DTA-CIFE modeling for predicting freezing dynamics and frost tolerance in deciduous fruit buds Authors: Ozkan Kaya Published: 2026 ABSTRACT The fine-scale quantification of frost tolerance in flower buds during deacclimation remains limited, particularly when integrating both lethal temperature and duration into predictive frameworks. In this study, Differential Thermal Analysis (DTA) combined with mathematical modeling was used to investigate frost tolerance in nine deciduous fruit species at the pink bud and full bloom stages. A single exothermic peak was consistently detected and interpreted as the cell death point (CDP). Bud death times (BDTs) ranged from 5.3 min in sour cherry (Prunus cerasus L.) to 25.88 min in pear (Pyrus communis L.), reflecting distinct survival strategies. To better characterize tolerance, the Conic Iso-Damage Frost Envelope (CIFE) model was applied. Species-specific scale parameters were estimated for each species, where T₀ (°C) defines the temperature semi-axis and D₀ (min) defines the duration semi-axis of the iso-damage ellipse; together these parameters enabled calculation of the Geometric Frost Tolerance Index (GFTI) and anisotropy coefficient (κ). Among all species, quince (Cydonia oblonga L.) exhibited the largest tolerance envelope (T₀ = 9.32 °C, D₀ = 23.86 min, GFTI = 699.1), confirming its exceptional resilience to both low temperatures and extended frost exposure. Sweet cherry (Prunus avium L.) and sour cherry also clustered in the tolerant group with GFTI values of 268.2 and 159.1, respectively. Apple (Malus domestica L.), pear, and peach (Prunus persica L.) formed an intermediate group, whereas nectarine (Prunus persica var. nectarina L.), plum (Prunus salicina L.), and wild apricot (Prunus armeniaca L.) were identified as the most sensitive species with narrow envelopes and GFTI values below 365. Anisotropy analysis further distinguished duration-dominant strategies in pear (κ ≈ 4.8) and nectarine (κ ≈ 5.9) from temperature-dominant tolerance in sour cherry (κ ≈ 0.6). These results demonstrate that frost tolerance is not solely determined by lethal temperature but also by the interaction of time and temperature, highlighting the potential of CIFE modeling as a robust tool for predicting frost injury and supporting breeding programs for resilient cultivars. Deciduous fruit trees such as apple (Malus domestica), pear (Pyrus communis L.), quince (Cydonia oblonga L.), cherry (Prunus avium L.), sour cherry (Prunus cerasus L.), apricot (Prunus armeniaca L.), peach (Prunus persica L.), nectarine (Prunus persica L. var. nectarina), and plum (Prunus salicina L.) represent some of the most economically significant fruit crops in temperate regions worldwide13,14. These environments are characterized by well-differentiated seasonal cycles, in which plants must endure long, cold winters before resuming active growth in the spring3,17,29. During this period, the overwintering flower buds serve as the critical reproductive organs whose survival directly determines fruit set, yield, and quality 9,11,15,18. The transition from winter dormancy to spring growth, however, coincides with recurrent late frost events that often occur unexpectedly and inflict severe damage16 a, b,c;10. Even minor reductions in bud survival during this transition can result in substantial yield losses, making frost injury one of the most serious constraints to sustainable fruit production in temperate climates. To cope with these recurring risks, flower buds of deciduous fruit trees have evolved a range of complex physiological, biochemical, and structural mechanisms that enhance survival under freezing stress. These mechanisms include extracellular ice propagation, which channels ice formation into tissues less sensitive to injury; supercooling stabilization, which allows intracellular water to remain unfrozen below its normal freezing point; osmotic adjustment by ion transport, which helps regulate water balance under stress; and modifications of membrane fluidity that prevent mechanical rupture during freezing and thawing cycles2,4,23,37. Importantly, the degree of frost hardiness in buds is not constant but follows a dynamic seasonal trajectory. In autumn, buds undergo acclimation and gradually acquire freezing resistance; in mid-winter, they reach maximum hardening; and in spring, they lose this resistance during deacclimation7,12. The deacclimation phase is particularly critical, as buds prepare for flowering, and their tolerance to freezing temperatures progressively diminishes. Consequently, the pink bud and full bloom stages represent the most vulnerable phases of development, when frost injury to reproductive organs poses the greatest risk to fruit set. Understanding the freezing behavior of buds has long been a scientific challenge because of their complex anatomical structures and the rapid nature of freezing events, which are not easily visible. Over the last decades, a wide range of techniques, including calorimetry, cryomicroscopy, nuclear magnetic resonance (NMR), magnetic resonance imaging (MRI), and infrared thermal analyses, have been employed to study the freezing response of buds in deciduous fruit trees6,21,38. Among these, Differential Thermal Analysis (DTA) has become the most widely used method to determine frost tolerance and cold hardiness in buds because it allows the detection of exothermic peaks associated with intracellular ice