An inexpensive method for remotely monitoring nest activity

Abstract

ABSTRACT In studies of avian nest success, investigators often face the difficult task of periodically checking nest status while at the same time limiting observer influence on nest survival. Remotely monitoring nests using temperature data loggers is one method that allows for continuous data capture regarding nest status (i.e., active vs. inactive) without the negative effects associated with repeated nest checks. We used small temperature data loggers (Thermochron iButtons) to remotely monitor nests of Long-billed Curlews (Numenius americanus) in northeastern Nevada. Data loggers programmed to record temperature at 10-min and 20-min intervals were placed in curlew nests. Data loggers were set to collect data throughout the nesting cycle to determine onset of incubation and timing of nest failure. On average, Long-billed Curlews began incubating approximately 3 d after the first egg was laid and onset of incubation coincided with the laying of the third egg. iButtons allowed us to determine when incubation was terminated in 17 of 23 unsuccessful Long-billed Curlew nests, including 13 of 17 depredated nests. The presence of iButtons in Long-billed Curlew nests did not affect daily survival rate, egg hatchability or rate of nest abandonment. iButtons are an efficient and practical means for remotely monitoring nests of large egg-laying birds, such as the Long-billed Curlew. En estudios de éxito de anidamiento, los investigadores muchas veces se encuentran con la dificultad de examinar el nido, tratando de limitar la infuencia del observador en la sobrevivencia de las aves. Uno de los métodos de monitoreo remoto es el uso de bitácoras electrónicas de temperatura que permiten tomar datos de forma contínua en un nido (e.g., activo vs inactivo) sin los efectos negativos asociados con el examen periodico del mismo. Utilizamos bitácoras electrónicas de temperatura (Thermochron iButtons) para monitorear nidos de chorlo (Numenius americanus) en el norte de Nevada. El instrumento se programó para tomar la temperatura dentro del nido a intérvalos de 10 a 20 minutos. El instrumento se utilizó para tomar datos a través del ciclo de anidamiento del ave para determinar el comienzo de la incubación y el momento en que el nido fuera abandonado. En promedio los chorlos comienzan a incubar aproximadamente tres dias después de puesto el primer huevo, lo que coincide con la puesta del tercer huevo. iButtons permitió determinar el final del periodo de incubacion de 13 de 17 nidos depredados. La presencia del instrumento en los nidos de chorlo no afecto la tasa diaria de sobrevivencia, tampoco el eclosionamiento o la tasa de abandono. Las bitácoras utilizadas son eficientes y además una forma práctica de monitorear a remoto los nidos de aves grandes tales como el chorlo estudiado. Studies of avian nesting biology and nest success often require researchers to make repeated nest visits. Frequent nest visitation for monitoring purposes, however, may be detrimental because it can reduce nest attendance by parents (Verboven et al. 2001), increase likelihood of nest abandonment (Ellison and Cleary 1978, Götmark 1992), increase egg mortality (Robert and Ralph 1975), and increase the probability of nest predation (Lenington 1979, Westmoreland and Best 1985, Major 1990). To counteract observer influences on nest survival, researchers commonly visit nests infrequently (e.g., every 3 d or weekly) to minimize nest disturbance. Yet, when time-specific effects on nest survival are present, longer between-visit intervals mean less precise information (Shaffer 2004). Daily records of nest status (i.e., active or not active), on the other hand, allow investigators to determine the exact day of nest failure and obtain precise information regarding time-specific effects. Recently, a small (diameter 11.5 mm, width: 6 mm, weight: 3.3 g), rugged, self-contained temperature data logger (Thermochron iButton, Maxim Integrated Products, Inc., Sunnyvale, CA) has been shown to be effective at documenting incubation behavior by recording temperature fluctuations in nests (Badyaev et al. 2003, Cooper and Mills 2005, Cooper et al. 2005). iButtons are preprogrammed to record time-stamped temperature readings (accuracy ± 1°C) at set intervals ranging from once every minute to once every 255 min. Perhaps the greatest advantage of iButtons compared to other temperature data loggers is cost. Thermochron iButtons can be purchased in bulk for as little as U.S. $14 per unit whereas traditional data loggers typically cost more than U.S. $100 per unit. The hardware needed to program iButtons through a personal computer is also inexpensive (approximately U.S. $32) and software is provided at no charge by the manufacturer. Previous studies using iButtons focused on incubation patterns of uniparental, intermittent incubators during focal intervals (e.g., over a 24 h cycle) and not the duration of the nesting cycle. Furthermore, studies using iButtons have focused on species that lay eggs smaller than the iButtons. We examined the effectiveness of using Thermochron iButtons to monitor nests over the entire nesting period in a species that lays