the VPS array determined group and 19.2 ± 1.0 min for simulated trout in the high that the two transmitters separated in space (Figure 3). residency group. Like location error, time error was af- Simulations showed that double-tagging fish can improve fected by both distance travelled from spawning site and estimates of time and location of spawning compared swim speed. Time error increased with increasing trans- to use of only an oviduct transmitter. Using changepoint mission delay (Figure 4; general linear model: t = 4.99, analysis [29] on simulated fish tracks, the time and loca- P <0.001). The effect of swim speed on time error, how- tion at which the two transmitters separated (that is, ever, was more complicated and was dependent on dis- expulsion of oviduct tag during spawning) could be pre- tance travelled from spawning site (Figure 4; general dicted from the change in distance between the two trans- linear model: t = 4.00, P <0.001). mitters over time (Figure 4) or by the change in relative positioning probability between the two transmitters over Detecting change in positioning probability between paired time (Figure 5). Of the two methods, change in relative transmitters positioning probability performed best (Figures 4 and 5). Change in relative positioning probability proved to be On average, location error of estimated transmitter separ- more reliable than change in distance between transmit- ation was nearly two times greater (70.1 m vs. 36.7 m) ters for estimating the time and location of transmitter when using change in distance between transmitters than separation (that is, spawning). On average, point of trans- when using change in relative positioning probability. mitter separation was estimated to have occurred within Similarly, average time error of estimated transmitter sep- the first two to three fish transmitter transmissions after aration was more than four times greater (10.7 min vs. 2.4 actual separation. Unlike when using change in distance min) when using change in distance between transmitters to estimate transmitter separation, time error did not than when using change in relative positioning probability. vary with distance travelled from spawning site (Figure 5; The accuracy of estimates of time and location of trans- general linear model; t = -1.27, P = 0.21) or swim speed mitter separation also varied greatly with characteristics of (Figure 5; general linear model; t = -0.52, P = 0.60) - mean fish tracks (that is, distance travelled from spawning site time error ranged from 2.4 min to 2.5 min across all four (low, medium, medium-high, and high residency groups) residency groups (that is, low, medium, medium-high, and and swim speed) and transmitter delay. high), and from 2.3 min to 2.5 min across all levels of −1 swim speed (that is, 0.25, 0.50, 0.75, and 1.00 m · s ). As Detecting change in distance between paired tags expected, however, time error did increase significantly When using change in distance between transmitters to with increasing transmission delay (Figure 5; general linear estimate time and location of transmitter separation model: t = 17.01, P <0.001). Location error was more vari- (that is, spawning), location errors ranged from 0.1 m able than time error and was affected by distance travelled to 825.2 m. Location error was smaller for fish that from spawning site, transmission delay, and swim speed remained near the spawning site (high residency group) (Figure 5; general linear model: t = -16.11, P <0.001, than for fish that moved farther from the spawning site t = -6.34, P <0.001, and t = -6.34, P <0.001, respectively). Binder et al. Animal Biotelemetry 2014, 2:14 Page 6 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 3 Examples of simulated lake trout tracks (random walk models) and associated transmitter tracks for low-residency (that is, σ = 5; A, B) and high-residency (that is, σ = 90; C, D) lake trout. Simulations (6,400 in total) were used to evaluate the potential for use of a paired-transmitter experimental design to estimate the time and location of oviduct transmitter expulsion. Black tracks in (A) and (C) depict the true path of trout, while red and blue tracks in (B) and (D) depict the reconstructed path of the fish transmitter and oviduct transmitter, respectively, based on triangulated positions from a hypothetical positional array. Note the difference in scale between the two trout tracks. Location errors ranged from 0.2 m to 995.5 m and in- is the first to evaluate the potential for use of oviduct creased with increasing distance travelled from spawning transmitters combined with positional telemetry to ac- site. For example, mean (±SE) location error was 80.7 ± curately estimate time and location of spawning in 2.5 m for simulated trout in the low residency group and fishes. We tested three assumptions of using oviduct 11.1 ± 0.2 m for simulated trout in the high residency transmitters to determine time and location of egg de- group. Location error also increased with increasing trans- position in lake trout and found that: (1) most oviduct mission delay and swim speed, but in both cases the size transmitters remained within lake trout until spawning of the effect was dependent on distance travelled from (that is, few were expelled prematurely), (2) most ovi- spawning site (Figure 5; general linear model: t = 25.38, duct transmitters were expelled with eggs during manual P <0.001 and t = 20.88, P <0.001 for interaction with trans- stripping, and (3) oviduct transmitters were difficult to mission delay and swim speed, respectively). Effect size of position once expelled into substrate, but timing and lo- swim speed was also dependent on transmission delay cation of transmitter expulsion (indicative of spawning) (Figure 5; general linear model: t = 9.80, P <0.001). could be determined with reasonable accuracy in simula- tions using positional telemetry and a paired-transmitter design where one transmitter is expelled with eggs dur- Discussion ing spawning and the other remains in the fish. Use of oviduct-inserted transmitters to determine time Transmitter retention and subsequent artificial expul- and location of egg deposition represents an important sion rate in lake trout was comparable to those observed step forward in the use of electronic transmitters to in previous studies that have used oviduct transmitters in study the reproductive ecology of fishes [30]. Our study Binder et al. Animal Biotelemetry 2014, 2:14 Page 7 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 4 Evaluation of use of change in distance between fish transmitter and oviduct transmitter to estimate time and location of separation in a simulated paired-transmitter study. Panels display data for four different spawning site residency (that is, tendency to remain close to the spawning site) groups, which were simulated using different standard deviation of turn angle in random walk models (σ = 5, 20, 45, and 90 for low, medium, medium-high, and high residency groups, respectively). Top and bottom graphs in each panel display time error (min) and location error (m), respectively, for point estimates of transmitter separation calculated using changepoint analysis. Shading indicates the effective transmission delay (that is, mean time between transmissions) in seconds. other fishes. Eighty percent of females that ripened during likely they would have been recognized as such, and not our study retained the oviduct transmitter until spawning have confounded results. The cause of premature oviduct and then expelled it during manual stripping. Pierce et al. transmitter expulsion is not known. Expulsion may have [17] reported 60% and 90% expulsion in northern pike been a passive process, in that transmitters may have been (Esox lucius) and muskellunge (Esox masquinongy), re- expelled as a result of physical forces at play during nor- spectively, and Skovrind et al. [18] noted 92% expulsion in mal swimming behavior. Alternatively, the trout may have their study on European perch (Perca fluviatilis). Lake perceived the transmitter as a foreign object and actively trout have no true connection between the ovary and expelled it by muscular contractions or rubbing on the urogenital pore [27], which might reduce the likelihood bottom. Some fishes are capable of encapsulating trans- that oviduct transmitters would be expelled with the mitters, and expelling them through the intestine [31,32]. eggs. This condition (gymnovarian) is rare among teleost However, given the short time between tagging and pre- fishes, but occurs frequently among many of the most mature expulsions, these expulsions likely occurred via commonly studied fish species - that is, trout and salmon the urogenital pore. Either way, the portion of tagged (Salmonidae), freshwater eels (Anguillidae), and most trout that prematurely expelled oviduct transmitters non-teleost fishes (for example, sturgeons and lampreys) was small enough that, beyond a slight reduction in [28]. All premature expulsions of oviduct transmitters in sample size, impacts on data quality and interpretation our study occurred within the first 2 weeks after tagging, would have been minimal. well before the onset of the lake trout spawning season. If Most lake trout in our study that retained the trans- these premature expulsions had occurred in a field study, mitter until the end of the spawning period expelled the Binder et al. Animal Biotelemetry 2014, 2:14 Page 8 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 5 Evaluation of use of change in relative position probability (that is, change in ratio of time since last position) to estimate time and location of separation in a simulated paired-transmitter study. Panels display data for four different spawning site residency (that is, tendency to remain close to the spawning site) groups, which were simulated using different standard deviation of turn angle in random walk models (σ = 5, 20, 45, and 90 for low, medium, medium-high, and high residency groups, respectively). Top and bottom graphs in each panel display time error (min) and location error (m), respectively, for point estimates of transmitter separation calculated using changepoint analysis. Shading indicates the effective transmission delay (that is, mean time between transmissions) in seconds. oviduct transmitter with eggs when manually stripped. blockage [35] and maximize likelihood of transmitter However, use of manual stripping as a surrogate for egg expulsion during spawning. deposition during spawning is a source of bias in our Poor ability to detect and position transmitters depos- study. Although we did our best to approximate normal ited into coarse substrate reduces the usefulness of ex- egg deposition by holding the trout in a natural spawn- pelled oviduct transmitters for determining timing and ing posture (that is, head and tail flexed slightly up- location of spawning. Reef environments with complex wards) [33,34] during stripping, we do not know to what topography pose many challenges to acoustic telemetry, extent physical forces applied to the eggs and oviduct including high levels of noise, signal bending, echoes, transmitter during manual stripping represent those and signal blocking [36,37]. Line of sight between trans- present during natural egg deposition. If force applied mitter and receiver is critical for positioning, so receiver during manual stripping was greater than would be placement in the water column, relative to transmitters, present during spawning, our results could overestimate is an important consideration [37,38]. Line of sight may transmitter expulsion rate. However, because transmit- have improved if we had located receivers nearer to the ters were easily expelled, our judgment was that reason- surface, but improvement in line of sight may have able pressure was applied. Transmitter size (V6 or V7) come at the expense of increased noise due to wind and had no effect on transmitter retention or expulsion rates waves [36,39] and possibly lower position accuracy due in our study. Nonetheless, use of the smallest available to receiver movement. Line of sight may also have been transmitter that suits the study system and objectives is improved by using more receivers and spacing them recommended to minimize the possibility of oviduct closer together. Binder et al. Animal Biotelemetry 2014, 2:14 Page 9 of 14 http://www.animalbiotelemetry.com/content/2/1/14 In instances where oviduct transmitters are difficult function of transmitter delay and positioning probability, to position once expelled into substrate, a paired- and is independent of the behavior of the fish, except transmitter design appeared to accurately estimate time in cases where the fish behaves in a manner that reduces and location of spawning. Our decision to use change- the positioning probability of the transmitter (for ex- point analysis [29] to determine when the two transmit- ample, takes refuge in vegetation [40]). ters separated (that is, began to track differently) was Transmitters with short delay between transmissions based on a need for objectivity and statistical defensibil- produced more accurate estimates of time and location ity. Initially, we attempted to use rule-of-thumb bio- of transmitter separation, but the effect size of transmit- logical filters based on expected behavior of the fish (for ter delay on accuracy of estimates of time and location example, point at which the distance between fish trans- of transmitter separation was dependent on the behavior mitter and oviduct transmitter was greater than would of the fish. A tradeoff exists between transmitter delay be expected given the mean swimming speed of the and the number of transmitters an acoustic array can ac- fish). However, results were subjective, and for many commodate at one time. If an acoustic receiver detects a trout tracks, rule-of-thumb filters failed even to identify signal from a second transmitter while decoding the sig- a point of transmitter separation (T Binder, unpublished nal from the first, the receiver cannot separate the two results). In contrast, changepoint analysis produced a signals and the result is a code collision, which results in point estimate of transmitter separation for all 6,400 neither transmission being decoded (that is, no detec- trout tracks. One advantage of the changepoint proced- tion) [41]. The probability of code collisions increases as ure was the change in positioning probability of expelled transmitter delay decreases and number of transmitters oviduct transmitters could be used to our