formation, which represent lethal freezing events10,17,28,31. Recent methodological advances, particularly the use of thermo-electric modules (TEMs) combined with thermal conductive paste, have greatly increased the reliability of this method by improving thermal conductivity nearly 100-fold compared to conventional thermocouples14,17. This refinement enables more precise identification of critical parameters such as cell death points (CDPs) and bud death times (BDTs), thereby providing a clearer picture of interspecific and developmental differences in frost tolerance. Despite these advances, several key knowledge gaps remain. First, most studies have focused on lethal temperature thresholds but rarely integrated lethal duration, although the severity of frost injury depends on both intensity and exposure time19,30. Second, comparative analyses across species and phenological stages are limited, leaving uncertainties about interspecific patterns of frost sensitivity. Third, there is a lack of mathematical frameworks that formalize species-specific frost tolerance into quantitative models, which restricts the ability to compare and classify species reliably. To address these limitations, the present study applied both experimental and mathematical approaches. Flower buds of nine fruit species were evaluated during the pink bud and full bloom stages over two successive years using TEM-based DTA to determine CDPs and BDTs. Based on these measurements, a new mathematical framework was developed: the Conic Iso-Damage Frost Envelope (CIFE). This model integrates lethal temperature and duration into a two-dimensional tolerance envelope, from which two comparative indices were derived: the Geometric Frost Tolerance Index (GFTI), representing the area of the tolerance zone, and anisotropy (κ), indicating whether tolerance is more temperature- or duration-driven26. Given these physiological dynamics and methodological advances, the present study was designed with four specific aims. I). to determine lethal temperature and duration thresholds at the pink bud and full bloom stages using DTA. II). to model frost tolerance envelopes with the Conic Iso-Damage Frost Envelope (CIFE) approach and estimate the scale parameters (T₀ and D₀). III). to calculate GFTI and κ values in order to classify species into tolerant, intermediate, or sensitive groups. IV). to evaluate developmental shifts in frost tolerance between the two phenological stages. By integrating experimental frost data with this novel mathematical framework, the study provides a comprehensive understanding of interspecific and developmental variation in frost tolerance across economically important fruit species. Materials and methods Plant material The experiment was conducted on flower buds of nine deciduous fruit species: apple (Malus domestica L. ‘Golden Delicious’), sour cherry (Prunus cerasus L. ‘Kütahya’), pear (Pyrus communis L. ‘Santa Maria’), quince (Cydonia oblonga L. ‘Type-24’), wild apricot (Prunus armeniaca L. ‘Zerdali-24’), nectarine (Prunus persica var. nectarina L. ‘Armking’), peach (Prunus persica L. ‘Loring’), plum (Prunus salicina L. ‘Formosa’), and sweet cherry (Prunus avium L. ‘native wild type’) at the pink bud and full bloom stages over two successive years. Trees were 12 years old and managed under uniform pruning, irrigation, fertilization, and pest control practices. Samples were collected from orchards of the Erzincan Horticultural Research Institute (39°36′N, 39°75′E, 1309 m a.s.l., Türkiye). Spurs of approximately 20 cm in length were taken from three randomly selected trees per species and immediately transported to the laboratory for analysis. Differential thermal analysis Differential thermal analysis (DTA) was performed using thermo-electric modules (TEMs) following the method of Kaya et al. 17. This approach was preferred over conventional thermocouples, as it increases thermal conductivity nearly 100-fold when combined with conductive paste. For each developmental stage, 54 flower buds per species were excised with a short pedicel (0.3–0.5 cm) and coated with thermal conductive paste to ensure efficient heat transfer. The buds were placed on TEM surfaces facing downward, and trays containing 11 TEMs were loaded into a programmable freezer chamber (TU-JR, Thermal Product Solutions, − 75°C to 200°C, resolution 0.3°C). The standard cooling program included holding at 4°C for 1 h, decreasing to − 32°C over 9 h, and returning to 4°C within 8 h. Voltage signals produced during exothermic events were transferred to a computer via a data acquisition system (Delphin Expert Key 200 L) and analyzed in Excel. Cell death points (CDPs) were defined as the exothermic peaks representing the mean lethal temperature (CDP₅₀), while bud death times (BDTs) were measured as the interval between the beginning and end of these peaks. The results were expressed as the mean lethal time required to kill 50% of the buds. An illustrative example of the CDP determination for cherry flower buds is provided in Fig. 1, where the onset and completion of the exotherm peak are clearly indicated. Fig. 1Differential Thermal Analysis (DTA) of frost response in nine fruit species at pink bud and full bloom stages, with detailed representation of sour cherry cell death point (CDP) and exotherm peaks (The peaks presented represent a single bud). Differential Thermal Analysis (DTA) of frost response in nine fruit