the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 24 Mar 2026
Mapping Plant Growth Index (PGI) over Australia from 1990 to 2024
Abstract. Australia spans nearly the full spectrum of global bioclimatic zones, from tropical savannas to arid deserts and alpine environments. Understanding how climate constrains vegetation growth across this gradient is essential for interpreting ecosystem dynamics and informing land-management decisions. We introduce a Plant Growth Index (PGI), a continental-scale metric derived from meteorological data from the Bureau of Meteorology Atmospheric high-resolution Regional Reanalysis for Australia (BARRA-R), European Space Agency (ESA) Plant Functional Type (PFT) layers, and C3/C4 grass fractions estimated from NASA Moderate Resolution Imaging Spectroradiometer (MODIS) enhanced vegetation index (EVI). The Plant Growth Index captures year-to-year variation in vegetation status, water availability, and climatic conditions. Spatially, values of the PGI are highest in tropical and subtropical regions and lowest in arid deserts. Benchmarking against gross primary productivity (GPP) from the Terrestrial Ecosystem Research Network (TERN) OzFlux network, the Normalised Difference Vegetation Index (NDVI), the Standardised Precipitation-Evapotranspiration Index (SPEI), and Australian Grassland and Rangeland Assessment by Spatial Simulation (Aussie-GRASS) indicates that PGI broadly reflects regional vegetation productivity patterns. The PGI provides a reproducible, continental-scale tool for ecological modelling, rangeland monitoring, and climate-impact studies and can be accessed at https://doi.org/10.5281/zenodo.18762343 (Retkute et al., 2026).
- Preprint
(6548 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 09 May 2026)
- RC1: 'Comment on essd-2026-161', Anonymous Referee #1, 06 Apr 2026 reply
-
RC2: 'Comment on essd-2026-161', Anonymous Referee #2, 08 Apr 2026
reply
In the manuscript “Mapping Plant Growth Index (PGI) over Australia from 1990 to 2024”, Retkute et al. designed the Plant Growth Index (PGI), which is primarily constrained by meteorological conditions across different plant functional types, and used PGI to characterize vegetation productivity patterns across the Australian continent. The topic is important, and the results and figures are generally clear and easy to follow. However, in its current form, the manuscript is not yet suitable for publication in Earth System Science Data because the methodological novelty and robustness remain insufficiently demonstrated.
Major comments:
- PGI framework
A major methodological concern is that the proposed PGI framework may not be suitable for representing vegetation productivity across the strong bioclimatic heterogeneity of Australia. As the authors themselves note, Australia spans a wide range of climate and ecosystem types, from tropical systems to temperate woodlands, arid deserts, and alpine environments. The dominant controls on plant growth are therefore unlikely to be spatially uniform. For example, tropical ecosystems may be more strongly constrained by energy availability, whereas inland arid systems are primarily water-limited.
In this context, defining PGI as the simple product of the Light Index, Temperature Index, and Moisture Index imposes a common structural assumption across all regions. Even if each component is normalized to a 0–1 range, the framework still assumes that these three controls can be integrated in the same multiplicative way across ecosystems with fundamentally different limiting mechanisms. This assumption is not sufficiently justified in the manuscript.
Importantly, the authors’ own evaluation results appear to support this concern. The reported correlations between PGI and GPP vary substantially across sites, ranging from r = 0.08 to 0.94, with a median of r = 0.66. In several ecosystems, especially the Wet Tropics (r = 0.12–0.25) and some coastal and semi-arid regions, the relationship is weak, indicating that PGI does not consistently capture productivity dynamics across biomes. These results suggest that the performance of PGI is highly ecosystem-dependent, rather than robust at the continental scale. This weakens the claim that PGI can serve as a general indicator of vegetation productivity patterns across Australia.
The manuscript should provide a much stronger ecological justification for using a uniform multiplicative framework, and should more explicitly discuss the limitations of applying the same index structure across water-limited, energy-limited, and seasonally constrained ecosystems. Additional biome-specific evaluation or sensitivity analysis would also be needed to demonstrate robustness. The authors could also highlight the novelty regarding detailed design of TI that incorporates PFT, and be more specific in recommending the reliability of PGI, i.e., whether they are most reliable in grasslands or across different biomes and climate zones.
