Good question. Several obstacles to widespread adoption come to my mind:

[1] Studies using specialized telemetry (heart-rate, microphone, etc.) have difficulty--at this time, at least--making the link between mechanisms and the kind of large-scale ecological processes that many are interested in (e.g patterns of selection, population viability, etc.). Many telemetry solutions, such as transmitters used for simple location data, are probably the exception since they can allow researchers to be informed about dispersal patterns or juvenile survival, etc.

[2] A corollary of #1 is the cost of specialized transmitters, which fortunately continues to decrease. Studying some processes like selection demand large sample sizes; thus, the costs of doing business using specialized telemetry would be prohibitive.

[3] Large data: This is a strange obstacle, but I think we could argue that the presence of large data and the act of data mining has downplayed the importance of mechanistic studies (for which specialized telemetry could be used). I'm seeing more and more graduate students leaning on public data (previously published dataset in the public domain, species distributions) rather than focusing on empirical data collection from natural systems.

Anyway, I just whipped out these thoughts. I will be very interested in hearing more from others.

Another issue is the validation studies that need to be carried out to verify that the output of such systems matches actual physiological or behavioural data for each species. It would be nice to presume that simply fitting a system to an animal would generate a load of data that can be analysed straight away, but without proper validation, results can't really be published.

This is a great and valid question, and J. Patrick Kelley's answer brings up good points. I will reiterate a few of those (I apologize for the repetition), as well as hopefully add some relevant thoughts. Please note that due to my background, I am speaking from the aquatic/marine side of things.

- To extend J. Patrick Kelley's first point - A lot of initial telemetry work was very descriptive (i.e. they only provide estimates of movement, dispersal, behaviours), without testing any specific hypotheses. Descriptive science in ecology is often looked down upon, as many feel that to advance the science we need to be experimental and hypothesis-based. While I generally agree with this sentiment, we cannot forget the importance of natural history of organisms as well. Back to telemetry - I think we are slowly seeing a shift to where telemetry is being used to answer specific hypotheses, and furthermore, is starting to be used in experimental settings (which I think is the real strength of telemetry - that we can start to use environment as a living laboratory). The progression has been slow but I believe it's real and improving. In addition, we are starting to see more papers combine approaches (telemetry and biopsies, telemetry and remote sensing/GIS, telemetry and experimental treatments, etc).

- I think a part of the issue has also been how novel the types of data are; although sample sizes of individuals (the sampling unit in most analyses) are small, the number of data points can be astronomical depending on the type of tag used and the duration. It has taken awhile for researchers to respond with appropriate analyses to extract as much pertinent information and knowledge from the data. See the numerous recent papers on state space models (SSM) by Breed and Jonsen that have been applied to telemetry.

- This is related to the above, but it always takes awhile for researchers to understand the limitations and potential pitfalls of their approach. More researchers are now understanding the influence of array design, tag limitations, and data quality issues that are common with telemetry, and are designing their studies as such to improve quality.

- Another huge gap that needs to be narrowed is the issue of tagging effects. Researchers using telemetry need to do as much as possible to assure that the tagged organism acts in the same way as an untagged organism would. I generally feel that this argument is overblown, as every technique we use in science can impact an organism's behaviours (i.e. mesocosms, any laboratory work that involves holding organisms, smaller external tags) and in these fields are rarely noted or studied, but it still remains a significant gap that we should attempt to address.

- Even with all of this, I would say a lot of folks have been really successful at using telemetry to answer important questions AND to publish in high-impact journals. I do think it's improving, and more and more labs that I know are adding the method to their "toolbox." Below are just some examples from the past couple of years in what I would consider to be high impact journals (although some are a bit too new for IFs possibly).

Maxwell, S. M., et al. (2013). Cumulative human impacts on marine predators. Nature communications, 4.

Papastamatiou, Yannis Peter, et al. "Telemetry and random walk models reveal complex patterns of partial migration in a large marine predator." Ecology (2013).

Block, B. A., et al. "Tracking apex marine predator movements in a dynamic ocean." Nature 475.7354 (2011): 86-90.