advantage within detection range of the receiver increases [36]. and, in our simulation exercise, change in relative posi- The impact of long transmitter delays on point estimates tioning probability turned out to be the most accurate of transmitter separation was greater in fish with low method for estimating the time and location of transmit- spawning site residency, indicating that use of a small ter separation. Ultimately, choice of changepoint method number of transmitters with a short delay is preferred in (that is, change in distance between transmitters vs. these cases. In contrast, because transmitter delay had change in relative positioning probability of the trans- less impact on estimates of time and location of trans- mitters) will depend on how well the acoustic array can mitter separation in fish that showed high spawning site position oviduct transmitters after expulsion. For ex- residency, one might choose a long transmitter delay ample, in our simulated tracks, positioning probability of (for example, 5 min) to reduce the frequency of trans- oviduct transmitters decreased 90% after expulsion mitter code collisions and allow for more transmitters to reflecting the poor ability of the acoustic array to pos- be deployed than if a short transmitter delay was chosen. ition transmitters in the substrate during our field study. Ultimately, appropriate choice of transmitter delay and However, in situations where positioning probability of number of transmitters deployed will depend on trade- oviduct transmitters decreases only a small amount after offs between the desired precision and accuracy of expulsion, change in distance between transmitters spawning time and location estimates and the behavior would likely produce more accurate point estimates of of the fish. transmitter separation than change in relative position- ing probability. Conclusions Accuracy of estimated time and location of transmitter Overall, our results suggest that use of oviduct transmit- separation varied greatly with distance travelled from ters with positional telemetry represents a promising spawning site (that is, spawning site residency) and swim new technique for studying spawning behaviors in lake speed of the fish. In general, time error decreased and trout. Despite the absence of a direct connection be- location error increased as distance travelled increased tween the ovary and urogenital pore in lake trout, the from the spawning site. This relationship occurred be- majority of oviduct transmitters implanted into the body cause fish that travelled further from the spawning site cavity near the oviduct remained within the fish until tended to move a greater linear distance from the site of spawning and were then expelled with eggs during man- transmitter separation during each transmission interval ual stripping. Use of a traditional single-transmitter ovi- than fish that tended to remain closer to the spawning duct tagging study design [16-18] in conjunction with site. Time error in estimates of transmitter separation positional telemetry would not be successful in lake calculated using the change in relative positioning prob- trout when spawning occurs over rough stony substrates ability method was an exception to the above-stated because oviduct transmitters would be difficult to pos- trend, in that time error was resistant to differences in ition. Most transmitters in the substrate continued to be swimming speed and spawning site residency. The cause detected on some receivers, but at a low rate. Thus, de- for the difference was the fact that positioning rate is a pending on the size of the study area and the required Binder et al. Animal Biotelemetry 2014, 2:14 Page 10 of 14 http://www.animalbiotelemetry.com/content/2/1/14 resolution of estimated time and location of spawning, 13°C) water from Lake Huron. Trout were offered a diet manual tracking would be an alternative option to deter- of 13 mm brood stock fish pellets until mid-September, mine location. However, greater effort would be required when the trout ceased feeding. to locate transmitters using manual tracking than if the Trout were tagged with oviduct transmitters on 17 VPS array could be used successfully. We found that use September 2013. To test the hypothesis that insertion of of a paired-transmitter experimental design in conjunc- transmitters through the oviduct does not affect retention tion with positional telemetry would be a better solution and expulsion of transmitters, relative to those that have when oviduct transmitters are expelled into substrates been surgically implanted, we inserted transmitters via the (for example, cobble or mud) that act as barriers to oviduct in half the fish (n = 24; hereafter referred to as acoustic signal reception. In our simulations, we were ‘inserted’) and surgically implanted transmitters just anter- able to use the decreased positioning probability of ex- ior to the genital pore in the rest (n = 24; hereafter referred pelled oviduct transmitters to estimate time and location to as ‘implanted’). Half of each tagging group received V6 of transmitter separation (that is, spawning) with reason- dummy transmitters (VEMCO, Nova Scotia, Canada; able accuracy. 16.5 × 6 mm, 1 g in air) and the remaining trout received Use of oviduct transmitters and positional telemetry to V7 dummy transmitters (VEMCO; 20 × 7 mm, 1.6 g in determine time and location of spawning has the poten- air). Dummy transmitters contained a passive integrated tial to be a useful tool for studying reproductive ecology transponder (PIT) tag so that individual fish could be across a broad range of fish species. Beyond the obvious identified. All trout were anesthetized in 40 L of clove oil benefits of reduced manual labor and increased preci- solution (0.8 mL/L of 1:9 clove oil:ethanol solution) and sion and accuracy of estimated spawning locations [14], then placed dorsal side down on a v-board prior to under- remote monitoring of spawning behavior via positional going the tagging procedure. Gills were perfused with aer- telemetry may offer opportunities to study spawning in ated lake water throughout the procedures. In trout fishes that have previously proven inaccessible [15]. For receiving the transmitter via oviduct-insertion, transmit- example, observations of lake trout spawning have been ters were gently inserted through the oviduct and pushed rare because it occurs mainly at night, during a time of approximately 5 cm deep using a 10 cm piece of sterilized year when weather patterns are often unpredictable and 6.5 mm diameter Tygon tubing. The procedure was per- dangerous for field work [26]. Nonetheless, effective use formed by a single researcher and took, on average, 20 to of oviduct transmitters to determine time and location 30 s. In trout receiving surgically implanted transmitters, a of spawning depends on the assumptions that oviduct small incision (1.5 to 2.0 cm) was made slightly off the transmitters will remain within the fish until spawning ventral midline, approximately 5 cm anterior to the genital and then be expelled with eggs at the spawning site, pore. After the transmitter was inserted into the body cav- and that oviduct transmitters can be located accurately ity, the incision was closed using two simple, interrupted after expulsion. Failure to meet these assumptions sutures (Ethicon, Inc.; 3-0 polydioxanone monofilament). will, at best, reduce sample size (for example, if transmit- Surgeries were performed by a single surgeon and took, ters are not expelled), and at worst, could result in erro- on average, 2 to 3 min. Following tagging, trout were neous conclusions (for example, if transmitters are returned to the holding tank to recover from anesthesia. expelled prematurely). Therefore, regardless of whether The holding tank was searched daily, September to researchers use positional telemetry or manual tracking, November, for expelled oviduct transmitters, as well as we recommend that feasibility studies be conducted expelled loose eggs that would indicate trout were ovu- to test assumptions prior to using oviduct transmitters lating. During daily searches, the fish were assessed for in a new species. signs of odd behavior (for example, failure to respond to the researcher’s movements or inability to maintain Methods equilibrium) and signs of poor health (for example, der- Retention and expulsion of oviduct-inserted transmitters mal discoloration or presence of fungus). If a transmitter We tested the assumptions that oviduct-inserted trans- was found, the associated PIT tag number, date, and mitters are retained until spawning and then expelled time were recorded. Lake trout did not spawn naturally with the eggs (Assumptions 1 and 2) in a laboratory under laboratory conditions, so manual stripping was tagging study. Forty-eight hatchery-raised, female lake used as a surrogate to determine if transmitters would trout (Seneca strain) from Sullivan Creek National be expelled with eggs. Our assumption was that, while Fish Hatchery (Brimley, MI, USA) were transferred not identical to natural spawning, the manual stripping to Hammond Bay Biological Station (Millersburg, MI, procedure was representative of the processes that occur naturally during egg expulsion. Manual stripping was USA) on 6 August 2013. The trout were housed in a 6,800 L rectangular (6.7 × 1.7 × 0.9 m) holding tank, sup- conducted on 25 November 2013, to coincide with the plied with chilled (water temperature between 7°C and end of the Lake Huron lake trout spawning period. The Binder et al. Animal Biotelemetry 2014, 2:14 Page 11 of 14 http://www.animalbiotelemetry.com/content/2/1/14 ovulating trout were removed from the holding tank one transmitters. One of each transmitter type was sus- at a time and anesthetized as before to prevent excessive pended 1 m above the substrate to act as a control, stress. Anesthetized trout were held dorsal side up with representing a transmitter that was still within the trout. heads and tails pointed slightly upward, as occurs during The two remaining transmitters of each type were natural egg deposition [33,34]. Their bellies were then placed in the cobble substrate (5 to 20 cm in diameter) massaged from anterior to posterior until the oviduct by scuba divers to represent transmitters that had been tag was expelled, or the trout stopped expelling eggs. expelled during spawning. Divers ensured that the trans- Differences in retention of oviduct transmitters were mitters were placed in the space between rocks, as this compared using time-to-event analysis (R package ‘sur- is most likely how the negatively-buoyant transmitters vival’; α = 0.05), which allowed us to not only compare the would settle when expelled naturally during a spawning proportion of individuals that retained the transmitter, event. Test transmitters were left in place for 20 min but also the temporal pattern of premature expulsion. (allowing for a total of 20 transmissions from each trans- The proportion of lake trout that expelled the oviduct mitter) before being retrieved and moved to the next test transmitter during manual stripping was compared among transmitter site. At the end of the study, receivers were groups using Fisher’s Exact Test (R package ‘stats’; retrieved and downloaded, and log files from the re- α = 0.05), and the number of stripping motions required ceivers were sent to VEMCO for postprocessing using to expel the transmitters was compared using a t-test their proprietary hyperbolic positioning algorithms [42]. (R package ‘stats’; α =0.05). Probabilities of detecting and positioning control and oviduct transmitters were compared across transmitter Estimating time and location of spawning using oviduct type using linear mixed models (R package ‘lme4’; α = transmitters 0.05). Treatment (control transmitter or oviduct transmit- Inferences from a single oviduct transmitter in each fish ter) and transmitter type (V6 or V7) were fixed effects and The ability of the VEMCO Positioning System (VPS) to transmitter identification code and array number were accurately position expelled oviduct acoustic transmit- random effects. Variables related to poor detectability of ters in substrate (Assumption 3) was investigated in a transmitters in the substrate were also investigated using a field study at known lake trout spawning reefs in the linear mixed model (R package ‘lme4’; α =0.05). Distance Drummond Island Lake Trout Refuge, northern Lake between transmitter and receiver, total area of obstruction Huron, North America (Figure 1). Three small VPS ar- between transmitter and receiver, tag type (V6 or V7), and rays (approximately 100 × 100 m) were constructed on treatment (control transmitter or oviduct transmitter) three separate near-shore (0.6 to 1.3 km from shoreline) were fixed effects; transmitter identification code and re- spawning sites. The three sites ranged in depth from ap- ceiver serial number were random effects. Area of obstruc- proximately 1.5 to 7 m and consisted of rocky substrate. tion, the total cross-sectional area of substrate located at The approximate 100 m spacing of receivers was based higher altitude than the line of sight between transmitter on recommendations from the manufacturer (VEMCO) and receiver, was calculated using high-resolution (1 m on the maximum detection range of the less powerful horizontal, 10 cm vertical) multibeam sonar bathymetry 180 kHz transmitters. Each array consisted of four (Seabat 7101 system, Teledyne RESON Inc.). receiver locations and a centrally-located synchronization transmitter site, which held transmitters used to synchronize Pairing oviduct transmitters with larger transmitters to the clocks on receivers. Because the V6 and V7 acoustic track fish after spawning transmitters transmit at different frequencies, we installed The feasibility of using transmitter separation in a paired- two acoustic receivers (VR2W-69 kHz and VR2W-180 transmitter design to accurately estimate time and loca- kHz) at each receiver location on a single mooring. The tion of oviduct transmitter expulsion (Assumption 3) was upper receiver was positioned upside down so that the tested using random walk models to simulate a variety of top-mounted hydrophones on the two receivers were possible lake trout spawning behaviors [43]. The paired- located as closely as possible to one another. transmitter design called for each simulated trout to be Five test transmitter locations were chosen for each implanted with two transmitters; a small oviduct transmit- VPS array. The first was at the center of the array. The