Australian Macadamia Orchards Predominantly Function as Carbon Sinks

Abstract

The need to develop practices enabling deep and continued reductions in greenhouse gases (GHG) is arguably the greatest challenge facing humanity in the 21st century. Here, we quantify GHG emissions of macadamia enterprises in Australia and then contrast potential abatement realized by practice change. We show that nitrogenous (N) fertilizer accounted for more than half of net farm emissions, followed by fuel and electricity. Soil organic carbon accrual dictated carbon removals and GHG emissions intensity. Many enterprises applied N fertilizer at rates higher than recommended best practices, suggesting that reduced fertilizer quantities may sustainably reduce GHG emissions without impacting yields. We conclude that many macadamia farms are net carbon sinks, which contrasts with other agricultural sectors that are often sources of GHGs. We illustrate how the adoption of bespoke interventions, such as optimization of N fertilizer use, improvement of intrarow ground cover, and avoidance of tillage, can abate enterprise emissions while also improving food security, enterprise prosperity, and environmental stewardship.

1. Introduction

The agriculture, forestry, and land use sectors contributed approximately 5.9 ± 4.1 Gt CO₂ of global greenhouse gas (GHG) emissions per year between 2010 and 2019. Some scholars opine that GHG emissions, through improved carbon removals, methane avoidance, and emissions mitigation, could diminish by up to −12.5 ± 3.2 Gt CO₂ per yr. − While many studies have been undertaken with the aspiration of developing practices, skills, and/or technologies for reducing GHG emissions, most have focused on the livestock, cropping, and/or land use change sectors; , few have examined the mitigation potential associated with the horticultural sector.

Enterprises that cultivate fruit and nut trees conceivably have significant advantages over other agricultural sectors in terms of GHG emission mitigation due to the ability of perennial trees to sequester carbon in woody vegetation. Indeed, a meta-analysis showed that the global average carbon footprint of beef was 26.6 kg carbon dioxide equivalent (CO₂eq) per kilogram (kg) of bone-free meat, whereas that of tree nuts was as low as 1.20 kg CO₂eq per kg of raw product. This suggests a greater propensity for tree-nut enterprises to achieve low- and negative-carbon products relative to other agricultural industries.

Macadamia nut trees (Macadamia integrifolia) were first discovered in Australia in the late 1850s and have since gained prominence, with contemporary Australian macadamia farm gate value worth more than $AU286 M per yr. Despite having existed for more than a century, little is known about why tree nut emission intensities vary so widely across production systems and agroecological regions. For example, the emissions intensity (EI) of almonds, hazelnuts, and pistachios grown in Italy has been quantified at 2.3, 1.2, and 2.5 t per tonne of CO₂eq, respectively. After accounting for carbon sequestered in nut shells and in tree vegetation, however, net emissions have decreased to 1.9, 0.5, and 1.7 t CO₂eq per tonne of raw product, respectively. Similar results have been found in Spain, where GHG emissions from almond production systems varied 4-fold across management interventions. Taken together, these studies imply significant scope for mitigation via practice change.

Key factors influencing enterprise GHG emissions of tree production systems are thought to arise from fertilizer, fuel, and electricity use. , Practice changes aimed at reducing such inputs have been shown to reduce gross GHG emissions, but, again, the extent of mitigation has been context-specific, depending on enterprise, location, environment, and current and previous management history. − Studies exploring approaches for emissions reduction and removals from fruit or nut trees have shown that shifting from conventional to organic farming reduced gross emissions by 56% in Mediterranean contexts, while transitioning from intensive tillage to no tillage reduced net emissions in almond production under semiarid Mediterranean conditions by more than 60%.

Given that GHG emissions vary as a function of agroecological region and production system, we contend that neither the carbon footprints nor the GHG emissions mitigation potential of macadamia orchards can be inferred from previous studies. As such, the objectives of this study were to (i) quantify net emissions and EI (GHG per unit nut-in-shell; NIS) and (ii) identify interventions for sustainably reducing GHG emissions of macadamia farms without impacting productivity (food security) or environmental stewardship.

2. Methods

2.1. Study Overview

We used real data from 23 macadamia farm enterprises in the Bundaberg region of Queensland, one of the largest macadamia production zones in Australia (Figure ). Farm enterprise data comprised 48 datasets of information pertaining to orchard and land use, farm management, yields, soil organic carbon (SOC), and other biophysical data (Table ). Orchards examined were part of the Ground-Truth Australian Orchards (GTAO) carbon project, a highly rigorous and systematically verified international carbon standard. Farm areas ranged from 2 to 2,060 ha (nut-bearing orchards) and from 12 to 1,303 ha for young orchards prior to the first nut harvest.

