• Energy Research
• 2025-2025close
• GB close
• ES close
• IT close
• English close

[This dataset contains all data used for Studies 2 (qualitative), 3 (quantitative survey) and 4 (longitudinal) in my PhD research.]<br>Thesis abstract:This thesis explores the potential positive impact of artificial intelligence (AI) technology on sustainability in and outside of the tourism industry through four studies. Study 1 introduced the AI4GoodTourism framework, emphasising the need for sustainability inclusion</em> and tourist involvement</em> to achieve a successful sustainability transition. Five themes were identified through a systematic review: intelligent automation to enhance tourist experience, preserve heritage, promote quality of life, measure tourist experience, and preserve the environment. The latter theme was the least explored scholarly topic. Study 2 conceptualised a conversational AI chatbot to promote pro-environmental behaviour spillover among tourists visiting the Gili Islands, Indonesia. A theoretical model was proposed, highlighting factors influencing chatbot usage and spillover effects. Study 3 identified relationships between factors from Study 2, revealing that factors such as performance expectancy, timing, </em>and credibility</em> significantly influenced people’s intention to use the proposed chatbot technology. A significant relationship was established between people’s intentions to use the chatbot and environmentally friendly transport. Scenario-based experiments showed that using the chatbot with educational information on sustainability was sufficient to trigger behaviour change. Study 4 explored the underlying mechanism of pro-environmental behaviour spillover through human-chatbot interactions using flashback nudging. A longitudinal experiment involving the Gili tourists demonstrated that flashback nudging delivered through chatbot technology strengthened their environmental self-identity, leading to significant differences in self-reported pro-environmental behaviour between treatment and control groups. In conclusion, the thesis demonstrates that AI technology, designed with high sustainability inclusion, can positively impact sustainability through tourists’ marginal contributions. The proposed AI4GoodTourism framework and the conceptualised chatbot technology, especially with flashback nudging, show potential for facilitating pro-environmental behaviour spillovers among tourists. All four studies in this thesis highlight the importance of prioritising sustainability in AI innovations for the tourism industry, offering insights for future AI development and adoption to support the global sustainability agenda.

1H and 13C nuclear magnetic resonance spectra were recorded on a Bruker Avance III 400 or Magritek Spinsolve 60 spectrometer at 293 K. Chemical shifts are reported as δ in parts per million (ppm) and referenced to the chemical shift of the residual solvent resonances (CDCl3: 1H: δ = 7.26 ppm, 13C: δ = 77.16 ppm). Polymer molecular weight and dispersity were determined using a Malvern Viscotek GPCmax size exclusion chromatograph instrument fitted with a Viscotek TDA 305 detector unit equipped with refractive index and light scattering detectors. Samples were dissolved in tetrahydrofuran at a concentration of approximately 1 mg mL-1 and eluted through a guard column and two Agilent PLGel 5 µm mixed C columns (300 x 7.5 mm) at a flow rate of 1 ml.min-1; the elution pathlength was heated to 30 °C for the duration. Molecular weights were calibrated against known poly(methyl acrylate) standards. Differential scanning calorimetry was conducted using a TA Instruments Discovery 2500. Samples were analysed in non-hermetic aluminium pans and compared against an empty reference pan of the same type. Loaded sample masses were between 3 and 10 mg. Samples were subjected to two complete heat/cool cycles from -50 °C to 150 °C (-85 °C to 150 °C for lower Tg samples) and both heating and cooling rates were set at 10 °C min-1. UV/Vis transmittance and absorption spectra were measured with a PerkinElmer Lambda 750 spectrophotometer. Transmittance spectra of films were measured using wavelength scan with a resolution of 1 nm at a scan speed of 267 nm/min and a slit width of 2 nm. Samples were directly mounted to the sample holder. Solution spectroscopy was carried out on solutions in THF in quartz SUPRASIL® cuvettes (10 mm pathlength). Absorption spectra of luminophore solutions were taken using a wavelength scan with a resolution of 0.5 nm at a scan speed of 141.20 nm/min and a slit width of 2 nm. A reference sample of THF in an identical cuvette was used to apply a 100% transmission correction. Steady-state PL spectroscopy was performed on a Fluorolog-3 spectrophotometer (Horiba Jobin Yvon). Solid-state emission spectra were recorded using the front-face configuration. Solution emission spectra were recorded using the right-angle configuration, over 10 averaged scans. The excitation and emission slits were adjusted so that the maximum PL intensity was within the range of linear response of the detector and were kept the same between samples if direct comparison between the emission intensity was required. Emission and excitation spectra were corrected for the wavelength response