ter that would be expelled with the eggs during spawning, remaining four were scattered throughout the array so and a larger fish transmitter that would remain within that they were located at random distances from the the fish after spawning. Time and location at which the center of the array. Three transmitters of each type (V6 two transmitters began to behave differently (point of and V7) were deployed at each site. Transmitters were transmitter separation) was used to estimate time and programmed to transmit with a fixed delay of 1 min and location of spawning. initialization of transmitters was staggered by 20 s to en- Each random walk model (referred to hereafter as ‘trout sure that no signal code collisions would occur between track’) consisted of 43,200 steps, each 1 m in length (total Binder et al. Animal Biotelemetry 2014, 2:14 Page 12 of 14 http://www.animalbiotelemetry.com/content/2/1/14 track = 43.2 km). Steps of the trout track were straight positioning the oviduct transmitter in the substrate from lines, with headings sampled from a random distribution this point forward was reduced to 10% (that is, only with mean turn angle equal to 0 and standard deviation σ. every 10th transmission could be positioned). For simplicity, the models assumed that trout remained In total, 100 trout tracks were simulated for each com- within the VPS array for the entire time series. However, bination of σ, swim speed and transmission delay (6,400 in practice, the only requirements are that the oviduct trout tracks in total). We used the R statistical package transmitter be expelled in the array and then the trout ‘changepoint’ to estimate time and location of transmit- remain within the array long enough to determine that ter separation for each simulated trout track [29]. Chan- the two transmitters had separated. Numerous possible gepoint analysis, an emerging tool in movement ecology spawning behaviors were simulated to determine how dif- [44,45], uses maximum likelihood methods to determine ferent behaviors, spanning a range of species, could affect points in a time series where the mean, variance, or mean ability to accurately estimate time and location of spawn- and variance of a response variable change. In our ana- ing using a paired-transmitter design. To determine how lyses, we tested for a single changepoint, and visual in- degree of spawning site residency affected our ability to spection of results indicated that the ‘changepoint’ accurately estimate the time and location of transmitter function ‘cpt.meanvar’ (that is, changepoint of mean and separation (that is, spawning location) four different values variance; number of changepoints (Q) = 1, test.stat = ‘Nor- were used for σ (5, 20, 45, and 90; hereafter referred to as mal’) was best able to identify when the two transmitters low, medium, medium-high, and high residency groups, separated. respectively). Distance travelled from spawning site was We tested two separate methods for estimating time inversely related to σ (Figure 3A and C). For example, sim- and location of transmitter separation. The first method ulated trout with σ = 5 (low residency group) travelled, on was to test for a change in the distance between the two average, 17 times further from the spawning site than sim- transmitters. For each time in the fish transmitter track, ulated lake trout with σ = 90 (high residency group). Mean we used linear interpolation to estimate the location of (±SE) distance between all positions for an individual and the oviduct transmitter at that time. The linear distance its actual spawning site was 1,976 ± 536 m, 493 ± 141 m, between the fish transmitter and oviduct transmitter was 223 ± 63 m, and 116 ± 33 m for simulated tracks with σ = calculated and changepoint analysis was used on those 5, 20, 45, and 90, respectively. We also varied mean swim values to determine when in the time series the mean −1 speed of individual lake trout between 0.25 m · s and and variance of the distance between the two transmit- −1 −1 1.00 m · s (0.25, 0.50, 0.75, or 1.00 m · s )to cover a ters changed. The second method took advantage of the range of realistic swimming speeds. reduction in positioning probability of expelled oviduct Transmitter positions, such as those that would be tri- transmitters that we observed in our field trial. For each angulated using the VPS system, were overlaid on each time in the fish transmitter track, we calculated the ratio trout track. The result was two transmitter tracks, one of the time since the previous position for the oviduct for the fish transmitter and another for oviduct transmit- transmitter and the time since the previous position for ter (Figure 3B and D). Transmitters were programmed the fish transmitter (that is, time since last oviduct trans- to transmit with mean transmission delay of 60, 180, mitter position / time since last fish transmitter pos- 300, or 420 s. The exact delay of each transmission ition). Changepoint analysis was then used to determine was sampled from a uniform distribution with bounds when in the time series the mean and variance of that delay - delay/2 and delay + delay/2. These delays were ratio changed. Change in ratio of positioning probability assumed to represent effective transmitter delays, which was used rather than simply testing for a change in incorporated not only the nominal delay of the transmit- the positioning probability of the oviduct transmitter it- ters as programmed by the manufacturer, but also any self because positioning probability can vary naturally due to environmental variability. The ratio method can variation due to environmental variables like wave noise and signal code collisions with other nearby transmit- accommodate this variation because any environmental ters. For the sake of realism, we incorporated positioning variation should affect both transmitters in a similar man- ner. Both methods outlined above can return estimates of error into each of the transmitter positions by sampling from a bivariate normal distribution with mean equal to time and location of transmitter separation if the oviduct the true location (x and y coordinates) of the trout at transmitter continues to be positioned (even at a lower rate). However, only the latter method (change in relative time t (based on trout track) and a standard deviation of 15 m. At the halfway point of each trout track, the ovi- positioning probability) can return an estimate of time duct transmitter was expelled into the substrate as a and location of transmitter separation if the oviduct trans- mitter ceases to be positioned after expulsion. simulated spawning event. Based on our field test of the ability of the VPS system to position transmitters ex- The output of the changepoint analysis was an esti- pelled into the substrate, we assumed the probability of mate of the location in the fish transmitter track time Binder et al. Animal Biotelemetry 2014, 2:14 Page 13 of 14 http://www.animalbiotelemetry.com/content/2/1/14 series where the two transmitters separated (that is, 2. Vitousek PM, D’Antonio CM, Loope LL, Rejmánek M, Westbrooks R: Introduced species: a significant component of human-caused global began to behave differently). Two metrics were calcu- change. New Zeal J Ecol 1997, 21:1–16. lated to evaluate the ability of the paired-transmitter 3. Hermoso V, Clavero M, Blanco-Garrido F, Prenda J: Invasive species and design to accurately estimate the time and location of habitat degradation in Iberian streams: an analysis of their role in freshwater fish diversity loss. Ecol Appl 2011, 21:175–188. oviduct expulsion: (1) difference in time (hereafter ‘time 4. Eby LA, Crowder LB, McClellen CM, Peterson CH, Powers MJ: Habitat error’) between estimated transmitter separation and degradation from intermittent hypoxia: impacts on demersal fishes. true time of spawning (that is, the midway point of Mar Ecol Prog Ser 2005, 291:249–261. 5. Casatti L, Langeani F, Ferreira CP: Effects of physical habitat degradation the trout track), and (2) distance between estimated lo- on the stream fish assemblage structure in a pasture region. Environ cation of transmitter separation and true location of Manage 2006, 38:974–982. spawning (hereafter ‘location error’). A series of general 6. Wilson SK, Fisher R, Pratchett MS, Graham NAJ, Dulvy NK, Turner RA, Cakacaka A, Polunin NVC, Rushton SP: Exploitation and habitat linear models (R package ‘stats’; α = 0.05) were used to degradation as agents of change within coral reef fish communities. determine the effects of σ, transmission delay and swim Glob Change Biol 2008, 14:2796–2809. speed on time and location error of estimated transmit- 7. Roessig JM, Woodley CM, Cech JJ Jr, Hansen LJ: Effects of global climate change on marine and estuarine fishes and fisheries. Rev Fish Biol Fisher ter separation. Effects were modeled separately using 2004, 14:251–275. both the change in distance between transmitter method 8. Ficke AD, Myrick CA, Hansen LJ: Potential impacts of global climate and change in relative positioning probability method. change on freshwater fisheries. Rev Fish Biol Fisher 2007, 17:581–613. 9. Farrell AP, Hinch SG, Cooke SJ, Patterson DA, Crossin GT, Lapointe M, Because the effect of spawning site residency on accur- Mathes MT: Pacific salmon in hot water: applying aerobic scope models acy of transmitter separation estimates was of interest, and biotelemetry to predict the success of spawning migrations. the continuous variable mean distance travelled from Physiol Biochem Zool 2008, 81:697–709. 10. Goodyear CS, Edsall TA, Ormsby Dempsey DM, Moss GD, Polanski PE: Atlas of spawning site (that is, mean distance between all posi- the spawning and nursery areas of Great Lakes fishes, Volume 5. Washington, tions in a given trout track and its true spawning loca- DC: Lake Huron. U.S. Fish and Wildlife Service; 1982. FWS/OBS-82/52. tion) replaced σ in the models. Interactions between 11. Eiler JH, Nelson BD, Bradshaw RF: Riverine spawning by sockeye salmon in the Taku River, Alaska and British Columbia. Trans Amer Fish Soc 1992, 121:701–708. distance travelled from spawning site and transmission 12. Bégout Anras ML, Cooley PM, Bodaly RA, Anras L, Fudge RJP: Movement delay, distance travelled from spawning site and swim and habitat use by whitefish during spawning in a Boreal lake: speed, and transmission delay and swim speed were also integrating acoustic telemetry and geographic information systems. Trans Amer Fish Soc 1999, 128:939–952. included in the models. 13. Flavelle LS, Ridgway MS, Middel TA, McKinley RS: Integration of acoustic Competing interests telemetry and GIS to identify potential spawning areas for lake trout The authors declare that they have no competing interests. (Salvelinus namaycush). Hydrobiologia 2002, 483:137–146. 14. Espinoza M, Farrugia TJ, Webber DM, Smith F, Lowe CG: Testing a new acoustic telemetry technique to quantify long-term, fine-scale movements Authors’ contributions of aquatic organisms. Fish Res 2011, 108:364–371. TB participated in all aspects of study conception, experimental design, data 15. Biesinger Z, Bolker BM, Marcinek D, Grothues TM, Dobarro JA, Lindberg WJ: acquisition and analysis, and drafted the manuscript. CH participated in data Testing autonomous acoustic telemetry positioning system for fine-scale analysis and manuscript preparation. SM participated in study conception, space use in marine animals. J Exp Mar Biol Ecol 2013, 448:46–56. data collection, and manuscript preparation. HT participated in data 16. Pierce RB: Oviduct insertion of radio transmitters as a means of locating collection, analysis, and manuscript preparation. CK participated in study northern pike spawning habitat. N Am J Fish Manage 2004, 24:244–248. conception, experimental design, and manuscript preparation. All authors read and approved the final manuscript. 17. Pierce RB, Younk JA, Tomcko CM: Expulsion of miniature radio transmitters along with eggs of muskellunge and northern pike - a new method for locating critical spawning habitat. Environ Biol Fish 2007, 79:99–109. Acknowledgements 18. Skovrind M, Christensen EAF, Carl H, Jacobsen L, Møller PR: Marine We thank E. Larson, L. Lesmeister, D. Operhall, and B. Lamoreux for their spawning sites of perch Perca fluviatilis revealed by oviduct-inserted technical assistance in the laboratory and field. This work was funded by the acoustic transmitters. Aquat Biol 2013, 19:201–206. Great Lakes Fishery Commission by way of Great Lakes Restoration Initiative 19. Andrews KS, Tolimieri N, Williams GD, Samhouri JF, Harvey CJ, Levin PS: appropriations (GL-00E23010-3). This paper is contribution 5 of the Great Comparison of fine-scale acoustic monitoring systems using home Lakes Acoustic Telemetry Observation System (GLATOS) and contribution range size in demersal fish. Mar Biol 2011, 158:2377–2387. number 1875 of the USGS Great Lakes Science Center. Any use of trade, 20. Hansen MJ: Lake trout in the Great Lakes: Basinwide stock collapse and product, or firm names is for descriptive purposes only and does not imply binational restoration. In Great Lakes fisheries policy and management: endorsement by the U.S. Government. a binational perspective. Edited by Taylor WW, Ferreri CP. East Lansing, MI: Michigan State University Press; 1999:417–454. Author details Department of Fisheries and Wildlife, Michigan State University, Hammond 21. Muir AM, Krueger CC, Hansen MJ: Re-establishing lake trout in the Bay Biological Station, 11188 Ray Rd, Millersburg, MI 49759, USA. United Laurentian Great Lakes: past, present, and future. In Great Lakes fisheries States Geological Survey, Hammond Bay Biological Station, 11188 Ray Rd, policy and management: a binational perspective. 2nd edition. Edited by Millersburg, MI 49759, USA. Center for Systems Integration and Taylor WW, Lynch AJ, Leonard NJ. East Lansing, MI: Michigan State Sustainability, Michigan State University, 1405 South Harrison Road, 115 University Press; 2013:533–588. Manly Miles Building, East Lansing, MI 48823, USA. 22. Ruzycki JR, Beauchamp DA, Yule DL: Effects of introduced lake trout on native cutthroat trout in Yellowstone Lake. Ecol Appl 2003, 13:23–37. Received: 29 May 2014 Accepted: 11 August 2014 23. Koel TM, Bigelow PE, Doepke PD, Ertel BD, Mahony DL: Nonnative lake Published: 4 September 2014 trout results in Yellowstone cutthroat decline and impacts to bears and anglers. Fisheries 2005, 30:10–19. References 24. Martinez PJ, Bigelow PE, Deleray MA, Fredenberg WA, Hansen BS, Horner 1. Mills EL, Leach JH, Carlton JT, Secor CL: Exotic species and the integrity of NJ, Lehr SK, Schneidervin RW, Tolentino SA, Viola AE: Western lake trout the Great Lakes. Bioscience 1994, 44:666–676. woes. Fisheries 2009, 34:424–442. Binder et al. Animal Biotelemetry 2014, 2:14 Page 14 of 14 http://www.animalbiotelemetry.com/content/2/1/14 25. Marsden JE, Casselman JM, Edsall TA, Elliott RF, Fitzsimons JD, Horn WH, Manny BA, McAughey SC, Sly PG, Swanson BL: Lake trout spawning habitat in the Great Lakes - a review of current knowledge. J Great Lakes Res 1995, 21(Suppl 1):487–497. 