1.

Location of farms examined in the present study.

1. Annual Average Farm Enterprise Data from 23 Macadamia Orchards between 2018 and 2023 (NIS = Nut-in-Shell).

Inventory categories	Mature orchards	Young orchards (<4 years old)
No. of orchards	14	9
No. of data sets	34	14
Avg annual cultivated area (ha)	336 ± 79	290 ± 118
Annual yield per ha (tonne NIS)	3.3 ± 0.2	0
Area under native vegetation (ha)	203.37 ± 63.9	8.6 ± 5.9
Avg annual temperature (°C)	22.1 ± 0.1	21.7 ± 0.2
Avg annual max temperature (°C)	27.5 ± 0.1	27.0 ± 0.2
Avg annual min temperature (°C)	16.6 ± 0.1	16.3 ± 0.2
Avg annual rainfall (mm)	1,048 ± 113	1,090 ± 164
Avg N fertilizer use (kg N/ha)	205 ± 35	84 ± 29
Lime (kg/ha)	207 ± 82	0
Agrichemicals (kg or liter per ha)	25.8 ± 2.4	7.4 ± 1.1
On-farm diesel (liter/ha)	391 ± 41	142 ± 24
On-farm petrol (liter/ha)	5.1 ± 1.1	4.4 ± 2.8
Irrigation electricity (kW/ha)	990 ± 104	283 ± 70
Annual SOC accrual (%)	1.17 ± 0.06	1.23 ± 0.08

^aAgrichemicals include insecticides, fungicides, herbicides, and adjuvants.

^bSOC was measured for the uppermost 30 cm of soil over an average duration of 5 years across farms.

We conducted life cycle assessments (LCA) to quantify GHG emissions from production to the farm gate associated with macadamia production. We then conducted a sensitivity analysis of several driving variables to elicit determinants of net farm GHG emissions and emissions intensities (Figure ).

2.

Conceptual overview of the study, including processes for computing macadamia GHG emissions, sources and sinks of GHG, and functional units quantified. Sources and sinks of GHG (non-CO₂ gases CH₄ and N₂O) were converted to CO₂eq to elucidate net farm GHG emissions.

2.2. Climate of the Bundaberg Region

The Köppen-Geiger climate classification of the Australian Bundaberg region is characterized by warm summers and humid, mild winters. Annual mean maximum and minimum temperatures are around 27 °C in summer and 16 °C in winter (Australian Bureau of Meteorology [ABM]). The long-term average annual rainfall of the region is around 1,095 mm (standard deviation 355 mm), with most precipitation occurring in January and February. The Bundaberg terrain is primarily flat and alluvial. Tenosols, Rudosols, Vertosols, and Sodosols predominate as soil types along the alluvial plains (predominantly clays and clay-loams that crack when dry and are of high fertility), whereas Podosols are more abundant in coastal areas.

2.3. Macadamia Enterprise GHG Emissions

We conducted an LCA to quantify total GHG emissions and the carbon footprint per unit nut-in-shell (NIS) following the guidelines of the International Organization for Standardization (ISO). We first prepared an inventory of inputs and outputs, then evaluated the environmental impacts stemming from their use.

2.3.1. System Boundary and Functional Units Quantified

An LCA system boundary sets out processes included in the analysis, defined herein as from macadamia inception to farm gate (Figure ). We included prefarm emissions from fertilizers, agrochemical production, and transportation; farm emissions from fuel and energy for farming machinery and irrigation activities; and direct and indirect field emissions from fertilizers, lime, and crop residue management. Carbon removals from soils within and outside macadamia orchards were considered within the system boundary (Figure ).

A functional unit (FU) is a standard reference used to measure the performance of a system, defined here as the marketable product, i.e., one tonne of macadamia NIS. Macadamia nut producers generally sell dehusked nuts to processors, as the hard shell of macadamias is difficult to open. In addition to the mass-based FU, we computed a land-based FU per unit area of the macadamia orchard. This allowed us to contrast net farm emissions and production efficiencies, including those from young orchards in which the nuts had yet to be harvested. Peer-reviewed literature suggests that by integrating land-based functional units in LCA, a more comprehensive evaluation of agricultural practices can be carried out, capturing both the efficiency of resource use and the broader environmental implications of the production methods adopted.