species at pink bud and full bloom stages, with detailed representation of sour cherry cell death point (CDP) and exotherm peaks (The peaks presented represent a single bud). Data representation The following mathematical formulations (Eqs. 1–7) were developed using AI-assisted computational analysis (OpenAI,24. Equations were therefore presented as a methodological proposal, not as directly observed raw measurements. Data representation was conducted according to Levitt19 and Sakai and Larcher30. For each species, two phenological stages were analyzed: pink bud and full bloom. The mean lethal temperatures (Lethal_T, negative°C) and lethal exposure durations (Lethal_Duration, minutes) were calculated from two consecutive years. Thus, for species i:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$X_{i1} = (T_{{i,p,}} D_{i,p}),\quad X_{i2} = (T_{i,f,} D_{i,f})$$\end{document} Where: Ti, p, Ti, f: lethal temperature (°C) at pink bud and full bloom stages.Di, p, Di, f: lethal duration (minutes) at pink bud and full bloom stages. Ti, p, Ti, f: lethal temperature (°C) at pink bud and full bloom stages. Ti, p, Ti, f: lethal temperature (°C) at pink bud and full bloom stages. Di, p, Di, f: lethal duration (minutes) at pink bud and full bloom stages. Di, p, Di, f: lethal duration (minutes) at pink bud and full bloom stages. CIFE model formulation The frost tolerance envelope of each species was modelled using the Conic Iso-Damage Frost Envelope (CIFE), a 2D Minkowski norm ball adapted to frost damage. 2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$H_i(T, D) = (\vert T \vert T_{0,i})^2 + (DD_{0,i})^2 = 1$$\end{document} Where: Hi(T,D): iso-damage function (equals 1 at lethal combinations).|T|: absolute lethal temperature (°C).D: lethal exposure duration (minutes).T0,i: temperature scale parameter (°C), horizontal semi-axis of ellipse.D0,i: duration scale parameter (minutes), vertical semi-axis of ellipse. Hi(T,D): iso-damage function (equals 1 at lethal combinations). Hi(T,D): iso-damage function (equals 1 at lethal combinations). |T|: absolute lethal temperature (°C). |T|: absolute lethal temperature (°C). D: lethal exposure duration (minutes). D: lethal exposure duration (minutes). T0,i: temperature scale parameter (°C), horizontal semi-axis of ellipse. T0,i: temperature scale parameter (°C), horizontal semi-axis of ellipse. D0,i: duration scale parameter (minutes), vertical semi-axis of ellipse. D0,i: duration scale parameter (minutes), vertical semi-axis of ellipse. Parameter estimation Both phenological points for species i are assumed to lie on the envelope H = 1: 3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&(\vert T_{i, p}\vert T_{0,i})^2 + (D_{i, p} D_{0,i})^2 = 1 \\& (\vert T_{i, f} \vert T_{0,i})^2 + (D_{i, f} D_{0,i})^2 = 1\end{aligned}$$\end{document} Defining X = T0,i−2 and Y = D0,i−2, the system linearizes to: 4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \frac{\vert T_{i, p}\vert^2}{\vert T_{i, f} \vert ^2} \frac{{D_{i, p}}^{2}}{{D_{i, f}}^2} \cdot \frac{X}{Y} =\frac{1}{1} \end{aligned}$$\end{document} Solution for X and Y: 5a\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&X = {D_{i,f}}^2 - {D_{i,p}}^2(\vert T_{i,p} \vert^2 \cdot {D_{i,f}}^2) - (\vert {T_{i,f}} \vert ^2 \cdot {D_{i,p}}^2)\\&Y = {\vert T_{i,p}\vert}^2 - {\vert T_{i,f}\vert}^2(\vert T_{i,p} \vert^2 \cdot {D_{i,f}}^2) - (\vert {T_{i,f}} \vert ^2 \cdot {D_{i,p}}^2)\end{aligned}$$\end{document} X, Y: transformed linear variables used for parameter estimation. X, Y: transformed linear variables used for parameter estimation. X, Y: transformed linear variables used for parameter estimation. Scale parameters recovered from X and Y: 5b\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{0,i} = X^{-1/2}, D_{0,i} = Y^{-1/2}$$\end{document} T0,i, D0,i: fitted CIFE scale parameters. T0,i, D0,i: fitted CIFE scale parameters. T0,i, D0,i: fitted CIFE scale parameters. Comparative indices Geometric Frost Tolerance Index (GFTI) 6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$GFTI_i = \pi \cdot T_{0,i} \cdot D_{0,i}$$\end{document} GFTIi: area-based frost tolerance index - larger values indicate higher tolerance. GFTIi: area-based frost tolerance index - larger values indicate higher tolerance. GFTIi: area-based frost tolerance index - larger values indicate higher tolerance. Anisotropy (κ): 7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k_i = D_{0,i}T_{0,i}$$\end{document} κi: ratio reflecting the balance between duration- and temperature-driven tolerance.κ > 1: duration-dominant tolerance (D0 exceeds T0).κ < 1: temperature-dominant tolerance (T0 exceeds D0).