Matich, Philip, and Michael R. Heithaus. "Multi‐tissue stable isotope analysis and acoustic telemetry reveal seasonal variability in the trophic interactions of juvenile bull sharks in a coastal estuary." Journal of Animal Ecology 83.1 (2014): 199-213.

Hazen, Elliott L., et al. "Predicted habitat shifts of Pacific top predators in a changing climate." Nature Climate Change (2012).

Rechisky EL, et al (2013) Influence of multiple dam passage on survival of juvenile Chinook salmon in the Columbia River estuary and coastal ocean . Proceedings of the National Academy of Sciences of the United States of America 110(17):6883-6888,

Bestley, Sophie, et al. "Integrative modelling of animal movement: incorporating in situ habitat and behavioural information for a migratory marine predator." Proceedings of the Royal Society B: Biological Sciences 280.1750 (2013).

Rechisky EL, et al. (2012) Estuarine and early-marine survival of transported and in-river migrant Snake River spring Chinook salmon smolts. Nature Scientific Reports. 2:448.

Nathan nailed the response to this question, much better than I did. To add on to Nathan's point about limitations about extracting pertinent information from large data, the other issue is data management. How to store the terabytes of information that may come in from a group of birds each outfitted with microphone transmitters. Currently, the model of data workflow is (1) collect, (2) store/backup, (3) process, (4) analyze. Intelligent and adaptive workflows would ideally process and analyze the data as it streams in. This would make the data immediately actionable.

Really great points, Nathan. John's point about making sure that specialized telemetry data matches physiological data is a really good one. This has been an obvious issue for linking field-collected heart-rate in birds to field metabolic rate, as validation studies are lacking or relatively anemic.

Firstly the the expensive equipment (including transmitters, receivers, antennas, softwate etc.) is probably the most important factor influencing relatively low number of radio-telemetry studies.

On the other hand, there is also the risk of uncertain outcome, because of possible transmitter lost, animal mortality or higher motion activity than expected.

In addition, if you observe long periodic migration you have to solve problem with extreme personal costs as day night searching, long transfers etc.

Finally, it is expansive, laborious and thrilling method which can provide exceptional data. But only if you are really ready to do it.

MB

I think some of the issues relate to taxon group, Many taxa are either too small to carry existing transmitters available or an attachment/insertion technology for them has not been developed. For example, one step is the size revolution as the development of hearing aids, which brought battery sizes down, but no technological drives have promoted reducing them further. I would seem that the next step in technology would be a passive transponder that cover return a signal from a powerful receiver that could pick it up at very large ranges.

However, I know the sample size issue mentioned by Furey is a deterrent to many. No question on can get a lot of data on a small number of animals, but translate to population-level phenomena remains largely out of reach.

Many good points. My take on why telemetry rarely makes it into high impact journals - It primarily occurs in the wild, which is an uncontrolled environment, so it's hard to disentangle the effects of many proximate factors on animal behaviour. It is also applied to larger animals over periods of days, or maximum a few years, and therefore fails to track them over time scales (lifetime, multigenerational) relevant to many fundamental questions e.g. tracking fitness. It is also very costly, and this type of research funding is typically sourced from more applied interest groups e.g. hydro companies. Lastly, there is a very high level of mathematical skill and knowledge required for thorough analysis of these types of data, which deters many ecologists. This is largely why telemetry has been primarily descriptive in the past, but this is of course changing.

Although I have limited experience with using telemetry to study animals other than fish, I can say that telemetry holds great potential to provide insight into fish behavior. This being said, I still managed to find 10 limitations to its application…

1) Transmitter size precludes the study of larval fishes.

2) Battery life of the transmitter is often too short.

3) Pulse rates can be increased to increase battery life, but this will sacrifice detection ability and resolution, especially at smaller time and spatial scales.

4) Uncertain effects of transmitters on behavior, growth, and survival, the severity of which often depends on the taxa or environment.

5) Detection ability can be difficult especially in open and marine systems, which limits one’s inferences.

6) Difficulty in determining a tagged fish from a predator that has eaten a tagged fish.

7) Algorithms to determine false positive detections are labor intensive and can vary in performance.