26. Muir AM, Blackie CT, Marsden JE, Krueger CC: Lake charr Salvelinus namaycush spawning behaviour: new field observations and a review of current knowledge. Rev Fish Biol Fisher 2012, 22:575–593. 27. Henderson NE: The urinary and genital systems of trout. J Fish Res Board Can 1967, 24:447–449. 28. Ganguly S: Reproductive biology in fishes and its diversified physiological mechanisms. Int J Sci Env Tech 2013, 2:401–403. 29. Killick R, Eckley I: changepoint: an R package for changepoint analysis. R package version1.1.1; 2014 [http://CRAN.R-project.org/package=changepoint]. 30. Crossin GT, Cooke SJ, Goldbogen JA, Phillips RA: Tracking fitness in marine vertebrates: a review of current knowledge and opportunities for future research. Mar Ecol Prog Ser 2014, 496:1–17. 31. Chisholm IM, Hubert WA: Expulsion of dummy transmitters by rainbow trout. Trans Amer Fish Soc 1985, 114:766–767. 32. Marty GD, Summerfelt RC: Pathways and mechanisms for expulsion of surgically implanted dummy transmitters from channel catfish. Trans Amer Fish Soc 1986, 115:577–589. 33. Esteve M, McLennan DA, Gunn J: Lake trout (Salvelinus namaycush) spawning behaviour: the evolution of a new female strategy. Environ Biol Fish 2008, 83:69–76. 34. Binder TR, Thompson HT, Muir AM, Riley SC, Marsden JE, Bronte CR, Krueger CC: New insight into the spawning behavior of lake trout, Salvelinus namaycush, from a recovering population in the Laurentian Great Lakes. Environ Biol Fish 2014, doi:10.1007/s10641-014-0247-6. 35. Peake S, McKinley RS, Beddow TA, Marmulla G: New procedure for radio transmitter attachment: oviduct insertion. N Amer J Fish Manage 1997, 17:757–762. 36. Simpfendorfer CA, Heupel MR, Collins AB: Variation in the performance of acoustic receivers and its implication for positioning algorithms in a riverine setting. Can J Fish Aquat Sci 2008, 65:482–492. 37. Welsh JQ, Fox RJ, Webber DM, Bellwood DR: Performance of remote acoustic receivers within a coral reef habitat: implications for array design. Coral Reefs 2012, 31:693–702. 38. Heupel MR, Semmens JM, Hobday AJ: Automated acoustic tracking of aquatic animals: scales, design, and deployment of listening station arrays. Mar Freshwater Res 2006, 57:1–13. 39. Clements S, Jepsen D, Karnowski M, Schreck CB: Optimization of an acoustic telemetry array for detecting transmitter-implanted fish. N Amer J Fish Manage 2005, 25:429–436. 40. Zamora JCL, García-Berthou E: Preliminary telemetry data on the movement patterns and habitat use of European catfish (Silurus glanis) in a reservoir of the River Ebro, Spain. Ecol Freshwater Fish 2007, 16:450–456. 41. Voegeli FA, Lacroix GL, Anderson JM: Development of miniature pingers for tracking Atlantic salmon smolts at sea. Hydrobiologia 1998, 371/ 372:35–46. 42. Smith F: Understanding HPE in the VEMCO positioning system (VPS). Halifax, NS: VEMCO; 2013. 43. Codling EA, Plank MJ, Benhamou S: Random walk models in biology. J Roy Soc Interface 2008, 5:813–834. 44. Gurarie E, Andrews RD, Laidre KL: A novel method for identifying behavioural changes in animal movement data. Ecol Lett 2009, 12:395–408. 45. Madon B, Hingrat Y: Deciphering behavioral changes in animal movement with a “multiple change point algorithm-classification tree” framework. Front Ecol Evol 2014, 2:1–9. Submit your next manuscript to BioMed Central and take full advantage of: doi:10.1186/2050-3385-2-14 Cite this article as: Binder et al.: Use of oviduct-inserted acoustic transmitters and positional telemetry to estimate timing and location of • Convenient online submission spawning: a feasibility study in lake trout, Salvelinus namaycush. • Thorough peer review Animal Biotelemetry 2014 2:14. • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Animal Biotelemetry Springer Journals

Use of oviduct-inserted acoustic transmitters and positional telemetry to estimate timing and location of spawning: a feasibility study in lake trout, Salvelinus namaycush

Download PDF
14 pages

Loading next page...
 
/lp/springer-journals/use-of-oviduct-inserted-acoustic-transmitters-and-positional-telemetry-21XHtpDJAE

References (44)

Publisher
Springer Journals
Copyright
Copyright © 2014 by Binder et al.; licensee BioMed Central Ltd.
Subject
Life Sciences; Animal Systematics/Taxonomy/Biogeography
eISSN
2050-3385
DOI
10.1186/2050-3385-2-14
Publisher site
See Article on Publisher Site

Abstract

Background: Oviduct-inserted transmitters have shown promise for determining precise location of spawning in fishes. Use of traditional manual tracking to locate expelled oviduct transmitters is laborious and accurate estimates of time of transmitter expulsion require frequent surveys. We tested the feasibility of using oviduct-inserted transmitters with positional telemetry to estimate time and location of spawning in lake trout (Salvelinus namaycush). Three assumptions were tested: (1) oviduct transmitters remain within fish until spawning, (2) oviduct transmitters are expelled with the eggs during spawning, and (3) time and location of oviduct transmitter expulsion can be accurately determined. Results: In the laboratory, 39 of 44 (89%) lake trout retained an oviduct transmitter until end of the spawning period and all premature transmitter expulsions occurred before maturation. Natural spawning in the laboratory was not feasible; however, of 35 ripe trout that retained transmitters, 31 (89%) expelled their transmitter with eggs when subjected to manual stripping. Ability to position transmitters with a telemetry array at known spawning sites in Lake Huron (North America) was poor when oviduct transmitters were placed in the substrate compared to transmitters suspended 1 m above substrate - 78% of transmitters in substrate could not be positioned. However, in simulations, time and location of spawning were determined with reasonable accuracy by double-tagging trout with one transmitter that is expelled with the eggs during spawning while another transmitter remains in the fish. Accuracy of estimated time and location of transmitter spatial separation varied with distance traveled from spawning site and swimming speed, and was dependent on transmission delay. Conclusions: Our results satisfied the three assumptions of oviduct tagging and suggested that oviduct transmitters can be used with positional telemetry to estimate time and location of spawning in lake trout and other species. In situations where oviduct transmitters may be difficult to position once expelled into substrate, pairing oviduct transmitters with a normal-sized fish transmitter that remains in the fish is recommended, with spawning inferred when the two tags separate in space. Optimal transmitter delay will depend on expected degree of spawning site residency and swim speed. Keywords: Acoustic telemetry, Changepoint analysis, Gymnovarian, Spawning behavior, Transmitter retention, Detection probability * Correspondence: tr.binder@gmail.com Department of Fisheries and Wildlife, Michigan State University, Hammond Bay Biological Station, 11188 Ray Rd, Millersburg, MI 49759, USA Full list of author information is available at the end of the article © 2014 Binder et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Binder et al. Animal Biotelemetry 2014, 2:14 Page 2 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Background The main disadvantage of using oviduct transmitters An understanding of the reproductive ecology of fishes in conjunction with mobile telemetry is the amount of is becoming increasingly important in light of wide- labor required to frequently manual track transmitters spread ecosystem alterations, such as introduction of [14,15]. The effort is compounded by the fact that ac- exotic species [1-3], critical habitat degradation [4-6], curacy of estimated time and location of transmitter and climate change [7-9]. Fishery managers need to have expulsion will be proportional to the amount of effort an understanding of spawning habitat, and the physio- expended in locating transmitters - for example, accur- chemical and biological variables that promote success- acy of the estimated time of spawning will be greater ful reproduction. Historically, spawning sites have been if manual tracking is done daily than if it is done once identified by surveying (for example, visually, by fishing, per week. Positional telemetry using fixed receivers is or hydroacoustics) for aggregations of ripe adult fish an attractive alternative to mobile telemetry when the [10], which have then allowed researchers to focus on general spawning area (scale of kilometers) is known specific areas and to deduce spawning locations at scales because positional telemetry provides nearly continuous of hundreds of meters to kilometers. More recently, ad- tracking data 24 h per day and requires little manual vances in biotelemetry have allowed researchers to track labor beyond deployment and retrieval of the receiver the fine-scale movements of individuals during spawning array [14,15,19]. In addition, depending on array design, season [11-13]. The current state-of-the-art in animal accuracy of position estimates from positional telemetry tracking is positional telemetry - use of three or more should be better than those obtained from mobile stationary receivers to triangulate the precise location of telemetry [14]. a transmitter-implanted fish to within a few meters of its Regardless of tracking method, use of oviduct-implanted true location [14,15]. In many cases, use of positional transmitters to estimate time and location of egg depos- telemetry could allow researchers to infer spawning loca- ition requires three assumptions: (1) oviduct transmitters tions at scales of tens to hundreds of meters. Nonetheless, remain within fish until spawning (that is, no premature the above techniques have two distinct disadvantages. expulsion), (2) oviduct transmitters are expelled with the First, estimation of the precise time of spawning is not eggs during spawning, and (3) time and location of ovi- possible. Second, the above techniques rely on the as- duct transmitter expulsion can be accurately determined. sumption that presence of a fish at a given location during The biggest unknown with respect to use of oviduct trans- the spawning season indicates that an individual spawned mitters in conjunction with positional telemetry is the at that location. Therefore, additional techniques outside likelihood that the telemetry system will be able to calcu- of those described above (for example, sampling egg de- late positions of oviduct transmitters after they have been position) are required to determine time of and location expelled onto or into the substrate. Nonetheless, all three of spawning. A possible solution to these limitations is use assumptions should be addressed before oviduct transmit- of oviduct-inserted transmitters to estimate the precise ters can be deemed a useful tool for studying spawning time and location of egg deposition; the idea being that behaviors in a given species. an oviduct transmitter will be expelled from the fish Lake trout (Salvelinus namaycush) are an ecologically with eggs during spawning, and then located resting and economically important fish species in the Laurentian on the spawning substrate. The technique was first Great Lakes, where they have been the focus of intensive demonstrated by Pierce [16], who inserted miniaturized restoration efforts [20,21], and in parts of western United radio transmitters into the oviduct of northern pike States, where they are invasive [22-24]. Lake trout typically (Esox lucius) to identify critical shoreline spawning spawn during autumn, and unlike most salmonines, they habitat. Since then, the technique has been used suc- tend to spawn in open-lake environments [25,26]. The cessfully in muskellunge, Esox masquinongy [17], and reproductive biology of lake trout presents two unique challenges to use of oviduct transmitters in conjunction European perch, Perca fluviatilis [18]. In all three of those studies, mobile tracking was used to determine with positional telemetry. First, unlike the majority of the location of expelled transmitters - Pierce and col- teleost fishes (including northern pike, muskellunge, and European perch), lake trout are gymnovarian, which leagues [16,17] searched daily for expelled transmitters and Skovrind et al. [18] searched for expelled transmit- means that no direct connection exists between the ovary ters twice per week. Because expelled transmitters were and urogenital pore. Instead, eggs are first released loose into the body cavity and then expelled through the oviduct not recovered in these studies, it is not possible to assess the accuracy of transmitter expulsion locations deter- [27,28]. Second, lake trout often spawn on cobble mined by mobile tracking. However, Skovrind et al. [18] shoals with clean, deep interstitial spaces [25,26]. Expelled estimated that accuracy of estimated transmitter loca- oviduct transmitters would likely settle in the spaces tions in their acoustic telemetry study was between 10 between rocks, where their acoustic signals may be and 125 m dependent on the acoustic environment. blocked and, therefore, less likely to be positioned. In Binder et al. Animal Biotelemetry 2014, 2:14 Page 3 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Table 1 Summary of results from a laboratory study of this study, we used a series of field and laboratory retention and expulsion of oviduct transmitters in female experiments and random walk models to test the lake trout three assumptions of using oviduct transmitters (that Stage Inserted Implanted is, transmitter retention until spawning, transmitter expulsion during spawning, and ability to accurately Initial sample size (N) 22 22 determine time and location of transmitter expulsion) Retained oviduct transmitter until end 18 (82%) 21 (95%) of spawning period to study the spawning behavior of lake trout. Our goal was to assess the potential for using oviduct transmit- Became ovulatory (ripened) 17 (94%) 18 (86%) ters in conjunction with positional telemetry to study Percent of ripe