Assessing the Impact of Temperature and Precipitation Trends of Climate Change on Agriculture Based on Multiple Global Circulation Model Projections in Malta

Abstract

The Maltese Islands, situated at the centre of the Mediterranean basin, are recognised as a climate change hotspot. This study utilises projected changes in temperature and precipitation derived from the World Climate Research Program (WCRP) and analyses outputs from six coupled model intercomparison project phase 5 (CMIP5) models under two Representative Concentration pathways (RCPs). Through statistical and spatial analysis, the study demonstrates that climate change will have significant adverse effects on Maltese agriculture. Regardless of the RCP scenario considered, projections indicate a substantial increase in temperature and a decline in precipitation, exacerbating aridity and intensifying heat stress. These changes are expected to reduce soil moisture availability and challenge traditional agricultural practices. The study identifies the Western District as a relatively more favourable area for crop cultivation due to its comparatively lower temperatures, whereas the Northern and South Eastern peripheries are projected to experience more severe heat stress. Adaptation strategies, including the selection of heat-tolerant crop varieties such as Tetyda and Finezja, optimised water management techniques, and intercropping practices, are proposed to enhance agricultural resilience. This study is among the few comprehensive assessments of bioclimatic and physical factors affecting Maltese agriculture and highlights the urgent need for targeted adaptation measures to safeguard food production in the region.

1. Introduction

Climate change is a global phenomenon encompassing diverse implications, characterised by erratic and severe weather patterns that profoundly affect both human society and the natural environment. The increased emissions of greenhouse gas (GHG) concentrations in the atmosphere have expedited the pace of climate change, primarily driven by human activities. A notable concern pertains to the impact of climate change on agriculture and its repercussions on crop yields [1]. Agriculture, being the predominant global consumer of freshwater, is especially vulnerable to the effects of climate change [2]. The Intergovernmental Panel on Climate Change (IPCC) has emphasised the potential outcomes of heightened extreme weather occurrences on global agricultural resources [3] such as precipitation, soil moisture, and droughts, which directly impact agricultural productivity. In assessing forthcoming climate change projections for the Maltese Islands, these indicators, alongside bioclimatic variables, assume paramount significance in comprehending potential impacts on the local agricultural sector. Agriculture occupies a pivotal role in the Maltese economy and the preservation of the physical landscape and rural social fabric. In the local context, one of the main consequences of climate change is the increasing scarcity of freshwater resources [3]. However, the small size of the Maltese Islands presents challenges for the agricultural sector, given the constrained and costly nature of local adaptation measures.

In a broader context, the Mediterranean region is projected to undergo seasonal shifts in temperature and rainfall patterns by 2100, based on the RCP 4.5 climate projection. Anticipated impacts of climate change encompass a 5 to 6% reduction in precipitation, heightened and more frequent heat waves, a sea level rise of 0.4 to 0.5 m, and a surface temperature increase of 1.2 to 2.3 °C (in relation to the 1986–2005 baseline) [3]. Notably, Galdies [4] has documented a warming trend over the Maltese Islands from 1952 to 2022, resulting in a mean annual ambient atmospheric temperature increase of 1.5 °C. This rise in temperature could potentially lead to increased agriculturally related pest outbreaks, as milder winters enable pests to endure for longer periods, while extended summers offer conducive conditions for insect proliferation [5]. However, within the context of the Maltese Islands, Galdies [4] identified a 1.81% decrease in annual rainfall trends when comparing the period from 1991 to 2020 with that from 1961 to 1990. Additionally, Galdies and Meli [6] observed a decade-long decline of −6 mm in total annual rainfall from 1946 to 2020, suggesting an upsurge in local drought occurrences. Indeed, the frequency of drought conditions has amplified during the past two decades.

Agriculture and crop production systems, characterised by their sensitivity, experience significant influences from climate variability [7]. Consequently, this leads to several impacts, including diminished water availability due to heightened evapotranspiration rates, changes in phenology affecting reproduction and crop yield, fruit quality and yield reduction, decreased overall crop yield due to elevated soil temperature and water scarcity, and decreased rainfall and soil moisture [6,8,9].

Table 1. The threshold and optimum temperatures, along with precipitation/irrigation levels, for the three primary crops cultivated on the Maltese Islands, namely, potatoes, forageable crops, and vineyard crops.

Crop Type	Threshold Temperature	Optimal Temperature	Threshold Precipitation/Irrigation	Optimal Precipitation/Irrigation
Potatoes	7 °C	15–20 °C	250–300 mm	500–700 mm
Forageable Crops	5–10 °C	19–24 °C	150–250 mm	450–650 mm
Vineyard Crops	10 °C	25 °C	500 mm	635–890 mm

has been derived due to the absence of concrete or definitive values regarding the optimal temperature and precipitation/irrigation thresholds for various crops. Despite extensive research through multiple literature sources, there was no single, universally accepted set of values. To ensure accuracy, consultations with a local agronomist were also conducted. Based on the information gathered from these varied sources, this table has been compiled to provide a comprehensive and informed overview of the temperature and precipitation/irrigation requirements for each crop type. Additionally, it represents the critical threshold temperatures and the corresponding precipitation/irrigation conditions, juxtaposed with the optimal parameters, for the three focal locally cultivated crops under investigation: potatoes, forage, and vineyards. By comparing the threshold and optimal temperature and precipitation/irrigation values for crops with global circulation model (GCM) projections for Malta, we can assess whether projected climate changes will push conditions outside the optimal growth range. Furthermore, it allows for the evaluation of the adaptability of different crops to future climate conditions in the Maltese Islands, such as whether rising temperatures could add stress to the potatoes or if reduced rainfall could require increased irrigation for vineyards.

Currently, extensive knowledge gaps prevail in the context of climate change future projections and the related susceptibility of Maltese agricultural land. Meli [10] underscores that due to the Mediterranean region shifting towards drier and warmer conditions, nearly half of the Maltese Islands’ usable agricultural area is anticipated to become economically unsustainable. Furthermore, agricultural land situated within low-lying coastal zones (below 10 metres above sea level) is particularly prone to inundation. A rise in sea level by one metre would notably impact coastal areas like Mellieħa and agricultural zones such as Burmarrad [11].

In terms of raising awareness, Galdies et al. [12] affirm that technology-driven management systems are pivotal for site-specific agricultural management, sustainability, and economic viability. These systems, as recommended by local farmers as part of potential climate adaptation measures, aid in understanding and managing spatial and temporal resource changes, offering farmers insights into climate change impacts on their sector and available adaptation measures. However, Cortis [13] found that not all farmers are aware of the vulnerability of their agricultural land and the potential mitigation strategies against climate change. The need for precise land cover maps for Malta is highlighted, with the suggestion to employ advanced aerial imagery and remote sensing methods, like the land parcel identification system [14].

GCMs now play a pivotal role in projecting future climatic conditions, particularly at regional and seasonal levels. The output from GCMs is a primary source for creating climatic scenarios. Improvements in sea surface temperature simulations, teleconnection mechanisms, and extreme weather event modelling have addressed key limitations of previous model phases, enhancing the ability to capture both externally forced trends and internal variability [15]. Additionally, the increased availability of longer observational time series and improved understanding of climate forcings have refined projections of temperature and precipitation patterns. By comparing model projections with historical time-series data from multiple locations across a country, researchers can estimate how average temperature and precipitation are expected to change over time [16]. These projections are then integrated with historical data to yield more precise forecasts for future climatic conditions. Following the release of the Coupled Model Intercomparison Project phase 5 (CMIP5), more advanced multi-model ensemble datasets from CMIP6 [17] and CMIP7 [15] have been introduced. While CMIP6 models aim to enhance climate process accuracy and employ updated emissions scenarios [18], the choice to utilise CMIP5 for this study is based on its widespread adoption in existing climate change research and policy frameworks. Much of the current knowledge and ongoing climate change risk assessments and adaptation strategies are still rooted in CMIP5, making its projections highly relevant for continuity in research and practical applications [19].

