Instructor Notes

Class Announcements: remind students regarding any issues with upcoming tutorials, files, packages, etc.

Class Discussion

Pre-question discussion:

Start discussion by seeing if the class has experience with phenology. This can vary from class to class whether the concept was completely unfamiliar or you have a group of experts and will help you guage whether you need to spend more time on basic concepts are can push quickly to deeper concepts. Start by asking class how many of them have read or thoughts about phenology. Ask them about phenology change over time. Ask about what phenology changes they have been exposed to before. Half the class will have some familiarity with phenology.

Discussion Questions

What is phenology?

  • Timing of activity - seasonal scale, within a year typically not across years.

This paper is very focused on plants, but what are some examples of phenology for nonplants?

  • Breeding season onset and stop
  • Migration days
  • Emergence from hibernation

What are some of the ways that phenology and climate are linked?

  • Increasing temperature changes cues for species
  • Increasing temperature can drive increased rates of development
  • In tropical or desert ecosystems, seasonality may be linked with water availability more than temperature fluctuations (rainy vs dry season)

How big have the shifts in phenology been thus far?

  • 2.3-5.1 days per decade

What kind of data do you need to detect and forecast that magnitude of change?

  • high frequency data

Think back to our discussion on data types - what types of data give us that scale of resolution? What challenge do they identify later in the paper that inhibits using that type of data collection to the study of phenology?

  • We’re typically talking satellite data, but this often has low spatial resolution and/or low data frequency relative to the scale of phenology (2-week intervals for the processed data)

Do you feel that its important or possible to forecast shifts that precisely?

  • This is an opinion question designed to get them thinking about data quality issues (and uncertainty in data) and how that might influence our ability to predict different types of phenomena

Phenology has been studied extensively, but where are most of the long-term records from? What biases does this insert into our understanding of phenology?

  • Temperate zones
  • Often managed plants (cherry trees, vineyards)
  • Biases us towards the northern hemisphere (generally). Towards temperature-based phenology change. Cultivated species may not have the same pressures and constraints on their timing as wild species.

The paper dedicated a section to open discussion about which family of models (theoretical, statistical, process oriented) is suitable for forecasting phenology. Each family has pros and cons. Describe each of these family of models and how they differ from each other?

  • Theoretical: based on theoretical trade offs of the costs and benefits of producing leaves to optimize resource acquisition.
  • Statistical: Based on relationships in empirical data with climate factors. These are statistical fits to the data (i.e. linear regressions).
  • Process-based: formally describe known cause-effect relationships between biological processes and environmental factors.
  • Note to instructor: spending some time on this is good because these concepts will come up repeatedly in the course. This is just a first exposure to get them exposed to the concept of models with different approaches/goals.

Which approach do you feel may be best? To answer which questions?

  • An opinion question to get them thinking about the strengths and weaknesses of the different approaches.
  • Statsitical: don’t need to understand details of a system or all the processes and how they interact (which is the state we’re in for most ecological systems) but reliant on past relationships. If processes change, then what?
  • Theoretical: Usually extremely simplified models of systems, but we may have data for some of the parameters. Should be capturing most important processes, but their simplified nature may miss important interactors.
  • Process-based: Students will often get hung up on differences between theoretical vs. process-based models. Process-based is based on important theoretical/conceptual processes, but often with more details and interactions with other processes. Pointing students towards island biogeography theory, lotka-volterra equations, and similar theoretical simplifications and discussing whether those simplifications would work for their systems and what they might want to add to generate a process based model.

If time here are some additional items to discuss:

How often have you read the word ‘uncertainty’ in the paper?

  • Once - wrt. Pollen transport.

Given the scale of the phenology response being measured, how important could uncertainty be?

  • Given that we’re detecting a few days per decade, seems like uncertainty could potentially be important to account for.

What sources of uncertainty can you think of that could be important for modeling or forecasting phenology?

  • in the data - for example, the remote sensing data which has uncertainty due to the smoothing/filtering effect
  • Observation/sampling uncertainty: over long time periods (in field records) could be changes in who thinks what signals indicate the onset of a phenology stage.
  • in our models: imperfections in our models, lack of details that could influence predictions on the scale of days/decade