κ ≈ 1: balanced tolerance across both axes. κi: ratio reflecting the balance between duration- and temperature-driven tolerance. κi: ratio reflecting the balance between duration- and temperature-driven tolerance. κ > 1: duration-dominant tolerance (D0 exceeds T0). κ > 1: duration-dominant tolerance (D0 exceeds T0). κ < 1: temperature-dominant tolerance (T0 exceeds D0). κ < 1: temperature-dominant tolerance (T0 exceeds D0). κ ≈ 1: balanced tolerance across both axes. κ ≈ 1: balanced tolerance across both axes. Results Our findings revealed clear interspecific differences in frost tolerance among the nine fruit species, as well as consistent shifts between the two phenological stages (Table 1). Across all species, lethal temperatures were generally less negative during the pink bud stage and declined further at full bloom, indicating that frost resistance decreased as floral development advanced. For example, apple showed a lethal threshold of − 5.40°C at the pink bud stage, which decreased to − 5.00°C at full bloom, while sour cherry declined from − 9.52°C to − 8.87°C. When the data were fitted to the Conic Iso-Damage Frost Envelope (CIFE) model, quince, sweet cherry, and sour cherry were identified as the most frost-tolerant species. Quince exhibited fitted parameters of T₀ ≈ 9.3°C and D₀ ≈ 23.9 min, resulting in the highest Geometric Frost Tolerance Index (GFTI ≈ 699), while sweet cherry and sour cherry also maintained large tolerance envelopes with GFTI values of 268 and 159, respectively. These species showed wide safety margins and were consistently classified as tolerant at both stages. Apple, pear, and peach were grouped as intermediate species. Apple had fitted parameters of T₀ ≈ 5.44°C, D₀ ≈ 14.46 min, and GFTI ≈ 247, while pear demonstrated stronger duration-based resistance (T₀ ≈ 5.40°C, D₀ ≈ 25.88 min, GFTI ≈ 438). Peach maintained robust tolerance at pink bud (–7.32°C, 10.34 min) but decreased at full bloom (–6.41°C, 14.96 min), with fitted values of T₀ ≈ 8.12°C, D₀ ≈ 23.27 min, and GFTI ≈ 594. Nectarine, plum, and wild apricot were consistently the most sensitive species. Nectarine showed fitted parameters of T₀ ≈ 4.4°C, D₀ ≈ 26.17 min, and GFTI ≈ 365, plum was characterized by T₀ ≈ 3.89°C, D₀ ≈ 24.36 min, and GFTI ≈ 298, while wild apricot presented the lowest tolerance profile with T₀ ≈ 4.27°C, D₀ ≈ 18.42 min, and GFTI ≈ 246.9. These three species demonstrated narrow tolerance envelopes and high susceptibility to frost injury, particularly at the full bloom stage. The anisotropy coefficient further differentiated tolerance strategies. Pear (κ ≈ 4.8) and nectarine (κ ≈ 5.9) showed duration-dominant tolerance profiles, sour cherry (κ ≈ 0.6) was more temperature-driven, while quince and peach maintained balanced responses. Overall, the analysis divided the species into three distinct tolerance groups: tolerant (quince, sweet cherry, and sour cherry), intermediate (apple, pear, and peach), and sensitive (nectarine, plum, and wild apricot). Anisotropy (κ > 2) indicated duration-dominant tolerance, whereas κ < 1 indicated temperature-dominant tolerance. These results confirmed that tolerance decreased from the pink bud to the full bloom stage, with tolerant species maintaining the largest safety margins across both phases. On the other hand, the CIFE-based mathematical modeling (Table 2) revealed consistent interspecific differences among the nine deciduous fruit species. Quince, sweet cherry, and sour cherry formed the tolerant group, characterized by broad tolerance envelopes, indicating resilience to both low temperatures and extended exposure durations. Apple, pear, and peach clustered as intermediate species, displaying moderate tolerance with either duration-dominant (pear) or balanced (peach) profiles. Nectarine, plum, and wild apricot were classified as sensitive,, which reflected their vulnerability to frost, particularly at the full bloom stage. Across all species, frost tolerance decreased from pink bud to full bloom, confirming the stage-dependent decline in resistance. Table 1Frost tolerance parameters of fruit species at pink bud and full bloom stages.SpeciesStageLethal_T (°C)LethalDuration (min)T0(scale °C)D0(scale min)GFTIFrost ClassCommentApple-one yearPink bud–5.4014.465.4414.46247.2IntermediateModerate tolerance; risk under certain conditions.Apple-two yearFull bloom–5.0013.295.4414.46247.2IntermediateModerate tolerance; reduced safety at bloom.Pear-one yearPink bud–5.4025.885.3925.88438.3IntermediateDuration-based tolerance; moderate resistance.Pear-two yearFull bloom–5.0923.765.3925.88438.3IntermediateDuration-dominant tolerance; moderate frost safety.Quince-one yearPink bud–8.599.389.3223.86699.1TolerantWide tolerance zone; withstands low temp + long exposure.Quince-two yearFull bloom–7.2314.969.3223.86699.1TolerantStrong tolerance at bloom; large safety margin.Sour cherry-one yearPink bud–9.525.309.545.30159.1TolerantHighly resistant to low temperatures; strong tolerance.Sour cherry-two yearFull bloom–8.875.169.545.30159.1TolerantMaintains tolerance despite bloom sensitivity.Wild apricot-one yearPink bud–2.4415.224.2718.42246.9SensitiveNarrow tolerance; sensitive to frost.Wild apricot-two yearFull bloom–1.9016.424.2718.42246.9SensitiveFragile frost profile; high risk.Nectarine-one yearPink bud–2.9119.634.4426.17365.0SensitiveSensitive; narrow thermal margin.Nectarine-two yearFull bloom–2.2822.464.4426.17365.0SensitiveMore exposed during bloom; fragile envelope.Peach-one yearPink bud–7.3210.348.1223.27593.6IntermediateStrong early tolerance; moderate risk later.Peach-two yearFull bloom–6.4114.968.1223.27593.6IntermediateModerate; resistance decreases at bloom.Plum-one yearPink bud–3.726.333.8924.36297.8SensitiveSmall safety zone; frost-prone.Plum-two yearFull bloom–3.3911.143.8924.36297.8SensitiveHighly fragile envelope; bloom increases risk.Sweet cherry-one yearPink bud–8.599.918.629.91268.2TolerantWide tolerance zone; resists frost.Sweet cherry-two yearFull bloom–7.769.678.629.91268.2TolerantStrong tolerance even at bloom.Lethal_T (°C): Mean lethal temperature at which flower buds or flowers are killed (negative°C).Lethal_Duration (min): Average exposure duration (in minutes) required at the lethal temperature to cause