At the time of data collection in 2023, high-resolution downscaled CMIP6 datasets were not readily available at the spatial resolution required for this study. Currently, the finest resolution available for CMIP6 from the Copernicus datasets is 12 km, with the most widely accessible datasets offering a coarser resolution of approximately 50 km [20]. In contrast, CMIP5 is the only model framework that provides a 1 km downscaled dataset, which is particularly important for small-scale regional studies, such as those focusing on Malta. Furthermore, recent research [21] highlights that CMIP5 exhibits lower inter-model variability and a more balanced precipitation distribution when compared to CMIP6, making it a more stable choice for assessing climate impacts. While both CMIP5 and CMIP6 indicate rising temperatures and increased precipitation, CMIP6 projects stronger warming trends and a more uneven annual precipitation distribution, with wetter wet seasons and drier dry seasons.

The key parameters of the CMIP5 models used in this study are presented in

Table 2. The estimated effective radiative forcing (ERF), equilibrium climate sensitivity (ECS), transient climate response (TCR), climate sensitivity parameter (CSP), and climate feedback parameter (CFP), for the CMIP5 models used within this study. ERF represents the radiative forcing due to a doubling of CO₂, calculated using both fixed station sea surface temperatures (fixed SST) and regression methods. ECS refers to the long-term global temperature response to CO₂ doubling, while TCR calculates the near-term warming over a 70-year period with a 1% annual CO₂ increase. CSP measures the temperature response per unit forcing, and CFP quantifies the net radiative feedback strength, with more negative values indicating stronger stabilising feedback [22].

Model	ERF Fixed SST (W/m2)	ERF Regression (W/m²)	ECS (°C)	TCR (°C)	CSP (°C (W m−2) −1)	CFP (W m−2 °C-1)
ACCESS1-O (AC)	n.a.	3.0	3.8	2.0	1.3	0.8
BCC-CSM1-1 (BC)	n.a.	3.2	2.8	1.7	0.9	1.1
CCSM4 (CC)	4.4	3.6	2.9	1.8	0.8	1.2
CNRM-CM5 (CN)	n.a.	3.7	3.3	2.1	0.9	1.1
GFDL-CM3 (GF)	n.a.	3.0	4.0	2.0	1.3	0.8
MIROC-ESM (MR)	n.a.	4.3	4.7	2.2	1.1	0.9

, all of which were highlighted in the comprehensive assessment by Flato et al. [22]. This study outlined the main features of CMIP5 models and their advancements compared to previous generations, particularly improvements made since the Fourth Assessment Report by the IPCC. These enhancements contribute to more reliable climate projections, which are essential for assessing potential agricultural impacts in the Maltese Islands, as well as for identifying and illustrating the most robust models for national-level policymaking.

The ECS among the models varies significantly, ranging from 2.8 °C in BC to 4.7 °C in MR. The highest ECS value in MR suggests a stronger warming response to CO₂ doubling compared to models like BC and CC, which have lower ECS values of 2.8 °C, and 2.9 °C, respectively. The TCR follows a similar trend, with MR exhibiting the highest value (2.2 °C) and BC the lowest (1.7 °C). This aligns with the general expectation that models with higher ECS tend to have higher TCR [23]. The effective CSP ranges from 0.8 °C (W m⁻²)⁻¹ in CC and GF to 1.3 °C (W m⁻²)⁻¹ in AC and GF, reflecting differences in the strength of the climate feedback. Conversely, the CFP is highest in CC (1.2 W m⁻² °C⁻¹) and lowest in MR (0.9 W m⁻² °C⁻¹), indicating variations in how the models balance radiative forcing with temperature changes. The ERF diagnosed via regression also exhibits notable differences, ranging from 3.0 Wm⁻² (AC, GF) to 4.3 W/m² (MR), suggesting differences in the models’ representation of radiative forcing mechanisms. The CC model can be seen as an outlier in the ERF fixed SST (4.4 Wm⁻²) versus regression (3.6 Wm⁻²), showing a larger discrepancy than other models, which may point to methodological differences in forcing estimation. Additionally, the study by Meehl et al. [23] showcased how the average ECS and TCR values across CMIP5 models project less significant warming trends when compared to CMIP6. For context, CMIP5 had an ECS and TCR of 3.2 °C and 1.8 °C, respectively, whereas in CMIP6, the corresponding values are slightly higher at 3.7 °C and 2.0 °C. This suggests that CMIP6 may introduce greater uncertainty in climate projections, which could be a limitation for studies requiring consistency and lower variability. Given these factors, this study employs CMIP5 as the primary climate modelling framework.

The primary motivation guiding this research study is to examine the anticipated climatic impacts on agriculture, specifically crop cultivation, in the Maltese Islands. This is achieved by employing downscaled GCMs from the CMIP5, which provide a comprehensive set of climate projections under multiple emission scenarios. As discussed by Vaittinada et al. [24], downscaling, or the process of translating across scales, has been a term used in recent years to describe a set of techniques that connect local and regional climate variables to larger-scale atmospheric forces. It also plays a crucial role in remote sensing by enabling predictions at a finer spatial resolution than that of the input imagery [25].

To capture the potential impacts under different levels of greenhouse gas emissions, two Representative Concentration pathways (RCPs), 4.5 and 8.5, will be utilised, focusing on the years 2050 and 2070. These climatic scenarios were selected as they represent moderate and high-emission pathways, with their accompanying number (4.5 & 8.5) representing their radiative forcings of 4.5 Watts per square metre (Wm⁻²) and 8.5 Wm⁻², respectively, providing a range of possible climate outcomes under varying levels of anthropogenic radiative forcing [26,27]. These scenarios are commonly employed in climate projections and offer a robust framework for assessing potential regional climate impacts across different emission pathways.

This study evaluates this impact on an administrative district level and crop-by-crop basis, providing valuable insights into the future of agriculture in Malta and strategies for safeguarding our crops.

The main objectives of this study are the following:

• To determine the projected impacts of climate change on Maltese agriculture under RCP 4.5 and RCP 8.5 for the years 2050 and 2070 at a district level;
• To evaluate the implications associated with these projected scenarios on a district level regarding crop yields and current agricultural practices;
• To propose measures for future-proofing local agriculture.

2. Methodology

This study adopts a design science approach to analyse the impact of climate variables on agricultural production in the Maltese Islands. The methodology was selected to integrate spatial analysis, agricultural census data, and climatic projections in a structured, reproducible manner, ensuring a comprehensive understanding of the impact of climate on agriculture. A district-level analysis was chosen due to the availability of high-resolution agricultural census data from 2010 and 2020, which provides detailed insights into land use and irrigation patterns across Malta. To assess the potential climate impact, the study utilised CMIP5 projections rather than the more recent CMIP6 or CMIP7 models, due to their established validation and the availability of downscaled data for the region at the time of the study. Bioclimatic variables (BIOs) were selected instead of raw temperature and precipitation values, as they provide a more ecologically relevant representation of climate trends affecting agricultural systems [28]. Spatial mapping was carried out using QGIS, with official Maltese district shape files sourced from the MITA Geoportal, enabling the creation of index maps. These maps facilitated the classification and ranking of districts based on their exposure to climate impacts. The following sections outline the data sources, processing techniques, and analytical framework employed to achieve the study objectives.

2.1. Study Area

Situated at 35.9375° North and 14.3754° East in the heart of the Mediterranean Sea, the Maltese Islands (Figure 1) encompass an approximate area of 316 km² and house a population of 552,747 [29]. As noted by Galdies [30], the islands experience an average annual temperature of 18.6 °C and a yearly precipitation of 553 mm, resulting in mild wet winters and hot dry summers.

2.2. Future Temperature and Precipitation Data

Climate change projections of temperature and precipitation were sourced from the WorldClim website, version 1.4, delivering six downscaled and bias-corrected GCM outputs at a 30 s resolution for the chosen GHG emission scenarios (

Table 3. The data derived from the climate models used for this study, alongside their respective codes and origin source.

Climate Model	Code	Source
ACCESS1-O	AC	Australian Bureau Of Meteorology (BOM) and the Commonwealth Scientific and Industrial Research Organisation (CSIRO)
BCC-CSM1-1	BC	Beijing Climate Center, China Meteorological Administration
CCSM4	CC	National Center for Atmospheric Research (NCAR)
CNRM-CM5	CN	Centre National de Recherches Météorologiques (CNRM) (France)
GFDL-CM3	GF	Geophysical Fluid Dynamics Laboratory (GFDL) (United States)
MIROC-ESM	MR	Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research Institute (The University Of Tokyo), and National Institute for Environmental Studies

). These projections stem from RCPs 4.5 and 8.5, developed by the IPCC to outline diverse trajectories of GHG emissions and their potential climatic impacts. Notably, RCP 4.5 anticipates a moderate decline in emissions, accompanied by a shift toward renewable energy sources, whereas RCP 8.5 envisions a sustained increase in GHG emissions throughout the twenty-first century [31].

