1 of 19

A look at ecological condition in Fynbos

Jasper A Slingsby 1,2, Adam M Wilson 3,

Brian Maitner 3, Glenn R Moncrieff 4,2,5

^^^Others

1 Biological Sciences and SEEC, University of Cape Town

2 South African Environmental Observation Network, Fynbos Node

3 University at Buffalo, NY

4 The Nature Conservancy

5 SEEC, Statistical Sciences, University of Cape Town

2 of 19

Satellite monitoring of Fynbos?

NDVI of a healthy Fynbos pixel can be anywhere between 0 and 1!!!???

Fire

Seasonality

Regrowth

Postfire recovery trajectory

(satellite NDVI)

3 of 19

Satellite monitoring of Fynbos?

Model:

Blue = forecast mean

Grey = 95% posterior Pr

Red = observed NDVI

​

Identify anomalies:

Observed - Expected

​

Fire

Seasonality

Regrowth

Postfire recovery trajectory

(satellite NDVI)

Wilson et al. 2015 PNAS

Slingsby et al. 2021 ISPRS

4 of 19

Change detection!

Too green:

  • invasion by alien trees

​

​

https://doi.org/10.1016/j.isprsjprs.2020.05.017

5 of 19

Change detection!

Not green enough:

  • vegetation clearing or mortality

https://doi.org/10.1016/j.isprsjprs.2020.05.017

6 of 19

Change detection!

Not green enough:

  • retarded postfire growth

https://doi.org/10.1016/j.isprsjprs.2020.05.017

7 of 19

NASA funded

ecoforecasting project

​

Operational near-real-time change detection for Fynbos

​

Identifies deviations from normal:

  • Plant invasions, mortality, vegetation clearing, slow growth, etc

www.emma.eco

8 of 19

New technologies? Imaging Spectrometry & Lidar

Gulfstream V

Flights in Oct/Nov 2023!

www.bioscape.io

Flights October - November 2023

LVIS

Land, Vegetation, and Ice Sensor (Light ranging altimeter)

HyTES

Hyperspectral Thermal Emission Spectrometer

PRISM

Portable Remote Imaging Spectrometer

AVIRIS-NG

Airborne Visible-Infrared Imaging Spectrometer (NextGen)

RGB-IR/Lidar

SAEON Shallow Marine and Coastal Research Infrastructure (SMCRI)

9 of 19

New technologies? Imaging Spectrometry & Lidar

Source: Lucas van den Bosch, https://commons.wikimedia.org/wiki/File:Mono,_Multi_and_Hyperspectral_Cube_and_corresponding_Spectral_Signatures_modified.svg, CC BY-SA 4.0

Single Band Multispectral “Hyperspectral”

Imaging Spectroscopy Lidar

Source: Purkis and Klemas

10 of 19

Ecosystem/cover mapping

  1. Train a spectral library of class types
  2. Use a machine learning algorithm to predict the (mixed) composition of each pixel
  3. Output classification and fraction cover layers

​

Hi-res RGB

Spectral library of class profiles

Classification

11 of 19

Ecosystem/cover mapping

Outputs:

  • classification
  • fraction cover layers
    • Should allow us to capture early invasions (small or scattered individuals)

​

12 of 19

13 of 19

14 of 19

15 of 19

Trait shifts and past impacts?

“Secondary natural”?

​

Can we infer past impacts with these new technologies?

  • E.g. subtle differences in plant functional traits?

Phytoplankton biodiversity and HABs

Source: K. Kovach

Plant functional traits

Source: NASA GSFC

16 of 19

Other issues in Fynbos?

Altered fire regimes?

Slingsby et al. 2020 GCB

17 of 19

Drivers

Indicators

Habitat loss

​

Land use

​

Climate impacts

​

Invasive species

​

Altered fire regimes

Land cover classification

​

Trait maps? Time series

​

Time series, indicator species

​

Trait maps, direct obs

​

Models, time-series,

indicator species

Nature's contributions to people?

Water production, Carbon storage/assimilation, Etc

​

Source: Turner 2014

18 of 19

Threats

Monitoring

Forecasts

​

Decision

makers

​

Management

Policy

Vegetation

Populations

Water Supply

Satellites

Sensors

Fieldwork

Impacts of decisions

Identify information needs

Scenarios

Ecoinformatics pipeline

Habitat degradation or loss

Altered fire

regimes

Invasive species

Climate change

Identify threats

Slingsby et al. 2023 MEE

19 of 19

Thanks!

​

​

Funders:

​

​

Management authorities:

​

​

​

www.emma.eco

plantecolo.gy

bioscape.io