Skip to main content
NCBI home page
Plant Diversity logoLink to Plant Diversity
. 2026 Mar 14;48(5):943–951. doi: 10.1016/j.pld.2026.03.008

Scale-dependent evidence for the abundant-center hypothesis and underlying mechanisms

Jingyue Huang a, Dingliang Xing a,b,⁎
PMCID: PMC13615776  PMID: 42801109

Abstract

Whether species are more abundant near the center of their distribution has been intensively debated. Previous studies have largely focused on local sampling units, often overlooking the inherent scale-dependence of ecological patterns. Here, we examine how sampling grain influences the relationship between species abundance and distance to a species’ distributional centroid, using inventory data of 212 tree species from the USDA Forest Inventory and Analysis program (FIA). We found scale-dependent support for the abundant-center hypothesis. At the fine spatial grain of individual FIA plots, positive and negative abundance–distance relationships were roughly balanced, with the mean regression slope approaching zero. This pattern reversed at coarser grains (> 0.5°), where the odds of detecting a negative trend more than doubled and the mean regression slope became significantly negative. Additionally, we identified niche overlap as a key factor constraining species abundance at smaller grains, with its impact diminishing at larger scales. Our results underscore the necessity of incorporating biotic interactions in fine-grain species abundance and distribution modeling. Conversely, traditional climate envelope models based on abiotic variables may effectively capture abundance patterns at coarser grains. This scale-explicit perspective helps reconcile previous inconsistent findings and advances our understanding of how ecological processes structure species distributions across scales.

Keywords: Abundant-center hypothesis, Climate envelope modeling, Interspecific interactions, Niche centrality, Species abundance

Highlights

  • •

    A scale-explicit perspective reconciles conflicting empirical evidence for the abundant-center hypothesis.

  • •

    Negative abundance–distance relationships emerge at coarser grains but are masked by niche overlap at fine grains.

  • •

    Sampling grain dictates patterns: abiotic factors dominate at coarse scales, while biotic factors constrain local abundance.

1. Introduction

How abundant a species is, and how widely distributed that species is are two key aspects of a species population that underlie the contribution of species to patterns of community structure and biodiversity (Andrewartha and Birch, 1954; Sagarin et al., 2006). Furthermore, these two population-level patterns—abundance and distribution—are often thought to be linked. Specifically, species often tend to be most abundant at the center of their geographic and/or environmental ranges, with abundance declining towards the edges (Maguire, 1973; Brown, 1984). This ‘abundant-center hypothesis’ posits that environmental conditions are optimal near the centroid of a species distribution (Brown et al., 1995), and provides a potential mechanistic explanation for an important macroecological pattern on the connection between abundance and distribution. If true, the abundant-center hypothesis can be used to develop new ecological and evolutionary hypotheses (Eckert et al., 2008), to identify causes of species range limits (Sexton et al., 2009), and to provide useful guidance for potential management interventions of species of interest, such as those that are rare or threatened (Yañez-Arenas et al., 2012), or those that are invasive (Jarnevich et al., 2021). In addition, the abundant-center hypothesis is often integrated with the outputs of climate envelope modeling to predict species’ responses to climate change, assuming that higher climatic suitability corresponds to greater species abundance (VanDerWal et al., 2009).

Recent years have seen a surge in studies testing the abundant-center hypothesis, driven in part by the increasing availability of high-resolution, broad-scale data, and more importantly, by the growing applications of climate envelope modeling in a rapidly changing world (Peterson et al., 2011; Dallas and Hastings, 2018; de la Fuente et al., 2021). However, empirical support for the abundant-center hypothesis remains mixed. While some studies report strong evidence (VanDerWal et al., 2009; Martínez-Meyer et al., 2013; Weber et al., 2017; Osorio-Olvera et al., 2020; Yancovitch Shalom et al., 2020), many others find weak or no support (Sagarin and Gaines, 2002; Pironon et al., 2017; Dallas et al., 2017; Freeman and Beehler, 2018; Santini et al., 2019; Waldock et al., 2019; Sporbert et al., 2020).

One potential explanation for these conflicting results lies in differences in the spatial grain of abundance sampling. Most empirical studies testing this hypothesis have used abundance data collected at small sampling grains — such as ∼25 m2 vegetation plots for herbaceous plants and shrubs (Sporbert et al., 2020), 672 m2 USDA Forest Inventory and Analysis plots for woody plants (Dallas et al., 2017), and 500 m2 Reef Life Survey sites for reef fishes (Waldock et al., 2019). However, species are generally spatially aggregated (Condit et al., 2000). When abundance is sampled at such fine grains and only non-zero samples are used (i.e., ignoring absences either intentionally or due to sampling constraints, which is a common practice in previous tests of the hypothesis; but see Weber et al., 2017), the resulting data may distort underlying abundance gradients. This distortion weakens the correlation between abundance and geographic distribution. Unlike fine-grained sampling, larger spatial grains smooth out the high stochasticity inherent in aggregated populations and effectively integrate absences, rendering the underlying abundance gradient more detectable (Wiens, 1989).