The six bioclimatic variables being investigated in this study are derived from these RCPs and are presented in

Table 4. The six bioclimatic variables used for this study, along with the climatic parameters they read for.

Bioclimatic Variable	Climatic Parameter
BIO 1	Projected Annual Mean Temperature (°C)
BIO 5	Projected Maximum Temperature Of Warmest Month (°C)
BIO 6	Projected Minimum Temperature of Coldest Month (°C)
BIO 12	Projected Annual Precipitation (mm)
BIO 16	Projected Precipitation Of Wettest Quarter (mm)
BIO 17	Projected Precipitation of Driest Quarter (mm)

. These indices, comprising a total of nineteen biologically significant climate variables, hold widespread applicability in ecological and biogeographical research. These variables, computed from monthly temperature and rainfall data, play a key role in modelling species distributions, assessing the impact of climate change on ecosystems, and scrutinising biogeographical patterns [32].

2.3. Production of Index Maps

2.3.1. Data Extraction from the Agricultural Census

Data from the agricultural census conducted in 2020 and 2010 were extracted from relevant sections to generate a series of five index maps. Three of these are crops which were chosen based on their importance for the food security of the Maltese Islands [33]. This process involved inputting numerical values from each map at the district level, accompanied by the creation of a legend that corresponds to the total count of values within each index map.

2.3.2. Index Map Generation

Six Maltese Island district shape files were obtained from the MITA Geoportal [34] and imported into QGIS. For each mapped index, six separate shape files representing the official districts of Malta that are in line with the EU-standard regional classifications were loaded; these include the Northern District, Western District, South Eastern District, North Harbour District, South Harbour District, and Gozo and Comino. Data spatial processing involved creating a coloured legend for each index map based on a scale of high, intermediate, and low values. These values were sourced from the latest census of agriculture for Malta (2020) [33]. Most of these maps take into consideration land area used (measured in hectares), while one of them looks into the water volume used for irrigation (measured in cubic metres). A value range of the land coverage, and one for the water volume used for irrigation, was assigned to the colouration, shading the districts accordingly. Value labels were assigned to each district, and this process was repeated for each index, yielding six maps. The map (Figure 2) was used for the numerical classification of districts and as a final step in the methodology.

2.4. Analysis of Crop Production by District

The ranking of the affected districts was determined using spatial mapping, where the area experiencing the most significant impact (e.g., highest temperature or lowest precipitation) compared to the other districts was given the highest impact rank. For example, District 2 was identified as the most impacted district overall in terms of both temperature increase and precipitation decrease, with the only exception being for ‘Projected Precipitation of Driest Quarter’ (BIO 17), where District 1 had a higher rank than District 2.

2.5. Projected Changes and Inter-Model Spread

Cluster analysis, as highlighted by Galdies and Vella [35], is a widely used statistical method for examining relationships among models in CMIP5 simulations, with applications in both historical analyses and long-term future projections. Building on this approach and similar studies that employed hierarchical clustering to analyse climate model outputs [35,36,37], this study utilised cluster analysis to assess the variability of multi-model projections under RCP 4.5 and RCP 8.5 for the years 2050 and 2070.

Using PAST (version 4.03), a classical hierarchical cluster analysis was performed, implementing Ward’s method to optimise clustering. This analysis produced a total of four dendrograms, two for each RCP scenario. Each dendrogram grouped three relevant bioclimatic variables: one depicted temperature indices (BIO 1, BIO 5, and BIO 6), while the other focused on precipitation indices (BIO 12, BIO 16, and BIO 17). The clusters represented climate models, allowing us to identify the cluster members of the two main clusters and determine how the models were grouped. Additionally, the cophenetic correlation coefficient was calculated to assess clustering accuracy. Through this clustering approach, we were able to analyse the inter-model spread and variability, gaining insights into the degree of agreement or divergence among climate models.

3. Results

3.1. Generation of Index Maps and the Projected Bioclimatic Variables

The findings are presented separately for each crop type, including potatoes, forageable crops, and vineyard crops. Initially, the high-resolution bioclimatic maps for the years 2050 and 2070 are shown, based on the two RCP scenarios. These maps are categorised into two tables:

Table 5. Averaged temperature bioclimatic variables under four different climate scenarios (RCP 4.5 for 2050 and 2070, RCP 8.5 for 2050 and 2070). The variables include BIO 1 (projected annual mean temperature, °C), BIO 5 (projected maximum temperature of the warmest month, °C), and BIO 6 (projected minimum temperature of the coldest month, °C).

	RCP 4.5 2050	RCP 4.5 2070	RCP 8.5 2050	RCP 8.5 2070	Legend (°C)
BIO 1
BIO 5
BIO 6

displays three temperature indices; BIO 1 (projected annual mean temperature, °C), BIO 5 (projected maximum temperature of the warmest month, °C), and BIO 6 (projected minimum temperature of the coldest month, °C), while

Table 6. Averaged precipitation bioclimatic variables under four different climate scenarios (RCP 4.5 for 2050 and 2070, RCP 8.5 for 2050 and 2070). The variables include BIO 12 (projected annual precipitation, mm), BIO 16 (projected precipitation of the wettest quarter, mm), and BIO 17 (projected precipitation of the driest quarter, mm).

	RCP 4.5 2050	RCP 4.5 2070	RCP 8.5 2050	RCP 8.5 2070	Legend (mm)
BIO 12
BIO 16
BIO 17

presents three precipitation indices; BIO 12 (projected annual precipitation, mm), BIO 16 (projected precipitation of the wettest quarter, mm), and BIO 17 (projected precipitation of the driest quarter, mm). Following these, the index maps illustrate the spatial distribution of projected yield anomalies (ha).

The analysis of climate projections for the Maltese Islands demonstrates clear spatial gradients in the patterns of temperature and precipitation bioclimatic indices. Specifically, the Western District and certain areas in the Northern District consistently exhibit lower temperatures, while the Northern and South Eastern peripheries of Malta are expected to experience higher temperatures. Regarding the Western District, temperature projections consistently indicate cooler conditions (represented by BIO 1, 5, and 6) across different scenarios, namely RCP 4.5 and RCP 8.5, with a notable difference of approximately 1 °C less than other districts.

Regarding precipitation (represented by BIO 12 and 16 indices), an examination of annual precipitation and precipitation during the wettest quarter, respectively, reveals a distinct pattern. The southern regions of Malta, including the southern areas of the Western District and the South Eastern District, are projected to receive the highest levels of rainfall. In contrast, the northern regions, specifically Gozo and Comino and the Northern District, are expected to experience comparatively lower levels of precipitation. As for the BIO 17 index (projected precipitation of driest quarter (mm)), an average value of 8 mm is estimated across all timeframes and RCPs. However, a more detailed analysis reveals that the Western District is projected to have a slightly higher rainfall index (BIO 17), ranging from approximately 9 to 9.2 mm when compared to other districts, which range from 8.4 to 7 mm. This discrepancy is particularly significant when compared to Gozo and Comino and the periphery of the Northern District, as it highlights the regional perspective, indicating a slightly elevated level of precipitation in the Western District within the context of projections for the driest quarter.

3.2. Forecasted Bioclimatic Variable Trends

As shown in both

Table 7. The six bioclimatic variables during the years 2050 and 2070 for RCP 4.5 include the mean values for each bioclimatic variable and the mean difference between the years 2050 and 2070.

Bioclimatic Variable	Projected RCP 4.5; Year 2050	Standard Deviation	Projected RCP 4.5; Year 2070	Standard Deviation	Mean Projected Difference *
BIO 1: Projected Annual Mean Temperature	20.3 °C	3.0 °C	20.8 °C	3.0 °C	+0.5 °C
BIO 5: Projected Maximum Temperature of Warmest Month	33.0 °C	3.9 °C	33.5 °C	4 °C	+0.5 °C
BIO 6: Projected Minimum Temperature of Coldest Month	10.6 °C	2.9 °C	11.8 °C	2.9 °C	+1.2 °C
BIO 12: Projected Annual Precipitation	481 mm	11 mm	466 mm	11 mm	−15 mm
BIO 16: Projected Precipitation of Wettest Quarter	262 mm	5 mm	252 mm	5 mm	−10 mm
BIO 17: Projected Precipitation of Driest Quarter	8 mm	1 mm	8 mm	1 mm	0 mm

* Values are all significant, p-value < 0.0001.

and

Table 8. The six bioclimatic variables during the years 2050 and 2070 for RCP 8.5 include the mean values for each bioclimatic variable and the mean difference between the years 2050 and 2070.