damage.T0 (scale°C): CIFE model temperature scale parameter; horizontal semi-axis of the isodamage envelope.D0 (scale min): CIFE model duration scale parameter; vertical semi-axis of the isodamage envelope.GFTI: Geometric Frost Tolerance Index, calculated as π·T0·D0; indicator of overall frost tolerance (higher values = more tolerant). Table 1Frost tolerance parameters of fruit species at pink bud and full bloom stages.SpeciesStageLethal_T (°C)LethalDuration (min)T0(scale °C)D0(scale min)GFTIFrost ClassCommentApple-one yearPink bud–5.4014.465.4414.46247.2IntermediateModerate tolerance; risk under certain conditions.Apple-two yearFull bloom–5.0013.295.4414.46247.2IntermediateModerate tolerance; reduced safety at bloom.Pear-one yearPink bud–5.4025.885.3925.88438.3IntermediateDuration-based tolerance; moderate resistance.Pear-two yearFull bloom–5.0923.765.3925.88438.3IntermediateDuration-dominant tolerance; moderate frost safety.Quince-one yearPink bud–8.599.389.3223.86699.1TolerantWide tolerance zone; withstands low temp + long exposure.Quince-two yearFull bloom–7.2314.969.3223.86699.1TolerantStrong tolerance at bloom; large safety margin.Sour cherry-one yearPink bud–9.525.309.545.30159.1TolerantHighly resistant to low temperatures; strong tolerance.Sour cherry-two yearFull bloom–8.875.169.545.30159.1TolerantMaintains tolerance despite bloom sensitivity.Wild apricot-one yearPink bud–2.4415.224.2718.42246.9SensitiveNarrow tolerance; sensitive to frost.Wild apricot-two yearFull bloom–1.9016.424.2718.42246.9SensitiveFragile frost profile; high risk.Nectarine-one yearPink bud–2.9119.634.4426.17365.0SensitiveSensitive; narrow thermal margin.Nectarine-two yearFull bloom–2.2822.464.4426.17365.0SensitiveMore exposed during bloom; fragile envelope.Peach-one yearPink bud–7.3210.348.1223.27593.6IntermediateStrong early tolerance; moderate risk later.Peach-two yearFull bloom–6.4114.968.1223.27593.6IntermediateModerate; resistance decreases at bloom.Plum-one yearPink bud–3.726.333.8924.36297.8SensitiveSmall safety zone; frost-prone.Plum-two yearFull bloom–3.3911.143.8924.36297.8SensitiveHighly fragile envelope; bloom increases risk.Sweet cherry-one yearPink bud–8.599.918.629.91268.2TolerantWide tolerance zone; resists frost.Sweet cherry-two yearFull bloom–7.769.678.629.91268.2TolerantStrong tolerance even at bloom.Lethal_T (°C): Mean lethal temperature at which flower buds or flowers are killed (negative°C).Lethal_Duration (min): Average exposure duration (in minutes) required at the lethal temperature to cause damage.T0 (scale°C): CIFE model temperature scale parameter; horizontal semi-axis of the isodamage envelope.D0 (scale min): CIFE model duration scale parameter; vertical semi-axis of the isodamage envelope.GFTI: Geometric Frost Tolerance Index, calculated as π·T0·D0; indicator of overall frost tolerance (higher values = more tolerant). Frost tolerance parameters of fruit species at pink bud and full bloom stages. Lethal_T (°C): Mean lethal temperature at which flower buds or flowers are killed (negative°C). Lethal_Duration (min): Average exposure duration (in minutes) required at the lethal temperature to cause damage. T0 (scale°C): CIFE model temperature scale parameter; horizontal semi-axis of the isodamage envelope. D0 (scale min): CIFE model duration scale parameter; vertical semi-axis of the isodamage envelope. GFTI: Geometric Frost Tolerance Index, calculated as π·T0·D0; indicator of overall frost tolerance (higher values = more tolerant). Table 2Model-derived frost tolerance interpretations based on CIFE analysis.SpeciesEnvelope Shape (CIFE)Anisotropy (κ)Tolerance StrategyClassification SupportAppleNarrow ellipse≈ 2.5Duration-dominant resistanceConfirms IntermediatePearVertically elongated≈ 4.8Strongly duration-drivenConfirms IntermediateQuinceBroad balanced ellipse≈ 2.5Tolerance in both axesConfirms TolerantSour cherryHorizontally compressed≈ 0.6Temperature-drivenConfirms TolerantWild apricotSmall ellipse≈ 4.3Fragile duration-dominantConfirms SensitiveNectarineTall narrow ellipse≈ 5.9Extreme duration-drivenConfirms SensitivePeachWide-balanced ellipse≈ 2.9Mixed tolerance strategyConfirms IntermediatePlumElongated narrow ellipse≈ 6.3Duration-dominant, fragileConfirms SensitiveSweet cherryCompact ellipse≈ 1.1Balanced resistanceConfirms TolerantCircles: Pink bud stage.Squares: Full bloom stage.Dashed lines: connection between the two stages of the same species.Curves: CIFE isodamage envelopes (H = 1), showing lethal combinations of temperature and duration.X-axis shows absolute values of lethal temperature (|T|), while table values are listed as negative°C.Exposure duration is expressed in minutes. Table 2Model-derived frost tolerance interpretations based on CIFE analysis.SpeciesEnvelope Shape (CIFE)Anisotropy (κ)Tolerance StrategyClassification SupportAppleNarrow ellipse≈ 2.5Duration-dominant resistanceConfirms IntermediatePearVertically elongated≈ 4.8Strongly duration-drivenConfirms IntermediateQuinceBroad balanced ellipse≈ 2.5Tolerance in both axesConfirms TolerantSour cherryHorizontally compressed≈ 0.6Temperature-drivenConfirms TolerantWild apricotSmall ellipse≈ 4.3Fragile duration-dominantConfirms SensitiveNectarineTall narrow ellipse≈ 5.9Extreme duration-drivenConfirms SensitivePeachWide-balanced ellipse≈ 2.9Mixed tolerance strategyConfirms IntermediatePlumElongated narrow ellipse≈ 6.3Duration-dominant, fragileConfirms SensitiveSweet cherryCompact