Below is a report I wrote for a two-month internship with a graduate student at Boise State University. I worked in the field conducting owl, woodpecker, and snag surveys in the Boise National Forest.
Owl and woodpecker research in the Boise National Forest, Idaho
June 2, 2011
In the spring of 2010, I assisted Micah Scholer with field work for his graduate thesis. Micah is a second year graduate student in the raptor biology program at Boise State University. He designed an experiment to determine whether diurnal bird species, specifically woodpeckers, can help predict the occurrence of nocturnal flammulated owls (Otus flammeolus). His research also sought to improve the accuracy of models for predicting the occurrence of flammulated owls. The ecological importance of woodpecker and flammulated owl occurrence lies in the cavities created by woodpeckers which can be used in subsequent years by different vertebrates, including flammulated owls. Most of my internship hours were spent in the field, working on three aspects of data collection: flammulated owl surveys, woodpecker surveys and snag surveys. The rest of my hours were spent on data entry.
This research took place throughout the Boise National Forest, located in southern Idaho.
The following methods refer only to the 2010 field season.
Surveys were conducted in 67 sites throughout the forest. Each site was marked with a central GPS point, and the entire site spanned out in a circle from this point. Each circular site had a 400-meter radius, making the area of each site 50.2 hectares. All point-count locations were separated by at least 1 kilometer to help ensure that the same individuals were not being counted at more than one survey site.
Woodpecker Point-count Surveys
Each site was surveyed three times for woodpeckers. The five woodpecker species detected for were hairy woodpeckers (Picoides villosus), Lewis’ woodpeckers (Melanerpes lewis), northern flickers (Colaptes auratus), pileated woodpeckers (Dryocopus pileatus) and red-naped sapsuckers (Sphyrapicus nuchalis). Surveys began as early as a half hour before sunrise, and ended before 11 am. Each survey was 12.5 minutes long. The first five minutes of the survey were spent silently listening for birds. Then one of five bird calls was played for fifteen seconds, followed by 30 seconds of silent listening, then another 15 seconds of calls from the same species, followed by another 30 seconds of silent listening. This pattern was repeated for each species. A FoxPro wildlife caller was used to broadcast the bird calls. The order in which the bird calls were played was randomized using a dice to determine one of five possible sequences of calls. Birds were detected visually and audibly.
At each survey, we recording average wind speed, temperature, barometric pressure, humidity, cloud cover, noise, and the presence and type of precipitation. Wind speed, temperature, barometric pressure and humidity were measured using a Kestrel 3500 Weather Meter. We measured noise with a decibel meter.
At each survey, the observer’s name was noted. The observer at each survey affected the detection probability because some observers are more experienced at identifying bird calls than others. For example, Micah is better at identifying a woodpecker drumming from a long distance than I am. Because I performed some of the surveys alone, this difference in detection probability is factored into the statistical analysis.
For each individual woodpecker detected during surveys, we recorded the approximate distance of the bird from the survey point, as well as the method of detection. Method of detection indicated how we observed the bird, whether by hearing their call or spotting them visually. Only individuals within a 400 meter radius of the survey point were recorded.
Owl Point-count Surveys
Flammulated owls were surveyed for during June 2010. All sites were surveyed once for flammulated owls. Previous field experiments in the study area confirmed that the probability of detection is very high during the month of June (Barnes and Belthoff 2008).
Each survey began at least 30 minutes after sunset. Each point-count began with a three minute listening period. This was followed by a 30-second broadcast of flammulated owl calls, a one-minute listening period, and an additional 30-second broadcast. The surveys concluded with a two-minute listening period.
At each survey, we recording average wind speed, barometric pressure, humidity, temperature, noise, and presence and type of precipitation. Wind speed, temperature, barometric pressure and humidity were measured using a Kestrel 3500 Weather Meter. Noise was measured using a decibel meter. The name of the observer was also noted.
We only surveyed for flammulated owls during periods of no precipitation. If it began raining during a survey, we usually finished the survey, but returned to repeat the survey when it was not raining.
We conducted snag surveys to determine if the area within a 400 meter radius around a point contained suitable nesting and feeding sites for owls and woodpeckers.
At each point, we randomly chose a compass bearing. From that bearing, three others were calculated, each within a different quadrant defined by the cardinal directions. The first randomly chosen bearing was assigned to a 400 meter transect, and the others were assigned to a 300 meter, a 200 meter and a 100 meter transect. All transects were 10 meters wide. Carrying a compass, DBH (diameter at breast height) tape and a GPS unit, we walked each transect for the required length. We noted any snag within the transect. For each snag, we recorded its DBH and distance from the survey site. We also stopped at every 50 meter interval to record our distance from the nearest snag.
We only recorded snags that were taller than two meters, greater than 15 cm in DBH, and capable of serving as a nesting tree for owls and woodpeckers. To be considered a capable nesting tree, the snag showed no signs of excessive deterioration, was not split from crown to base, and was able to stand without support from surrounding trees. We also surveyed several of the sites by walking 400-meter transects for all four of the transects of a single site. The results of this more robust survey (four 400-meter transects) were compared to the initial survey strategy (400, 300, 200 and 100-meter transects). Micah found that the difference in snag abundance between the two survey methods was not statistically significant.