trout that expelled 15 (88%) 16 (89%) reproductive ecology, not only in lake trout, but in other transmitter with eggs fishes as well. Percent of total trout that expelled 68% 73% transmitter with eggs Results Oviduct transmitters were either ‘inserted’ into the body cavity via the oviduct or ‘implanted’ surgically just anterior to the genital pore. Values are the Given the methodological focus of this study, we recom- number of lake trout that reached each stage and percentages in parentheses mend that readers review the Methods section at the are based on the number of lake trout that made it to the previous stage. end of the paper prior to reading Results. expel the oviduct transmitter. It is possible that these Retention and expulsion of oviduct-inserted transmitters trout would have become ovulatory after a few more In a laboratory study, 48 transmitters were inserted via days in the holding tank, but any results obtained would the oviduct (hereafter ‘inserted’) or surgically implanted have been confounded by previous handling. Therefore (hereafter ‘implanted’) into adult female lake trout. In- lake trout that were not ovulatory were excluded from sertion and implantation procedures had no observable analysis of transmitter expulsion. effect on the behavior or condition of lake trout during Transmitter expulsion rates during stripping were high the 10-week lab study; however, four trout died during for both the inserted and implanted groups (88% and the study (two from the inserted group: both V7 trans- 89%, respectively) and did not differ from one another mitters; two from the implanted group: one V6 transmit- (Fisher’s exact test: P >0.999). The number of stripping ter and one V7 transmitter). Mortalities occurred at 13, motions required to expel the transmitters ranged from 15, 36, and 55 days post tagging, but cause of death was 1 to 17 (mean (±SD) = 7.0 ± 3.8, median = 6) and was not apparent. Between the two different sized oviduct not different between the two treatment groups (t-test: transmitters tested (V6 and V7, VEMCO, Nova Scotia, t = 1.03, P = 0.31). In lake trout that did not expel the Canada), no transmitter-size effect was observed on re- transmitter during stripping, postmortem autopsies re- tention (time-to-event analysis: χ = 1.6, P = 0.21) or ex- vealed that the oviduct transmitter had moved anteri- pulsion (Fisher’s exact test: P >0.999), so data from the orly, away from the oviduct. two tag types were pooled for subsequent analyses to in- crease statistical power. Data from the four lake trout Estimating time and location of spawning using oviduct that died during the study were not included in analyses. transmitters Most lake trout in the inserted and implanted groups Inferences from a single oviduct transmitter in each fish retained their oviduct transmitters until the end of The ability of acoustic receivers to detect stationary spawning season (Table 1). Five (11.3%) trout expelled transmitters at known lake trout spawning reefs in the their transmitter prematurely. More trout in the inserted Drummond Island Refuge (northern Lake Huron, North group (4 of 22 fish) than the implanted group (1 of America; Figure 1) was reduced when transmitters were 22 fish) prematurely expelled their transmitter. The dif- manually placed in the substrate to simulate a transmit- ference was not significant (survival analysis: χ = 2.0, ter expelled during spawning versus a transmitter that P = 0.16), but because our sample size was relatively was suspended 1 m above the substrate, simulating a small, statistical power was marginal (for example, transmitter still within the fish (Figure 2A; linear mixed power to identify a 25% difference in premature expul- model: t = -13.91, P <0.001). Mean (±SE) detection prob- sion rate was 0.68). All five premature expulsions oc- ability (that is, number of transmissions detected by re- curred during the first 2 weeks after tagging, well before ceivers as a proportion of total transmissions multiplied maturity and the start of spawning season and ovulation. by the number of receivers) of control transmitters sus- Not all lake trout were ovulatory at the end of the pended 1 m above substrate was 0.74 ± 0.27 and 0.70 ± spawning season when eggs were manually stripped to 0.16 for V6 and V7 transmitters, respectively. For trans- determine if oviduct transmitters would be expelled with mitters located in the substrate, probability of detection the eggs. Three trout from the implanted group and one dropped to 0.25 ± 0.17 and 0.23 ± 0.14 for V6 and V7 from the inserted group did not shed eggs, nor did they transmitters, respectively. Detection probabilities did Binder et al. Animal Biotelemetry 2014, 2:14 Page 4 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 1 Map of Drummond Island Lake Trout Refuge field study site used for testing the ability of the VEMCO Positioning System (VPS) to position lake trout oviduct transmitters after they are expelled into substrate. Left: Location of the refuge in Lake Huron (North America; shown in orange). The blue star indicates the actual study site (45°55.840ˈ N, 83°39.572ˈ W). Right: Location of three small VPS arrays (at known lake trout spawning sites) overlaid on high-resolution multibeam sonar bathymetry. Blue squares indicate location of receivers and white circles represent test transmitter locations. One transmitter at each test transmitter location was suspended 1 m above the substrate to act as a control and two more were placed in the substrate. A synchronization transmitter was placed at the central test transmitter site to synchronize receiver clocks. not differ between V6 and V7 transmitters (linear mixed 0.03 ± 0.10, respectively. No positions were obtained for model: t = -15.56, P = 0.38). 78% (47 of 60) of transmitters in the substrate compared The ability of the VEMCO Positioning System (VPS) to only 10% (3 of 30) of suspended transmitters that to calculate positions for transmitters was reduced for returned no positions. V6 and V7 transmitters did transmitters placed in the substrate (Figure 2B; linear not differ in positioning probability (linear mixed model: mixed model: t = -15.56, P <0.001) in comparison to t = -0.44, P =0.66). suspended transmitters. The decrease in probability of The poor ability of the VPS to position transmitters in positioning (that is, number of positions obtained as a the substrate, though many were still being detected by proportion of total number of transmissions) was greater the array (albeit at a lower rate than for suspended than the decrease in probability of detection described transmitters), occurred because transmissions from above (Figure 2A and B). Mean (±SE) positioning prob- transmitters planted in the substrate were less likely to ability for suspended transmitters was 0.69 ± 0.35 and be detected simultaneously by three or more receivers. 0.64 ± 0.27 for V6 and V7 transmitters, respectively. The detection failure by multiple receivers was due, in For transmitters placed in the substrate, mean (±SE) part, to lower elevation of transmitters in substrate com- probability of positioning decreased to 0.04 ± 0.10 and pared to suspended tags and complex rough bathymetry Figure 2 Probability of detecting (A) and positioning (B) V6 and V7 oviduct transmitters (VEMCO, Halifax, Canada) following expulsion into substrate at known spawning sites in the Drummond Island Lake Trout Refuge, Lake Huron. V6 transmitters operated at 180 kHz and V7 transmitters operated at 69 kHz. Control transmitters were suspended 1 m above the substrate. Transmitter type had no significant effect on probability of detection or positioning (linear mixed model: P <0.001); however, oviduct transmitters in the substrate were significantly less likely to be detected and positioned than control transmitters (linear mixed model; P <0.001 for both detection and positioning). Seventy-eight percent of oviduct transmitters in the substrate were not positionable. Binder et al. Animal Biotelemetry 2014, 2:14 Page 5 of 14 http://www.animalbiotelemetry.com/content/2/1/14 of the cobble substrate in the spawning area, both of (low residency group) (Figure 4; general linear model: which restricted line of sight between transmitters and t = -5.512, P <0.001). For example, mean (±SE) location receivers. Probability of detection was negatively associ- error was 10 times greater for simulated trout in the low ated with the area obstructing line of sight (cross-sec- residency group than that for simulated trout in the high tional area of substrate located above line of sight residency group (186.1 ± 3.7 m location error vs. 15.8 ± between transmitter and receiver) between transmitters 0.3 m location error). Location error also increased with and receivers (linear mixed model: t = 2.094, P = 0.037). increasing transmission delay, but size of the effect Probability of detection for both suspended transmitters was also greater for fish that moved farther from the and transmitters in substrate was also negatively associ- spawning site (Figure 4; general linear model: t = 45.940, ated with distance between transmitter and receiver P = <0.001). In general, location error increased with (linear mixed model: t = -8.051, P <0.001). swim speed, but size of this effect varied with both dis- tance travelled from spawning site and transmission Pairing oviduct transmitters with larger transmitters to delay (Figure 4; general linear model: t = 33.6, P <0.001 track fish after spawning and t = 10.3, P <0.001 for interaction with distance trav- In light of the poor performance of the VPS array with elled from spawning site and transmission delay, respect- respect to positioning transmitters expelled into the sub- ively). Time errors ranged from 0.0 min to 569.2 min, strate, an alternative paired-transmitter approach was but in contrast to location error, decreased with distance tested via simulation of fish double-tagged with a small travelled from spawning site (Figure 4; general linear oviduct transmitter (to be expelled during spawning) model: t = -5.51, P <0.001). Mean (±SE) time error was and a larger transmitter that is retained after spawning. 6.4 ± 0.1 min for simulated trout in the low residency Spawning was inferred when the VPS array determined group and 19.2 ± 1.0 min for simulated trout in the high that the two transmitters separated in space (Figure 3). residency group. Like location error, time error was af- Simulations showed that double-tagging fish can improve fected by both distance travelled from spawning site and estimates of time and location of spawning compared swim speed. Time error increased with increasing trans- to use of only an oviduct transmitter. Using changepoint mission delay (Figure 4; general linear model: t = 4.99, analysis [29] on simulated fish tracks, the time and loca- P <0.001). The effect of swim speed on time error, how- tion at which the two transmitters separated (that is, ever, was more complicated and was dependent on dis- expulsion of oviduct tag during spawning) could be pre- tance travelled from spawning site (Figure 4; general dicted from the change in distance between the two trans- linear model: t = 4.00, P <0.001). mitters over time (Figure 4) or by the change in relative positioning probability between the two transmitters over Detecting change in positioning probability between paired time (Figure 5). Of the two methods, change in relative transmitters positioning probability performed best (Figures 4 and 5). Change in relative positioning probability proved to be On average, location error of estimated transmitter separ- more reliable than change in distance between transmit- ation was nearly two times greater (70.1 m vs. 36.7 m) ters for estimating the time and location of transmitter when using change in distance between transmitters than separation (that is, spawning). On average, point of trans- when using change in relative positioning probability. mitter separation was estimated to have occurred within Similarly, average time error of estimated transmitter sep- the first two to three fish transmitter transmissions after aration was more than four times greater (10.7 min vs. 2.4 actual separation. Unlike when using change in distance min) when using change in distance between transmitters to estimate transmitter separation, time error did not than when using change in relative positioning probability. vary with distance travelled from spawning site (Figure 5; The accuracy of estimates of time and location of trans- general linear model; t = -1.27, P = 0.21) or swim speed mitter separation also varied greatly with characteristics of (Figure 5; general linear model; t = -0.52, P = 0.60) - mean fish tracks (that is, distance travelled from spawning site time error ranged from 2.4 min to 2.5 min across all four (low, medium, medium-high, and high residency groups) residency groups (that is, low, medium, medium-high, and and swim speed) and transmitter delay. high), and from 2.3 min to 2.5 min across all levels of −1 swim speed (that is, 0.25, 0.50, 0.75, and 1.00 m · s ). As Detecting change in distance between paired tags expected, however, time error did increase significantly When using change in distance between transmitters to with increasing transmission delay (Figure 5; general linear estimate time and location of transmitter separation model: t = 17.01, P <0.001). Location error was more vari- (that is, spawning), location errors ranged from 0.1 m able than time error and was affected by distance travelled to 825.2 m. Location error was smaller for fish that from spawning site, transmission delay, and swim speed remained near the spawning site (high residency group) (Figure 5; general linear model: t = -16.11, P <0.001, than for fish that moved farther from the spawning site t = -6.34, P <0.001, and t = -6.34, P <0.001, respectively). Binder et al. Animal Biotelemetry 2014, 2:14 Page 6 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 3 Examples of simulated lake trout tracks (random walk models) and associated transmitter tracks for low-residency (that is, σ = 5; A, B) and high-residency (that is, σ = 90; C, D) lake trout. Simulations (6,400 in total) were used to evaluate the potential for use of a paired-transmitter experimental design to estimate the time and location of oviduct transmitter expulsion. Black tracks in (A) and (C) depict the true path of trout, while red and blue tracks in (B) and (D) depict the reconstructed path of the fish transmitter and oviduct transmitter, respectively, based on triangulated positions from a hypothetical positional