large eggs. Introducing foreign objects into or around bird nests for remote-monitoring purposes may increase the conspicuousness of nests to potential nest predators or cause nest abandonment (Pietz and Granfors 2000, Renfrew and Ribic 2003). Artificial objects, such as a camera near the nest, could also have the opposite effect of increasing nest survival (Thompson et al. 1999, Herranz et al. 2002), perhaps by keeping specific predators from approaching the nest. Such confounding factors can lead to erroneous conclusions regarding nest success. Evaluating the effects of remote monitoring equipment on nest survival is therefore critical for extrapolating results to a broader context. Here we report the use of iButtons as a means for monitoring nests of Long-billed Curlews (Numenius americanus). The goals of our study were to (1) examine the effectiveness of iButtons for monitoring nests, (2) determine the onset of incubation, (3) determine the time of nest predation, and (4) examine the possible effects of iButtons on nest survival. Study site and nest searching Our study was conducted in 2004 and 2005 on two large cattle ranches (∼1600 ha) in northern Ruby Valley, Nevada (40°41′52″N, 115°14′00″W; elevation 1800 m). These ranches consist of irrigated hayfields and adjacent herbaceous rangeland habitat that support large numbers of nesting Long-billed Curlews. We initiated nest searches in April, shortly after curlews arrived at the site, and continued searching for nests through mid-June. Nest searches involved observing curlews in the act of nest building and observing individuals walking to nests and assuming an incubation posture. After discovery, nests were monitored from a distance every 3–4 d until failure or hatching. In addition, nests were visited periodically when we attempted to trap adults. For each nest we calculated nest initiation date (NID). When nests were found with one egg, we assumed that day to be the date of initiation. For nests found with 2–3 eggs, we estimated NID by counting back and assuming egg-laying intervals of 1.5 d (see Discussion). When nests were found after clutch completion, eggs were floated to determine stage of development (Westerskov 1950, Hays and Lecroy 1971) and estimate clutch completion date. We then counted back, assuming 5 d for egg laying, to obtain a NID estimate. iButton installation and data capture We used Thermochron iButtons (Model DS1921G) to monitor 44 Long-billed Curlew nests during 2004 (N= 21) and 2005 (N= 23). iButtons were placed in nests opportunistically when nests were discovered. As such, the timing of iButton installation varied from shortly after the first egg was laid to 2 weeks into incubation (mean nest age when iButton installed = 5.47 ± 0.83 [SE] d). However, in 31 of 44 nests, iButtons were installed prior to clutch completion (i.e., before the fourth egg was laid). We used two methods to install iButtons. In 2004 unaltered iButtons were placed in the nest bowl and lightly covered with existing nest materials (grasses and forbs). However, several iButtons installed in this manner were removed either by the incubating parents or a nest predator. Once removed from nests, iButtons could not be recovered and data were lost. In 2005 preprogrammed iButtons were glued to the top of 25-cm long aluminum stakes and covered with 1-mm thick black polyester. A small length of black nylon thread was used to secure the polyester to the iButton and stake. Stakes were then pushed by hand into the center of the nest until the covered iButton was flush with the bottom of the nest. No nest materials were used to cover iButtons in 2005. This method was effective in retaining iButtons after predation events and all iButtons installed in 2005 were recovered. To minimize the number of nest visits and disturbance, we programmed iButtons to collect data less frequently than in previous studies (Badyeav et al. 2003, Cooper and Mills 2005). In 2004 iButtons were programmed to collect data every 10 min. At this frequency iButtons collected temperature data for 14.2 d before memory was exhausted. Because it takes 33 d from the laying of the first egg until hatching (5 d for egg laying and 28 d for incubation), an iButton placed in a nest during egg laying that survived until hatching needed to be replaced twice to cover the entire life of the nest. In 2005 we programmed iButtons to collect data every 20 min, allowing iButton memory to last 28.4 d. At this frequency iButtons had to be replaced at most only once to cover the entire life of a nest. iButtons were exchanged opportunistically and usually when we attempted to trap adults at nests. One limitation of the data collection frequencies we used is that it did not allow for the detection of periods when an incubating bird left the nest for only a few minutes. This was a necessary trade-off to limit disturbance at nests while collecting data throughout incubation. Analysis After retrieval from nests, iButton data text files were converted into Microsoft Excel format, plotted as a line graph, and assessed visually. Temperature readings at each nest were compared to measures of ambient temperature derived from an iButton placed on the ground at a central location