2.3.2. Farm Enterprise Data

Data from each macadamia enterprise between 2018 and 2023 were collated and used for analysis. Orchards consisted of young and mature trees; nine orchards were between 1 and 4 years old and had not begun commercial nut harvests. Fourteen orchards had trees 5–30 years of age and harvested and sold nuts. In aggregate, these comprised a total arable land area of 8,851 ha, of which 6,839 ha were planted with macadamia crops. In addition to the orchard area, farms consisted of an average area of 147 ± 47 ha under native vegetation (Table ).

2.3.3. Greenhouse Gas Emissions

Global warming potential (GWP) was quantified over 100 years using Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5) emissions factors of 1, 265, and 28 for carbon dioxide (CO₂), nitrous oxide (N₂O) and methane (CH₄), respectively IPCC, expressed in tonnes of CO₂eq Although the IPCC has published updated emissions factors in the Sixth Assessment Report (AR6), the Australian National Greenhouse Gas Inventory (ANGHGI) currently uses emissions factors from the IPCC Fifth Assessment Report (AR5). To maintain consistency with the national reporting framework, we adopted AR5 emissions factors. Computations of GHG emissions followed ISO 14064-2. , Direct and indirect GHG emissions from inorganic fertilizers, agrochemicals, fuel, and electricity were calculated following guidelines of the International Organization for Standardization. Upstream (embedded) emissions were calculated using the Ecoinvent database, v3.8. Daily rainfall and temperature data from 2018 to 2023 were acquired from a national climate database (https://www.longpaddock.qld.gov.au/silo/).

Emission factors pertaining to farm activities were based on the ANGHGI, maintained by the Australian Department of Climate Change, Energy, Environment and Water (DCCEEW). Direct emissions of N₂O from nitrogen (N) fertilizer application were calculated as the product of fertilizer applied, a direct emissions factor, and a conversion factor. A direct emissions factor of 0.0085 g of N₂O–N per gram of N applied, applicable to horticultural systems in Queensland, was adopted following the ANGHGI. A molecular weight of 1.57 was used to convert N₂O–N to N₂O. Indirect emissions were calculated using the same framework, accounting for N losses due to leaching and runoff with a regional emissions factor of 0.293 and volatilization followed by atmospheric deposition, consistent with the ANGHGI. Upstream Scope 3 emissions associated with fertilizer production were calculated using values derived from the Ecoinvent database v3.8.

Emissions arising from the production and use of agrichemicals were included under upstream emissions. Emissions linked with the manufacture of inputs were computed based on the mass of chemicals used, their active ingredient concentration, and energy content. The GHG emissions from the use of fuel were categorized as either on-farm consumption (tractor operations, such as spraying and irrigation; Scope 1) or supply chain emissions associated with the transport of inputs (Scope 3). Total GHG emissions from fuel use were calculated using fuel quantities consumed and fuel type, following the National Inventory DCCEEW. Transport emissions were calculated as a function of the product of mass and the distance the input was moved. It was assumed that diesel was the primary fuel, with an average consumption rate of 0.55 L/km for trucks carrying 20–40 t payloads. Emissions were calculated using standard diesel combustion emission factors of 69.9 kg CO₂eq per gigajoule of diesel energy for CO₂, 0.1 kg CO₂eq per gigajoule for CH₄ and 0.2 kg CO₂eq per gigajoule for N₂O following the ANGHGI.

The GHG emissions from the purchase of grid electricity (Scopes 2 and 3) were calculated using regional or state-specific emissions factors adopted from the ANGHGI. We adopted an emissions factor for grid electricity from Queensland of 0.73 kg CO₂eq per kWh and 0.07 kg CO₂eq per kWh for Scopes 2 and 3, respectively. GHG emissions from the use of lime were determined following the ANGHGI. Emissions were calculated by multiplying the total mass of raw carbonate material used with the emissions factor according to the type of carbonate and the percentage of carbonate in the material used. Farm CO₂ emissions from limestone or gypsum application were quantified as direct emissions. Direct emission factors for the calcination of carbonates were 0.440 t CO₂ per tonne of limestone, 0.522 t CO₂ per tonne of magnesium carbonate, and 0.477 t CO₂ per tonne of dolomite, following the ANGHGI.