array. Note the difference in scale between the two trout tracks. Location errors ranged from 0.2 m to 995.5 m and in- is the first to evaluate the potential for use of oviduct creased with increasing distance travelled from spawning transmitters combined with positional telemetry to ac- site. For example, mean (±SE) location error was 80.7 ± curately estimate time and location of spawning in 2.5 m for simulated trout in the low residency group and fishes. We tested three assumptions of using oviduct 11.1 ± 0.2 m for simulated trout in the high residency transmitters to determine time and location of egg de- group. Location error also increased with increasing trans- position in lake trout and found that: (1) most oviduct mission delay and swim speed, but in both cases the size transmitters remained within lake trout until spawning of the effect was dependent on distance travelled from (that is, few were expelled prematurely), (2) most ovi- spawning site (Figure 5; general linear model: t = 25.38, duct transmitters were expelled with eggs during manual P <0.001 and t = 20.88, P <0.001 for interaction with trans- stripping, and (3) oviduct transmitters were difficult to mission delay and swim speed, respectively). Effect size of position once expelled into substrate, but timing and lo- swim speed was also dependent on transmission delay cation of transmitter expulsion (indicative of spawning) (Figure 5; general linear model: t = 9.80, P <0.001). could be determined with reasonable accuracy in simula- tions using positional telemetry and a paired-transmitter design where one transmitter is expelled with eggs dur- Discussion ing spawning and the other remains in the fish. Use of oviduct-inserted transmitters to determine time Transmitter retention and subsequent artificial expul- and location of egg deposition represents an important sion rate in lake trout was comparable to those observed step forward in the use of electronic transmitters to in previous studies that have used oviduct transmitters in study the reproductive ecology of fishes [30]. Our study Binder et al. Animal Biotelemetry 2014, 2:14 Page 7 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 4 Evaluation of use of change in distance between fish transmitter and oviduct transmitter to estimate time and location of separation in a simulated paired-transmitter study. Panels display data for four different spawning site residency (that is, tendency to remain close to the spawning site) groups, which were simulated using different standard deviation of turn angle in random walk models (σ = 5, 20, 45, and 90 for low, medium, medium-high, and high residency groups, respectively). Top and bottom graphs in each panel display time error (min) and location error (m), respectively, for point estimates of transmitter separation calculated using changepoint analysis. Shading indicates the effective transmission delay (that is, mean time between transmissions) in seconds. other fishes. Eighty percent of females that ripened during likely they would have been recognized as such, and not our study retained the oviduct transmitter until spawning have confounded results. The cause of premature oviduct and then expelled it during manual stripping. Pierce et al. transmitter expulsion is not known. Expulsion may have [17] reported 60% and 90% expulsion in northern pike been a passive process, in that transmitters may have been (Esox lucius) and muskellunge (Esox masquinongy), re- expelled as a result of physical forces at play during nor- spectively, and Skovrind et al. [18] noted 92% expulsion in mal swimming behavior. Alternatively, the trout may have their study on European perch (Perca fluviatilis). Lake perceived the transmitter as a foreign object and actively trout have no true connection between the ovary and expelled it by muscular contractions or rubbing on the urogenital pore [27], which might reduce the likelihood bottom. Some fishes are capable of encapsulating trans- that oviduct transmitters would be expelled with the mitters, and expelling them through the intestine [31,32]. eggs. This condition (gymnovarian) is rare among teleost However, given the short time between tagging and pre- fishes, but occurs frequently among many of the most mature expulsions, these expulsions likely occurred via commonly studied fish species - that is, trout and salmon the urogenital pore. Either way, the portion of tagged (Salmonidae), freshwater eels (Anguillidae), and most trout that prematurely expelled oviduct transmitters non-teleost fishes (for example, sturgeons and lampreys) was small enough that, beyond a slight reduction in [28]. All premature expulsions of oviduct transmitters in sample size, impacts on data quality and interpretation our study occurred within the first 2 weeks after tagging, would have been minimal. well before the onset of the lake trout spawning season. If Most lake trout in our study that retained the trans- these premature expulsions had occurred in a field study, mitter until the end of the spawning period expelled the Binder et al. Animal Biotelemetry 2014, 2:14 Page 8 of 14 http://www.animalbiotelemetry.com/content/2/1/14 Figure 5 Evaluation of use of change in relative position probability (that is, change in ratio of time since last position) to estimate time and location of separation in a simulated paired-transmitter study. Panels display data for four different spawning site residency (that is, tendency to remain close to the spawning site) groups, which were simulated using different standard deviation of turn angle in random walk models (σ = 5, 20, 45, and 90 for low, medium, medium-high, and high residency groups, respectively). Top and bottom graphs in each panel display time error (min) and location error (m), respectively, for point estimates of transmitter separation calculated using changepoint analysis. Shading indicates the effective transmission delay (that is, mean time between transmissions) in seconds. oviduct transmitter with eggs when manually stripped. blockage [35] and maximize likelihood of transmitter However, use of manual stripping as a surrogate for egg expulsion during spawning. deposition during spawning is a source of bias in our Poor ability to detect and position transmitters depos- study. Although we did our best to approximate normal ited into coarse substrate reduces the usefulness of ex- egg deposition by holding the trout in a natural spawn- pelled oviduct transmitters for determining timing and ing posture (that is, head and tail flexed slightly up- location of spawning. Reef environments with complex wards) [33,34] during stripping, we do not know to what topography pose many challenges to acoustic telemetry, extent physical forces applied to the eggs and oviduct including high levels of noise, signal bending, echoes, transmitter during manual stripping represent those and signal blocking [36,37]. Line of sight between trans- present during natural egg deposition. If force applied mitter and receiver is critical for positioning, so receiver during manual stripping was greater than would be placement in the water column, relative to transmitters, present during spawning, our results could overestimate is an important consideration [37,38]. Line of sight may transmitter expulsion rate. However, because transmit- have improved if we had located receivers nearer to the ters were easily expelled, our judgment was that reason- surface, but improvement in line of sight may have able pressure was applied. Transmitter size (V6 or V7) come at the expense of increased noise due to wind and had no effect on transmitter retention or expulsion rates waves [36,39] and possibly lower position accuracy due in our study. Nonetheless, use of the smallest available to receiver movement. Line of sight may also have been transmitter that suits the study system and objectives is improved by using more receivers and spacing them recommended to minimize the possibility of oviduct closer together. Binder et al. Animal Biotelemetry 2014, 2:14 Page 9 of 14 http://www.animalbiotelemetry.com/content/2/1/14 In instances where oviduct transmitters are difficult function of transmitter delay and positioning probability, to position once expelled into substrate, a paired- and is independent of the behavior of the fish, except transmitter design appeared to accurately estimate time in cases where the fish behaves in a manner that reduces and location of spawning. Our decision to use change- the positioning probability of the transmitter (for ex- point analysis [29] to determine when the two transmit- ample, takes refuge in vegetation [40]). ters separated (that is, began to track differently) was Transmitters with short delay between transmissions based on a need for objectivity and statistical defensibil- produced more accurate estimates of time and location ity. Initially, we attempted to use rule-of-thumb bio- of transmitter separation, but the effect size of transmit- logical filters based on expected behavior of the fish (for ter delay on accuracy of estimates of time and location example, point at which the distance between fish trans- of transmitter separation was dependent on the behavior mitter and oviduct transmitter was greater than would of the fish. A tradeoff exists between transmitter delay be expected given the mean swimming speed of the and the number of transmitters an acoustic array can ac- fish). However, results were subjective, and for many commodate at one time. If an acoustic receiver detects a trout tracks, rule-of-thumb filters failed even to identify signal from a second transmitter while decoding the sig- a point of transmitter separation (T Binder, unpublished nal from the first, the receiver cannot separate the two results). In contrast, changepoint analysis produced a signals and the result is a code collision, which results in point estimate of transmitter separation for all 6,400 neither transmission being decoded (that is, no detec- trout tracks. One advantage of the changepoint proced- tion) [41]. The probability of code collisions increases as ure was the change in positioning probability of expelled transmitter delay decreases and number of transmitters oviduct transmitters could be used to our advantage within detection range of the receiver increases [36]. and, in our simulation exercise, change in relative posi- The impact of long transmitter delays on point estimates tioning probability turned out to be the most accurate of transmitter separation was greater in fish with low method for estimating the time and location of transmit- spawning site residency, indicating that use of a small ter separation. Ultimately, choice of changepoint method number of transmitters with a short delay is preferred in (that is, change in distance between transmitters vs. these cases. In contrast, because transmitter delay had change in relative positioning probability of the trans- less impact on estimates of time and location of trans- mitters) will depend on how well the acoustic array can mitter separation in fish that showed high spawning site position oviduct transmitters after expulsion. For ex- residency, one might choose a long transmitter delay ample, in our simulated tracks, positioning probability of (for example, 5 min) to reduce the frequency of trans- oviduct transmitters decreased 90% after expulsion mitter code collisions and allow for more transmitters to reflecting the poor ability of the acoustic array to pos- be deployed than if a short transmitter delay was chosen. ition transmitters in the substrate during our field study. Ultimately, appropriate choice of transmitter delay and However, in situations where positioning probability of number of transmitters deployed will depend on trade- oviduct transmitters decreases only a small amount after offs between the desired precision and accuracy of expulsion, change in distance between transmitters spawning time and location estimates and the behavior would likely produce more accurate point estimates of of the fish. transmitter separation than change in relative position- ing probability. Conclusions Accuracy of estimated time and location of transmitter Overall, our results suggest that use of oviduct transmit- separation varied greatly with distance travelled from ters with positional telemetry represents a promising spawning site (that is, spawning site residency) and swim new technique for studying spawning behaviors in lake speed of the fish. In general, time error decreased and trout. Despite the absence of a direct connection be- location error increased as distance travelled increased tween the ovary and urogenital pore in lake trout, the from the spawning site. This relationship occurred be- majority of oviduct transmitters implanted into the body cause fish that travelled further from the spawning site cavity near the oviduct remained within the fish until tended to move a greater linear distance from the site of spawning and were then expelled with eggs during man- transmitter separation during each transmission interval ual stripping. Use of a traditional single-transmitter ovi- than fish that tended to remain closer to the spawning duct tagging study design [16-18] in conjunction with site. Time error in estimates of transmitter separation positional telemetry would not be successful in lake calculated using the change in relative positioning prob- trout when spawning occurs over rough stony substrates ability method was an exception to the above-stated because oviduct transmitters would be difficult to pos- trend, in that time error was resistant to differences in ition. Most transmitters in the substrate continued to be swimming speed and spawning site residency. The cause detected on some receivers, but at a low rate. Thus, de- for the difference was the fact that positioning rate is a pending on the size of the study area and the required Binder et al. Animal Biotelemetry 2014, 2:14 Page 10 of 14 http://www.animalbiotelemetry.com/content/2/1/14 resolution of estimated time and location of spawning, 13°C) water from Lake Huron. Trout were offered a diet manual tracking would be an alternative option to deter- of 13 mm brood stock fish pellets until mid-September, mine location. However, greater effort would be required when the trout ceased feeding. to locate transmitters using manual tracking than if the Trout were tagged with oviduct transmitters on 17 VPS array could be used successfully. We found that use September 2013. To test the hypothesis that insertion of of a paired-transmitter experimental design in conjunc- transmitters through the oviduct does not affect retention tion with positional telemetry would be a better solution and expulsion of transmitters, relative to those that have when oviduct transmitters are expelled into substrates been surgically implanted, we inserted transmitters via the (for example, cobble or mud) that act as barriers to oviduct in half the fish (n = 24; hereafter referred to as acoustic signal reception. In our simulations, we were ‘inserted’) and surgically implanted transmitters just anter- able to use the decreased positioning probability of ex- ior to the genital pore in the rest (n = 24; hereafter referred pelled oviduct transmitters to estimate time and location to as ‘implanted’). Half of each tagging group received V6 of transmitter separation (that is, spawning) with reason- dummy transmitters (VEMCO, Nova Scotia, Canada; able accuracy. 