on the study site (2004) or from a RAWS weather station located approximately 60 km from the study site (2005). This station is in the same valley and at the same elevation as our study site and experiences similar weather patterns. Because the centrally located ambient-temperature iButton on the site was destroyed by ranching equipment in 2005, we used weather station data for comparison purposes. All ambient temperature readings were collected at hourly intervals. We conducted observations from a distance for all nests with iButtons throughout the incubation period. For each observation we recorded time of day, duration of observation, and whether a curlew was at the nest. These data were then matched with time-stamped data obtained from the iButtons to give us temperature readings at the nest during known on and off bouts. Using this information, combined with measures of ambient temperature, we assigned nests as being attended or not attended during each time-stamped temperature reading. Timing of onset of incubation was determined by following iButton nest temperature plots and ambient temperature plots forward through time until the two signals diverged. Prior to incubation, nest attentiveness consists of sporadic, short duration on bouts. Temperature signals prior to incubation, therefore, closely follow ambient temperature. When incubation began, nest temperature became relatively constant, whereas ambient temperature continued to fluctuate with the daily cycle. Timing of termination of incubation, or nest failure, was assessed by following iButton nest temperature plots and ambient temperature plots backwards through time until the two signals diverged. This yielded the last moment the nest was being incubated. Temperature signals from nest iButtons and ambient temperature readings after this event were similar and both followed the cyclical daily fluctuations in ambient temperature. To test for potential effects of iButtons on nest survival, we compared daily survival rates of nests with iButtons to control nests without iButtons. We included only nests that were depredated, abandoned, trampled by cattle, or where eggs hatched. Nests with iButtons that were flooded (N= 3), destroyed by farm equipment (N= 1), or that failed due to trapping activities (N= 1) were not included in survival analysis. Similarly, control nests in these categories were not considered. All nests that received iButtons and met the above criteria were included in survival analysis, regardless of whether the iButton was recovered or not. We used nest survival analysis in Program MARK (White and Burnham 1999, Dinsmore et al. 2002) to model daily survival rates of Long-billed Curlew nests. We considered eight a priori models: (1) a model with constant daily survival rate (DSR), S(·); (2) a model where DSR varied by the treatment effect of iButton presence/absence, S(iButton); (3) a model where DSR varied by year, S(year); (4) a model where DSR varied by NID, S(NID); (5) a model where DSR varied by year and iButton presence/absence, S(year + iButton); (6) a model where DSR varied by a year and iButton presence/absence interaction, S(year + iButton + year×iButton); (7) a model where DSR varied by iButton presence/absence and NID, S(iButton + NID); and (8) a model where DSR varied by an iButton presence/absence and NID interaction, S(iButton + NID + iButton × NID). We selected the best fit models using Akaike's Information Criterion corrected for small sample size (AICc; Akaike 1973). Although most iButtons were installed in early season curlew nests and in nests found within a few days of initiation, we continued to discover curlew nests throughout the breeding season. As a result curlew nests without iButtons had greater mean NID and age at discovery than curlew nests with iButtons. To control for potential difference in nest survival associated with these discrepancies, we included as controls only nests for which NID and nest age at discovery values were within the range of values of nests with iButtons. Another priority was to maintain equal nest visitation rates among iButton nests and control nests to avoid confounding the effect of iButtons on nest survival with number of visits to the nest. To maintain equal nest visitation between iButton nests and control nests, iButtons were installed and replaced during nest visitation events that took place at all nests regardless of treatment category (e.g., initial finding of the nest and adult trapping attempts). We used logistic regression to test whether assignment of nests to the treatment and control groups controlled for specific variables by setting the response variable as iButton nest or control nest and introducing predictor variables of NID, nest age at discovery, and number of nest visits per number of exposure days. Significant predictor variables in the logistic model would indicate a bias in the assignment of iButton and control nests which could potentially influence any differences in nest survival between groups. Logistic regression analysis was performed using PROC LOGISTIC (SAS Institute 2004) with significance level set at α= 0.05. Values are presented as mean ± 1 SE. We recovered iButton temperature