2.3.4. Carbon Sequestration in Soils

SOC was assessed on an annual and site-specific basis, with samples analyzed following the measurement-based soil carbon method articulated by the Australian Government Clean Energy Regulator (CER). Annual SOC data were used to assess changes in SOC stocks and related emissions during the evaluation period. SOC stocks were estimated for the entire harvested area of the orchard, following the guidelines of the ANGHGI. Longitudinal flux in SOC stock (ΔC _soil30) during the reporting period was calculated using measurements from the top 0.3 m of soil. The results were presented either as tonnes of carbon per hectare per year or as kg CO₂eq per unit of product:

$$\mathrm{\Delta }{C}_{\mathrm{soil30}}=\sum _{i=1}^n(\mathrm{\Delta }{C}_{\mathrm{soil}i}-{\mathit{CNSF}}_i)$$ 1

where ΔC _soil30 is the change in SOC in the top 0.3 m layer from the baseline year to the reporting year. ΔC _soili refers to the SOC change during the reporting period for each cropping area; n is the total number of cropping areas, and CNSF _i indicates the total amount of organic carbon input from all nonsynthetic fertilizers applied to the production area during the reporting period. SOC for a given (C _soil) was determined by computing a weighted average of SOC results from soil testing (CF) for the cropping area (i), with variation depending on soil bulk density (BD):

$$C_{\mathrm{soil}}=\sum _{i=1}^n({CF}_i\times {BD}_i\times V_i)/V$$ 2

where CF _i is the carbon fraction in the cropping area (percentage divided by 100); BD _i is bulk density (kg per cm³) in the cropping area i; and V denotes the volume of soil within a 0.3-m soil layer (m³). The number of soil tests required for a reporting year was based on the size of the cropping areas. The procedure for determining soil sampling locations can vary according to the production system and changes in farming practices. Following official guidance on soil sampling procedures provided by the Australian Government Clean Energy Regulator (see above), farm areas were stratified into sampling zones according to proximally sensed indices correlated to SOC. These zones established stratification using a combination of BD, elevation and normalized difference vegetation index (NDVI) such that (i) variability within strata/zones was minimized (ii) variability between strata/zones was maximized and (iii) variables for stratification strongly correlate to SOC change. A minimum of three soil cores per stratum were sampled.

2.3.5. Carbon Sequestration in Vegetation

Sequestration in native vegetation was quantified using remote sensing for woodlands, native grasses, improved pasture, and riparian vegetation. Vegetated areas were delineated, excluding infrastructure such as buildings, dams, and roads, and conservative sequestration rates were applied to calculate potential carbon sequestration within these vegetated systems. Carbon sequestration within the production system was allocated according to cultivation area (ha) and converted to per unit of product (tonne NIS) following the ISO 14044 standard for LCA. A value of three tonnes CO₂eq per ha per yr measured by Murphy et al. based on destructive sampling of macadamia orchards, was adopted for carbon sequestration values in macadamia biomass.

2.4. Impact of Practice Change and Climate on GHG Emissions

Systematic statistical procedures to analyze the effects of climate and management variables on emissions intensity (EI) were conducted using R software version 4.3.1 (2023-06-16). For preliminary analysis, means and standard deviations were computed using Microsoft Excel to characterize data sets, with scatterplots and correlation matrices presented (Figure S1 and Table S1).

2.4.1. Correlation Analysis

A correlation matrix was constructed to explore the relationships between climatic variables and management variables. These were correlated with the total GHG emissions as the response variable.

2.4.2. Multiple Linear Regression Analysis

Stepwise multiple linear regression analysis (MLR) was carried out using the Akaike Information Criterion (AIC) “Mass” package in R. The model was optimized by iteratively adding or removing variables following the stepwise regression approach. The regression model was validated based on the Breusch–Pagan and Shapiro–Wilk tests for homoscedasticity and normality of residuals. , Diagnostic plots were examined for homoscedasticity and normality, including residuals vs fitted values, histograms of residuals, and Q–Q plots (Figure S2). Multicollinearity among predictors was quantified with the Variance Inflation Factor (VIF), considering a threshold of 10 as indicating high multicollinearity.

2.5. Sensitivity Analyses

Sensitivity analyses were conducted to quantify how incremental changes in farm inputs and/or carbon sequestration impacted overall GHG emissions. We utilized LCA results pertaining to emissions and reductions by evaluating the effects of systematically perturbing five key contributors to net EI.