Meanwhile, species’ environmental distributions are typically modeled using climate, topographic, and soil variables at much coarser spatial grains—often > 1 km2 (Smith and Santos, 2020). Local abundance, reflecting a species’ realized niche (Blonder, 2016; Laughlin and McGill, 2024), is influenced by multiple factors beyond the broad-scale environmental gradients captured in such models (Vellend, 2010; Xing et al., 2014; Mertes and Jetz, 2018). For example, biotic interactions can constrain local species abundance, particularly in otherwise highly suitable environments (Braz et al., 2020). Environmental heterogeneity may increase species richness by providing suitable conditions for species with diverse ecological requirements, while simultaneously reducing the average population size of individual species (Allouche et al., 2012; Ben-Hur and Kadmon, 2020). Historical climate change can also shape contemporary abundance patterns. Past climatic events, such as post-glacial recolonization, may leave persistent lag effects through slow dispersal and the legacy of glacial refugia, thereby decoupling current abundance from present-day environmental suitability (Williams et al., 2004; Copenhaver-Parry et al., 2017). Together, these processes can reduce species abundance in otherwise suitable localities, further obscuring the relationship between abundance and distribution.

Thus, sampling abundance at larger spatial grains may yield insights into the abundance–distribution relationship that are obscured at finer grains. Indeed, among the studies that show support for the abundant-center hypothesis, several estimated abundances at spatial grains larger than grains used for environment variables. For instance, Osorio-Olvera et al. (2020) and Williams et al. (2021) showed strong evidence for the hypothesis using bird abundance estimates from survey routes spanning 50 km.

Here, we develop a synthetic perspective on the relationship between abundance and distribution, both in geographical and environmental space (Fig. 1). If the abundant-center hypothesis is not true, then species abundance is completely independent of the sampling location within its range, regardless of the sampling grain size (hypothesis H0). Alternatively, species abundance can be related to distribution in different ways. First, species abundance is highest near its range center, but the evidence is only detectable with larger sampling grain size (hypothesis H1). At small grains, species abundance can be either unpredictable or related to local biotic factors rather than the sampling position (Prediction 1.1; McGill, 2012; Dallas and Hastings, 2018; Sporbert et al., 2020). At large grains, since the signature of local species interactions diminishes (Soberón and Nakamura, 2009) and the environmental conditions better reflect the species’ ecological requirements, a negative abundance–distance relationship is expected (Prediction 1.2). Second, the location of the highest species abundance within the range (termed the abundant core; Fristoe et al., 2023) deviates from its range center, yet a clear abundance gradient exists within the species’ range—where abundance declines outward from the abundance core (hypothesis H2). For example, physiological performance curves for some vascular plants are asymmetric, with optimal performance occurring closer to the extreme values of the environmental variable (Oksanen and Minchin, 2002). In this case, (1) at small grains, species abundance is expected to be independent of the sampling location due to the influence of local factors (Prediction 2.1); (2) at large grains, the abundance–distance relationship is expected to exhibit a unimodal on the side closer to the abundance core, and a negative correlation on the side farther from it (Prediction 2.2).

Fig. 1.

Fig. 1

Open in a new tab

Conceptual diagrams comparing competing hypotheses about abundance–distribution relationships. Darker shading in a, c, and e indicates areas of higher species abundance. Sampling scales are color-coded, with blue and red colors representing large and small grain sizes, respectively. (a and b) The null hypothesis (H0, rejection of abundant-center hypothesis) proposes no correlation between species abundance and distance-to-centroid across all sampling scales. (c and d) Alternative hypothesis H1 demonstrates scale-dependent abundance–distance relationships (Prediction 1.2), where diminished correlations at finer sampling scales are partially attributable to biotic constraints (Prediction 1.1). (e and f) Alternative hypothesis H2 incorporates the offset between abundance cores (peak abundance areas) and range centers. When analyzed using conventional range center definitions, samples from the proximal side of abundance cores exhibit unimodal abundance–distance patterns, while maintaining H1’s scale-dependent predictions.

We tested these hypotheses using inventory data of 212 tree species from the extensive USDA Forest Inventory and Analysis (FIA) program in the United States, by estimating species’ mean abundance at six sampling grain sizes and quantifying distance of the sampling grains to the range centroid in both geographic and environmental spaces. We find that the strength of abundance–distance relationships depends on the spatial grain at which abundance is measured. At large grains, abundance can be explained by sampling position within a species’ range, while small-grain abundance is mostly independent of sampling location but more strongly influenced by niche overlap. These findings are strengthened by the incorporation of other covariates, such as environmental heterogeneity and historical climate change, which also contribute significantly to the prediction of species abundance.

2. Materials and methods

2.1. Abundance data

We obtained tree species data from the USDA Forest Inventory and Analysis (FIA) program (https://www.fia.fs.usda.gov/). FIA plots are distributed across the forested portions of the United States, with a sampling intensity of approximately one plot per 2500 ha. Each FIA plot consists of four circular subplots (each 168 m2), within which all stems with a diameter at breast height (DBH) ≥ 12.7 cm were recorded. We selected the most recent inventories (from 1997 to 2021) from 75,393 FIA plots that are natural and undisturbed, meet the national design standards, and occur within the contiguous U.S. (excluding Alaska, Hawaii and U.S. overseas territories; Fig. S1). We excluded species that were exceptionally rare and only occurred in < 10 plots. Following previous studies, we also removed species exhibiting potential northern range truncation within the FIA domain (Dallas et al., 2017). Specifically, species were excluded if their abundance at the northern range margin exceeded the mean abundance across occupied FIA plots. A total of 212 native species were finally included in our analysis.

