Kyra Bankhead, undergraduate student
1 April 2022
This month has been a waiting game in the preparation of sorting out my data analysis. Grace Freeman has been my savior in helping me untangle my data at the waterfront and marina, and meeting with her through this process has been extremely helpful. As it stands right now, it seems “day of year” and “noise level” are significant as fixed factors and site as a random factor. The interaction between noise and tide is also significant. This is all in a GAMM based on a negative binomial family fit using REML. I am excited to have a final meeting with her once she has finished working on getting the most parsimonious model.
April will be a very busy month, with scholar's week and the society of marine mammal conference coming up. I hope I will be very close to getting my manuscript done so that I can have everything wrapped up to present my research at both events. I am also applying to graduate school at OSU, and plan to work in Dr. Cantor’s research lab. I’ve been very busy trying to get funding for grad school so this will add to the busy April load. I’m very excited for all that is to come for my last quarter at WWU.
Madison Gard, undergraduate student
1 April 2022
Spring is here at last - I’m loving the warming weather, extra sun, and cherry blossoms! Over spring break, I was lucky enough to visit Cholla Bay, Mexico and soak up the sun while spending quality time with family. Now I’m feeling recharged and ready for everything that spring quarter brings :)
In the lab we’ve had some promising results for the manuscript. One plate was run with the California sea lion samples, the new Taqman assays, and adjusted settings on the qPCR instrument and every sample was detected how it should! We’ve had better results with the most recent Steller sea lion sample plates as well. It’s looking like we have our methods together now to begin writing that section of the manuscript. This also means we can start running plates with Zoë’s thesis samples for some fresh data that may validate differential prey consumption of male and female Stellers! Exciting things!
Writing wise - I’ve worked on an outline for my introduction and drafted an abstract to register for the NWSSMM conference. I’ll also need to get my poster together for the upcoming conferences soon. I’m hoping to have a first draft of my introduction and methods completed by the end of April.
In my classes, I presented in the 3rd Latin American Studies Forum at WWU on March 3rd about the Trans Community Network activist group in Bogotá, Colombia for my Culture of Latin America course. I also took a lot away from Dr. Sobocinski’s Biostatical Analysis course. It was neat to learn so much about statistical tests and which experimental designs they apply best to. I’m excited yet nervous for this next set of courses and their challenges, while looking forward to spring weather and getting some quality hikes in!
Zoë Lewis, graduate student
1 April 2022
Well, finally Maddie and I have validated our sex determining probes, and we are starting to identify the sex of some of the wild samples for my project! After much trial and error, we discovered that the samples we received from VANAQUA had some issues, and discovered that these Steller sea lion samples that we had been trying to use to validate our probes were not working due to DNA quality. As soon as we tried these samples with the California sea lion scats, we found that both the Harbor seal probe and the Steller sea lion probe work fantastic on the California sea lion scats! At first, we were a bit bummed to reinvent the wheel, but this has ended up being a blessing in disguise… after comparing the Steller sea lion probe with the Harbor seal probe on wild samples, we learned that the newly returned probe detects a higher number of samples with y, resulting in more male scats. This is really great news; with this improved method, I have higher confidence that we aren’t accidentally calling male scats female scats, due to the lack of signal from the SRY probes. So, our rigorous work and validation is starting to pay off!
In between running lab experiments, I’ve received my hard parts analysis, and have been working hard to draft up my results without the sexing yet. Although this is preliminary, since sex will be an important component of my modeling, I am excited to have a clear picture of what my statistics will look like, and already have code built to analyze my dataset once it is complete. Along with the hard parts analysis came a few more questions; there were only a few salmonid samples with otoliths, and thus the method developed in Thomas et al. 2017 will not work for determining the proportion of juveniles vs. adults within our sample. Further, given the demographics of salmonids in the region, I’m digging into the otolith identification methods a bit deeper. The salmonid population on the coast is full of salmon of all species, ages, and life stages. It is hard to guarantee that “smaller” otoliths indicate a “younger” fish in these circumstances… so I want to make sure I really understand these age assignments if I am going to include them in my project.