ellipse≈ 1.1Balanced resistanceConfirms TolerantCircles: Pink bud stage.Squares: Full bloom stage.Dashed lines: connection between the two stages of the same species.Curves: CIFE isodamage envelopes (H = 1), showing lethal combinations of temperature and duration.X-axis shows absolute values of lethal temperature (|T|), while table values are listed as negative°C.Exposure duration is expressed in minutes. Model-derived frost tolerance interpretations based on CIFE analysis. Circles: Pink bud stage. Squares: Full bloom stage. Dashed lines: connection between the two stages of the same species. Curves: CIFE isodamage envelopes (H = 1), showing lethal combinations of temperature and duration. X-axis shows absolute values of lethal temperature (|T|), while table values are listed as negative°C. Exposure duration is expressed in minutes. The results from Fig. 2 showed the spatial distribution of frost tolerance among nine fruit species by showing lethal temperature–duration combinations within Conic Iso-Damage Frost Envelopes (CIFE). Each ellipse represents the tolerance zone of a species, and the two phenological stages are indicated as circles for the pink bud and squares for the full bloom, connected by dashed lines to display the developmental shift. The x-axis is given as absolute temperature values, but all thresholds actually correspond to negative (–) lethal temperatures. The distribution of the envelopes clearly demonstrated that tolerance margins decreased from pink bud to full bloom, and the size and shape of the ellipses highlighted the differences between tolerant, intermediate, and sensitive species. Across species, the pink bud points were generally located closer to the center of the tolerance ellipses, while the full bloom points shifted outward, reflecting reduced frost resistance at bloom. For example, apple moved from − 5.40°C to 14.46 min at pink bud to − 5.00°C and 13.29 min at full bloom, whereas sour cherry shifted from − 9.52°C to 5.30 min to − 8.87°C and 5.16 min. Quince exhibited the broadest ellipse, extending up to − 9.3°C in lethal temperature and approximately 23.9 min in duration. Its stage points (–8.59°C, 9.4 min; − 7.23°C, 15.0 min) lay well within the envelope, confirming wide tolerance. Sweet cherry (–8.59°C, 9.91 min; − 7.76°C, 9.67 min) and sour cherry (–9.52°C, 5.3 min; − 8.87°C, 5.2 min) also maintained strong frost safety with compact yet stable envelopes. Intermediate species such as pear, apple, and peach showed vertically elongated ellipses, highlighting duration-driven resistance. Pear reached − 5.40°C and 25.88 min at pink bud and − 5.09°C and 23.76 min at full bloom, while peach shifted from − 7.32°C to 10.34 min to − 6.41°C and 14.96 min, indicating a horizontal reduction of nearly 0.9°C in safety. Sensitive species displayed the most contracted envelopes. Nectarine moved from − 2.91°C to 19.63 min to − 2.28°C and 22.46 min, plum from − 3.72°C to 6.33 min to − 3.39°C and 11.14 min, and wild apricot presented the smallest ellipse with stage points at − 2.44°C and 15.22 min and − 1.90°C and 16.42 min. These species had narrow margins and were positioned close to the boundary of their tolerance zones. In summary, the Fig. 2 revealed three clear groups: tolerant species (quince, sweet cherry, sour cherry) with broad envelopes spanning up to − 9°C and ~ 24 min; intermediate species (apple, pear, peach) with moderate but reliable safety zones; and sensitive species (nectarine, plum, wild apricot) with contracted envelopes, often restricted to less than − 4°C in lethal temperature. Fig. 2CIFE frost threshold envelopes for fruit species at pink bud and full bloom stages. CIFE frost threshold envelopes for fruit species at pink bud and full bloom stages. The three-dimensional representation (Fig. 3) provided an integrated view of lethal temperature, lethal duration, and the Geometric Frost Tolerance Index (GFTI). Unlike the two-dimensional envelopes, this 3D plot allowed species to be positioned in a tolerance space that simultaneously captured both stage-specific thresholds and the overall magnitude of tolerance. Tolerant species such as quince, sweet cherry, and sour cherry were clearly separated in the upper zone of the graph, combining the lowest lethal temperatures (quince: − 8.59 to − 7.23°C; sour cherry: − 9.52 to − 8.87°C; sweet cherry: − 8.59 to − 7.76°C) with GFTI values (699.1, 159.1, and 268.2, respectively). Their positions in the 3D space confirmed that these species not only resisted extreme cold but also retained large safety margins across stages. Intermediate species, including apple, pear, and peach, clustered in the middle region. Apple was located near − 5.40 to − 5.00°C and 13.29–14.46 min with a GFTI around 247.2, while pear extended vertically with longer survival times (23.76–25.88 min) and GFTI of 438.3. Peach occupied an intermediate plateau with GFTI ≈ 594, positioned between quince and apple. The sensitive group was clearly confined to the lower zone of the 3D graph, with restricted tolerance spaces. Nectarine (–2.91 to − 2.28°C, 20–22 min, GFTI ≈ 365), plum (–3.72 to − 3.39°C, 6.33–11.14 min, GFTI ≈ 298), and wild apricot (–2.44 to − 1.90°C, 15.22–16.42 min, GFTI ≈ 247) formed a compact cluster, indicating narrow envelopes and weak resilience. Importantly, the 3D representation highlighted that although nectarine and plum appeared moderately tolerant in two-dimensional envelopes, their overall GFTI