At the scale of the individual FIA plot, we defined species abundance as the number of individuals in the entire FIA plot. To estimate abundance at broader spatial grains, we divided the contiguous United States into 0.125°, 0.25°, 0.5°, 1°, and 2° grid cells. For each sampling grain (i.e., grid cell), we calculated the mean abundance of a species across all FIA plots within the intersection of that grid cell and a convex hull depicting the species’ geographic range (see below; Fig. 2a). We excluded grid cells in our subsequent analyses for a species when > 50% of their area was outside the geographic range as defined by the convex hull.

Fig. 2.

Fig. 2

Open in a new tab

A demonstration of how sampling grain size could influence the relationship between species abundance and distance to geographic range center, using species Carya alba. (a) Blue points show standard FIA plots where the species was observed. Grey points show plots where the species was absent. Red star shows the geographic range center. The entire region was divided into grid cells of a certain size (2° × 2° shown) and for each cell, mean abundance was calculated by averaging observed abundances of all FIA plots within the cell and within the geographic range (convex hull). Grid cells where more than 50% area is outside the convex hull or the mean abundance is zero are excluded in subsequent analyses. (b) Scatterplots showing the relationships between log-transformed abundance and distance to geographic range center for Carya alba at six different grain sizes. Lines represent linear regression.

To protect privacy of landowner and plot integrity, the coordinates of FIA plots are fuzzed and swapped within the same county (Woudenberg et al., 2010). However, this plot location uncertainty is typically < 1.6 km, which is negligible compared to the range of even the most narrowly distributed species, and thus unlikely to affect our results.

2.2. Estimating distance-to-centroid and the abundance core

For each sampling grain, we quantified the distance-to-centroid for each abundance estimate using two methods: geographic space and environmental space. In geographic space, we defined the species range as a convex hull fitted to the species’ occurrences in FIA plots using the “chull” function in the R package grDevices, and the centroid as the mean of the vertices of this convex hull (see e.g., Dallas et al., 2017; Osorio-Olvera et al., 2020; Chevalier et al., 2021). We calculated the Haversine distance (i.e., Euclidean distance accounting for the curvature of the Earth) from each FIA plot to the centroid and used it to predict species abundance at the finest grain. For coarser grains, we defined the distance from a grid cell to the centroid as the average Haversine distance of all FIA plots within that grid cell to the centroid.

We also calculated distances-to-centroid in environmental space (Dallas et al., 2017). To do so, we first performed a principal component analysis (PCA) on the 19 BioClim variables and elevation data using the function “rasterPCA” in the R package RStoolbox. The data were downloaded from the WorldClim v.2.1 database at a resolution of 30 arc second (Hijmans et al., 2005). Next, we used the first two principal components, which together accounted for 74% of the climatic variance, to define species’ climate envelope by delineating a convex hull in the two-dimensional PCA space based on the loadings from the species’ occurrences in FIA plots. We defined the centroid as the mean of the vertices of this convex hull and the distance-to-centroid as the Euclidean distance from the loading of an FIA plot to the centroid (for the FIA plot grain) or the average Euclidean distance of the loadings of all FIA plots within a grid cell to the centroid (for larger grains).

For each species, we used abundance data to determine the location of the highest abundance in both geographic and environmental spaces. The abundance core in geographic space was calculated as the centroid of FIA plots where species abundance exceeds the 50th percentile (Fristoe et al., 2023), using the “gCentroid” function in the R package rgeos. We then translated the geographic locations of the top half FIA plots of each species into environmental space and computed their average environmental condition to define the abundance core in environmental space.

2.3. Quantifying niche overlap with co-occurring species

To test the hypothesis that species with similar environmental requirements may negatively affect each other’s abundance, we calculated a niche overlap index. When two species exhibit high niche overlap, they are likely to share similar resources as well as common enemies. Consequently, a reduction in the abundance of one species may arise either from interspecific competition with the other species (Beaudrot et al., 2019) or from increases in the abundance or activity of shared enemies driven by the other species (i.e., apparent competition mediated by shared enemies; Holt, 1977). In other words, a high degree of niche overlap is presumed to increase the likelihood of negative biotic interactions, thereby potentially constraining the abundance of a focal species. We defined the pairwise niche overlap as the intersection area of two species’ convex hulls in environmental space (Blonder et al., 2014; Mammola, 2019). This metric captures the similarity of their environmental requirements. For each focal species within a given grain, we then calculated a summed niche overlap index by summing the overlap values with all other co-occurring species.

Crucially, we acknowledge that this index quantifies similarity in occupied environmental conditions and is not a direct measure of realized interspecific interactions. Co-occurrence in similar environments does not necessarily equate to ecological interactions (Blanchet et al., 2020). Therefore, we interpret this variable not as a proxy for interaction itself, but as a measure of the potential for such interactions to occur due to shared environmental affinities. We note that there are different methods to quantify niche overlap. Our results and conclusion remain robust when a different overlap metric was employed (Schoener’s D; Schoener, 1970; Broennimann et al., 2012).