of the system and the intensity of the lamp profile over the excitation range, respectively, using correction factors supplied by the manufacturer. Photoluminescence quantum yields (ΦPL) were measured using a Quanta-phi integrating sphere (Horiba Jobin Yvon) mounted on the Fluorolog-3 spectrophotometer. The UC emission and phosphorescence spectra, threshold intensity (I_th), UC quantum yield (UC) and lifetime measurements were performed using an FLS1000 time-correlated single photon counting (TCSPC) spectrometer (Edinburgh Instruments Ltd.). The samples were excited with a 532 nm laser (MGL-III-532, 200mW). To determine I_th, the laser power was adjusted using a Thorlabs PM100A Power Meter Console combined with a S120VC Si photodiode power sensor (range: 200-1100 nm) before the measurement, across the 5 to 8000 mW cm-2. The ΦUC was measured with an integrating sphere (SNS125 5-inch sphere, three windows, International Light Technologies). The sample was placed at the center of the sphere using a sample holder. A baffle is placed in front of the observation window, which blocks any scattering and reflection of the laser from the sample surface. The angle of the sample holder is adjustable. The normal direction of the sample holder is 22.5˚ to the excitation beam line, which leads the reflection of the laser to the inner surface of the sphere. The laser power was measured with a photodiode before each ΦUC measurement. Both the emission of the sample (380-500 nm) and scattering of the laser beam (530-534 nm) were measured. A neutral density filter (O.D.=3.0) was placed before the excitation beam for the scattering intensity measurements. Six data sets were collected to calculate the ΦUC of each sample: 1. sample in the path of the beam – “in fluorescence”; 2. sample in scattering; 3. sample facing away from beam – “out of fluorescence”, 4. sample out of scattering; 5. empty sphere fluorescence; 6. empty sphere scattering. Fluorescence decay measurements were performed using the multi-channel scaling (MCS) method on a the FLS1000 TCSPC spectrometer. The emission decay was recorded using a photomultiplier tube (PMT-980) equipped with TCC2 counting electronics. For the upconversion lifetime measurements, a wavelength of 440 nm was selected, and a short-pass filter (cut-off at 500 nm, Thorlabs) was placed in front of the detector. For the phosphorescence lifetimes, a wavelength of 660 nm was selected, and a long-pass filter (cut-off 550 nm, Thorlabs) was used. The instrument response function (IRF) was measured using Ludox® colloidal silica solution (a SiO2 particle suspension solution) and using a neutral density filter (O.D.=3) to attenuate the laser intensity. The pulse repetition rate was adjusted to ensure the full decay was detected within the time window. Data-fitting was carried out by tail fitting to each emission decay trace using a multiexponential decay function within the FAST software package (Edinburgh Instruments Ltd.). The goodness of fit was evaluated using the reduced chi-square statistics (χ2) and the randomness of the residuals. Please also see the readme file for more details on data collection and file organisation.

Perovskite-based material La₀.₂Sr₀.₂₅Ca₀.₄₅MₓTi₁₋ₓO₃₋δ (M=Fe, Co, Ni) was synthesised successfully using optimised Pechini method. SOFC system was fabricated using the synthesised material as the anode, YSZ as the electrolyte, and LSM as the cathode. Initial evaluations were conducted with hydrogen as the fuel. Electrochemical switching in humidified hydrogen is found to significantly improve the performance of the cells. Hydrogen SOFC with La₀.₂Sr₀.₂₅Ca₀.₄₅Co₀.₀₂₅Fe₀.₀₂₅Ti₀.₉₅O₃₋δ anode is found to show low polarisation resistance (3.23 Ω/cm²) and high maximum power density (227 mW/cm²). HDCFCs were setup with the fuel mixture containing eutectic K₂CO₃/Li₂CO₃ and activated charcoal. When purged with N₂, performance of HDCFCs was found inferior to that of hydrogen SOFCs. Better performance was observed with CO₂ as the purging gas. For example, in CO₂, HDCFC with La₀.₂Sr₀.₂₅Ca₀.₄₅Co₀.₀₂₅Fe₀.₀₂₅Ti₀.₉₅O₃₋δ anode showed a polarization resistance of 2.31 Ω/cm² and power density of 99.6 mW/cm². To explore the waste-to-energy applications of HDCFCs, medium density fibreboard was pyrolysed, forming a biochar with high oxygen content. Using this biochar as the fuel in the HDCFCs, unexpected OCV loss was observed for high operation temperatures when purged with N₂. With the limitation in testing temperature, worse performance was observed. CO₂ purging maintained OCV at higher temperatures, with the La₀.₂Sr₀.₂₅Ca₀.₄₅Co₀.₀₂₅Fe₀.₀₂₅Ti₀.₉₅O₃₋δ anode showing a polarisation resistance of 9.79 Ω/cm² and power density of 11.6 mW/cm². Abnormal shape in the I-V curve was observed for HDCFCs and some of the SOFCs tested. In attempt to understand this observation, EIS at different applied voltages were obtained, various conditions were applied for the I-V scans, carbonate concentrations in the fuel mixture were tested and different voltages were applied to the HDCFCs while the off-gases were analysed using gas chromatography. Redox behaviour of anode materials under different applied voltages and the change in CO and carbonate concentrations are suggested to be related to the abnormal curve.