- Lag effect and fire disturbance
A second major concern is that the PGI framework only considers contemporaneous meteorological conditions and therefore cannot capture important lagged ecological responses or major disturbance effects. In many Australian ecosystems, vegetation productivity is not determined solely by current-day light, temperature, and moisture conditions. Instead, it is often influenced by antecedent environmental conditions, including prior rainfall, soil moisture storage, plant physiological carryover, and broader lag effects that can persist from weeks to seasons. A substantial body of literature (e.g. https://doi.org/10.1029/2022JG007144, https://doi.org/10.5194/bg-19-1913-2022 )has shown that such lagged effects can be as important as, or even more important than, concurrent climate drivers in regulating vegetation productivity.
In addition, the framework does not account for fire disturbance, which is a major driver of vegetation dynamics in Australia, particularly in northern savanna ecosystems. Fire strongly affects canopy structure, biomass, phenology, and post-disturbance recovery, and can therefore decouple observed productivity from the meteorological conditions represented in PGI. Ignoring fire disturbance is likely to reduce the ecological realism of the index in fire-prone regions and may partly explain some of the weaker PGI-GPP relationships reported across sites.
These omissions raise further concerns about the robustness of PGI as a continent-wide representation of vegetation productivity. The authors should more clearly acknowledge that PGI reflects only immediate meteorological suitability, and should discuss the implications of excluding lag effects and disturbance processes. Ideally, they should also evaluate how these factors may affect PGI performance across different biomes.
- Light and soil moisture acclimation
Another important methodological concern is the inconsistent treatment of plant acclimation across the three PGI components. For the Temperature Index (TI), the manuscript adopts plant functional type (PFT)-specific thermal response parameters, including Tmin, Topt, and Tmax. This implicitly acknowledges that plant responses to temperature are not universal, but differ among vegetation types and may shift through physiological acclimation and ecological differentiation. This is a reasonable and ecologically meaningful choice.
However, a similar level of ecological realism is not applied to the Light Index (LI) or Moisture Index (MI). This is problematic because acclimation to light and water availability is also fundamental to plant growth and productivity. Both light and moisture responses, like temperature responses, often involve an optimum range rather than a simple monotonic “more is better” relationship. For light, photosynthesis commonly saturates beyond a certain level, and the most favourable radiation environment depends on canopy structure, shade tolerance, and acclimation state https://doi.org/10.1038/s41559-020-1258-7 . For soil moisture, both insufficient and excessive water availability can constrain growth, and the effective optimum may differ among ecosystems and vegetation types https://doi.org/10.1038/s41467-024-54156-7 . Therefore, maximum light availability or maximum soil moisture should not automatically be interpreted as the most favourable conditions for plant growth.
The current framework appears conceptually unbalanced. The TI formulation recognizes that vegetation-specific responses matter and explicitly represents an optimum-based response curve, yet LI and MI appear to rely on a much more generalized representation of light and moisture limitation. If the authors argue that temperature responses require PFT-specific parameterization because of acclimation and ecological differences, then it is unclear why comparable acclimation, optimum ranges, or trait-based differentiation are not considered for light and moisture. This asymmetry weakens the ecological consistency of the PGI framework and raises concerns about whether PGI provides an ecologically balanced representation of plant growth constraints across Australia.
Minor comments:
L75: The factor should be 0.0036 rather than 0.036.
L100: The manuscript should explain why the ESA CCI PFT product was selected, rather than other possible alternatives such as IGBP-based PFT classifications or the Australian NVIS dataset. Because PFT information directly affects the TI parameterization and hence PGI, this is an important methodological choice that deserves clearer justification.
L130: The authors should clearly specify which GPP partitioning product was ultimately used for the flux tower validation (e.g., LT/DT, or SOLO).
L150: The symbol used for Pearson’s correlation coefficient is usually r rather than ρ. Using ρ may cause confusion, as it is often reserved for the population correlation coefficient or Spearman’s rank correlation.
Citation: https://doi.org/10.5194/essd-2026-161-RC2 -
RC3: 'Comment on essd-2026-161', Anonymous Referee #3, 09 Apr 2026
reply
Retkuke et al. introduce a retrospective and forecast-aligned Plant Growth Index (PGI), a gridded metric that estimates interannual vegetation status in the form of vegetation growth and water availability over continental Australia. The PGI is calculated using meteorological inputs from light, temperature, and moisture. Benchmarking was performed against various productivity-relevant metrics. The authors imply PGI’s utility in monitoring various ecosystems and in climate modeling based on benchmarking results. Overall, the paper is structured well, easily readable, and addresses an important issue. However, the paper is not ready for publication in Earth System Science Data without major revisions. The authors need to address major concerns regarding the formulation and justification of the PGI equation and its components and additionally analyze the strength of PGI against productivity-relevant metrics on seasonal timescales relevant to vegetation status.