3. Results

3.1. Emissions Intensity Excluding Sequestration

Emissions intensity of orchards producing nuts averaged 1.8 ± 0.2 t of CO₂eq per tonne of NIS (Table S2). Significant variation in emissions intensity (excluding sequestration) existed between farms, ranging from 0.56 t of CO₂eq tonne of NIS to 4.5 t of CO₂eq per tonne of NIS (Figure ). Farm M25 had the lowest emissions intensity (0.6 t per tonne NIS), followed by farm M2 (0.7 tCO₂/t NIS), while M18 had the highest EI (4.5 tCO₂eq). Application of N fertilizer constituted a primary source of farm GHG emissions, contributing 58% of total emissions (Figure ). Average fertilizer use for nut-bearing orchards was 205 ± 35 kg N/ha. Emissions from farm fuel use (20%) were another major source, mainly comprising diesel use (e.g., 55 L/ha in M18 to 1347 L/ha in M11). Average electricity use accounted for 16% of total emissions. Annual farm electricity consumption (primarily for irrigation) ranged between 0 kW/ha (M40/M41) to 2,355 kW/ha (M16). Annual mean gross emissions across farms were 4.3 ± 0.4 t of CO₂eq per ha, including emissions from nut-bearing orchards (5.4 ± 0.4 t) and young orchards (1.8 ± 0.3 t). The highest gross emissions occurred for M11 (10.8 t of CO₂eq/ha), while M25 had the lowest (1.8 t of CO₂eq/ha).

3.

Net GHG emissions intensity per unit nut-in-shell (NIS). Net emissions were computed by subtracting total carbon sinks (total sink) from total emissions (total GHG). Sinks included SOC, macadamia biomass carbon, and native vegetation (negative values); emission sources included N fertilizer, farm fuel use, irrigation electricity use, agrochemicals, lime, and transport (positive values). Boxes demonstrate the interquartile range of emissions across the dataset; central line represents the median value, and dots represent outliers.

4.

Proportional contributions of farm inputs to average farm GHG emissions. Total GHG emissions were derived from total N fertilizer application, farm fuel use, electricity use for irrigation, agrochemicals, and lime application.

3.2. Carbon Sequestration

Mean annual SOC sequestration in the uppermost 30 cm across orchards was 8.7 ± 2.2 t CO₂eq per ha per yr, with 10 ± 3.0 t CO₂eq per ha per yr in nut-bearing orchards (3.1 ± 0.8 t CO₂eq per tonne NIS) and 3.4 ± 1.8 t CO₂eq per ha per yr in young orchards. Average annual sequestration by native vegetation across enterprises was 0.3 ± 0.1 t of CO₂eq per ha from an average of 146 ± 0.05 ha under native vegetation (Figure S3). Biomass carbon sequestration in tree wood averaged 1.0 ± 0.1 t of CO₂eq per tonne of NIS.

3.3. Net Farm GHG Emissions Including Carbon Sequestration

Net farm GHG emissions, including all sources of sequestration, were −8.9 ± 2.8 t CO₂eq per ha per yr in nut-bearing orchards and −4.6 ± 1.8 t CO₂eq per ha per yr in young orchards. There was substantial variation in net emissions per unit of cropped area between nut-bearing orchards. Farm M6 exhibited net emissions as low as −74 t CO₂eq per ha, while farm M31 had emissions of up to 3.3 t CO₂eq per ha. For young orchards, we found that farm M49 had the lowest net emissions (−23 t of CO₂eq per ha), while farm M28 had the highest emissions (1.3 t of CO₂eq per ha). Emissions intensity of nut-bearing orchards was −2.5 ± 0.7 t of CO₂eq per tonne of NIS, varying from −16.8 to 1.9 t of CO₂eq per tonne of NIS (Figure ). Orchard M6 had the lowest net EI (−16.8 tCO₂eq per tonne of NIS), while orchard M46 had the highest net EI (1.9 t of CO₂eq per tonne of NIS) (Figure ).

5.

Emissions intensity (tonnes of CO₂eq per tonne of NIS) for commercial nut-producing macadamia farms. Bars represent individual farms disaggregated by emissions and emissions removal sources: N fertilizer use, farm fuel use, farm lime use, emissions due to transport of inputs, SOC sequestration, macadamia biomass carbon, and native vegetation carbon. Black dots represent net emissions intensity.

3.4. Impact of Climate and Management Practices on Emissions Intensity

Total GHG emissions were positively skewed (data not shown) and were thus log-transformed to meet prerequisite requirements for multiple linear regression analysis (Total_GHG_log). Pearson’s correlation analysis revealed a significant positive correlation between the use of electricity for irrigation, N fertilizer, and petrol use (Figure ). A moderate negative correlation was observed between temperature and rainfall, indicating lower temperatures in environments with higher rainfall. Total_GHG_log was positively correlated with N fertilizer, diesel, and electricity use.

6.

A Pearson correlation matrix showcasing relationships among predictor variables and log-transformed total GHG emissions. Cells with values closer to ±1.0 denote stronger correlations between each pair of variables.