2.4. Environmental heterogeneity and historical climate change

We defined contemporary climate and elevation data as above, and sourced paleoclimatic variables at the Last Glacial Maximum (LGM) from the CHELSA database at a 30 arc second resolution (Karger et al., 2023). To address uncertainty in past climate simulations, we used the average value from three available simulations for the LGM: MPI-ESM-P (Max-Planck-Institute Earth System Model), MIROC-ESM (Model for Interdisciplinary Research on Climate-Earth System Model), and CCSM4 (Community Climate System Model v.4).

Because the heterogeneity in temperature, precipitation, and elevation influences the distribution of woody plants (Qian and Kissling, 2010; Wang et al., 2011; Stein and Kreft, 2015), we estimated heterogeneity of these variables for grid cells ranging from 0.125° to 2° (individual FIA plots were too small to measure heterogeneity) using the R package raster. We calculated environmental heterogeneity as the range of mean annual temperature (bio1), annual precipitation (bio12), and elevation within each grid cell. Since these three heterogeneity measures were highly correlated, we performed a principal component analysis (PCA) on them for each grain size (Fig. S2) and used the first principal component (> 80.4% of the total variance) as a composite metric for environmental heterogeneity.

To measure past climate change, we calculated the difference in mean annual temperature and precipitation between the present and the LGM, defined as the temperature anomaly and precipitation anomaly. We calculated these anomalies for each FIA plot at the finest grain via the R package raster, and then the average anomaly within each grid cell. Due to the strong correlation between temperature and precipitation anomalies, we focused solely on the temperature anomaly for analysis.

2.5. Statistical analysis

To evaluate how sampling grain affects the abundance–distance relationship, we performed linear regressions for each species at six spatial grains (from individual FIA plots to 2°; Fig. 2b). These models related the natural log-transformed species abundance to the distance-to-centroid in both geographic and environmental space. We used the slope, P-value, and R2 from each model to assess the direction, significance, and strength of the relationship, respectively. For analyses at grains coarser than individual plots, the number of plots per grid cell was non-uniform, potentially introducing heteroscedasticity. We therefore applied weighted least squares regression, using the number of plots in each cell as the weight. This approach yielded qualitatively similar results to unweighted regressions, though unweighted models identified slightly fewer significant negative relationships (Fig. S3).

To further investigate the potential effects of other covariates on abundance, we used multiple regression models. For each species, the natural log-transformed abundance was used as the response variable, with distance-to-centroid, niche overlap, their interaction, environmental heterogeneity, and temperature anomaly as predictors. The interaction between niche overlap and distance-to-centroid was included because biotic interactions are expected to be more limiting when the abiotic environment is less stressful (Bertness and Callaway, 1994; Crotty and Bertness, 2015; Braz et al., 2020). We included all predictors at all grain sizes, except environmental heterogeneity, for which we did not have heterogeneity values at the finest FIA plot scale. We determined the final selection of variables for each model through stepwise model selection using the function “stepAIC” in the R package MASS. To improve normality and linearity, we log-transferred the niche overlap. To simultaneously assess the effects of all predictors on species abundance, we scaled all predictor variables to a mean of zero and a standard deviation of one for each species.

To test whether peak abundance was shifted from the geometric range center towards the abundance core, we tested for an asymmetric unimodal abundance distribution. For each species, we bisected its range with a line passing through the range center and perpendicular to the axis connecting the center to the abundance core. This line separated all plots into two groups based on their spatial position relative to the abundance core: the “core-ward” group (closer to the abundance core) and the “distal” group (farther from the abundance core). For the “core-ward” group of plots, we fitted both linear and quadratic regression models to the abundance–distance data, using the Akaike Information Criterion (AIC) to select the best model. For the “distal” group, we fitted a linear model. We then counted the number of species exhibiting the hypothesized asymmetric pattern: a unimodal relationship on the core-ward side and a negative relationship on the distal side. All analyses were conducted in R v.4.2.3 (R Core Team 2023).

3. Results

3.1. The effect of sampling grain on abundance–distance relationships

Our analyses rejected the hypothesis H0 of species abundance being randomly distributed across their range and revealed a scale-dependent relationship between species abundance and distance-to-centroid, both in geographic and environmental space (Fig. 3). At the finest sampling grain (individual FIA plots), the pattern was ambiguous. The mean regression slope was not significantly different from zero in either geographic (−0.0001, 95% CI [−0.0002, 0]) or environmental space (−0.0210, 95% CI [−0.0474, 0.0054]; Fig. 3a). Correspondingly, positive and negative abundance–distance slopes were nearly equally represented among the 212 species, with the odds of finding a negative slope—the ratio of the number of species with negative to positive slopes—being statistically insignificant (geographic odds [95% CI]: 1.30 [0.99, 1.71]; environmental odds: 1.26 [0.96, 1.65]; Fig. 3c and Table S1).

Fig. 3.

Fig. 3

Open in a new tab

Distribution of summary statistics of the relationship between species abundance and distance-to-centroid across six sampling grains. The distance-to-centroid was quantified in both geographic space (left panels) and environmental space (right panels). Boxes in (a) and (b) show median and interquartile range. Blue and pink colors in (c) show, respectively, proportions of negative and positive relations. ∗ P < 0.05, • P < 0.10, ns P > 0.10.