8) A solid knowledge of computer programming languages for data manipulation and statistical modeling are a must.

9) Implantation of transmitters for large studies ( hundreds to thousands of tagged fish) are extremely labor intensive. which can be prohibitive.

10) The equipment is undoubtedly expensive.

Owen, these lists may be good to consolidate into a master document, as I think some of these considerations will be important for future readers of RG and elsewhere. Maybe put a link to the document in your projects page?

This is a really interesting and valuable discussion - it's great to get so many people's opinions.

In my opinion, one of the major obstacles to using telemetry data in a wider ecological sense is the (necessarily) small sample sizes and the difficulty in generalising the results. I think that the resources required to produce a study that would be taken up by high-impact journals are often unrealistic/unattainable. High-impact journals appear most interested in multi-species studies and meta-analyses, whereas telemetry studies are *usually* based on a single species (although this is changing now that more data is available from studies that have already been published). They also have small sample sizes, often because of the difficulty in accessing animals to tag, or the cost of the devices and field efforts.

Additionally, a further hindrance to increasing sample sizes is the rarity of the species being studied. Often telemetry studies are focused on threatened or endangered species. My experience is working in a colony of critically endangered seabirds, where we were extremely aware of the potential impacts on the birds, so kept the sample sizes reasonably small. We always left sections of the colony totally undisturbed, and also had a some control birds who were handled but not tagged. With a small population, you soon run out of study animals when taking this cautionary approach!

Great idea of Patrick's to try and collate this info somewhere.

I think sufficient sample size in telemetry studies is not always clear. I recently received a review comment on a paper I submitted that used data on 5 acoustic-tagged lake trout. However, over 20 years of abundance estimates indicate the lake has ~250 lake trout, so we sampled about 2% of the population and tracked them for over two consecutive years. In contrast, I recently published a paper that used thousands of radio-tagged juvenile salmon that were tracked for

Great idea Patrick, I'll collate the lot.

Really enjoying reading everyone's thoughts, and thanks for embracing this discussion as you have.

I think the point about sample sizes is a very good one, which in turn is a product of cost and difficulty in deploying the necessary number of devices (but as John mentioned, what is 'adequate'?).

I think we are starting to see a bit of a culture change in Biotelemetry now where researchers are a bit more willing to share data, so that they can be collated and used to answer bigger questions in ecology. In my opinion, if the field is to advance it will take more studies that adopt this strategy (just think how much data is sitting in some peoples hard drives waiting to be mined). I think that's what makes initiatives such as MoveBank so exciting.

Yes, good points all over. Having worked with fish telemetry all my career, I don't really think there are inherent obstacles for employing telemetry or publishing the results in whatever journal you like. There may be, however a slight tendency that descriptive studies appear less frequently than large meta-data studies, However, in my group we have published descriptive papers based on a few tagged individuals in Science. I tend to focus most on the point of evaluation of tagging effects as Nathan wrote:

"- Another huge gap that needs to be narrowed is the issue of tagging effects. Researchers using telemetry need to do as much as possible to assure that the tagged organism acts in the same way as an untagged organism would. I generally feel that this argument is overblown, as every technique we use in science can impact an organism's behaviours (i.e. mesocosms, any laboratory work that involves holding organisms, smaller external tags) and in these fields are rarely noted or studied, but it still remains a significant gap that we should attempt to address".

Right now there is a debate/dispute going on in two journals (Fisheries Research and J. of Fish and Wildlife Management) on the neccessety of aquiring a high level of asepsis for fish surgery. It is mainly a discussion between veterinareans (Dr. Mulcahy and researchers (myself), quite entertaining.

Usually, in Bio-telemetry the hypothesis just can be constructed after the observed behaviors. This idiosyncrasy requires of new approaches such us unsupervised classifications in order to asses what factors can influenced the observed behaviors. It is then after this classification when the hypothesis should be formulated.

However, this approach is not new since naturalist have been doing the same over centuries.

Studies of animal behavior should be well grounded (a priori) in evolution by natural selection. By doing so, I believe much of the concern about post-hoc formulation of hypotheses can be avoided, and should only aid in the publication within high profile journals because base principles are adhered to.