The stepwise regression approach, which combines forward selection and backward elimination, gradually eliminates variables that do not significantly impact total GHG emissions. The final model is presented in eq

$${\mathrm{Ttl\_GHG}}_{\log }={\beta }_{0}0+{\beta }_1\times \mathrm{N\_fertilizer}+{\beta }_2\times \mathrm{Diesel}+{\beta }_3\times \mathrm{Electricity}+{\beta }_4\times \mathrm{Agrichemicals}+\epsilon$$ 3

Where,

Total _ GHG _log = natural logarithm of total GHG emissions

β _i =coefficients calculated by the model

N_fertiliser = N fertilizer used on farm

Diesel = diesel used on farm

Electricity = electricity used for irrigation

Agrichemicals = combined insecticides, herbicides, fungicides, and adjuvants used for plant protection

∈ = represents error.

The final model retained N fertilizer use (N_fertiliser), diesel use for farm activities (Diesel), electricity used for irrigation (Electricity) and agrichemicals (Agrochemicals) as significant predictors of log total GHG. The analysis demonstrated a significant positive relationship between the use of N fertilizer, diesel, and irrigation electricity (Table ), whereas the negative association between agrichemicals and total GHG was not significant.

2. Multiple Linear Regression Model Coefficients and Statistics,

Variable	Coefficient (Estimate)	Standard Error	t value	p-value	Significance
Intercept	6.57 × 1000	1.37 × 10–01	47.817	<0.001	***
N_fertiliser	2.21 × 10–03	1.08 × 10–03	2.053	0.049	*
Agrochemicals use	–6.15 × 10–06	4.58 × 10–06	–1.342	0.19
Diesel	2.82 × 10–03	7.50 × 10–04	3.755	<0.001	***
Electricity	1.02 × 10–03	2.68 × 10–04	3.82	<0.001	***

^aN fertiliser use (N_fertiliser), diesel use (Diesel) and electricity use for irrigation (Electricity) were significant predictors of total GHG (Ttl_GHGlog).

^b*p < 0.05, ***p < 0.001. Model residual standard error was 0.34 on 29 degrees of freedom, R ² of 0.64, adjusted R ² of 0.59 and F-statistic of 13 (p < 0.001).

Model diagnostics showed that statistical assumptions were preserved (VIF values were below the threshold of 10, indicating minimal multicollinearity; Table S3). The Breusch–Pagan test (BP = 3.6, df = 4, p = 0.45) and Shapiro–Wilk test (W = 0.94, p = 0.06) did not reveal heteroscedasticity or departure from normality, respectively. Residual plots confirmed normally distributed residuals (Figure S2).

3.5. Interventions for Reducing GHG Emissions

The sensitivity analysis of how various management interventions impact net farm GHG emissions indicated that SOC sequestration had the greatest impact, with an increase or decrease of 30% in SOC reducing or increasing net emissions intensity by 22%, respectively. N fertilizer was also a key driver of net GHG emissions. Net emissions intensity was less sensitive to the use of electricity, irrigation, fuel, and carbon sequestration in macadamia vegetation.

4. Discussion

Here, our objective was to determine the GHG emissions and carbon removals associated with commercial enterprises in a key macadamia production region of Australia. We showed that nitrogen fertilizer, diesel, and electricity were key GHG drivers, accounting for 58%, 20%, and 16% of CO₂eq emissions per tonne of NIS. We documented an average EI of 1.8 ± 0.2 t CO₂eq per tonne NIS. Total emissions were 5.4 ± 0.4 and 1.8 ± 0.3 t CO₂eq per ha for mature and young orchards, respectively.

4.1. Commercial Macadamia Enterprises Tend to Be Carbon Sinks

A fundamental insight was that a majority of macadamia orchards studied here were carbon sinks. Using LCA from cradle to gate, we revealed that net GHG emissions for these orchards averaged −7.6 ± 2.1 t CO₂eq per ha per yr (range −74.1 to +6.0 t CO₂eq per ha per yr), equivalent to −2.5 ± 0.7 t CO₂eq per tonne NIS (range −16.8 to 1.9 t CO₂eq per tonne NIS). These results suggest that aggregate GHG emissions across farms studied were negative, which would mitigate global warming. These results emanate from the fact that annual GHG emissions were often less than the combined carbon sequestration in soils and woody biomass of macadamia and native vegetation, which averaged −8.9 ± 2.8 t of CO₂eq per ha per year across farms. Our results parallel those of Ledo, Heathcote, Hastings, Smith, and Hillier, who opine that assimilated carbon more than counterbalanced emissions from inputs when carbon sequestration in plant biomass and residue was considered. Collectively, our results suggest that commercial macadamia enterprises are more likely to be net carbon sinks than carbon sources. This contrasts with broadacre cropping and livestock sectors, which are predominantly net sources of GHG emissions. ,, This is ultimately good news for Australian macadamia producers and the nut production industries more generally.