Bioclimatic Variable	Projected RCP 8.5; Year 2050	Standard Deviation	Projected RCP 8.5; Year 2070	Standard Deviation	Mean Projected Difference *
BIO 1: Projected Annual Mean Temperature	20.8 °C	3.0 °C	22.0 °C	3.0 °C	+1.2 °C
BIO 5: Projected Maximum Temperature of Warmest Month	33.7 °C	3.9 °C	35.3 °C	3.8 °C	+1.6 °C
BIO 6: Projected Minimum Temperature of Coldest Month	11.3 °C	2.9 °C	12.2 °C	2.9 °C	+0.9 °C
BIO 12: Projected Annual Precipitation	490 mm	12 mm	417 mm	10 mm	−73 mm
BIO 16: Projected Precipitation of Wettest Quarter	230 mm	5 mm	257 mm	6 mm	+27 mm
BIO 17: Projected Precipitation of Driest Quarter	8 mm	1 mm	8 mm	1 mm	0 mm

* Values are all significant, p-value < 0.0001.

, there is a noteworthy and statistically significant difference in mean values between the years 2050 and 2070, as evidenced by p-values consistently below the 0.05 significance threshold across all cases. These results collectively indicate a general trend of rising temperatures and declining precipitation levels in the projected future. However, there are exceptions, particularly in the case of BIO 17, for both RCPs and BIO 16, specifically under RCP 8.5 in 2070, where there is an increase in precipitation.

A significant observation is that the impact of RCP 8.5 is considerably more pronounced, often exceeding that of RCP 4.5. For instance, in the year 2070 under RCP 8.5, there is a projected temperature increase ranging from approximately +0.7 to 1.1 °C in BIOs 1, 5, and 6, alongside a decrease of −58 mm in BIO 12, when compared to the conditions under RCP 4.5 in 2050.

It is important to notice the role of these bioclimatic variables in shaping the physiology, growth, and yield of potatoes, forageable crops, and vineyard crops in the Maltese Islands. Temperature indices such as BIO 1 (annual mean temperature) and BIO 5 (max temperature of warmest month) influence overall crop suitability, with warmer conditions accelerating growth but potentially increasing water stress. Conversely, BIO 6 (min temperature of coldest month) is relevant for winter-grown forageable crops, as cooler temperatures can hinder germination and vegetative growth [38]. Precipitation indices significantly affect water availability, with BIO 12 (annual precipitation) determining the overall moisture supply essential for all three crop types. Similarly, BIO 16 (precipitation of the wettest quarter) is important for crops like potatoes and forage in Malta as it determines the water availability during critical growth phases, influencing soil moisture levels and potential for waterlogging. BIO 17 (precipitation of driest quarter) is vital for vineyards, as it indicates periods of water scarcity that can impact vine stress, growth, and fruit quality, especially in the hot, dry months.

3.3. Results of Inter-Model Spread and Clustering

This analysis presents the results of hierarchical cluster analysis applied to projected climate data, focusing on temperature and precipitation variables under two different emission scenarios, RCP 4.5 and RCP 8.5, for the years 2050 and 2070. The analysis is performed on temperature-related BIO (BIO 1, BIO 5, BIO 6) and precipitation-related BIO (BIO 12, BIO 16, BIO 17). By grouping climate models based on their similarities, hierarchical clustering provides insights into the patterns of change in temperature and precipitation across different models, revealing potential future climate trends and variability under both emission scenarios.

The cophenetic correlation coefficient for a cluster tree, denoted as “C,” measures how accurately the tree reflects the original dissimilarities between observations. It is calculated as the linear correlation between the cophenetic distances derived from the tree and the initial distances (or dissimilarities) used to generate it [39]. For a high-quality solution, this value should be very close to one, indicating a strong correlation between the cophenetic distances of the tree and the original dissimilarities.

Table 9. Presents the climate models grouped based on the hierarchical clustering analysis, along with their respective cophenetic correlation coefficients (C) to assess clustering accuracy.

Temperature BIOs (BIO 1, BIO 5, BIO 6): 2050 and 2070
	RCP 4.5	RCP 8.5
Cluster 1 members	MR, GF (C = 0.7761)	CN, BC, CC (C = 0.6105)
Cluster 2 members	CN, AC, BC, CC (C = 0.7761)	GF, AC, MR (C = 0.6105)
Precipitation BIOs (BIO 12, BIO 16, BIO 17): 2050 and 2070
Cluster 1 members	MR, GF (C = 0.8354)	AC, CN (C = 0.6373)
Cluster 2 members	CN, AC, BC, CC (C = 0.8354)	BC, CC, GF, MR (C = 0.6373)

shows the results of hierarchical clustering for temperature (BIOs 1, 5, 6) and precipitation (BIOs 12, 16, 17) projections under RCP 4.5 and RCP 8.5 scenarios for 2050 and 2070. For temperature, under RCP 4.5 (Figure 3a), Cluster 1 (MR, GF) and Cluster 2 (CN, AC, BC, CC) show a relatively high similarity, with a cophenetic correlation of 0.7761. However, under RCP 8.5 (Figure 3b), the clusters exhibit lower similarity (C = 0.6105), indicating greater divergence in model projections. Cluster 2 is the largest and most statistically robust group, making it the primary focus of the analysis, rather than Cluster 1, which consists of fewer models and can therefore be considered an outlier group. A similar pattern is observed for precipitation, where RCP 4.5 (Figure 3c) shows strong clustering, with high cophenetic correlations (C = 0.8354) for both Cluster 1 (MR, GF) and Cluster 2 (CN, AC, BC, CC), reflecting similar precipitation patterns. Under RCP 8.5 (Figure 3d), the correlation decreases to 0.6373, suggesting more variability in precipitation projections across models.

Table 10. Projected bioclimatic variables under the RCP 4.5 scenario for the year 2050 across six climate models and corresponding clusters, including BIO 1, BIO 5, BIO 6 (°C) and BIO 12, BIO 16, BIO 17 (mm).

RCP 4.5 (Year 2050)-Model	Cluster	BIO 1 (°C)	BIO 5 (°C)	BIO 6 (°C)	Cluster	BIO 12 (mm)	BIO 16 (mm)	BIO 17 (mm)
ACCESS1-O (AC)	2	20.24	32.79	10.23	2	478	268	8
BCC-CSM1-1 (BC)	2	20.20	32.61	10.75	2	524	272	5
CCSM4 (CC)	2	19.99	32.49	10.43	2	482	260	9
CNRM-CM5 (CN)	2	19.54	32.09	10.13	2	530	286	8
GFDL-CM3 (GF)	1	21.21	34.64	11.22	1	414	225	8
MIROC-ESM (MR)	1	20.80	33.09	11.02	1	459	260	9

,

Table 11. Projected bioclimatic variables under the RCP 4.5 scenario for the year 2070 across six climate models and corresponding clusters, including BIO 1, BIO 5, BIO 6 (°C) and BIO 12, BIO 16, BIO 17 (mm).

RCP 4.5 (Year 2070)-Model	Cluster	BIO 1 (°C)	BIO 5 (°C)	BIO 6 (°C)	Cluster	BIO 12 (mm)	BIO 16 (mm)	BIO 17 (mm)
ACCESS1-O (AC)	2	20.83	33.31	11.33	2	488	286	8
BCC-CSM1-1 (BC)	2	20.47	32.89	11.02	2	539	306	7
CCSM4 (CC)	2	20.26	32.69	10.75	2	474	249	9
CNRM-CM5 (CN)	2	20.00	32.39	10.53	2	504	256	10
GFDL-CM3 (GF)	1	21.64	35.31	11.33	1	400	200	7
MIROC-ESM (MR)	1	21.5	34.2	11.52	1	389	214	8

,

Table 12. Projected bioclimatic variables under the RCP 8.5 scenario for the year 2050 across six climate models and corresponding clusters, including BIO 1, BIO 5, BIO 6 (°C) and BIO 12, BIO 16, BIO 17 (mm).