Major Comments
- The equations for PGI and its components introduce confusion. I’m lacking a physical and conceptual understanding of equation design. The paper mentions that their methods are adapted from a previously used scalar index framework. Because this paper is introducing a new gridded dataset based on these equations, it would be useful to expand on the reasons and physical meanings of each equation and assumption, with added citations wherever necessary, throughout the methods. For example:
- Why is the clear-sky reference the 95th percentile of daily solar exposure; was this cited in another paper for LI or is it introduced for the first time here?
- The paper mentions that top layer soil moisture (0-10 cm) is used as a proxy for root-zone soil moisture in the MI calculation. However, soil moisture behavior differs through the vertical profile, and so do rooting depths. What root depths are being approximated? Is surface soil moisture alone used as a proxy, or was root-zone soil moisture estimated with another equation? ERA5-Land has different volumetric soil water layers that can be used for this purpose if BARRA-R doesn’t contain deeper soil moisture observations, but overall, it’s important to incorporate the dynamical nature of soil moisture somehow as water availability may differ over different PFTs and depths.
- The design of the PGI equation requires a more substantive description. In the way it's designed, what is the physical and conceptual meaning of how the variables are arranged and why?
- There are a couple of concerns regarding PGI seasonality in benchmarking that need to be addressed:
- In the benchmarking results against GPP, the authors state that strong relationships were observed in cool regions with clear seasonal cycles. However, Pearson’s correlation can be high between two distinct curves with different magnitudes and distributions if both exhibit a clear seasonal cycle. It might be helpful for the study to benchmark across distinct seasons rather than the entire time series to see if PGI has a stronger footprint in certain parts of the year and why or why not.
- The study makes the assertion that PGI effectively captures vegetation productivity responses during transition seasons. The results do not show or quantify this, as benchmarking was performed across the entire time series, not distinct seasons. The authors should consider defining dry-to-wet and wet-to-dry transition periods respective to the sample sites using known methods of defining seasonal transition timing (e.g. https://doi.org/10.1002/2016JD025428). This is important to do across transitions because during these periods, vegetation moves from water/energy limitations to energy/water limitations, respectively. Benchmarking across distinctive seasons (previous response) and transition periods (this response) would make the study more robust and interpretable across different bioclimatic gradients, vegetation-relevant growth periods, and seasonal climate risk assessments.
Minor Comments
- General: When a supplementary figure is mentioned, it is typically structured as “SI Figure S#.” However, the SI figures are labeled as “Figure A#.” Make sure to double check that the letterings are consistent throughout the paper.
- [Figure 2c]: Consider adding a letter to the different case sites as each is a different sub-plot. It can get confusing labeling all five examples as Figure 2c rather than Figure 2c-g.
- L65-L70: There is a closing parenthesis “)” without an accompanying opening parenthesis “(.”
- L120-125: Consider briefly adding the re-gridding approach used to scale data to the BARRA-R 12km (or 0.1o) resolution (e.g., bilinear, linear?). Additionally, if any temporal gap filling was performed to make the datasets consistent with each other, please expand on why or why not.
- L190-195: Some of the available flux tower observations are short (e.g., Fletcherview in Figure 2c seems to have data from 2022-2025). This makes comparing PGI's correlational strength against other sites difficult as temporal span differs. In the written results it might be useful to put the timespan of available flux observations for each site next to their correlation coefficient to help readers with interpretability.
Citation: https://doi.org/10.5194/essd-2026-161-RC3 -
RC4: 'Typographical Correction to RC3', Anonymous Referee #3, 09 Apr 2026
reply
I would like to submit a minor typographical correction to my comment (RC3: 'Comment on essd-2026-161'). In my response, the last name of the corresponding author was typed as "Retkuke" when the correct spelling is "Retkute."