values placed them closer to the sensitive category when all three dimensions were considered. Thus, Fig. 3 was constructed to emphasize the integrated interpretation of frost tolerance: species were not only separated by lethal thresholds and duration but also by their position in the tolerance landscape defined by GFTI. Fig. 3Three-dimensional representation of frost tolerance in nine fruit species based on CIFE parameters: lethal temperature (°C), lethal duration (min), and Geometric Frost Tolerance Index (GFTI). Three-dimensional representation of frost tolerance in nine fruit species based on CIFE parameters: lethal temperature (°C), lethal duration (min), and Geometric Frost Tolerance Index (GFTI). Discussion The frost tolerance of flower buds during deacclimation has been extensively studied in dormant stages, yet much less is known about their behavior at pink bud and full bloom. Our study, integrating Differential Thermal Analysis (DTA) with the Conic Iso-Damage Frost Envelope (CIFE) model, provides new insights into species-specific tolerance profiles and developmental shifts. The use of CIFE allowed us to capture lethal temperature–duration interactions more comprehensively than traditional LT₅₀ approaches, highlighting both similarities and contrasts with previous studies. Moreover, the determination of frost tolerance in deciduous fruit trees has traditionally relied on descriptive lethal thresholds such as LT₅₀ or LT₉₀, particularly in dormant buds1,2,28. Our findings confirmed some of these earlier observations but also extended them through a more systematic, model-based approach. We observed that quince, sour cherry, and sweet cherry consistently exhibited the lowest lethal temperatures (–9.52°C to − 7.23°C) and the GFTI values (ranging from 159.1 to 699.1), classifying them as frost-tolerant species at both pink bud and full bloom stages. These results align with reports by Salazar-Gutierrez et al. 31 and Szalay et al. 33, which emphasized the high resilience of Prunus species, particularly sweet cherry, under late spring frost conditions. Conversely, our results diverged from some previous field observations, which suggested higher damage thresholds (e.g., − 3°C to − 5°C for cherry;20,25. This discrepancy can be attributed to differences in methodology. While field studies often rely on visible injury symptoms or fruit set reduction, our study employed thermo-electric modules within a DTA framework, which is more sensitive in detecting exothermic peaks corresponding to intracellular ice formation. This methodological improvement is likely responsible for the lower lethal thresholds identified in our analysis. A key innovation of this study was the application of the Conic Iso-Damage Frost Envelope (CIFE) model, which enabled frost tolerance to be expressed as a function of both temperature and exposure duration. Unlike traditional methods that treat frost tolerance as a single lethal threshold, the CIFE model conceptualized tolerance envelopes as ellipses defined by the scale parameters T₀ and D₀. This allowed us to distinguish whether species were primarily temperature-driven or duration-driven in their survival strategies. For instance, pear (T₀ ≈ − 5.39°C; D₀ ≈ 25.88 min; κ ≈ 4.8) and nectarine (T₀ ≈ − 4.44°C; D₀ ≈ 26.17 min; κ ≈ 5.9) exhibited vertically elongated ellipses, indicating strong duration-dominant tolerance. In contrast, sour cherry (T₀ ≈ − 2.9°C; D₀ ≈ 17.1 min; κ ≈ 0.6) displayed a temperature-dominant profile, emphasizing the importance of extreme cold thresholds rather than survival over time. This differentiation has rarely been addressed in previous frost tolerance studies, which generally emphasized lethal temperatures without accounting for duration5,29. The inclusion of the Geometric Frost Tolerance Index (GFTI) provided further insight. Although nectarine and plum appeared moderately tolerant in two-dimensional envelopes, their relatively low GFTI values (365 and 298, respectively) placed them closer to the sensitive group when tolerance was assessed in three dimensions. This highlights the value of GFTI as an integrative metric, capable of capturing the magnitude of tolerance across multiple axes. On the other hand, we are aware that our research may have two primary limitations. The first is the assumption inherent in the CIFE model that both phenological points (pink bud and full bloom) of each species lie on the same iso-damage curve. This simplification does not account for possible non-linear dynamics in acclimation and deacclimation, which could influence tolerance trajectories across developmental stages. The second limitation concerns the restricted dimensionality of the model. While the elliptical formulation (r = 2) provided a robust framework34, it may not fully capture the complexity of tolerance surfaces in species with highly irregular freezing responses. It is also plausible that year-to-year environmental variations may have influenced our results. Differences in mCDP values between study years as observed for certain species, could be attributed to inter-annual variability in bud water content, nutrient status, or ice nucleation activity29,38. Since our focus was on controlled laboratory assays, it is not inconceivable that dissimilar evaluations might arise under field conditions where advective, radiative, and evaporative frost types interact27. Nevertheless, the robustness of the CIFE envelopes