On to April, which will include lots of conference prep and a pretty complete results draft by the end of this month!
Holland Conwell, undergraduate student
1 April 2022
March flew by! This last month, I discovered 26 samples from Madelyn Voelker’s original dataset that did not have DNA diet fractions summing to 1, as I had expected. I met up with Madelyn to discuss these anomalies, and although she thought this could be due to the deletion, at some point, of rare prey taxa, we did not determine the cause for sure. Instead, per Dietmar’s recommendation, I will be going through these 26 samples by hand and checking each DNA diet fraction with Austen Thomas’s original dataset.
I also deleted all sites with no sexing information, which turned out to be quite a few. Additionally, during my meeting with Madelyn, I asked about some strange mismatches within the duplicated Baby Island data. Unfortunately, Madelyn also was not quite sure why this might be, but she suggested opting for the Baby Island data that did not come from her dataset for analysis. It’s looking like once I remove that Baby Island duplicate and go through the previously-mentioned, strange 26 samples from Madelyn’s dataset, I’ll hopefully be done with major data cleaning!
In the meantime, I’ve been working on building my data analysis codes in RStudio with my current version of the dataset. I chose to begin with a logistic regression in RStudio, using sex ratio as the designated response while looking at possible predictors. I tested ten models with different combinations of predictors (month, year, site, season, and interactions) then compared the AIC values from each model. Model 8 (month+site+year+month*site+month*site*year+month*year+site*year) ended up having the lowest AIC value and thus seemed to be the best fit model. I then calculated the accuracy of this test, getting 70% accuracy. Given that I still need to do some final data cleaning, I’m taking all results of this test with a grain of salt and anticipating that this accuracy may change the next time that I do a logistic regression. Not to mention, I’m certainly still workshopping this analysis with Zoe, who I owe many thanks to for insight and some great resources. All that said, I’m excited to be finally working on data analysis! I anticipate that once I polish up this first look at predictors of sex ratio, I’ll then move on to looking at how variables such as sex might be impacting diet proportions.
Kathleen McKeegan, graduate student
1 April 2022
March flew by! I don’t know where the time went. Throughout March, I focused on completing the Photo IDs for fall 2021. At this point, almost all the observation days have been completed, with just a few tricky days remaining. Once those last few days are completed, I can begin comparing seal presence and foraging success across years (2019 – 2021). I’m itching to get started on that part of my project, so that will hopefully be completed soon.
A few weeks ago, Alejandro made a connection with Dr. Elliser, the Founder and Research Director of Pacific Mammal Research in Anacortes, WA. She reached out because she and her team are refining the WildMe algorithm to work effectively at photo IDing harbor seals. To help train the AI system, I compiled high quality photos of 55 known individuals from Whatcom Creek across several days or years. Once the AI system is trained, we can start using it in our lab to help expedite the photo ID process. We’ve tried photo ID software in the past, including Discovery and WildID, which both worked but had various issues that lead to us abandoning the use of the system. I’m very hopeful that, with training and finessing, WildMe will be a great tool for our lab in future years!
Otherwise, I have been focusing on processing photos, working on data analysis (as always), and writing the first full draft of my results section. I was recently accepted to give a speed talk at the North American Conference for Conservation Biology in July and will be submitting an abstract to present at the Northwest Student Chapter for the Society of Marine Mammalogy virtual conference at the end of April. I’m starting to feel the anxiety that comes with wrapping up a project. I have about 5 months left in this program before I officially (hopefully) graduate, and I have a lot to do in that time! Luckily, I have an amazing team of undergraduates, fantastic lab partners (shout out to Zoë and Kate!), and an incredibly helpful advisor, so I know I’ll be able to finish on time. Hopefully I will have my results completed and an outline of my discussion section done by the next blog!