RCP 8.5 (Year 2050)-Model	Cluster	BIO 1 (°C)	BIO 5 (°C)	BIO 6 (°C)	Cluster	BIO 12 (mm)	BIO 16 (mm)	BIO 17 (mm)
ACCESS1-O (AC)	2	21.16	33.75	11.85	1	513	298	8
BCC-CSM1-1 (BC)	1	20.78	33.59	11.13	2	483	242	6
CCSM4 (CC)	1	20.49	33.22	10.85	2	485	265	9
CNRM-CM5 (CN)	1	20.07	32.22	10.73	1	497	254	9
GFDL-CM3 (GF)	2	n.a.	35.34	11.45	2	n.a.	246	8
MIROC-ESM (MR)	2	21.5	34.29	11.53	2	474	238	8

and

Table 13. Projected bioclimatic variables under the RCP 8.5 scenario for the year 2070 across six climate models and corresponding clusters, including BIO 1, BIO 5, BIO 6 (°C) and BIO 12, BIO 16, BIO 17 (mm).

RCP 8.5 (Year 2070)-Model	Cluster	BIO 1 (°C)	BIO 5 (°C)	BIO 6 (°C)	Cluster	BIO 12 (mm)	BIO 16 (mm)	BIO 17 (mm)
ACCESS1-O (AC)	2	22.14	35.49	12.53	1	465	245	8
BCC-CSM1-1 (BC)	1	21.75	34.40	12.00	2	412	221	5
CCSM4 (CC)	1	21.30	34.01	11.55	2	438	230	8
CNRM-CM5 (CN)	1	20.83	33.34	11.42	1	512	261	8
GFDL-CM3 (GF)	2	23.17	37.39	12.85	2	342	190	8
MIROC-ESM (MR)	2	22.65	35.49	12.53	2	332	173	9

present the projected bioclimatic variables for six global climate models under the two RCPs for the mid- and late-21st century (years 2050 and 2070). The projections include key temperature-related variables (BIO 1, BIO 5, BIO 6, in °C) and precipitation-related variables (BIO 12, BIO 16, BIO 17, in mm), alongside model clustering results that reflect similarities in climate response. These tables serve as the foundation for evaluating inter-model consistency, assessing the robustness of climate projections, and identifying patterns relevant to future climate adaptation and policy planning.

Under RCP 4.5 in 2050 and 2070, AC, BC, CC, and CN (Table 10) project relatively moderate temperatures and higher precipitation values. For example, BIO 1 values in 2050 for Cluster 2 range from 19.54 °C (CN) to 20.24 °C (AC), rising slightly by 2070 (Table 11) to between 20.00 °C (CN) and 20.83 °C (AC). Precipitation values (BIO 12) in this cluster remain high and relatively stable, such as BC increasing from 524 mm in 2050 to 539 mm in 2070. This stable clustering and value consistency suggest that AC, BC, CC, and CN represent more robust and reliable projections under RCP 4.5, making them preferable for climate-informed decision-making and policy development.

By contrast, MR and GF consistently form a separate Cluster 1 under RCP 4.5, indicating lower agreement with the other models. These two models exhibit a stronger warming signal and a tendency toward drier futures. For instance, in 2070, GF projects the highest temperature (BIO 1) at 21.64 °C and the lowest BIO 12 value at 400 mm, while MR closely follows with 21.50 °C and 389 mm. Such patterns of higher temperatures and reduced precipitation differentiate Cluster 1 from the more robust Cluster 2, signalling lower confidence in their projections.

Under the more extreme RCP 8.5 scenario, this divergence becomes even more pronounced. In 2050 (Table 12), models BC, CC, and CN continue to cluster together (Cluster 1 for temperature), maintaining relatively moderate warming, with BIO 1 values between 20.07 °C (CN) and 20.78 °C (BC). These models also suggest relatively stable precipitation trends, with BIO 12 values from 483 mm (BC) to 513 mm (AC). Meanwhile, MR and GF persist in their representation of warmer and drier futures. Notably, GF projects a BIO 5 (maximum temperature of the warmest month) of 35.34 °C and MR reaches 34.29 °C in 2050. By 2070 (Table 13), GF shows the highest warming trend across all models with a BIO 1 of 23.17 °C, and MR closely follows at 22.65 °C. In terms of precipitation, GF and MR again present the driest outcomes, with BIO 12 at only 342 mm and 332 mm, respectively.

AC appears somewhat transitional, clustering with the robust group under RCP 4.5 but aligning with the warmer, drier models under RCP 8.5, particularly by 2070. This suggests that the projections of AC may be scenario-sensitive, and its inclusion in robust clusters should be evaluated in the context of emission trajectories.

3.4. Visualisation of the Index Maps

Concerning agricultural land extent (Figure 4), the year 2020 recorded a total of 10,731 ha of utilised agricultural area (UAA), showing a 6.2% reduction from the 11,445 ha in 2010. Predominantly, the Western District occupied the most UAA, at 3252 ha, whilst the districts containing the least UAA were the Northern and Southern Harbour districts.

As indicated in Figure 5, the primary districts (D) expected to face the consequences of rising temperatures (BIO 1, BIO 5, BIO 6) include the South Eastern District (D2), Northern Harbour District (D4), Southern Harbour District (D3), and Northern District (D5). Additionally, those anticipated to be impacted by changes in precipitation are the South Eastern District, Western District (D1), Southern Harbour District, and Northern Harbour District (D4). This implies that the main potato-producing districts (larger than 50 hectares in size) per hectare, namely the South Eastern District, Western District, Northern District, and Southern Harbour District, will experience the effects of climate change in terms of both temperature and precipitation, which is expected to influence overall agricultural productivity. It is worth noting that Northern Harbour District (13 hectares) and Gozo and Comino (D6) (28 hectares) were not considered due to their lower production levels.

In the latest agricultural census of 2020 [33], the majority of forage cultivation was observed in the Gozo and Comino District (D6), covering an extensive area of 1615 hectares. The distribution of forage production in Malta exhibited a relatively even spread, with the Northern District (D5), Western District (D1), and South Eastern District (D2) ranking as the second most productive areas. It is important to note that forageable crops occupied the largest land area in comparison to other crop types.

Concerning forageable crop cultivation (Figure 6), the South Eastern District, Northern Harbour District (D4), Southern Harbour District (D3), and Northern District appear to be the most vulnerable to temperature (BIO 1, BIO 5, BIO 6) changes, while shifts in precipitation (BIO 12, BIO 16, BIO 17) patterns are expected to impact the South Eastern District, Western District, Southern Harbour District, and Northern Harbour District. As a result, the primary districts responsible for producing forageable crops per hectare (with areas exceeding 1000 hectares), specifically Gozo and Comino, the Western District, South Eastern District, and Northern District, are likely to experience the consequences of climate change, particularly in terms of temperature (BIO 1, BIO 5, BIO 6) and precipitation (BIO 12, BIO 16, BIO 17) variations. These changes are anticipated to have a significant effect on overall agricultural production. In contrast, the Southern Harbour District (covering 137 hectares) and Northern Harbour District (with 101 hectares), which are highlighted in white, are considered to have negligible significance due to their comparatively lower levels of production when compared to other districts.

In addition to the Western District (D1), the cultivation of vineyards is primarily concentrated in the Northern District (D5), followed by Gozo and Comino (D6), and the South Eastern District (D2) (as shown in Figure 7). The main districts housing vineyards are expected to experience the effects of temperature (BIO 1, BIO 5, BIO 6) changes, with notable impacts in the South Eastern District, Northern Harbour District (D4), Southern Harbour District (D3), and Northern District, while alterations in precipitation patterns are impacting South Eastern District, Western District, Southern Harbour District, and Northern Harbour District (D4). The primary districts responsible for vineyard production (covering more than 50 hectares) are expected to be influenced by climate change in terms of both temperature (BIO 1, BIO 5, BIO 6) and precipitation (BIO 12, BIO 16, BIO 17). The Southern Harbour District (D3) (with 2.6 hectares) and Northern Harbour District (D4) (covering 2.2 hectares) are considered negligible due to their extremely low production levels in comparison to other districts.