In contrast, as the sampling grain coarsened, a clear negative trend emerged and strengthened. At the 2° grid scale, mean regression slope became sharply negative in both geographic (−0.0014, 95% CI [−0.0018, −0.0010]) and environmental space (−0.73, 95% CI [−0.89, −0.58]; Fig. 3a). The proportion of negative relationships surged to 75% (geographic) and 83% (environmental), a dramatic increase from ∼56% observed at the finest scale (Fig. 3c). This shift corresponds to a 2.35-fold [1.45, 3.80] increase in the odds of detecting a negative trend in geographic space and a 3.91-fold [2.30, 6.66] increase in environmental space (Table S1). Consistently, the odds of finding a statistically significant negative relationship at > 0.5° grid cells were 1.6–3 times higher than at the finest scale (Table S2).

Meanwhile, the explanatory power of the models also increased with scale. The mean R2 (±SD) value rose to 0.15 (±0.17) in geographic space and 0.17 (±0.17) in environmental space at 2° grid scale, from close to zero at the finest scale (Fig. 3b). Overall, the negative abundance–distance relationships were slightly more pronounced in environmental space than in geographic space across all coarser grains (Fig. 3).

When we relaxed the strict centrality assumption of the traditional abundant-center hypothesis to allow cases in which species’ abundance peaks are displaced from the distribution centroid (hypothesis H2), we found a modest increase in support for abundance gradients. Specifically, the proportion of species supporting the abundant-center hypothesis increased by approximately 3% in both geographic and environmental space, with a more pronounced increase at finer grains (Fig. S4).

3.2. Determinants of species abundance beyond distance-to-centroid

Including niche overlap, environmental heterogeneity, and temperature anomaly improved the model fit to the abundance data across most species-grain size combinations (Fig. S5). These best-selected models also revealed an increased (by 3–7%) proportion of species showing significant negative partial abundance–distance relationships compared to the distance-only models at grain sizes < 0.5° (Fig. S6).

Among the tested predictors, niche overlap had the strongest standardized effect at smaller sampling grains, showing a negative association with local abundance for the majority of species (82–95% for grain sizes < 0.5°; Fig. 4). This indicates that niche overlap plays a more important role than distance-to-centroid in explaining small-grain abundances. When grain sizes exceeded 0.5°, the coefficient for niche overlap became not significant for over 63% species, and the average standardized coefficient for niche overlap approached zero (Fig. 4). At the same time, the average standardized coefficient for distance-to-centroid became increasingly negative (Fig. 4). Additionally, a notable proportion of species (16–20%), spanning grain sizes from FIA plots to 0.5° grid cells, exhibited a significant positive coefficient for the interactive effect of distance and niche overlap in the best-selected models, suggesting that the constraining effect of niche overlap on abundance is stronger near the center of a species’ environmental range. Excluding this interaction effect, along with the effect of niche overlap, from the best-selected models decreased the proportion of significant negative coefficients for distance by 6–8% across these grain sizes (Fig. S7).

Fig. 4.

Fig. 4

Open in a new tab

Standardized multiple regression coefficients for different predictors of species abundance across six sampling grains. Black points indicate significant coefficients (P < 0.05). Grey points show insignificant coefficients (P > 0.05). Violin plots show the distribution of the coefficients for each predictor, with boxes showing the median, the first and third quartiles.

Environmental heterogeneity was significantly correlated with abundance for 36–45% species, across the studied grain sizes. Species abundance decreased with increasing heterogeneity for most (64–73%) species (Fig. 4). Removing environmental heterogeneity from the best-selected models resulted in a slight decrease (∼1%) in the proportion of significant negative coefficients for distance. The temperature anomaly was significantly correlated with species abundance for 26–47% species, among which about 58% were positive (Fig. 4). Removing this variable from the best-selected models led to a slight decrease (∼1%) in the proportion of significant negative coefficients for distance.

4. Discussion

Overall, our findings provide support for the hypothesis that the relationship between species abundance and distance to the center of their geographic or environmental ranges is scale dependent (hypothesis H1), helping to reconcile the apparent contradiction found in previous studies. At the finest sampling grain (i.e., individual FIA plots), we observed no clear trend in the correlation between local abundance and distance-to-centroid. This aligns well with the findings of a number of studies that used small-scale data for abundance estimates like Dallas et al. (2017), Waldock et al. (2019), and Sporbert et al. (2020). This also confirms that local abundance patterns are hard to predict only from deterministic abiotic factors such as spatial proximity to the geographic or environmental range centroid (McGill, 2012; Ren et al., 2013).

As the sampling grain increases (e.g., > 0.5°), we found that 79% of species exhibited negative abundance–distance relationships in environmental space, with the mean regression slope becoming significantly negative. These results support the predictions of the abundant-center hypothesis and reject hypothesis H0 of random abundance distribution. Our findings are consistent with studies such as Osorio-Olvera et al. (2020) and Williams et al. (2021), who documented a negative correlation between species abundance and environmental suitability at coarse spatial resolutions. The stronger relationships we observed in environmental space than in geographic space suggest that climatic factors play a significant role in determining species abundance patterns across broader scales, consistent with the well-established view that climatic factors are the major determinant of large-scale plant distributions (Woodward, 1987; Davis and Shaw, 2001; Pearson and Dawson, 2003). These findings underscore the importance of considering scale when explaining and modelling abundance patterns and suggest that the abundant-center hypothesis may only hold at coarser spatial grains.

