Abstract: We build a general framework which establishes a one-to-one correspondence between species abundance distribution (SAD) and species accumulation curve (SAC). The appearance rates of the species and the appearance times of individuals of each species are modeled as Poisson processes. The number of species can be finite or infinite. Hill numbers are extended to the framework. We introduce a linear derivative ratio family of models, $\mathrm{LDR}_1$, of which the ratio of the first and the second derivatives of the expected SAC is a linear function. A D1/D2 plot is proposed to detect this linear pattern in the data. By extrapolation of the curve in the D1/D2 plot, a species richness estimator that extends Chao1 estimator is introduced. The SAD of $\mathrm{LDR}_1$ is the Engen's extended negative binomial distribution, and the SAC encompasses several popular parametric forms including the power law. Family $\mathrm{LDR}_1$ is extended in two ways: $\mathrm{LDR}_2$ which allows species with zero detection probability, and $\mathrm{RDR}_1$ where the derivative ratio is a rational function. Real data are analyzed to demonstrate the proposed methods. We also consider the scenario where we record only a few leading appearance times of each species. We show how maximum likelihood inference can be performed when only the empirical SAC is observed, and elucidate its advantages over the traditional curve-fitting method.
Introduction
In biodiversity studies, researchers have traditionally relied on two primary tools to characterize ecological communities: the Species Abundance Distribution (SAD) and the Species Accumulation Curve (SAC). The SAD provides a snapshot of community structure by recording the frequency of species with specific individual counts (e.g., how many species are represented by a single individual, two individuals, etc.). Conversely, the SAC tracks the discovery of new species over time or sampling effort.
While ecologists have long recognized an intuitive link between these two metrics, a rigorous mathematical bridge has remained elusive. Previous attempts often relied on specific parametric assumptions or treated the relationship as a simple curve-fitting exercise, which frequently failed to account for the interdependence of data points. Furthermore, traditional models struggled to accommodate communities with potentially infinite richness or the presence of "zero-rate" species—transient individuals or sequencing errors that appear only once. This paper introduces the Mixed Poisson Partition Process (MPPP) to provide a unified, statistically sound framework for these fundamental concepts.
A Unified Framework: The Mixed Poisson Partition Process
The core contribution of this research is the establishment of a formal one-to-one correspondence between SAD and SAC using a stochastic process. In the MPPP framework, species arrival rates are drawn from an intensity measure, and the arrival of individuals within each species is modeled as an independent Poisson process. This approach allows the authors to prove a fundamental theorem: an expected SAC is mathematically valid if and only if it is a Bernstein function passing through the origin.
As illustrated in Figure 1, the process begins by generating species with specific arrival rates ($\lambda$). Individuals of these species then appear over time $t$ according to their respective Poisson processes. This generation process simultaneously yields the empirical species accumulation curve $N_+(t)$ and the frequency of frequencies $N_k(t_0)$, effectively unifying the two perspectives into a single generative model.
The LDR₁ Family and D1/D2 Plots
Building on this framework, the authors introduce the "Linear Derivative Ratio" (LDR₁) family of models. In this family, the ratio of the first and second derivatives of the SAC is a linear function of time. This family is exceptionally flexible, encompassing widely used models such as the power law, the geometric distribution, and Engen’s extended negative binomial distribution.
To identify these patterns in real-world data, the paper proposes the D1/D2 plot. By plotting the ratio of the first and second derivatives of the empirical SAC, researchers can visually diagnose whether their data fits the LDR₁ model. A key insight from these plots is that a slope greater than 1 indicates that the expected species richness is likely infinite—a critical distinction for microbial and genomic studies.
Statistical Inference and Empirical Results
A major methodological shift proposed in this paper is the move away from least-squares curve fitting for SACs. The authors argue that traditional curve fitting is theoretically flawed because it ignores the interdependence of data points along the curve. Instead, they propose a Maximum Likelihood Estimation (MLE) approach based on the timing of species' first appearances. Simulation studies demonstrate that this MLE method consistently achieves a lower Root Mean Squared Relative Error (RMSRE) in extrapolation compared to traditional methods.
The framework was validated across four diverse datasets: swine feces genomic data, insurance accident claims, tomato flower cDNA, and North American bird abundance. The D1/D2 plots successfully identified infinite species richness in the microbial and genomic data, while correctly identifying the bird population as finite. Furthermore, the paper introduces a new species richness estimator, $\hat{E}^*(D)$, which extends the Chao1 lower bound by allowing rare species to follow a distribution rather than assuming equal abundance.
Conclusion and Future Directions
This framework provides a robust mathematical foundation for biodiversity analysis, successfully unifying SAD and SAC while offering superior tools for richness estimation and model selection. While the LDR₁ family is broad, the authors acknowledge that some complex datasets may require the more advanced Rational Derivative Ratio (RDR₁) family. Future work will focus on fully non-parametric settings to eliminate the need for model selection and investigating labor-saving sampling designs where recording stops after a species is observed a set number of times.
Summary updated