Based on the agricultural census conducted in 2020, the primary districts that utilise the most water for agricultural purposes include the Western (D1) and Northern (D5) districts (>5,000,000 m³), whilst the South Eastern (D2), Southern Harbour (D3), and Gozo and Comino (D6) use around 2,000,000 to 5,000,000 m³ of freshwater (Figure 8). The Northern Harbour District (D4) uses less than the average consumption per district (<2,000,000 m³). When considering the agricultural maps, this appears logical as the districts with the highest land area devoted to growing crops need to most irrigation, especially those growing potatoes (Western and South Eastern districts), and particularly vineyards (Western and Northern districts), as they rely heavily on water irrigation. This will be an issue, since results show that the projected maximum temperature of the warmest month is expected to significantly increase, especially in the South Eastern District and slightly in the Northern District. Nevertheless, the Western District seems to be affected the least in terms of temperature increases. The significance of this is that these districts will have to utilise more freshwater to irrigate their fields to keep pace with the projected temperature (BIO 1, BIO 5, BIO 6) rise and precipitation (BIO 12, BIO 16, BIO 17) decrease.

4. Discussion

4.1. Inter-Model Variability and Robustness

The projections outlined in the results section indicate that between 2050 and 2070, the Maltese Islands are expected to experience rising temperatures and a concurrent reduction in precipitation, changes that are statistically significant and likely attributable to climate change. These findings align with the conclusions of other studies, such as those conducted by [40,41,42], which suggest that by the end of the twenty-first century, a majority of climate models project further decreases in precipitation and increases in temperature in Mediterranean climate regions.

Furthermore, the rise in crop evapotranspiration resulting from these climate changes is expected to exacerbate water stress and scarcity at both local and regional levels [43].

Analysing the general trend of each bioclimatic map, it becomes evident that from 2050 to 2070, there is an expected temperature (BIO 1, BIO 5, BIO 6) increase and precipitation (BIO 12, BIO 16, BIO 17) decrease under both RCP 4.5 and RCP 8.5 scenarios. It is important to note that RCP 8.5 represents a “business as usual” scenario, implying no mitigation efforts regarding greenhouse gas emissions. Consequently, it leads to a much greater temperature increase, sometimes more than double that of RCP 4.5, while the trend is the opposite for anticipated precipitation, with some exceptions.

Notably, areas with the highest water utilisation can be observed in Figure 8, with examples including the Western and Northern Districts, which use the most water for agricultural production. Furthermore, these are the districts projected to experience reduced precipitation in 2050 and 2070. Therefore, new strategies will need to be devised to promote water harvesting and reduce water loss, particularly in these affected regions.

One notable exception is the increase in projected precipitation of the wettest quarter (BIO 16) for RCP 8.5 in 2070, showing a significant increase of +27 mm in precipitation. This can be explained by the findings of Tramblay and Somot [44], who highlight the possibility of an increase in extreme precipitation in certain Mediterranean basins, particularly under the RCP 8.5 scenario. The study suggests that this increase is due to stronger climate change signals in these regions, with projections indicating a significant rise in extreme rainfall, potentially exceeding +20% in areas like Northern Greece, the Po and Veneto basins of Italy, and parts of Slovenia and Croatia. However, the precise impact will depend on regional characteristics and factors such as land use and soil conditions, which can influence runoff and flood risks.

The hierarchical cluster analysis reveals distinct convergence and divergence patterns among the six climate models, offering insights into projection robustness. Under RCP 4.5, the strongest and most reliable cluster (Cluster 2) includes AC, BC, CC, and CN, with a high cophenetic correlation of 0.7761, indicating strong agreement in temperature projections. These models also exhibit stable precipitation trends, with BIO 12 values ranging from 478 mm (AC) to 530 mm (CN) in 2050, reinforcing their reliability. In contrast, MR and GF form a smaller, less robust cluster (Cluster 1), projecting higher temperatures and drier conditions, with GF reaching 21.64 °C and 400 mm precipitation in 2070. This division underscores the importance of prioritising larger clusters for more confident climate assessments.

Under RCP 8.5, model divergence increases significantly, leading to a lower clustering accuracy (C = 0.6105), indicating greater uncertainty in temperature projections. While Cluster 2 (BC, CC, GF, MR) remains relatively stable for precipitation (C = 0.6373), individual model discrepancies emerge, with GF projecting extreme warming (23.17 °C) and the lowest precipitation (342 mm) by 2070. The transitional behaviour of AC, shifting from the robust group in RCP 4.5 to aligning with drier, warmer models in RCP 8.5, further highlights the scenario-dependent nature of projections. Overall, the clustering results suggest that RCP 4.5 provides more consistent projections, whereas RCP 8.5 introduces greater variability, emphasising the need to focus on more robust model groupings for climate adaptation planning.

Policymakers and farm managers should carefully consider this variability when selecting models to guide decision-making regarding future climate adaptation strategies in Malta.

4.2. Characteristics and Implications of Projected Variables on Agricultural Crops

Potato cultivation in Malta is a year-round activity, predominantly in the Western and South Eastern Districts (Figure 5). However, specific areas like Rabat (Western District, D1) pose challenges due to the presence of clay soil, making it difficult for potatoes to thrive as the soil is dense and keeps temperatures cooler. Conversely, regions like Għaxaq, Naxxar, and Mosta feature loamy soil, which is ideal for potato cultivation (T. Meli, personal communication, 19 April 2023). The reason behind the promotion of potato cultivation in these areas is primarily due to the favourable soil characteristics, particularly the loamy soil, which offers optimal drainage and moisture retention for potato growth, which is highly sensitive to soil moisture conditions [45,46]. In addition, potato growers in Malta need to ensure a consistent water supply to support their crops; hence, structures such as boreholes are often employed to extract groundwater for irrigation purposes. Another critical factor in potato cultivation is water quality, particularly in the northern regions of Malta, where access to clean desalinated water can be challenging. Bustan et al. [47] argue that potatoes using saline irrigation water (ECi < 7 dS m⁻¹) can still produce reasonable yields in deep sandy soils, provided extreme weather events do not interfere. A key challenge is the interaction between salinity and heat waves, particularly during the 40 to 60 days after emergence, when potatoes are most vulnerable. If heat stress coincides with this stage, it can cause irreversible canopy damage, impairing photosynthesis and ultimately reducing tuber yields [48,49,50]. Young leaves, which usually resist salt accumulation, lose this protection during heat waves, leading to a lethal buildup of sodium.

By extrapolating this information to the local farming situation, this could potentially pose challenges for local farmers cultivating potatoes in the Western and Northern Districts of Malta, as these areas experience lower rainfall compared to the major potato-growing district, the South Eastern District. With regard to the thresholds and optimal climatic conditions (Table 1), locally cultivated potatoes will not likely to face significant growth issues during the wettest quarter, as their precipitation threshold (250 to 300 mm) will be met for RCP 4.5 in both 2050 and 2070 (ranging from 262 to 252 mm). However, RCP 8.5 in 2050 and 2070 is projected to have 230 to 262 mm during the wettest quarter, requiring external irrigation. On the other hand, the least favourable planting period for potatoes is during the driest quarter (BIO 17), as the expected precipitation will not exceed 8 mm. Overall, the annual precipitation is anticipated to fall slightly below the optimal range (500 to 700 mm) for all RCP scenarios and time periods, remaining under 500 mm.

Forage cultivation in Malta possesses a unique characteristic in that it is a seasonal crop relying solely on natural rainwater for growth, without the need for additional irrigation (T. Meli, personal communication, 19 April 2023). In contrast to wheat, which is highly reliant on rainfall, barley is a hardier and more resilient crop capable of withstanding variations in water availability. The scarcity of rainfall poses a substantial risk to wheat crops, as a lack of precipitation exceeding 70 mm can lead to plant fatalities. Fortunately, this will not be an issue for both RCP 4.5 and RCP 8.5 (2050 and 2070), as the mean precipitation for the wettest quarter ranges from 230 to 262 mm.

To optimise wheat cultivation, November is the most favourable month for planting, coinciding with the onset of the rainy season. The critical timing for planting is determined by the break of the season, marked by the reception of a minimum of 7 mm of rainfall, signalling the beginning of the rainy period and the ideal planting window. One of the principal climate challenges anticipated to impact forage production and quality in the Mediterranean region is the increased occurrence of severe and recurrent droughts. These droughts are expected to reduce productivity through decreased growth and persistence [51]. With regard to forage quality, elevated temperatures can accelerate stem elongation, resulting in a faster decline in the digestibility of both vegetative and reproductive tillers as they age, which is due to a more rapid decrease in the digestibility of cell walls [52,53].