Our results show that accounting for potential deviations of abundance peaks from the distribution centroid (hypothesis H2) only marginally (∼3%) increases the support for the abundant-center hypothesis (Fig. S4). This suggests that, although skewed abundance patterns occur in some species, relaxing the centrality assumption alone contributes relatively little compared to the effect of spatial scale and does not resolve the mixed empirical support for the abundant-center hypothesis reported in previous studies. Taken together, hypotheses H1 and H2 describe complementary but non-competing aspects of abundance–distribution relationships. While H2 addresses geometric deviations in the location of abundance peaks, H1 captures a more fundamental scale dependence in the detectability of abundance gradients. Our results therefore indicate that scale dependence, rather than off-center abundance cores per se, is the primary factor reconciling conflicting evidence for the abundant-center hypothesis.

Incorporating additional factors beyond distance-to-centroid, including niche overlap, environmental heterogeneity, and temperature anomalies, into our models significantly improved their explanatory power across all studied grain sizes (Fig. S5), highlighting the importance of considering these factors when modeling species abundance. We also found that the negative abundance–distance relationship can be detected after incorporating the significant covariates, reinforcing support for the abundant-center hypothesis and the importance of considering potential confounding factors in assessing this hypothesis.

The niche overlap between species may reflect potential interspecific competition for similar abiotic resources or apparent competition mediated by shared enemies. McGill (2012) found that the trade-off between competitive ability and environmental tolerance shapes species distribution and abundance. Bhattarai et al. (2017) reported that invasive plants increase pressure from shared enemies on native plants, leading to reduced fitness of the native species. In our study, the strong negative association between local abundance and niche overlap at smaller grains suggests that such potential interspecific interactions are likely affecting species abundance in these contexts. This finding aligns with a substantial body of literature establishing that biotic interactions drive local abundance (MacArthur, 1972; Soberón, 2007; McGill, 2010; Braz et al., 2020). Notably, competitive interactions may be more important when abiotic stress is less important (Bertness and Callaway, 1994; Crotty and Bertness, 2015), such as near the range center. Indeed, we observed higher variation in species abundance near the range center for some species, and detected a positive interaction between distance-to-centroid and niche overlap. This suggests that interspecific interactions may exert a stronger suppressive effect on abundance near the range center, thereby potentially weakening the expected negative abundance–distance correlation.

Environmental heterogeneity and paleo-climate change further complicate the relationship between abundance and distance-to-centroid. Local-scale environmental heterogeneity can override macroclimatic and geographic influences, leading to deviations from abundant-center expectations (Moutouama and Gaoue, 2022; Fristoe et al., 2023). Moreover, historical climate change leaves lasting imprints on species distributions (Eldridge and Delgado-Baquerizo, 2019; Heo et al., 2023). In regions that experienced unstable climates over glacial–interglacial cycles, species may show the disequilibrium between current climate conditions and species abundance patterns (Copenhaver-Parry et al., 2017). Species with narrow climate tolerance appear especially sensitive to such instability, showing reduced population sizes in areas that experienced large temperature anomalies (see Appendix S2).

The scale dependence of abundance–distance relationships has significant implications for climate envelope modeling, as projections of high-resolution species distributions based on fine-grained species abundance data and coarse-grained environmental suitability data may not accurately reflect true local abundance patterns (Elith and Leathwick, 2009; Dallas and Hastings, 2018; Santini et al., 2019; Lee-Yaw et al., 2021). Conversely, coarse-resolution (e.g., > 0.5° for woody plants) projections of climate envelope models are backed by the empirical abundance–distance relationships and should be used with confidence in applications. Our findings highlight the necessity of incorporating biotic interactions in modeling small-scale species distribution and abundance. Encouragingly, the fast development of joint species distribution modeling (jSDM) over the past decade can partly address this problem (Ovaskainen and Abrego, 2020).

In conservation practice, our findings emphasize an often overlooked yet critical issue: mismatches in spatial scale can lead to misinterpretations of true species abundance and distribution patterns, thereby undermining the effectiveness of invasive species management and endangered species protection (Mertes and Jetz, 2018). For example, when setting conservation priorities such as whether to focus on the central or peripheral populations, relying solely on fine-grain abundance data may allow local habitat heterogeneity or strong biotic interactions to obscure the true abundance pattern, ultimately leading to suboptimal allocation of conservation resources (McGill, 2012). Similarly, using data from a single spatial scale to predict invasion risk may overlook highly vulnerable populations and thus underestimate the overall risk that invasive pests and pathogens pose to ecosystems (Nunez-Mir et al., 2022). In contrast, a multi-scale framework can more accurately identify future ecological vulnerability zones and invasion hotspots, thereby guiding more targeted management and conservation actions.

While our study provides support for the abundant-center hypothesis, several caveats must be acknowledged. Recent analyses, such as those by Osorio-Olvera et al. (2020) and Chevalier et al. (2021) have pointed out methodological limitations that may affect the detectability and significance of abundance–distance relationships. Although our findings were robust across multiple grains and model configurations, unexplained variation in species abundance suggests that other processes, such as dispersal limitations (Dallas and Santini, 2020; Feng et al., 2021) and demographic stochasticity (Mori, 2019), may also play significant roles.