data from 36 of 44 Long-billed Curlew nests (13 of 21 nests in 2004 and 23 of 23 nests in 2005). Eight iButtons were removed from nests by curlews or nest predators in 2004 and no data were acquired. At two nests that survived until hatching in 2004, curlews removed replacement iButtons installed shortly before hatching. As a result, we did not obtain data for the last 8–9 d of incubation at these nests. At another nest predated in 2004, the initial iButton was recovered, the first replacement iButton was removed by the incubating birds, and a second replacement iButton was recovered. This resulted in a 7-d gap in the data at this nest. In 2005, all iButtons were recovered. Onset of incubation Of 36 recovered iButtons, 27 were installed prior to clutch completion, with 13 installed with one egg in the nest, eight with two eggs in the nest, and six with three eggs in the nest. Onset of incubation was determined for 23 nests by documenting the day when nest temperature became regular (i.e., nest temperature ceased to follow the cyclical pattern demonstrated by ambient temperature). The other four nests were either predated or flooded prior to clutch completion and onset of incubation. For nests where iButtons were installed when one egg was in the nest, incubation began about 3 d later (= 2.8 ± 0.3 d, range: 1–4 d, N= 11). For nests where iButtons were installed when two eggs were in the nest, incubation began approximately one and a half days later (= 1.6 ± 0.2 d, range: 1–2 d, N= 7). At two of these nests, we observed onset of incubation to coincide with the laying of the third egg. Finally, for nests where iButtons were installed when three eggs were in the nest, there was no delay in the onset of incubation at all but one nest (= 0.2 ± 0.2 d; Range: 0–1 d, N= 5). Timing of nest predation Of the 36 nests where iButtons were recovered, 17 were depredated, 12 hatched, three were flooded and abandoned, two were trampled by cattle, one was abandoned after excessive cattle disturbance, and one failed because of egg damage during a trapping attempt. Of 17 predated nests, we determined the time of nest predation for 13 by documenting the hour when incubation was terminated. For the remaining four nests, we were unable to determine when predation occurred because eggs were not being incubated. At two nests, clutches had not yet been completed when nest predation occurred. At the other two nests, the parents were intermittent incubators, probably because of rising water levels in and around the nest. iButton temperature data for one depredated nest is shown in Figure 1. Nest 003-2004 was depredated between 20:00 and 21:00 on 18 May 2004. At 20:30, the temperature in the nest dropped precipitously and mirrored ambient temperature until the iButton was recovered a few days later. Nest iButton temperature data (▪) and corresponding ambient temperature (□) for Nest 003-2004. See text for details. Of 13 predated nests where we could determine when incubation was terminated, seven (54%) were predated between 17:00 and 22:00 and five between 03:00 and 10:00. No nests with iButtons were predated between 10:00 and 17:00. In addition, iButtons revealed the time of nest failure for two nests trampled by cattle, the nest abandoned due to cattle disturbance, and one of the three flooded nests. Incubation had not yet begun in the other two flooded nests so we were unable to determine the time of failure. Nest survival and egg hatchability We modeled daily survival rate for 88 Long-billed Curlew nests discovered in 2004 and 2005 (39 with iButtons and 49 without). Although iButton data often yielded the exact day of nest failure, the same data were not available for nests without iButtons. As a result nest survival analysis was based only on data from nest checks or trapping attempts that were the same for iButton nests and control nests. Logistic regression analysis indicated that nests with iButtons and control nests without iButtons did not differ with respect to the examined variables (model likelihood ratio χ23= 2.94, P= 0.40; all individual predictor variables n.s.). Therefore, our assignment of nests to the iButton treatment and control treatment was unbiased and accurately tested for the effect of iButtons on nest survival. Nest survival was best explained by a model with a constant daily survival rate (DSR = 0.965 ± 0.006) that did not include an effect of iButton presence or absence (Table 1). Models incorporating an effect of iButton on nest survival received little support (∑w(iButton)≈ 0.35). The model consisting of a constant daily survival rate with an iButton effect (S(iButton)) was greater than two delta AICc values from the top model that did not contain an iButton effect (S(·)). A likelihood ratio test between these two models revealed that iButton presence did not significantly affect Long-billed Curlew nest survival (χ21= 0.0, P= 0.99). Estimates of daily survival rate of iButton nests and control nests under the model with an iButton effect (S(iButton)) were almost identical (0.9658 ± 0.006 and 0.9659 ± 0.006, respectively). The model incorporating a year × iButton interaction (S(year+iButton+year × iButton)) also received little support, suggesting