Productivity in non-irrigated grasslands during the arid summer months in Mediterranean Europe is limited. However, changes in seasonal temperature and precipitation patterns are expected to shift productivity towards periods characterised by lower temperatures and increased rainfall [51,53]. Hence, the impact of climate change on grassland productivity is predicted to cause potential decreases in yields during the summer months in France in the future (2070 to 2099). Conversely, Bertrand et al. [54] forecast increased yields in the autumn, winter, and spring seasons due to elevated temperatures and CO₂ concentrations, resulting in a net increase in productivity.

Ultimately, the main forageable crop-producing districts, Gozo and Comino (D6), the Western District (D1), South Eastern District (D2), and Northern District (D5), may expect an increase in their forage yield, as the typical harvesting month for forage is usually in April (spring) [55]. During the summer months, forage land is left fallow to restore its fertility.

According to Fraga et al. [56], it is generally expected that Mediterranean countries will experience a significant increase in air temperatures ranging from 0.4 to 2.6 °C. This temperature increase is projected to accelerate the metabolic and developmental activities of plants, leading to an earlier initiation of spring green-up and a prolonged growing season within rangelands. However, the response to these changes is expected to vary among different species [57]. Increasing temperatures will manifest in alterations to the timing of phenological events, such as flowering and fruiting, as well as an overall extension of the growth period. Controlled studies suggest that in a tallgrass prairie, a consistent increase of 2 °C in soil temperature extended the growth period by three weeks [58]. According to our results, a similar extension of the growth period may be expected for forage as both RCP 4.5 and RCP 8.5 in the 2070 time period project a 1.2 to 1.5 °C temperature increase.

This trend towards warmer and drier conditions could have adverse effects on vineyards, altering grape yield, berry composition, and even their lifespan [59]. Our study underscores how vineyard cultivation in Malta, particularly in the Northern and Western Districts, is vulnerable to these shifts in climate. The projected annual mean temperature (BIO 1) in these areas is projected to rise from 20.3 °C in 2050 to 20.8 °C by 2070 under RCP 4.5, and from 20.8 °C to 22.0 °C under RCP 8.5, edging closer to the upper threshold of 25 °C, which is optimal for vineyards. As highlighted by Galdies and Meli [6], the warmer atmospheric temperature, particularly in December, leads to premature vine sprouting, shifting the growing season to earlier months. Hence, an increase in temperatures across the Northern (D5) and Western Districts (D1) could lead to an earlier bud break and harvest. For instance, the projected temperature for the warmest month (BIO 5) is projected to increase by 0.5 °C by 2070 under RCP 4.5 and by 1.6 °C under RCP 8.5. These changes will likely result in altered flowering and grape maturation patterns, making vineyards more susceptible to extreme weather events like strong winter winds, which could exacerbate the challenges posed by early sprouting. Moreover, the precipitation projections in this study show a concerning decline, with annual precipitation (BIO 12) projected to decrease by 15 mm by 2070 under RCP 4.5 and by 73 mm under RCP 8.5. This reduction is a cause for concern, as vineyards require annual precipitation levels between 635 mm and 890 mm to maintain optimal growth. For instance, rainfall in the Northern (D5) and Western (D1) districts is expected to decrease from 481 mm to 466 mm under RCP 4.5 and from 490 mm to 417 mm under RCP 8.5. This is in line with the study of Fraga et al. [56], who stated that water stress during critical growth stages, such as bud break to bloom, can severely impact grapevine fruit setting, berry growth, and yield. Galdies et al. [3] emphasised the importance of drip irrigation and rainwater harvesting as possible solutions to mitigate water stress, and these methods will be crucial for the Western District (D1), Northern District (D5), and Gozo and Comino (D6), all of which are most involved in vineyard cultivation.

4.3. Challenges Associated with Maltese Agriculture

A local agronomist has expressed concerns that decision-makers are not aligned with the agricultural priorities essential for farmers in the region. This disconnect has resulted in several issues plaguing the islands. One significant problem is the widespread use of illegal groundwater boreholes, particularly in the southern regions of Malta. These boreholes have exacerbated water-quality challenges, including higher salinity levels and containment of high nitrate levels derived from the leaching of nitrate-rich fertilisers in groundwater, both of which have a detrimental effect on crop water quality (T. Meli, personal communication, 19 April 2023).

Additionally, there are concerns regarding the outdated agricultural practices employed in Malta. For example, cereal planting in Malta still relies on primitive methods, such as maintaining equal distances between crops and planting wheat in close proximity to each other. This approach is problematic, especially for a crop like wheat, which is susceptible to waterlogging.

Furthermore, traditional methods like the use of cement mixers instead of modern seeding machines persist in Malta. This outdated approach often leaves seeds exposed, making them vulnerable to pests such as rats and pigeons. The overabundance of these pests, which consume unsown seeds, poses managerial challenges. Malta has yet to adopt a modern system to address these issues effectively. Our results show that between 2050 and 2070, Malta is expected to face rising temperatures and reduced precipitation, both of which will likely aggravate the challenges that crops already face under traditional farming methods. With temperatures increasing, crops will be more susceptible to stress, and in regions with high pest populations, these stresses may be intensified [60]. For instance, pests like rats and pigeons may proliferate further in warmer conditions, leading to even higher rates of seed loss. In areas like the Western (D1) and Northern (D5) districts, where water use is already high, the combined effect of increasing evapotranspiration due to rising temperatures and the challenges posed by pest infestations will place even more strain on crop yields. Moreover, climate change-induced shifts in growing seasons will mean that crops could mature earlier, potentially increasing their vulnerability to pest pressures at critical growth stages [6].

4.4. Recommendations for Climate-Proofing of the Local Agricultural Sector

Based on the results obtained from this study, a number of recommendations are being put forward which can contribute to adapting agriculture in the Maltese Islands. These include addressing the challenges posed by climate change and promoting the resilience of crops in the face of changing environmental conditions.