16.5 × 6 mm, 1 g in air) and the remaining trout received Use of oviduct transmitters and positional telemetry to V7 dummy transmitters (VEMCO; 20 × 7 mm, 1.6 g in determine time and location of spawning has the poten- air). Dummy transmitters contained a passive integrated tial to be a useful tool for studying reproductive ecology transponder (PIT) tag so that individual fish could be across a broad range of fish species. Beyond the obvious identified. All trout were anesthetized in 40 L of clove oil benefits of reduced manual labor and increased preci- solution (0.8 mL/L of 1:9 clove oil:ethanol solution) and sion and accuracy of estimated spawning locations [14], then placed dorsal side down on a v-board prior to under- remote monitoring of spawning behavior via positional going the tagging procedure. Gills were perfused with aer- telemetry may offer opportunities to study spawning in ated lake water throughout the procedures. In trout fishes that have previously proven inaccessible [15]. For receiving the transmitter via oviduct-insertion, transmit- example, observations of lake trout spawning have been ters were gently inserted through the oviduct and pushed rare because it occurs mainly at night, during a time of approximately 5 cm deep using a 10 cm piece of sterilized year when weather patterns are often unpredictable and 6.5 mm diameter Tygon tubing. The procedure was per- dangerous for field work [26]. Nonetheless, effective use formed by a single researcher and took, on average, 20 to of oviduct transmitters to determine time and location 30 s. In trout receiving surgically implanted transmitters, a of spawning depends on the assumptions that oviduct small incision (1.5 to 2.0 cm) was made slightly off the transmitters will remain within the fish until spawning ventral midline, approximately 5 cm anterior to the genital and then be expelled with eggs at the spawning site, pore. After the transmitter was inserted into the body cav- and that oviduct transmitters can be located accurately ity, the incision was closed using two simple, interrupted after expulsion. Failure to meet these assumptions sutures (Ethicon, Inc.; 3-0 polydioxanone monofilament). will, at best, reduce sample size (for example, if transmit- Surgeries were performed by a single surgeon and took, ters are not expelled), and at worst, could result in erro- on average, 2 to 3 min. Following tagging, trout were neous conclusions (for example, if transmitters are returned to the holding tank to recover from anesthesia. expelled prematurely). Therefore, regardless of whether The holding tank was searched daily, September to researchers use positional telemetry or manual tracking, November, for expelled oviduct transmitters, as well as we recommend that feasibility studies be conducted expelled loose eggs that would indicate trout were ovu- to test assumptions prior to using oviduct transmitters lating. During daily searches, the fish were assessed for in a new species. signs of odd behavior (for example, failure to respond to the researcher’s movements or inability to maintain Methods equilibrium) and signs of poor health (for example, der- Retention and expulsion of oviduct-inserted transmitters mal discoloration or presence of fungus). If a transmitter We tested the assumptions that oviduct-inserted trans- was found, the associated PIT tag number, date, and mitters are retained until spawning and then expelled time were recorded. Lake trout did not spawn naturally with the eggs (Assumptions 1 and 2) in a laboratory under laboratory conditions, so manual stripping was tagging study. Forty-eight hatchery-raised, female lake used as a surrogate to determine if transmitters would trout (Seneca strain) from Sullivan Creek National be expelled with eggs. Our assumption was that, while Fish Hatchery (Brimley, MI, USA) were transferred not identical to natural spawning, the manual stripping to Hammond Bay Biological Station (Millersburg, MI, procedure was representative of the processes that occur naturally during egg expulsion. Manual stripping was USA) on 6 August 2013. The trout were housed in a 6,800 L rectangular (6.7 × 1.7 × 0.9 m) holding tank, sup- conducted on 25 November 2013, to coincide with the plied with chilled (water temperature between 7°C and end of the Lake Huron lake trout spawning period. The Binder et al. Animal Biotelemetry 2014, 2:14 Page 11 of 14 http://www.animalbiotelemetry.com/content/2/1/14 ovulating trout were removed from the holding tank one transmitters. One of each transmitter type was sus- at a time and anesthetized as before to prevent excessive pended 1 m above the substrate to act as a control, stress. Anesthetized trout were held dorsal side up with representing a transmitter that was still within the trout. heads and tails pointed slightly upward, as occurs during The two remaining transmitters of each type were natural egg deposition [33,34]. Their bellies were then placed in the cobble substrate (5 to 20 cm in diameter) massaged from anterior to posterior until the oviduct by scuba divers to represent transmitters that had been tag was expelled, or the trout stopped expelling eggs. expelled during spawning. Divers ensured that the trans- Differences in retention of oviduct transmitters were mitters were placed in the space between rocks, as this compared using time-to-event analysis (R package ‘sur- is most likely how the negatively-buoyant transmitters vival’; α = 0.05), which allowed us to not only compare the would settle when expelled naturally during a spawning proportion of individuals that retained the transmitter, event. Test transmitters were left in place for 20 min but also the temporal pattern of premature expulsion. (allowing for a total of 20 transmissions from each trans- The proportion of lake trout that expelled the oviduct mitter) before being retrieved and moved to the next test transmitter during manual stripping was compared among transmitter site. At the end of the study, receivers were groups using Fisher’s Exact Test (R package ‘stats’; retrieved and downloaded, and log files from the re- α = 0.05), and the number of stripping motions required ceivers were sent to VEMCO for postprocessing using to expel the transmitters was compared using a t-test their proprietary hyperbolic positioning algorithms [42]. (R package ‘stats’; α =0.05). Probabilities of detecting and positioning control and oviduct transmitters were compared across transmitter Estimating time and location of spawning using oviduct type using linear mixed models (R package ‘lme4’; α = transmitters 0.05). Treatment (control transmitter or oviduct transmit- Inferences from a single oviduct transmitter in each fish ter) and transmitter type (V6 or V7) were fixed effects and The ability of the VEMCO Positioning System (VPS) to transmitter identification code and array number were accurately position expelled oviduct acoustic transmit- random effects. Variables related to poor detectability of ters in substrate (Assumption 3) was investigated in a transmitters in the substrate were also investigated using a field study at known lake trout spawning reefs in the linear mixed model (R package ‘lme4’; α =0.05). Distance Drummond Island Lake Trout Refuge, northern Lake between transmitter and receiver, total area of obstruction Huron, North America (Figure 1). Three small VPS ar- between transmitter and receiver, tag type (V6 or V7), and rays (approximately 100 × 100 m) were constructed on treatment (control transmitter or oviduct transmitter) three separate near-shore (0.6 to 1.3 km from shoreline) were fixed effects; transmitter identification code and re- spawning sites. The three sites ranged in depth from ap- ceiver serial number were random effects. Area of obstruc- proximately 1.5 to 7 m and consisted of rocky substrate. tion, the total cross-sectional area of substrate located at The approximate 100 m spacing of receivers was based higher altitude than the line of sight between transmitter on recommendations from the manufacturer (VEMCO) and receiver, was calculated using high-resolution (1 m on the maximum detection range of the less powerful horizontal, 10 cm vertical) multibeam sonar bathymetry 180 kHz transmitters. Each array consisted of four (Seabat 7101 system, Teledyne RESON Inc.). receiver locations and a centrally-located synchronization transmitter site, which held transmitters used to synchronize Pairing oviduct transmitters with larger transmitters to the clocks on receivers. Because the V6 and V7 acoustic track fish after spawning transmitters transmit at different frequencies, we installed The feasibility of using transmitter separation in a paired- two acoustic receivers (VR2W-69 kHz and VR2W-180 transmitter design to accurately estimate time and loca- kHz) at each receiver location on a single mooring. The tion of oviduct transmitter expulsion (Assumption 3) was upper receiver was positioned upside down so that the tested using random walk models to simulate a variety of top-mounted hydrophones on the two receivers were possible lake trout spawning behaviors [43]. The paired- located as closely as possible to one another. transmitter design called for each simulated trout to be Five test transmitter locations were chosen for each implanted with two transmitters; a small oviduct transmit- VPS array. The first was at the center of the array. The ter that would be expelled with the eggs during spawning, remaining four were scattered throughout the array so and a larger fish transmitter that would remain within that they were located at random distances from the the fish after spawning. Time and location at which the center of the array. Three transmitters of each type (V6 two transmitters began to behave differently (point of and V7) were deployed at each site. Transmitters were transmitter separation) was used to estimate time and programmed to transmit with a fixed delay of 1 min and location of spawning. initialization of transmitters was staggered by 20 s to en- Each random walk model (referred to hereafter as ‘trout sure that no signal code collisions would occur between track’) consisted of 43,200 steps, each 1 m in length (total Binder et al. Animal Biotelemetry 2014, 2:14 Page 12 of 14 http://www.animalbiotelemetry.com/content/2/1/14 track = 43.2 km). Steps of the trout track were straight positioning the oviduct transmitter in the substrate from lines, with headings sampled from a random distribution this point forward was reduced to 10% (that is, only with mean turn angle equal to 0 and standard deviation σ. every 10th transmission could be positioned). For simplicity, the models assumed that trout remained In total, 100 trout tracks were simulated for each com- within the VPS array for the entire time series. However, bination of σ, swim speed and transmission delay (6,400 in practice, the only requirements are that the oviduct trout tracks in total). We used the R statistical package transmitter be expelled in the array and then the trout ‘changepoint’ to estimate time and location of transmit- remain within the array long enough to determine that ter separation for each simulated trout track [29]. Chan- the two transmitters had separated. Numerous possible gepoint analysis, an emerging tool in movement ecology spawning behaviors were simulated to determine how dif- [44,45], uses maximum likelihood methods to determine ferent behaviors, spanning a range of species, could affect points in a time series where the mean, variance, or mean ability to accurately estimate time and location of spawn- and variance of a response variable change. In our ana- ing using a paired-transmitter design. To determine how lyses, we tested for a single changepoint, and visual in- degree of spawning site residency affected our ability to spection of results indicated that the ‘changepoint’ accurately estimate the time and location of transmitter function ‘cpt.meanvar’ (that is, changepoint of mean and separation (that is, spawning location) four different values variance; number of changepoints (Q) = 1, test.stat = ‘Nor- were used for σ (5, 20, 45, and 90; hereafter referred to as mal’) was best able to identify when the two transmitters low, medium, medium-high, and high residency groups, separated. respectively). Distance travelled from spawning site was We tested two separate methods for estimating time inversely related to σ (Figure 3A and C). For example, sim- and location of transmitter separation. The first method ulated trout with σ = 5 (low residency group) travelled, on was to test for a change in the distance between the two average, 17 times further from the spawning site than sim- transmitters. For each time in the fish transmitter track, ulated lake trout with σ = 90 (high residency group). Mean we used linear interpolation to estimate the location of (±SE) distance between all positions for an individual and the oviduct transmitter at that time. The linear distance its actual spawning site was 1,976 ± 536 m, 493 ± 141 m, between the fish transmitter and oviduct transmitter was 223 ± 63 m, and 116 ± 33 m for simulated tracks with σ = calculated and changepoint analysis was used on those 5, 20, 45, and 90, respectively. We also varied mean swim values to determine when in the time series the mean −1 speed of individual lake trout between 0.25 m · s and and variance of the distance between the two transmit- −1 −1 1.00 m · s (0.25, 0.50, 0.75, or 1.00 m · s )to cover a ters changed. The second method took advantage of the range of realistic swimming speeds. reduction in positioning probability of expelled oviduct Transmitter positions, such as those that would be tri- transmitters that we observed in our field trial. For each angulated