This paper presents the findings of an indepth qualitative study of the most important forest logging companies and syndicates to explore the factors which influence forest exploitation and related businesses in the Congo Basin of Africa to act or not in environmentally sustainable ways. More specifically, the study explored the motivations, the benefits and the factors which facilitate or constrain sustainable behaviour amongst forest exploitation companies in Cameroon. Data analysis was undertaken using a holistic model consisting of institutional, economic and resource based factors. Economic motivations were the most cited factors driven by increased awareness and demands from clients. Interestingly, the most cited benefit from adopting environmentally sustainable behaviour related to gains in internal organisation, transparency and productivity within the company. The regulatory institutional environment was the most cited constraint because of illegality, weak law enforcement and corruption in the country’s forest sector followed by high costs of investment and unclear financial premiums from environmentally sourced timber. The policy implications are discussed.

# databases used for statistical analyses in manuscript WLB-2024-01345 [https://doi.org/10.5061/dryad.tht76hf7v](https://doi.org/10.5061/dryad.tht76hf7v) ## Description of the data and file structure **List of excel files used for GLMMs and other analyses in manuscript WLB-2024-01345.R1 – “Precipitation and female experience are major determinants of the breeding performance of Canarian houbara bustards”** ### Files and variables #### File: GLM2b2NestInitiatDate.xlsx **Description:** ** database for GLMM Nest Initiation Date (NIDF, see Supplementary Table S3)** ##### Variables * definitions as in other excel files #### File: GLM4aNestAttemptSuccess.xlsx **Description:** **database for GLMM Nest Attempt Success (see Supplementary Table S3)** ##### Variables * clutchOrder: order of the clutch (1 first, 2 second –replacement-, 3 third –replacement-clutch), chicksSurvived: chicks survived until productivity control (1= yes/0= no, see Methods), MeanTemp: during 23 days incubation + 2 months in nestings that have chicks on the control date, AvMaxTemp: average maximum temperature during 23 days incubation + 2 months in nestings that have chicks on the control date, AvMinTemp: average minimum temperature during 23 days incubation + 2 months in nestings that have chicks on the control date, pp: precipitation during 23 days incubation + 2 months in nestings that have chicks on the control date, other variables as defined in other excel files #### File: GLM4bFledSuccess.xlsx **Description:** **database for GLMM Fledging Success (see Supplementary Table S3)** ##### Variables * Variables:** **pp: precipitation during** **23 days since nesting start + 2 months in nestings that have chicks on the control date; 23+1 month, in nestings that do not have chicks on the control date, other variables as defined in other excel files #### File: GLM5ReClutchProb.xlsx **Description:** **database for GLMM Re-clutching Probability (see Supplementary Table S3)** ##### Variables * Variables: Reclutch: 1= has a replacement clutch/0= does not have a replacement clutch, DurationIncubation: duration of the incubation period (days), MeanTemp, AvMaxTemp, AvMinTemp, pp: measured over the incubation period, other variables as defined in other excel files #### File: Weighted\_precipitations.xlsx **Description:** **databases to calculate weighted precipitation amounts, periods of precipitation and nestings (see Methods for details)** ##### Variables * definitions as in other excel files #### File: GLM6a3FemaleProductivity.xlsx **Description:** **database for GLMM Productivity (see Supplementary Table S3)** ##### Variables * nClutches: number of clutches (1,2,3), NchicksSurvived (1,2 up to fledging), pp2: precipitation measured from one month before the first laying to the laying date of the last clutch, other variables as defined in other excel files #### File: GLM6bFemaleProductivity.xlsx **Description:** **database for GLMM Productivity (as GLM6a, but measuring precipitation over the same period for all years: from 1 September to 13 March [mean hatching start date of the latest year, which was 2022]; see Supplementary Table S3)** ##### Variables * as in GLM6a3FemaleProductivity.xlsx, but measuring precipitation over the same period for all years: from 1 September to 13 March [mean hatching start date of the latest year, which was 2022; other variables as defined in other excel files #### File: GLM7aLengthBreedSeason.xlsx **Description:** **database for GLMM Length of the Breeding Season (see Supplementary Table S3)** ##### Variables * daysBreeding: duration of the breeding season in days (see definition in Methods), temperature and precipitation (PP) measured from 1 month before the first day of incubation of that year in any female until the date of independence of the last chick (see Azar et al 2018: Total rainfall during the nesting period (the period between the first and last nest found each year). other variables as defined in other excel files #### File: GLM7bLengthBreedSeason.xlsx **Description:** **database