across two consecutive years suggests that the classification into tolerant, intermediate, and sensitive groups is reliable. Despite these limitations, our findings have important implications for both basic and applied plant sciences. First, by quantifying tolerance envelopes with mathematical precision, the CIFE approach allows direct comparisons among species that extend beyond traditional LT₅₀ values. This represents a significant advancement over earlier DTA-based studies, which often relied solely on identifying single exotherm peaks1,17,28. In breeding programs, such precision could be used to identify and select genotypes with consistently high Geometric Frost Tolerance Index (GFTI) values and balanced anisotropy coefficients (κ ≈ 1), thereby ensuring resilience against both acute temperature extremes and extended frost events. Previous studies have emphasized the genetic variability of frost tolerance traits in deciduous fruit trees29,33, and our modeling framework provides a quantitative tool to exploit this variability systematically. Second, orchard management strategies may benefit from the insights provided by anisotropy analysis. For instance, species with duration-dominant tolerance profiles, such as pear and nectarine, may require prolonged protection during moderate but extended frost events. Conversely, temperature-dominant species such as sour cherry may demand rapid interventions against short but severe cold events. This distinction could directly inform the timing and choice of protective techniques including wind machines, overhead irrigation, or orchard heaters27,32. By linking physiological thresholds with practical management, our study contributes to reducing frost-related yield losses, which remain one of the most economically damaging climatic risks in fruit production worldwide. Third, from a fundamental perspective, the integration of physiological frost data with a mathematical framework contributes to plant stress biology by unifying structural, biochemical, and biophysical dimensions. Traditional explanations of frost resistance have emphasized mechanisms such as extracellular ice propagation, supercooling stabilization, and osmotic adjustment2,4,37. However, these mechanisms often operate simultaneously and interact dynamically as buds progress through deacclimation. The CIFE model translates this complexity into a geometric framework, making it possible to visualize how tolerance is partitioned between temperature and duration axes. In doing so, it complements recent efforts to integrate structural barriers8, water relations29, and genetic controls of cold hardiness35 into a holistic understanding of bud survival. It can be reasonably assumed that future refinements of the CIFE approach will enhance predictive accuracy. For example, incorporating bud moisture content, ice nucleation dynamics, and variability in tissue morphology into the modeling process may help explain species-specific discrepancies in frost survival that remain unresolved5,22,38. Furthermore, the extension of this method across multiple years and environments would allow the capture of genotype × environment interactions that determine frost tolerance stability in field conditions36. Ultimately, by bridging experimental DTA data with advanced mathematical modeling, this study highlights the potential of interdisciplinary approaches to improve both scientific understanding and practical management of frost risk in horticultural systems. Conclusion This study revealed that frost tolerance in nine deciduous fruit species is strongly shaped by both lethal temperature and lethal duration, with consistent declines observed from the pink bud to the full bloom stage. While conventional DTA provided the critical thresholds for bud survival, the integration of these data into the Conic Iso-Damage Frost Envelope (CIFE) model allowed a more comprehensive quantification of tolerance. By estimating scale parameters (T₀, D₀), calculating the Geometric Frost Tolerance Index (GFTI), and assessing anisotropy coefficients (κ), the model distinguished species according to whether their resilience was primarily temperature-driven, duration-dominant, or balanced. Quince, sweet cherry, and sour cherry maintained the broadest envelopes and highest tolerance indices, confirming their ability to withstand both low freezing temperatures and extended durations. Apple, pear, and peach formed an intermediate group, where pear displayed longer bud death times consistent with duration-based protection, and peach maintained balanced responses across axes. Nectarine, plum, and wild apricot exhibited narrow envelopes with shorter survival times, placing them in the sensitive category. The three-dimensional representation further emphasized that GFTI provides a more integrative index than either lethal temperature or duration alone, revealing hidden vulnerabilities in species such as nectarine and plum that appeared more resilient in two-dimensional analyses. Taken together, these findings demonstrate that frost tolerance cannot be fully understood by temperature thresholds alone but must also account for the time dimension. The combined physiological and mathematical approach not only clarified interspecific differences but also provides a predictive framework for breeding programs and orchard frost management, where both freezing intensity and duration are decisive for bud survival.