using the VPS system, were overlaid on each time in the fish transmitter track, we calculated the ratio trout track. The result was two transmitter tracks, one of the time since the previous position for the oviduct for the fish transmitter and another for oviduct transmit- transmitter and the time since the previous position for ter (Figure 3B and D). Transmitters were programmed the fish transmitter (that is, time since last oviduct trans- to transmit with mean transmission delay of 60, 180, mitter position / time since last fish transmitter pos- 300, or 420 s. The exact delay of each transmission ition). Changepoint analysis was then used to determine was sampled from a uniform distribution with bounds when in the time series the mean and variance of that delay - delay/2 and delay + delay/2. These delays were ratio changed. Change in ratio of positioning probability assumed to represent effective transmitter delays, which was used rather than simply testing for a change in incorporated not only the nominal delay of the transmit- the positioning probability of the oviduct transmitter it- ters as programmed by the manufacturer, but also any self because positioning probability can vary naturally due to environmental variability. The ratio method can variation due to environmental variables like wave noise and signal code collisions with other nearby transmit- accommodate this variation because any environmental ters. For the sake of realism, we incorporated positioning variation should affect both transmitters in a similar man- ner. Both methods outlined above can return estimates of error into each of the transmitter positions by sampling from a bivariate normal distribution with mean equal to time and location of transmitter separation if the oviduct the true location (x and y coordinates) of the trout at transmitter continues to be positioned (even at a lower rate). However, only the latter method (change in relative time t (based on trout track) and a standard deviation of 15 m. At the halfway point of each trout track, the ovi- positioning probability) can return an estimate of time duct transmitter was expelled into the substrate as a and location of transmitter separation if the oviduct trans- mitter ceases to be positioned after expulsion. simulated spawning event. Based on our field test of the ability of the VPS system to position transmitters ex- The output of the changepoint analysis was an esti- pelled into the substrate, we assumed the probability of mate of the location in the fish transmitter track time Binder et al. Animal Biotelemetry 2014, 2:14 Page 13 of 14 http://www.animalbiotelemetry.com/content/2/1/14 series where the two transmitters separated (that is, 2. Vitousek PM, D’Antonio CM, Loope LL, Rejmánek M, Westbrooks R: Introduced species: a significant component of human-caused global began to behave differently). Two metrics were calcu- change. New Zeal J Ecol 1997, 21:1–16. lated to evaluate the ability of the paired-transmitter 3. Hermoso V, Clavero M, Blanco-Garrido F, Prenda J: Invasive species and design to accurately estimate the time and location of habitat degradation in Iberian streams: an analysis of their role in freshwater fish diversity loss. Ecol Appl 2011, 21:175–188. oviduct expulsion: (1) difference in time (hereafter ‘time 4. Eby LA, Crowder LB, McClellen CM, Peterson CH, Powers MJ: Habitat error’) between estimated transmitter separation and degradation from intermittent hypoxia: impacts on demersal fishes. true time of spawning (that is, the midway point of Mar Ecol Prog Ser 2005, 291:249–261. 5. Casatti L, Langeani F, Ferreira CP: Effects of physical habitat degradation the trout track), and (2) distance between estimated lo- on the stream fish assemblage structure in a pasture region. Environ cation of transmitter separation and true location of Manage 2006, 38:974–982. spawning (hereafter ‘location error’). A series of general 6. Wilson SK, Fisher R, Pratchett MS, Graham NAJ, Dulvy NK, Turner RA, Cakacaka A, Polunin NVC, Rushton SP: Exploitation and habitat linear models (R package ‘stats’; α = 0.05) were used to degradation as agents of change within coral reef fish communities. determine the effects of σ, transmission delay and swim Glob Change Biol 2008, 14:2796–2809. speed on time and location error of estimated transmit- 7. Roessig JM, Woodley CM, Cech JJ Jr, Hansen LJ: Effects of global climate change on marine and estuarine fishes and fisheries. Rev Fish Biol Fisher ter separation. Effects were modeled separately using 2004, 14:251–275. both the change in distance between transmitter method 8. Ficke AD, Myrick CA, Hansen LJ: Potential impacts of global climate and change in relative positioning probability method. change on freshwater fisheries. Rev Fish Biol Fisher 2007, 17:581–613. 9. Farrell AP, Hinch SG, Cooke SJ, Patterson DA, Crossin GT, Lapointe M, Because the effect of spawning site residency on accur- Mathes MT: Pacific salmon in hot water: applying aerobic scope models acy of transmitter separation estimates was of interest, and biotelemetry to predict the success of spawning migrations. the continuous variable mean distance travelled from Physiol Biochem Zool 2008, 81:697–709. 10. Goodyear CS, Edsall TA, Ormsby Dempsey DM, Moss GD, Polanski PE: Atlas of spawning site (that is, mean distance between all posi- the spawning and nursery areas of Great Lakes fishes, Volume 5. Washington, tions in a given trout track and its true spawning loca- DC: Lake Huron. U.S. Fish and Wildlife Service; 1982. FWS/OBS-82/52. tion) replaced σ in the models. Interactions between 11. Eiler JH, Nelson BD, Bradshaw RF: Riverine spawning by sockeye salmon in the Taku River, Alaska and British Columbia. Trans Amer Fish Soc 1992, 121:701–708. distance travelled from spawning site and transmission 12. Bégout Anras ML, Cooley PM, Bodaly RA, Anras L, Fudge RJP: Movement delay, distance travelled from spawning site and swim and habitat use by whitefish during spawning in a Boreal lake: speed, and transmission delay and swim speed were also integrating acoustic telemetry and geographic information systems. Trans Amer Fish Soc 1999, 128:939–952. included in the models. 13. Flavelle LS, Ridgway MS, Middel TA, McKinley RS: Integration of acoustic Competing interests telemetry and GIS to identify potential spawning areas for lake trout The authors declare that they have no competing interests. (Salvelinus namaycush). Hydrobiologia 2002, 483:137–146. 14. Espinoza M, Farrugia TJ, Webber DM, Smith F, Lowe CG: Testing a new acoustic telemetry technique to quantify long-term, fine-scale movements Authors’ contributions of aquatic organisms. Fish Res 2011, 108:364–371. TB participated in all aspects of study conception, experimental design, data 15. Biesinger Z, Bolker BM, Marcinek D, Grothues TM, Dobarro JA, Lindberg WJ: acquisition and analysis, and drafted the manuscript. CH participated in data Testing autonomous acoustic telemetry positioning system for fine-scale analysis and manuscript preparation. SM participated in study conception, space use in marine animals. J Exp Mar Biol Ecol 2013, 448:46–56. data collection, and manuscript preparation. HT participated in data 16. Pierce RB: Oviduct insertion of radio transmitters as a means of locating collection, analysis, and manuscript preparation. CK participated in study northern pike spawning habitat. N Am J Fish Manage 2004, 24:244–248. conception, experimental design, and manuscript preparation. All authors read and approved the final manuscript. 17. Pierce RB, Younk JA, Tomcko CM: Expulsion of miniature radio transmitters along with eggs of muskellunge and northern pike - a new method for locating critical spawning habitat. Environ Biol Fish 2007, 79:99–109. Acknowledgements 18. Skovrind M, Christensen EAF, Carl H, Jacobsen L, Møller PR: Marine We thank E. Larson, L. Lesmeister, D. Operhall, and B. Lamoreux for their spawning sites of perch Perca fluviatilis revealed by oviduct-inserted technical assistance in the laboratory and field. This work was funded by the acoustic transmitters. Aquat Biol 2013, 19:201–206. Great Lakes Fishery Commission by way of Great Lakes Restoration Initiative 19. Andrews KS, Tolimieri N, Williams GD, Samhouri JF, Harvey CJ, Levin PS: appropriations (GL-00E23010-3). This paper is contribution 5 of the Great Comparison of fine-scale acoustic monitoring systems using home Lakes Acoustic Telemetry Observation System (GLATOS) and contribution range size in demersal fish. Mar Biol 2011, 158:2377–2387. number 1875 of the USGS Great Lakes Science Center. Any use of trade, 20. Hansen MJ: Lake trout in the Great Lakes: Basinwide stock collapse and product, or firm names is for descriptive purposes only and does not imply binational restoration. In Great Lakes fisheries policy and management: endorsement by the U.S. Government. a binational perspective. Edited by Taylor WW, Ferreri CP. East Lansing, MI: Michigan State University Press; 1999:417–454. Author details Department of Fisheries and Wildlife, Michigan State University, Hammond 21. Muir AM, Krueger CC, Hansen MJ: Re-establishing lake trout in the Bay Biological Station, 11188 Ray Rd, Millersburg, MI 49759, USA. United Laurentian Great Lakes: past, present, and future. In Great Lakes fisheries States Geological Survey, Hammond Bay Biological Station, 11188 Ray Rd, policy and management: a binational perspective. 2nd edition. Edited by Millersburg, MI 49759, USA. Center for Systems Integration and Taylor WW, Lynch AJ, Leonard NJ. East Lansing, MI: Michigan State Sustainability, Michigan State University, 1405 South Harrison Road, 115 University Press; 2013:533–588. Manly Miles Building, East Lansing, MI 48823, USA. 22. Ruzycki JR, Beauchamp DA, Yule DL: Effects of introduced lake trout on native cutthroat trout in Yellowstone Lake. Ecol Appl 2003, 13:23–37. Received: 29 May 2014 Accepted: 11 August 2014 23. Koel TM, Bigelow PE, Doepke PD, Ertel BD, Mahony DL: Nonnative lake Published: 4 September 2014 trout results in Yellowstone cutthroat decline and impacts to bears and anglers. Fisheries 2005, 30:10–19. References 24. Martinez PJ, Bigelow PE, Deleray MA, Fredenberg WA, Hansen BS, Horner 1. Mills EL, Leach JH, Carlton JT, Secor CL: Exotic species and the integrity of NJ, Lehr SK, Schneidervin RW, Tolentino SA, Viola AE: Western lake trout the Great Lakes. Bioscience 1994, 44:666–676. woes. Fisheries 2009, 34:424–442. Binder et al. Animal Biotelemetry 2014, 2:14 Page 14 of 14 http://www.animalbiotelemetry.com/content/2/1/14 25. Marsden JE, Casselman JM, Edsall TA, Elliott RF, Fitzsimons JD, Horn WH, Manny BA, McAughey SC, Sly PG, Swanson BL: Lake trout spawning habitat in the Great Lakes - a review of current knowledge. J Great Lakes Res 1995, 21(Suppl 1):487–497. 26. Muir AM, Blackie CT, Marsden JE, Krueger CC: Lake charr Salvelinus namaycush spawning behaviour: new field observations and a review of current knowledge. Rev Fish Biol Fisher 2012, 22:575–593. 27. Henderson NE: The urinary and genital systems of trout. J Fish Res Board Can 1967, 24:447–449. 28. Ganguly S: Reproductive biology in fishes and its diversified physiological mechanisms. Int J Sci Env Tech 2013, 2:401–403. 29. Killick R, Eckley I: changepoint: an R package for changepoint analysis. R package version1.1.1; 2014 [http://CRAN.R-project.org/package=changepoint]. 30. Crossin GT, Cooke SJ, Goldbogen JA, Phillips RA: Tracking fitness in marine vertebrates: a review of current knowledge and opportunities for future research. Mar Ecol Prog Ser 2014, 496:1–17. 31. Chisholm IM, Hubert WA: Expulsion of dummy transmitters by rainbow trout. Trans Amer Fish Soc 1985, 114:766–767. 32. Marty GD, Summerfelt RC: Pathways and mechanisms for expulsion of surgically implanted dummy transmitters from channel catfish. Trans Amer Fish Soc 1986, 115:577–589. 33. Esteve M, McLennan DA, Gunn J: Lake trout (Salvelinus namaycush) spawning behaviour: the evolution of a new female strategy. Environ Biol Fish 2008, 83:69–76. 34. Binder TR, Thompson HT, Muir AM, Riley SC, Marsden JE, Bronte CR, Krueger CC: New insight into the spawning behavior of lake trout, Salvelinus namaycush, from a recovering population in the Laurentian Great Lakes. Environ Biol Fish 2014, doi:10.1007/s10641-014-0247-6. 35. Peake S, McKinley RS, Beddow TA, Marmulla G: New procedure for radio transmitter attachment: oviduct insertion. N Amer J Fish Manage 1997, 17:757–762. 36. Simpfendorfer CA, Heupel MR, Collins AB: Variation in the performance of acoustic receivers and its implication for positioning algorithms in a riverine setting. Can J Fish Aquat Sci 2008, 65:482–492. 37. Welsh JQ, Fox RJ, Webber DM, Bellwood DR: Performance of remote acoustic receivers within a coral reef habitat: implications for array design. Coral Reefs 2012, 31:693–702. 38. Heupel MR, Semmens JM, Hobday AJ: Automated acoustic tracking of aquatic animals: scales, design, and deployment of listening station arrays. Mar Freshwater Res 2006, 57:1–13. 39. Clements S, Jepsen D, Karnowski M, Schreck CB: Optimization of an acoustic telemetry array for detecting transmitter-implanted fish. N Amer J Fish Manage 2005, 25:429–436. 40. Zamora JCL, García-Berthou E: Preliminary telemetry data on the movement patterns and habitat use of European catfish (Silurus glanis) in a reservoir of the River Ebro, Spain. Ecol Freshwater Fish 2007, 16:450–456. 41. Voegeli FA, Lacroix GL, Anderson JM: Development of miniature pingers for tracking Atlantic salmon smolts at sea. Hydrobiologia 1998, 371/ 372:35–46. 42. Smith F: Understanding HPE in the VEMCO positioning system (VPS). Halifax, NS: VEMCO; 2013. 43. Codling EA, Plank MJ, Benhamou S: Random walk models in biology. J Roy Soc Interface 2008, 5:813–834. 44. Gurarie E, Andrews RD, Laidre KL: A novel method for identifying behavioural changes in animal movement data. Ecol Lett 2009, 12:395–408. 45. Madon B, Hingrat Y: Deciphering behavioral changes in animal movement with a “multiple change point algorithm-classification tree” framework. Front Ecol Evol 2014, 2:1–9. Submit your next manuscript to BioMed Central and take full advantage of: doi:10.1186/2050-3385-2-14 Cite this article as: Binder et al.: Use of oviduct-inserted acoustic transmitters and positional telemetry to estimate timing and location of • Convenient online submission spawning: a feasibility study in lake trout, Salvelinus namaycush. • Thorough peer review Animal Biotelemetry 2014 2:14. • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit

Journal

Animal Biotelemetry – Springer Journals

Published: Sep 4, 2014

There are no references for this article.