We showed that net emissions per unit orchard area varied greatly between mature nut-bearing and young farms, with the former being almost twice as great as the latter. Farms with the highest net emissions were primarily attributable to high extant SOC stocks and thus had less scope for SOC sequestration (e.g., Figure ). This suggests that SOC sequestration cannot continue indefinitely, − and when SOC sequestration rates diminish, net GHG emissions and EI tend to increase. These results underline the sequestration versus mitigation conundrum (i.e., carbon removals vs reduction of GHG emissions quantum) that challenges all land-use sectors: whereas longitudinal sequestration in soils and vegetation inevitably slows over time, there is often an opportunity to reduce, avoid, and mitigate emissions of GHG. Further, carbon sequestered can be lost (e.g., through soil cultivation or burning of wood), while avoided GHG emissionssuch as the use of less N fertilizertend to elicit permanent mitigation (avoided N₂O cannot reenter the atmosphere at a later stage). We therefore suggest that macadamia farmers, and indeed all land-use sectors, implement both mitigation and carbon removal technologies and practices, given that carbon sequestration eventually slows over time, whereas ceteris paribus mitigation of GHG sources accumulates linearly.

We found significant variation in SOC between young and mature trees, with the mature farms sequestering more than three times the SOC compared with young farms. With their extensive root systems and higher biomass, mature trees in older orchards contribute significantly to the SOC pool via surface and soil organic matter, root exudate deposition, leaf litter, woody debris, and recruitment of soil microbiota. This organic input is crucial for carbon incorporation into the SOC pool, although it is less pronounced in younger orchards due to lower biomass and root depth. Gross EI, excluding sequestration, varied widely across enterprises, ranging from 0.6 to 4.1 t of CO₂eq Similar variability has been reported in Californian almonds (0.6 to 3.1 t CO₂eq per tonne NIS), walnuts (0.0 to 1.3 t CO₂eq per tonne NIS), and pistachios (−0.1 to +2.5 t CO₂eq per tonne NIS). Differences in EIs within the same crop can be driven by multiple factors, including those pertaining to agronomy, environment, social issues, and regulatory factors. ,,,,

We applied a constant annual tree sequestration rate of 3 t CO₂eq/ha based on empirical measurements reported in peer-reviewed literature. As this constant rate may not capture the full spectrum of carbon sequestration across cultivars, tree age, soil conditions, management practices, growing regions, and interactions therein, we conducted a sensitivity analysis to examine how GHG emissions mitigation was perturbed by varying carbon sequestration rates (Figure ). This analysis suggested that emissions intensity was more sensitive to soil carbon sequestration and the use of nitrogen fertilizer on farms compared with carbon sequestration in macadamia wood. As such, we contend that our adoption of constant macadamia wood sequestration is a robust approach, given that it is based on empirical data measured in the study region. Future studies could account for variation in carbon sequestration with tree age, location, and orchard management practices should they aspire to greater sophistication than employed here.

7.

Sensitivity analysis of net emissions intensity to perturbation of GHG sources and carbon sinks.

4.2. Reducing N Fertilizer Could Reduce N₂O Emissions without Yield Penalty

Similar to horticultural studies hitherto, we found that N fertilizer was a fundamental determinant of EI. As such, reducing N fertilizer (in our study implemented at an average rate of 193 kg N per ha per yr) could reduce total GHG emissions in macadamia farms while lowering production costs, assuming that yields did not decline. We showed that a 30% reduction in N fertilizer could reduce farm-level GHG emissions by up to 17%, primarily due to the avoidance of N₂O emissions, consistent with previous research. In the same vein, we did not find a strong relationship between N fertilizer use and yield (Figure S4). Previous studies have shown that using more than 138 kg N per ha per year could even reduce macadamia yields, while lower fertilization (46 kg N per ha per yr) improved yield and kernel recovery. , The fact that most enterprises in the present study applied N fertilizers at far greater rates implies significant scope exists for sustainably reducing N application without loss in nut yield or quality, , although further investigation of soil nutrient status and plant N uptake is required to solidify this hypothesis.