for GLMM Length of the Breeding Season, same as GLM7aLengthBreedSeason.xlsx, but precipitation and temperature measured over an equal period for all years: from 1 September to 13 March (= average hatching starting date of the latest year, 2022) (see Supplementary Table S3)** ##### Variables * as in GLM7aLengthBreedSeason.xlsx, but precipitation and temperature measured over an equal period for all years: from 1 September to 13 March #### File: WeightedPrecipitationPeriods.xlsx **Description:** **database to calculate weighted precipitation periods and nestings (see Methods for details)** ##### Variables * as in other excel files #### File: GLM2cNestInitiatDate.xlsx **Description:** **database for GLMM Nest Initiation Date of First Clutches (NIDF2, see Supplementary Table S3)** ##### Variables * definitions as in other excel files #### File: GLM1bNestingRate.xlsx **Description:** **database for GLMM Nesting Rate (see Supplementary Table S3)** ##### Variables * indiv= individual female, year, femaleNests: 1=Yes/0=No, startNest = date when nesting started, pp30days: precipitation on the 30 days before (in mm), pp60ays: precipitation on the 60 days before (in mm), pp90days: precipitation on the 90 days before (in mm), TempMean30days: mean temperature on the 30 days before (in oC), TempMax30days: maximum temperature on the 30 days before (in oC), TempMin30days: minimum temperature on the 30 days before (in oC), TempMean60days: mean temperature on the 60 days before (in oC), TempMax60days: maximum temperature on the 60 days before (in oC), TempMean60days: mean temperature on the 60 days before (in oC), Weight: weight of the female (g), PC1p: Principal Component 1 of the PCA including weight, PC1sinP: Principal Component 1 of the PCA excluding weight, Breedingexperience: breeding experience of the female, as defined in Methods. #### File: GLM1cNestingRate2.xlsx **Description:** ##### Variables * indiv= individual female, year, femaleNests: 1=Yes/0=No, startNest = date when nesting started, pp30days: precipitation on the 30 days before (in mm), pp60ays: precipitation on the 60 days before (in mm), pp90days: precipitation on the 90 days before (in mm), TempMean30days: mean temperature on the 30 days before (in oC), TempMax30days: maximum temperature on the 30 days before (in oC), TempMin30days: minimum temperature on the 30 days before (in oC), TempMean60days: mean temperature on the 60 days before (in oC), TempMax60days: maximum temperature on the 60 days before (in oC), TempMean60days: mean temperature on the 60 days before (in oC), Weight: weight of the female (g), PC1p: Principal Component 1 of the PCA including weight, PC1sinP: Principal Component 1 of the PCA excluding weight, Breedingexperience: breeding experience of the female, as defined in Methods, pp_1sep_13mar: precipitation measured between 1st September and 13th March, T_1sep_13mar: temperature measured between 1st September and 13th March #### File: GLM2a2NestInitiatDate.xlsx **Description:** **database for GLMM Nest Initiation Date (NIDF, see Supplementary Table S3)** ##### Variables * ordinalDate: Ordinal date as defined in Methods, rest of variables: definitions as in other excel files #### File: GLM3HatchSuccess.xlsx **Description:** **database for GLMM Hatching Success (see Supplementary Table S3)** ##### Variables * : endNest: date when incubation finished, ppIncub: precipitation during incubation (23 days since start incubation), AvMaxTempIncub: average maximum temperature during incubation, AvMaxTempIncub: average maximum temperature during incubation, ppIncub: mean temperature during incubation, hatchSuccess: 1= incubation until hatching date is successful/ 0= incubation until hatching date is not successful, rest of variables: definitions as in other excel files ## Code/software data can be viewed using EXCEL; other files from the process of statistical analysis were obtained using package “lme4” (Bates et al. 2015) in R v.2.15.1 (R Development Core Team, 2015) Precipitation is one of the main triggers of reproduction in desert-breeding birds. The unpredictability of rainfall patterns in arid environments has led species to adapt their breeding effort to episodes of abundant food after rainfall. The response is not the same for all individuals in a population, and may vary especially with the age and experience of each female. Here we investigate the effects of precipitation, temperature, body size and breeding experience, among other variables, on reproductive parameters of 20 females of Canarian houbara bustard (Chlamydotis undulata fuertaventurae), an endangered desert bird endemic of the eastern Canary Islands. Precipitation and breeding experience were the main determinants of female breeding performance. Higher rainfall determined an increase in nesting rate, and earlier autumn rains caused an advancement of nesting to October, allowing the breeding season to be extended to eight months. This favoured an extraordinary increase in productivity in more rainy breeding seasons, with 15 times more females nesting in the two most rainy winters than in dry years. In addition, females with more breeding experience showed a higher tendency to breed, higher nest attempt and fledging success, and longer breeding season, which allowed them to rear more chicks. A female even double brooded successfully in the same season, which, considering that chicks remain with the mother for up to six months, indicates a great capacity to optimise reproductive investment, by adapting to highly variable rainfall regimes. In recent decades, the eastern Canary Islands have undergone a process of aridification, and climate models predict a medium-term increase in the frequency and duration of drought periods. Thus, Canarian houbaras are particularly vulnerable to climate change, so measures are urgently needed to reduce their mortality and improve the quality of their habitat, in order to favour their reproduction and prevent their extinction.  