CRediT authorship contribution statement

Jingyue Huang: Conceptualization; Methodology; Formal analysis; Validation; Writing – original draft; Writing – review and editing. Dingliang Xing: Conceptualization; Methodology; Writing – original draft; Writing – review and editing; Funding acquisition.

Data availability statement

Data and code are available on Zenodo at https://zenodo.org/records/17188563.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We thank Prof. Jonathan Chase for insightful comments and helpful edits of an earlier version of this manuscript. This study was supported by the National Nature Science Foundation of China (32471623 and 32101280) and the Innovation Program of Shanghai Municipal Education Commission (2023ZKZD36).

Footnotes

Peer review under the responsibility of Editorial Office of Plant Diversity.

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2026.03.008.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Multimedia component 1
mmc1.docx (7.6MB, docx)

References

  1. Allouche O., Kalyuzhny M., Moreno-Rueda G., et al. Area–heterogeneity tradeoff and the diversity of ecological communities. Proc. Natl. Acad. Sci. U.S.A. 2012;109:17495–17500. doi: 10.1073/pnas.1208652109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Andrewartha H.G., Birch L.C. University of Chicago Press; Chicago: 1954. The Distribution and Abundance of Animals. [Google Scholar]
  3. Beaudrot L., Acevedo M.A., Lessard J.-P., et al. Local temperature and ecological similarity drive distributional dynamics of tropical mammals worldwide. Global Ecol. Biogeogr. 2019;28:976–991. [Google Scholar]
  4. Ben-Hur E., Kadmon R. Heterogeneity–diversity relationships in sessile organisms: a unified framework. Ecol. Lett. 2020;23:193–207. doi: 10.1111/ele.13418. [DOI] [PubMed] [Google Scholar]
  5. Bertness M.D., Callaway R. Positive interactions in communities. Trends Ecol. Evol. 1994;9:191–193. doi: 10.1016/0169-5347(94)90088-4. [DOI] [PubMed] [Google Scholar]
  6. Bhattarai G.P., Meyerson L.A., Cronin J.T. Geographic variation in apparent competition between native and invasive Phragmites australis. Ecology. 2017;98:349–358. doi: 10.1002/ecy.1646. [DOI] [PubMed] [Google Scholar]
  7. Blanchet F.G., Cazelles K., Gravel D. Co-occurrence is not evidence of ecological interactions. Ecol. Lett. 2020;23:1050–1063. doi: 10.1111/ele.13525. [DOI] [PubMed] [Google Scholar]
  8. Blonder B. Do hypervolumes have holes? Am. Nat. 2016;187:E93–E105. doi: 10.1086/685444. [DOI] [PubMed] [Google Scholar]
  9. Blonder B., Lamanna C., Violle C., et al. The n-dimensional hypervolume. Global Ecol. Biogeogr. 2014;23:595–609. [Google Scholar]
  10. Braz A.G., de Viveiros Grelle C.E., de Souza Lima Figueiredo M., et al. Interspecific competition constrains local abundance in highly suitable areas. Ecography. 2020;43:1560–1570. [Google Scholar]
  11. Broennimann O., Fitzpatrick M.C., Pearman P.B., et al. Measuring ecological niche overlap from occurrence and spatial environmental data. Global Ecol. Biogeogr. 2012;21:481–497. [Google Scholar]
  12. Brown J.H. On the relationship between abundance and distribution of species. Am. Nat. 1984;124:255–279. [Google Scholar]
  13. Brown J.H., Mehlman D.W., Stevens G.C. Spatial variation in abundance. Ecology. 1995;76:2028–2043. [Google Scholar]
  14. Chevalier M., Broennimann O., Guisan A. Using a robust multi-settings inference framework on published datasets still reveals limited support for the abundant centre hypothesis: more testing needed on other datasets. Global Ecol. Biogeogr. 2021;30:2211–2228. [Google Scholar]
  15. Condit R., Ashton P.S., Baker P., et al. Spatial patterns in the distribution of tropical tree species. Science. 2000;288:1414–1418. doi: 10.1126/science.288.5470.1414. [DOI] [PubMed] [Google Scholar]
  16. Copenhaver-Parry P.E., Shuman B.N., Tinker D.B. Toward an improved conceptual understanding of North American tree species distributions. Ecosphere. 2017;8 [Google Scholar]
  17. Crotty S.M., Bertness M.D. Positive interactions expand habitat use and the realized niches of sympatric species. Ecology. 2015;96:2575–2582. doi: 10.1890/15-0240.1. [DOI] [PubMed] [Google Scholar]
  18. Dallas T., Decker R.R., Hastings A. Species are not most abundant in the centre of their geographic range or climatic niche. Ecol. Lett. 2017;20:1526–1533. doi: 10.1111/ele.12860. [DOI] [PubMed] [Google Scholar]
  19. Dallas T., Hastings A. Habitat suitability estimated by niche models is largely unrelated to species abundance. Global Ecol. Biogeogr. 2018;27:1448–1456. [Google Scholar]
  20. Dallas T.A., Santini L. The influence of stochasticity, landscape structure and species traits on abundant–centre relationships. Ecography. 2020;43:1341–1351. [Google Scholar]