Citation: https://doi.org/10.5194/essd-2026-161-RC4
- The equations for PGI and its components introduce confusion. I’m lacking a physical and conceptual understanding of equation design. The paper mentions that their methods are adapted from a previously used scalar index framework. Because this paper is introducing a new gridded dataset based on these equations, it would be useful to expand on the reasons and physical meanings of each equation and assumption, with added citations wherever necessary, throughout the methods. For example:
Data sets
Mapping Plant Growth Index (PGI) over Australia from 1990 to 2024 R. Retkute et al. https://doi.org/10.5281/zenodo.18762343
Model code and software
Mapping Plant Growth Index (PGI) over Australia from 1990 to 2024 R. Retkute et al. https://github.com/rretkute/Mapping-Plant-Growth-Index
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 108 | 30 | 6 | 144 | 6 | 8 |
- HTML: 108
- PDF: 30
- XML: 6
- Total: 144
- BibTeX: 6
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Retkute et al. introduce the Plant Growth Index (PGI), a continental-scale metric for Australia designed to quantify potential vegetation growth by integrating meteorological data from the BARRA-R reanalysis, ESA PFT layers, and MODIS EVI-derived C3/C4 grass fractions. The PGI is calculated as the product of light, moisture, and temperature indices to capture the spatiotemporal variability in vegetation status. The authors benchmark PGI against GPP, NDVI, SPEI, and Aussie-GRASS, demonstrating broad agreement and highlighting its utility for ecological modeling, rangeland monitoring, and climate-impact studies. Overall, the paper addresses a relevant topic, and the manuscript is generally well-structured. However, the theoretical justification for the index formulation, the lack of a rigorous baseline comparison, and the spatial scale of the validation present significant limitations. These aspects require substantial improvement before the manuscript can be recommended for publication.
Major Comments
My primary concern lies in the justification of the PGI’s added value. The authors validate the PGI by demonstrating its correlation with widely used proxies for vegetation growth and water availability (GPP, NDVI, SPEI, and Aussie-GRASS). While this confirms that the PGI behaves consistently with established metrics, it does not prove that the PGI is an improvement over them. The authors must provide evidence showing that the PGI performs better than, or at least provides distinct advantages over, previously used proxies. For example, does PGI predict GPP more accurately than SPEI across both spatial and temporal dimensions? Without a clear demonstration of its comparative advantage, the necessity of introducing a new index remains unsubstantiated.
The inherent scientific and ecological meaning of the PGI requires deeper discussion. Equation (1) relies on a simple multiplicative formulation, but this approach feels arbitrary and lacks sufficient physical or empirical justification. The authors should clearly define what "growth" specifically means in the context of this index. Furthermore, they need to justify why Equation (1) is the optimal formulation. I highly recommend benchmarking this equation against standard data-driven models. If the authors were to use the same meteorological inputs to train a multiple linear regression (LR) model or a machine learning algorithm (e.g., Random Forest) to predict short-term plant growth, would the PGI outperform them? If the PGI cannot provide superior or equal performance to basic empirical models, its utility is questionable. Finally, the manuscript lacks an analysis of how much each individual component (i.e., light, moisture and temperature indices) drives the final index. It would be helpful if the authors include a sensitivity analysis quantifying the relative contribution of each component in Equation (1) to the final PGI across different biomes or seasons.
The current validation is conducted primarily at the bioregion scale. This aggregated approach masks localized variability and fails to explicitly demonstrate the power of the PGI as a fine-scale proxy. To make a persuasive case for the index’s resolution, the authors must validate the performance of the PGI at finer scales (e.g., pixel-level comparisons), rather than solely relying on smoothed bioregional averages.
Minor Comments:
Equation (1): There is an extraneous comma in the equation that needs to be removed.
Table 1: Are the temperature thresholds obtained from the cited literature genuinely applicable to the highly adapted and often endemic vegetation of Australia? This requires an in-depth discussion to ensure the thresholds are ecologically valid for the study region.
Line 100: The temperature index is highly dependent on the classified vegetation types. What is the inherent accuracy of the ESA PFT classification product over Australia? The authors must discuss how classification errors propagate through the model and impact the reliability of the final PGI.
Line 135: The description of the Aussie-GRASS data is vague. Please explicitly detail which specific variables or outputs were extracted and utilized.
Line 195: The PGI time series exhibits significantly stronger temporal amplitude/variations compared to the observed GPP time series. This mismatch is a notable flaw. The authors should investigate whether the PGI formulation needs refinement to dampen this over-sensitivity, or explicitly explain why this divergence occurs physically.
Line 255: The discussion should be expanded to explicitly address the temporal variance discrepancy noted in my comment for Line 195.