that the method of iButton installation did not affect nest survival. We observed no effect of iButtons on egg hatchability (two-sample t-test: t= 0.85, P= 0.41), with 79% of eggs from successful nests with iButtons hatching compared to 86% from successful nests without iButtons. If a nest where only one of four eggs hatched because of disturbance caused by ranching activities is omitted, hatchability for nests with iButtons was 84%. iButtons provided an effective and safe way to monitor Long-billed Curlew nests and to ascertain onset of incubation, daily nest status, and timing of nest failure. Onset of incubation was determined for all nests where iButtons were installed prior to clutch completion that survived until clutch completion. In our study most Long-billed Curlews began incubating clutches after the third egg was laid and before the fourth egg was laid. Similar behavior has been reported in some species of sandpipers (Holmes 1973, Ashkenazie and Safriel 1979, Oring et al. 1986, Oring et al. 1988) and may also occur in others (Hagar 1966, Marks et al. 2002). iButton data also revealed the approximate egg-laying intervals of Long-billed Curlews. Incubation generally began with the third egg, and it took approximately 3 d (after the first egg is laid) and 1.5 d (after the second egg is laid) for incubation to begin, suggesting that curlews lay, on average, one egg every 1.5 d. This corresponds to an average egg-laying period of 4.5 d from the first egg to the fourth egg. Our data also showed that onset of incubation varied from 1 to 4 d after the first egg was laid, suggesting that females vary in egg-laying frequency and that some pairs may begin incubating prior to the laying of the third egg. As with periodic nest visits, iButton data cannot differentiate between abandonment that precedes nest predation and predation events. Curlews may have abandoned nests with predation following at a later time. Nest abandonment, however, is rare in Long-billed Curlews (Redmond and Jenni 1986), especially after clutch completion. Of four nests with iButtons that were abandoned in our study, three were flooded and the other disturbed by grazing cattle. For all nests in 2004 and 2005, abandonment for reasons other than cattle disturbance or flooding was rare (three of 137 nests), suggesting that predation caused termination of incubation in most cases. Compared to periodic nest visits, iButton data may increase the accuracy of assigning nest fates. For example, in Ruby Valley, farmers sometimes rake fields and, if present, curlew nests are invariably destroyed. During a standard nest check in 2005, a curlew nest was found to have been raked. However, data from the recovered iButton revealed that the nest had failed a day before raking occurred, probably due to predation. Thus, using only periodic nest visits, this nest would have been incorrectly classified as destroyed by raking. Installation of iButtons iButtons loosely placed in Long-billed Curlew nests were often removed from the nest by incubating birds or predators. We attached iButtons to metal stakes in 2005 and none were lost. Wrapping iButtons in black stocking material also made iButtons less visible and removed the need to cover shiny iButtons with nest materials. Regardless of installation method, iButtons had no apparent impact on nesting Long-billed Curlews. Thus, nest survival results obtained from nests with iButtons were indicative of all Long-billed Curlew nests in our study. In addition, no Long-billed Curlews abandoned nests because of iButtons and no eggs were damaged by iButtons. Our methods did not provide an accurate representation of nest temperature. Long-billed Curlew eggs are large (as much as 73 mm by 52 mm, unpubl. data). As a result, 6 mm tall iButtons at the bottom of a curlew nest were a few centimeters from the brood patches of incubating birds. Thus, temperatures recorded by iButtons were considerably less than the expected temperature of the eggs during incubation. If measures of nest temperature are needed, data loggers encapsulated in artificial eggs would be more suitable (e.g., Flint and MacCluskie 1995). We feel our methods were effective at determining presence or absence of an incubating bird, especially when used in combination with some measure of ambient temperature. However, the ambient temperature at our study site averaged 12°C from April to June. With such low ambient temperatures, the temperature increase associated with an incubating bird on the nest was striking, and aided our assignment of incubation activity. At locations where ambient temperature closely resembles iButton temperature readings in the nest, assignment of incubation status would be problematic. We thank the gracious landowners in Ruby Valley, Nevada for allowing us to conduct this study on their land. We also thank our many field technicians for their assistance in locating curlew nests. Funding was provided through the Nevada Arid Rangelands Initiative of the United States Department of Agriculture, the Nevada Department of Wildlife and the Nevada Agricultural Foundation. We thank C. Cooper and one anonymous reviewer for their insightful comments on a previous version of this manuscript.