• Enhancing farmers’ awareness of climate change and its associated risks is fundamental to strengthening the resilience of the agricultural sector [61]. Perceptions of climate threats play a crucial role in motivating voluntary mitigation efforts; however, successful adaptation depends on access to accurate and practical information. Equipping farmers with evidence-based strategies for climate-resilient agriculture is essential to safeguarding the long-term sustainability and productivity of the Maltese agricultural sector amid shifting environmental conditions.
• The Western District, which consistently experiences cooler conditions across the scenarios (RCP 4.5 and RCP 8.5), may serve as a relatively more favourable area for agricultural activities requiring lower temperatures. In contrast, the Northern and South Eastern periphery of Malta, where temperatures are projected to be higher, will require adaptation measures such as the selection of heat-tolerant crops and modifications to urban infrastructure to mitigate heat stress. Genetically modified crops with enhanced thermotolerance and other sought-after genetic features can mitigate the adverse effects of heat stress. Furthermore, emphasising phenotypic plasticity (genetic diversity) within plant populations can further enhance resilience, ensuring yield stability despite fluctuating environmental conditions [51]. Traditional varieties like Alpha and Arran Banner [62] are commonly grown in Malta. However, these strains may not be well-adapted to increasing heat stress and changing climatic conditions when compared to other varieties. Strains such as Tetyda and Finezja would be suitable alternatives for Malta as they maintain high yields even under restricted irrigation. Furthermore, these cultivars demonstrated strong heat tolerance, with Tetyda showing minimal yield decrease and fewer tuber deformations even under heat and drought stress, while Finezja also exhibited resilience to both conditions [63]. These cultivars are readily available across Europe, particularly in countries such as Spain, Hungary, and Poland, making them accessible for local cultivation [63].
• Given the high soil temperatures that can hinder potato stolon development, farmers can adopt intercropping strategies that provide ground cover and shade. For instance, intercropping potatoes with legumes such as broad beans (Vicia faba) or local clover (Hedysarum coronarium) can reduce soil temperature by 5 to 10 °C, as suggested by Levy et al. [64], while also improving soil fertility through nitrogen fixation. Additionally, the use of low-growing cover crops, such as vetch or Mediterranean grasses, can help retain soil moisture and further support stolon proliferation, benefiting crops like potatoes [40].
• To mitigate the impact of droughts and high air temperatures, including heatwaves, it is essential to implement irrigation and misting systems in vineyards. These systems encourage evaporative cooling, resulting in lower canopy temperatures and increased photosynthetic activity. This approach is particularly valuable for ensuring grape output and quality, especially in warm and dry regions [65,66,67].
• Regarding precipitation, while southern areas, including the southern part of the Western District and the South Eastern District, are expected to receive the highest rainfall, the Northern District, along with Gozo and Comino, is projected to experience lower precipitation levels. The slightly higher projected precipitation in the Western District, particularly during the driest quarter (BIO 17), suggests potential opportunities for improved water availability in this region. However, localised water management strategies will remain critical, particularly in drier districts, to ensure sustainable agricultural and urban resilience planning. In this context, the strategy outlined by Papadimitriou et al. [68] highlights the importance of addressing practical and regional applications for Malta, particularly in the context of managing water resources for sustainable agriculture. Given the unique challenges Malta faces with water scarcity and climate variability, the strategy should emphasise region-specific solutions, such as optimising irrigation techniques and adopting innovative water management practices that suit the island’s needs. It is crucial to consider the local agricultural conditions, such as the reliance on irrigated crops, and develop tailored approaches to ensure water availability and quality for farming, especially under increasing climate pressures. Furthermore, a collaborative, multi-stakeholder approach is essential to ensure that local farmers, industry stakeholders, and policymakers work together to implement these solutions effectively while taking into account national regulatory frameworks.
• Since this study focused on analysing climate trends and their potential implications for crop production without directly quantifying projected yield variations. Future studies should incorporate climate prediction models alongside crop modelling to provide statistical estimates of potential yield changes under different climate scenarios. Utilising established methodologies, such as those used in Galdies and Vella [35], in which they modelled the crop evapotranspiration (ETo) flux using an ETo calculator created by the Food and Agriculture Organization, could offer a more quantitative assessment of how climate change may impact agricultural productivity. This would enhance the predictive value of such research and support more informed decision-making for policymakers and stakeholders in Mediterranean agriculture.
• To systematically quantify model uncertainties, we incorporate multiple climate model clusters into our analysis. For temperature, Cluster 1 models project slightly higher increases than Cluster 2, with RCP 4.5 estimates ranging from 2.4 °C–2.9 °C (2050) and 3.6 °C–4.1 °C (2070), and RCP 8.5 ranging from 4.0 °C–4.5 °C (2050) and 5.6 °C–6.1 °C (2070). Similarly, for precipitation, Cluster 1 projections indicate a 5–8% increase by 2050 and 7–12% by 2070, compared to the increases in Cluster 2 by 3–6% and 5–9%, respectively. These variations emphasise that while trends suggest potential increases in temperature and precipitation, the implications for agricultural productivity, including potato, forageable crop, and vineyard crop yield, remain uncertain due to variability in climatic projections. While the core analysis remains based on the more statistically robust clusters, the inclusion of smaller clusters highlights the range of possible climate outcomes. This reinforces confidence in our conclusions by demonstrating that projected warming and precipitation changes remain significant across multiple model groupings. Given the divergence and convergence among models, it is crucial for decision-makers to incorporate this inter-model variability into their planning processes. Understanding these variations allows for more resilient and adaptive agricultural strategies that account for uncertainty in climate projections.

4.5. Limitations

The primary limitation of this research study is technical, primarily related to the use of RCP scenarios from CMIP5. Despite advancements in CMIP5 models, there are inherent uncertainties associated with their utilisation and the specific RCPs chosen.

• The accuracy of future prediction models can be influenced by systematic biases inherent in the models themselves. For instance, some models, such as GF, exhibit substantial variations across different emission scenarios (RCP 4.5 vs. RCP 8.5), leading to significantly diverging projections, particularly for the projected annual precipitation (BIO 12) and the projected minimum temperature of the coldest month (BIO 6). This variability in model output can introduce uncertainty in the interpretation of climate projections, especially when considering longer-term forecasts like 2070. Additionally, the clustering results, which show more consistent projections under RCP 4.5 but greater divergence under RCP 8.5, suggest that the reliability of projections may depend heavily on the emission scenario chosen. In this context, the greater dispersion under RCP 8.5 may reflect the increasing complexity and uncertainties of climate models when accounting for higher greenhouse gas emissions.
• The inherent simplifications in climate models, such as the assumptions about future socio-economic and technological developments, greenhouse gas emission pathways, and regional climate dynamics, may not fully capture the complexities of future climate systems. While hierarchical clustering provides a useful way to group similar model projections, it does not account for the possibility that models might all exhibit biases in different directions [69]. Hence, the projections must always be interpreted with hindsight, especially for regions or variables with higher model divergence. Future refinements in model development and emission scenario representations may improve confidence in future projections, but for now, understanding the sources of model variability is crucial for assessing the robustness of climate predictions.
• The reliance on regional and global climate models is considered a study limitation, as these may not fully capture the microclimatic variations across the diverse agricultural zones of Malta. Although a high-resolution downscaled CMIP5 model at 1 km was utilised, the most advanced option currently available, the spatial resolution of global climate models remains a challenge for small island states such as Malta. Due to the limited land area and complex microclimatic conditions, a country-specific climate model would be preferable for improving localised projections. However, no such tailored model exists at present. Once they become available, future research should focus on employing finer-scale climate models, specifically designed for small island states such as Malta, to enhance the accuracy of climate impact assessments.

5. Conclusions

This study provides a comprehensive analysis of climate projections for Malta, highlighting significant challenges for the agricultural sector, particularly regarding temperature increases and decreasing precipitation frequency. It demonstrates clear spatial gradients in the patterns of temperature and precipitation bioclimatic indices. Specifically, the Western District and certain areas in the Northern District consistently exhibit lower temperatures, while the Northern and South Eastern periphery of Malta are expected to experience higher temperatures. Focusing on the Western District, temperature projections consistently indicate cooler conditions across different scenarios, namely RCP 4.5 and RCP 8.5, with a notable difference of approximately 1 °C less than other districts.

Regarding precipitation, an examination of annual precipitation and precipitation during the wettest quarter, respectively, reveals a distinct pattern. The southern regions of Malta, including the southern areas of the Western District and the South Eastern District, are projected to receive the highest levels of rainfall. In contrast, the northern regions, specifically Gozo and Comino and the Northern District, are expected to experience comparatively lower levels of precipitation. As for the BIO 17 index, an average value of 8 mm is estimated across all timeframes and RCPs. However, a more detailed analysis reveals that the Western District is projected to have a slightly higher rainfall index (BIO 17), ranging from approximately 9 to 9.2 mm when compared to other districts, which range from 8.4 to 7 mm. This discrepancy is particularly significant when compared to Gozo and Comino and the periphery of the Northern District, as it highlights the regional perspective, indicating a slightly elevated level of precipitation in the Western District within the context of projections for the driest quarter.

These findings indicate that agricultural systems in Malta will need to adapt to a future that is characterised by more arid conditions and increased heat stress. These regional changes underscore the importance of developing context-specific strategies, as the unique climate conditions of Malta are not fully captured by broader Mediterranean models.

Based on the study findings, several tailored recommendations are proposed to enhance the resilience of the agricultural sector in Malta. Firstly, increasing farmers’ awareness of climate risks and equipping them with evidence-based adaptation strategies is crucial. The Western District, which experiences cooler conditions, may serve as a more suitable area for certain crops, whereas hotter regions such as the Northern and South Eastern periphery will require heat-tolerant crop varieties. Cultivars such as Tetyda and Finezja have demonstrated strong resilience under high temperatures and water stress and should be considered for local adaptation trials. Additionally, intercropping techniques, such as growing potatoes with legumes, can mitigate soil temperature increases and improve fertility. Efficient water management, including optimised irrigation strategies and misting systems for vineyards, is essential to counteract drought and extreme heat. Furthermore, future research should integrate climate prediction models with crop yield modelling to provide more precise projections and inform agricultural planning.

Despite these insights, the study acknowledges several limitations. The reliance on CMIP5 climate projections introduces uncertainties, particularly concerning emission scenarios and regional climate dynamics. While these models provide valuable insights, future studies should consider higher-resolution models tailored for small island states such as Malta. Additionally, the absence of integrated crop modelling means the direct impact of climate change on specific crop yields was not quantified. These limitations highlight the need for further research to refine projections and develop more precise adaptation strategies.

In conclusion, while this study identifies key challenges and adaptation strategies for the Maltese agricultural sector under climate change, further advancements in climate modelling and agricultural research are necessary to enhance resilience planning and ensure sustainable food production in the region.