Fertilizer timing, type, and source may provide other opportunities for increasing plant uptake of N fertilizer (e.g., split dosage per annum or use of slow-release fertilizers), reducing N₂O and ammonia emissions, and thus increasing N-use efficiency. It is important to note that planting density, soil type, and real-time nutrition status must also be considered. Other interventions, such as the use of biochar as a soil amendment, − irrigation automation and monitoring, , and reduced or avoided tillage, have also been recommended. , Adoption of precision techniques, such as N-inhibition compounds and slow-release fertilizers, shows promising prospects for GHG mitigation. − Other authors , caution that any practice change requires a holistic lens, considering all aspects of the production system. This is required to avoid emissions swapping, where one GHG increases with the mitigation of another, such that benefits in one dimension are not offset by detrimental changes in other dimensions. , Benefit-to-cost ratios are also fundamental in determining whether prospective interventions will be adopted for the long term, as farmers will not tend to change practices that reduce profitability.

Farm use of diesel and irrigation electricity was responsible for more than a third (36%) of total emissions. This highlights the potential of the sector to further lower its carbon footprint via avoiding fossil fuels and grid electricity. Opportunities include the implementation of renewable energy sources, such as waste biomass, solar power, and/or wind turbines. The use of renewable sources of energy for irrigation would seem a viable option to lower emissions associated with electricity use and may also constitute cost savings once initial capital expenditure is recouped.

4.3. Impact of Seasonal and Long-Term Climate

Various studies , have concluded that increased irrigation did not fully compensate for other detrimental climatic impacts on productivity. The low association between climate and yield in the present study could be attributed to the fact that macadamia trees are indigenous to northeastern Australia, which may confer inherent drought resilience and/or adaptation to local climatic conditions. Additionally, the soils of the Bundaberg region are highly fertile, often consisting of deep black cracking clays rich in organic matter. In contrast, soils in regions where negative impacts of climate on macadamia production have been observed tend to have low organic carbon and predominantly sandy-clay-loam topsoils of lower fertility − suggesting that environmental conditions were less conducive to macadamia production than those in the Bundaberg region examined here.

4.4. A Need for Integrated Economic, Social, and Environmental Assessments

Wider economic, social, environmental, and cultural factors play a considerable role in determining whether or not a low-emissions practice, skill, or technology is adopted, even when such technologies have been shown to substantively reduce GHG emissions. As such, the use of cost–benefit analysis, among other assessments, is suggested to determine the economic viability of any given practice change. , While investment costs associated with adopting sustainable practices like organic farming or renewable energy can be significant, long-term benefitssuch as improved soil health, increased biodiversity, and lower energy costsalong with incentives through premium pricing or carbon sales may yield significant payback in the future. At the same time, an improved understanding of socio-economic factors that influence the adoption of recommended management strategies is essential, for example, knowledge, skills, and barriers to adoption. ,

Moreover, the use of multiple metrics to compare the environmental impacts of various agricultural sectors could conceivably lead to more balanced conclusions compared with assessments based on only one metric (e.g., GHG emissions). For example, a global environmental impact analysis of manifold commodities revealed that the fruit and tree nut sector has a much lower environmental impact compared with other sectors, with the production of 100 g of nut protein requiring 8 m² of land, while the same amount of beef protein requires 20-fold more land. Per 100 g of protein, dried shell-free nuts emit 0.3 kg of CO₂eq, whereas bone-free beef meat emits around 50 kg of CO₂eq Water usage also varies widely, with beef requiring nearly 160 L compared with 28 L for nuts, depending on how water use is quantified. In terms of emissions reduction, more orchards (relative to other agricultural industries) may lower agricultural GHG emissions. However, climatic and pedological variables preclude the widespread expansion of perennial fruits and nuts; indeed, livestock production occurs in many regions that are unsuitable for plant-based enterprises. Even so, extending a production system to reduce emissions from another production system may not always be viable, as reduced yields could necessitate larger cultivation areas. One option could be enterprise diversification, where broadacre cereal cropping/livestock producers implement a commercial orchard on part of the farm to generate alternative financial income.

Further investigation of potential carbon removals associated with biowaste management may improve contemporary understanding of the drivers of GHG abatement. ,, For example, biogenic waste management in orchards through the conversion of biomass into biofuel, compost, animal feed, or biochar offers notable carbon sequestration and emissions reduction opportunities. − We conclude that most macadamia farms contribute not just to food security and enterprise prosperity but also to environmental stewardship via the mitigation of global warming.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c05777.

• Model diagnostics, assumptions, and correlation analysis, including diagnostic plots used to evaluate stepwise regression assumptions; additional results on carbon credits derived from native vegetation across farms; supplementary analysis of the interaction between N fertilizer use and crop yield (PDF)

The authors declare no competing financial interest.