During my time on this project, Micah experimented with two methods for measuring canopy cover. The initial strategy for measuring canopy cover was performed in the field. While walking transects for the snag surveys, we stopped every 25 meters to record the overhead canopy cover. If there was canopy cover directly above us, the point received a score of 1. If there was no canopy cover directly above us, the point received a score of 0. The faults of this method included surveyor bias, and the area surveyed represented a very small portion of the entire site.
The second strategy was to use GIS techniques to determine canopy cover. Micah downloaded satellite images of the areas being surveyed, then used a buffer tool to draw in the circular perimeter of each site. A “fishnet” was drawn over each circle, which was essentially a grid of 30 x 30-meter hollow squares. If vegetation occupied more than 50 percent of the square, that square would receive a score of 1. Squares with less that 50 percent vegetation were given a score of 0. We then added the number of squares that scored a 1, to give each site an overall score. Because each site contained the same number of squares, these overall scores could be compared between sites. The problem with this method was that the images were not always clear enough to distinguish between shadows and vegetation. Another problem was that vegetation in the photos was not always a clear indicator of canopy cover. It was often difficult to distinguish between ground cover and canopy level vegetation.
I was not involved with any data analysis in Micah’s project. I produced results from point data and a land cover image that Micah provided to me.
By conducting GIS data queries, I analyzed land cover of the survey sites used in the 2009-2010 field seasons. I calculated the percent of sites within each land cover type. I performed the same calculation for sites where a flammulated owl was detected. The goal of this analysis was to expose any land cover type which flammulated owls may be more likely to occupy.
Flammulated owls occupied a greater percent of sites dominated by ponderosa pine and douglas-fir trees than other land cover types. (Figure 2).
Figure 2. Percent of survey sites dominated by a particular land cover type, for all survey sites and sites where a flammulated owl was detected.
Working one on one with a graduate student gave me valuable insight into the process of designing and carrying out an experiment. Because I participated in this experiment in the spring of 2010, my involvement was limited to the data collection process. I had no real involvement in the analysis of the data, besides what Micah and I discussed during our hours of driving to field sites. Most of my insights on this internship focus on issues with data collection.
Within days of beginning the internship, I began to realize the difficulties of field research. With any field experiment, weather is sure to play a part. Beyond forcing the researcher to work in miserable conditions, I learned that weather plays a larger role in field experiments: it can drastically alter data, and can sometimes make data collection a near-impossible task. Micah had expected to complete most of the woodpecker surveys during May. May in Boise, ID is usually warm and sunny, making for great surveying weather. In 2010, southern Idaho experienced one of its rainiest Mays on record. We were often forced to cancel trips to the field due to excessive rain or snow. When June came around, we were incredibly busy, making up for lost time on woodpecker surveys as well as completing all the flammulated owl surveys.
Other difficulties with data collection included noise and wind. Most of our points were placed along remote forest service roads, which typically run through valleys and alongside streams. Doing field work during the spring melt-off made it very difficult to hear bird calls over raging streams. Besides the noise from streams, the acoustics within valleys often distorted the broadcast calls. Sometimes we were unsure if our calls could even be heard 400 meters away due to steep valley walls. Another issue with being situated along roads was noise from other vehicles. This was especially problematic at survey points along the road to a popular ski resort outside of Boise.
Flammulated owl surveys presented a unique set of difficulties. Throughout most of June, Micah noticed a decline in owls detected from the previous year. This was frustrating because we had planned to survey each site only once. Micah noted that there seemed to be fewer insects out compared to last year’s field season. We hypothesized that this was due to the unusually cold weather. The primary source of food for Flammulated owls is moths. If the cold weather was negatively impacting their food sources, the owls might not be as likely to respond to broadcasts because they are conserving energy.
Snag surveys had issues of their own, mostly due to variability in terrain. As I mentioned, most of our points were alongside streams and in steep valleys. We often had to cross streams and snow fields to complete transects. We also faced cliffs and incredibly thick underbrush in burn areas. All of these obstacles made it difficult to stay exactly on our bearing throughout an entire transect.
As I mentioned in the methods section, our procedures changed as research progressed. Micah experimented with changing the distance of transects for snag surveys and the method for quantifying canopy cover. I learned how difficult it can be to predict the best way to conduct an experiment before it even begins, and the frustration of sustaining the same procedures knowing those might not be the most robust or precise methods.
The unexpected obstacles that Micah faced throughout his research process added to the extent of the project. I learned that a research project can take two or three times as much time and work than originally predicted, when all the uncertainty is factored in.
Overall, I enjoyed the time I spent working on this project. Being directly involved in field research gave me a new understanding of the interconnectedness of published research projects. At times I got frustrated because our research covered such a small niche in the natural environment. Eventually, I realized that specialized research projects build the foundation for understanding connections within the broader environment. Ecology is the study of how species interact with each other and abiotic elements to maintain an ecological community. I have a much deeper appreciation for these connections, now that I have observed them first hand.
Barnes, K. P. and J. R. Belthoff. 2008. Probability of detection of flammulated owls using nocturnal broadcast surveys. Journal of Field Ornithology 79:321-328.
Scholer, M. N. and J. R. Belthoff. 2010. Occupancy and habitat associations of forest owls in the Boise National Forest, 2009 Interim Report.