How temperature regimes near the equinox synchronize spring biological events

Technical Summary: How Temperature Regimes Near the Equinox Synchronize Spring Biological Events

Authors: Jonathan Auerbach, Andrew Gelman, and E. M. Wolkovich
Date: April 26, 2026

---

1. Problem Statement

The timing of seasonal biological events (phenology), such as plant flowering, is widely believed to be governed by the thermal-sum model: an event is triggered once cumulative temperatures reach a specific threshold ($\tau$).

While climate change is clearly advancing the timing of spring, recent observations have presented a paradox:

1. Declining Sensitivity: The rate at which spring advances appears to be slowing down.
2. Declining Synchrony: The variance in the timing of these events is increasing in some regions (a "scattered spring"), making biological events less coordinated.

Existing literature has proposed numerous complex biological mechanisms (e.g., photoperiod, chilling requirements, or epigenetic factors) to explain these trends. This paper asks a fundamental question: Are these observations actually emergent properties of the basic thermal-sum model itself, or is the model fundamentally insufficient?

2. Methodology

The authors move away from complex biological assumptions and instead apply renewal theory—specifically the mathematics of stopped random walks—to the existing thermal-sum framework.

Mathematical Framework:
They model daily temperature $X_i$ as a combination of a deterministic trend and stochastic noise:
$$X_i = \alpha + \beta i + \epsilon_i$$
Where:

• $\alpha$: The baseline average daily temperature.
• $\beta$: The rate of temperature increase (warming trend).
• $\epsilon_i$: Independent noise with mean 0 and variance $\sigma^2$ (representing microclimate or idiosyncratic variation).

The authors identify two distinct temperature regimes based on the value of $\beta$:

1. Stationary Regime (Winter Temperatures, $\beta = 0$): Temperatures are relatively flat (e.g., near the winter solstice). Cumulative temperature grows linearly.
2. Increasing Regime (Spring Warming, $\beta > 0$): Temperatures increase at a constant rate (e.g., near the spring equinox). Cumulative temperature grows quadratically.

Validation:
The theory is tested against two datasets:

• Observational Data: 10,613 observations of common lilac (Syringa vulgaris) bloom dates across the US (1956–2025).
• Experimental Data: Controlled warming experiments on walnut tree (Juglans sp.) stems where temperature was held constant (simulating the stationary regime).

3. Key Contributions

• Theoretical Unification: The paper demonstrates that the "complexity" observed in phenological shifts (decelerating advancement and increasing variance) is a mathematical consequence of the transition between the two temperature regimes.
• The "Scattered Spring" Concept: They formally define how climate change can shift the period of temperature accumulation from the "Spring Warming" regime to the "Winter Temperatures" regime, leading to less synchronized biological events.
• Parsimony: They argue that many proposed biological complexities (like chilling or photoperiod) may already be implicitly captured by the stochasticity within the thermal-sum model, suggesting that simpler models may be more robust than previously thought.

4. Results

The mathematical derivations and empirical tests yield the following:

• Mean Timing (Advancement):
  • In the stationary regime, the expected bloom date advances linearly with $\alpha$.
  • In the increasing regime, it advances according to the square root of the warming rate ($\sqrt{2\tau/\beta}$).
  • This explains the nonlinear, diminishing rate of advancement observed as climates warm.
• Variance (Synchrony):
  • In the Spring Warming regime ($\beta > 0$): The variance is inversely proportional to the threshold ($\text{Var} \propto 1/\sqrt{\tau}$). As temperatures rise, the quadratic growth of cumulative heat "overwhelms" the noise, effectively synchronizing the plants.
  • In the Winter Regime ($\beta = 0$): The variance is directly proportional to the threshold ($\text{Var} \propto \tau$). Here, noise remains influential, and higher temperatures can lead to desynchronization.
• Empirical Fit: The lilac dataset and the walnut experiment both showed a near-perfect fit to the predicted nonlinear relationships for both mean and standard deviation.

5. Significance

This research has profound implications for ecology and climate science:

1. Ecological Resilience: It predicts that as spring moves earlier, ecosystems may face "scattered springs," where the lack of synchrony between plants and their pollinators could lead to trophic mismatches and ecosystem instability.
2. Modeling Standards: It challenges researchers to justify the use of highly complex phenological models, suggesting that the basic thermal-sum model, when viewed through the lens of renewal theory, is much more powerful and explanatory than previously realized.
3. Predictive Power: The framework allows scientists to predict how local climate shifts (whether warming is driven by higher winter baselines or steeper spring trends) will specifically impact the timing and coordination of local biological communities.