In this post, I’ll discuss the vegetation reconstruction of our project. Modeling is how we use data generated from plant labwork (pollen and macrofossil identification), and obtain estimates of plant abundance and biomass, and net primary productivity (NPP). Both of these outputs are important for the project’s next stage: modeling the food webs. Knowing vegetation abundance then tells us what plant food sources were available for small mammals, and energy available for the food web can be inferred from NPP. While both these models are later stages of the project (see below), there is much to do in the meantime to develop our best practices. We want to 1) ensure everything runs smoothly once plant lab analyses are complete, and 2) know how to best translate real data into inputs the model will understand.
The model I’m using to get these NPP and plant abundance estimates is a dynamic global vegetation model (DGVM) called LPJ-GUESS. DGVMs simulate vegetation interactions and change at local, regional and/or global scales. A common approach in DGVMs is grouping vegetation into plant functional types (PFTs), such as tropical broadleaved raingreen trees, C3 grass, boreal needleleaved trees, etc. LPJ-GUESS is a model that combines two things: biogeochemical processes (e.g. carbon uptake, environmental response, photosynthesis) of the PFTs, and population dynamics such as resource competition and mortality. Treating different processes as separate modules mimics real ecosystem complexity, and this approach gives flexibility to tackle projects at different scales.
Determining the best inputs for a model like this takes time, in order to generate the pollen and macrofossil lab data, and to find appropriate climate datasets. Paleotemperature, soil conditions, and atmospheric carbon dioxide (CO2) time series data can offer insight to past climate conditions. Often, the most complete paleoclimate datasets are not directly from the region studied. They also need to be recalculated so that the temporal resolution is consistent between them all. So, you might want run a DGVM with results every 1,000 years. But, the best available paleoclimate data are from samples taken every few hundred years. This means deciding the best approach for averaging that data. For CO2, it is common to use a dataset that represents average global concentrations, taken from one of the polar ice cores.
For vegetation inputs, we can easily categorize identified fossil pollen and macrofossils by PFT. Western blue-eyed grass (Sisyrinchium bellum) is a C3 grass, for example, and Juniperus spp. – a frequent occurrence in Rancho La Brea deposits – is a temperate needleleaf evergreen tree. One challenge is how to use these data to estimate plant proportions on the landscape. Most excavations come with a healthy dose of preservation bias, and we are aware that certain species might be over-represented in any given tar seep. A juniper tree that grew right next to an asphaltic deposit ~33,000 years ago, for example, would “oversupply” the deposit with its own scales and seeds over the course of its lifetime. Meanwhile, acorns and leaves from a live oak 50 yards away will end in the asphalt only on occasion. We have found evidence of both these plants, and their presence is important in visualizing past vegetation community structure (woodland, or more open?) so that we have the best estimate possible for vegetation abundance as our model input.
These plant samples and preliminary identifications are from material that dates to 50,000 – 30,000 years before present. Prior studies suggest that Southern California was cooler and wetter than modern conditions, though not as cold as the depths of the Last Glacial Maximum (26,000 – 19,000), when local mountains supported glaciers! The presence of Juniperus in La Brea deposits is consistent with a recent pollen study from nearby Lake Elsinore (Riverside County) that starts 32,000 years ago, so this confirms its presence at lower elevations. It’s also exciting that our use of a DGVM to reconstruct the vegetation community has not yet been employed for this time period in a region where winter-wet, summer-dry mediterranean conditions currently prevail. Knowing that this approach is unique offers enough motivation for the challenge of getting debugging a complex model, and knowing I can take an out-of-office break to “visit” with the remarkably preserved material from the past. For example, this woody branch. No identification yet, so it still holds its secrets – but it’s incredible to see intact bark, after being submerged in asphalt for ~30,000 years…followed by recent excavation, then chemical cleaning.
If you’d like to view these plant macrofossils virtually, check out our project iNaturalist page! Project members frequently add images, including a bunch last month.