We used 5-year breeding phenology and breeding success data from 20 female houbara bustards captured in Lanzarote and equipped with backpack-mounted GSM/GPRS data loggers. The influence of predictor variables on breeding parameters was modeled by means of generalized linear mixed models (GLMMs) using package “lme4” (Bates et al. 2015) in R v.2.15.1 (R Development Core Team, 2015).

Concrete is the world’s most used construction material. There are significant challenges with the decarbonisation of concrete, particularly cement, due to the release of carbon dioxide emissions during clinker production. Therefore, strategies to reduce embodied carbon in concrete buildings should aim to focus on material efficiency efforts first. Existing relationships between concrete and embodied carbon, such as increased concrete strength, increasing global warming potential (GWP), increased span length increasing GWP, and taller buildings generally having higher carbon have been investigated. However, some relationships are yet to be explored. These include the relationship of form and associated structural layouts to embodied carbon. One of the challenges of conducting embodied carbon assessments is the selection of material embodied carbon coefficients (ECCs). This is particularly true during early-stage design when the specific material selection is still unknown. A common early-stage technique for identifying embodied carbon reduction strategies is identifying carbon hotspots. However, these hotspots can be heavily influenced by the large variation in environmental product declarations (EPD) across manufacturers and differing product specifications. Without quantifying this uncertainty due to material ECC uncertainty, selecting the most likely lowest-impact design option is challenging. The most common approach for uncertainty propagation is to use Monte Carlo (MC) simulations. This propagation provides results as a distribution instead of a single, deterministic result. It is simple to compare single-value results to demonstrate the difference between design options and select the lowest-impact alternative. However, using uncertainty statistical methods to compare structural frame designs against each other and also against industry benchmarks is yet to be investigated. Therefore, comparative statistical methods are evaluated for their suitability in early-stage decision-making and, specifically, structural frame design comparisons. This thesis introduces a newly-proposed methodology that propagates uncertainties in material quantities and ECCs during early-design comparisons. The methodology incorporates a novel ranking step to identify key contributing materials, streamlining assessments by reducing time and focusing on material hotspots. A new uncertainty characterisation of construction materials is introduced, utilising statistical parameters from an industry material ECC database. The thesis later integrates quantity uncertainty by design stage with material ECC uncertainty for early-stage structural EC assessments, capturing incompleteness and variation due to quantity take-off methods and early-stage estimations. Additionally, the thesis introduces a new parametric tool for early-stage concrete frame designs. This tool incorporates form definition (for seven-shaped buildings), followed by a layout derivation for all possible solutions within a span range. Next, a C# script within the tool conducts structural RC design for multiple slab types, and finally, product-stage embodied carbon calculations with uncertainty are presented. An investigation into the influence of architectural form on structural frame layouts and resulting embodied carbon was conducted, considering seven equally-sized shaped forms for two plot sizes. In combining the uncertainty procedure and parametric form design tool, comparative statistical methods for evaluating uncertain results are tested for the first time in a building EC context. Lastly, the thesis concludes by proposing a novel application for comparing structural EC results against the SCORS rating system. By including uncertainty in early-stage building EC assessments, this thesis enables engineers to conduct more reliable and fair comparative EC studies. Relying on average material ECCs (with uncertainty) through early design stages also benefits practitioners by prioritising EC savings through demand reduction and material efficiency without relying on low-carbon products.