  21. Davis M.B., Shaw R.G. Range shifts and adaptive responses to Quaternary climate change. Science. 2001;292:673–679. doi: 10.1126/science.292.5517.673. [DOI] [PubMed] [Google Scholar]
  22. de la Fuente A., Hirsch B.T., Cernusak L.A., et al. Predicting species abundance by implementing the ecological niche theory. Ecography. 2021;44:1723–1730. [Google Scholar]
  23. Eckert C.G., Samis K.E., Lougheed S.C. Genetic variation across species' geographical ranges: the central–marginal hypothesis and beyond. Mol. Ecol. 2008;17:1170–1188. doi: 10.1111/j.1365-294X.2007.03659.x. [DOI] [PubMed] [Google Scholar]
  24. Eldridge D.J., Delgado-Baquerizo M. The influence of climatic legacies on the distribution of dryland biocrust communities. Glob. Change Biol. 2019;25:327–336. doi: 10.1111/gcb.14506. [DOI] [PubMed] [Google Scholar]
  25. Elith J., Leathwick J.R. Species distribution models: ecological explanation and prediction across space and time. Annu. Rev. Ecol. Evol. Syst. 2009;40:677–697. [Google Scholar]
  26. Feng X., Qiao H., Schrodt F. Accounting for dispersal using simulated data improves understanding of species abundance patterns. Global Ecol. Biogeogr. 2021;31:200–214. [Google Scholar]
  27. Freeman B.G., Beehler B.M. Limited support for the “abundant centre” hypothesis in birds along a tropical elevational gradient: implications for the fate of lowland tropical species in a warmer future. J. Biogeogr. 2018;45:1884–1895. [Google Scholar]
  28. Fristoe T.S., Vilela B., Brown J.H., et al. Abundant-core thinking clarifies exceptions to the abundant-center distribution pattern. Ecography. 2023;2 [Google Scholar]
  29. Heo N., Leopold D.J., Lomolino M.V., et al. Global and regional drivers of abundance patterns in the hart's tongue fern complex (Aspleniaceae) Ann. Bot. 2023;131:737–750. doi: 10.1093/aob/mcac129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hijmans R.J., Cameron S.E., Parra J.L., et al. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 2005;25:1965–1978. [Google Scholar]
  31. Holt R.D. Predation, apparent competition, and the structure of prey communities. Theor. Popul. Biol. 1977;12:197–229. doi: 10.1016/0040-5809(77)90042-9. [DOI] [PubMed] [Google Scholar]
  32. Jarnevich C.S., Sofaer H.R., Engelstad P. Modelling presence versus abundance for invasive species risk assessment. Divers. Distrib. 2021;27:2454–2464. [Google Scholar]
  33. Karger D.N., Nobis M.P., Normand S., et al. CHELSA-TraCE21k – high-resolution (1 km) downscaled transient temperature and precipitation data since the last glacial maximum. Clim. Past. 2023;19:439–456. [Google Scholar]
  34. Laughlin D.C., McGill B.J. Trees have overlapping potential niches that extend beyond their realized niches. Science. 2024;385:75–80. doi: 10.1126/science.adm8671. [DOI] [PubMed] [Google Scholar]
  35. Lee-Yaw J.A., McCune J.L., Pironon S., et al. Species distribution models rarely predict the biology of real populations. Ecography. 2021;6 [Google Scholar]
  36. MacArthur R.H. Harper & Row; New York: 1972. Geographical Ecology: Patterns in the Distribution of Species. [Google Scholar]
  37. Maguire B. Niche response structure and the analytical potentials of its relationship to the habitat. Am. Nat. 1973;107:213–246. [Google Scholar]
  38. Mammola S. Assessing similarity of n-dimensional hypervolumes: which metric to use? J. Biogeogr. 2019;46:2012–2023. [Google Scholar]
  39. Martínez-Meyer E., Díaz-Porras D., Peterson A.T., et al. Ecological niche structure and rangewide abundance patterns of species. Biol. Lett. 2013;9 doi: 10.1098/rsbl.2012.0637. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. McGill B.J. Trees are rarely most abundant where they grow best. J. Plant Ecol. 2012;5:46–51. [Google Scholar]
  41. McGill B.J. Matters of scale. Science. 2010;328:575–576. doi: 10.1126/science.1188528. [DOI] [PubMed] [Google Scholar]
  42. Mertes K., Jetz W. Disentangling scale dependencies in species environmental niches and distributions. Ecography. 2018;41:1604–1615. [Google Scholar]
  43. Mori A.S. Local and biogeographic determinants and stochasticity of tree population demography. J. Ecol. 2019;107:1276–1287. [Google Scholar]
  44. Moutouama J.K., Gaoue O.G. Altitude-mediated soil properties, not geography or climatic distance, explain the distribution of a tropical endemic herb. Ecol. Evol. 2022;12 doi: 10.1002/ece3.8572. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Nunez-Mir G.C., Walter J.A., Grayson K.L., et al. Assessing drivers of localized invasive spread to inform large-scale management of a highly damaging insect pest. Ecol. Appl. 2022;32 doi: 10.1002/eap.2538. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Oksanen J., Minchin P.R. Continuum theory revisited: what shape are species responses along ecological gradients? Ecol. Model. 2002;157:119–129. [Google Scholar]
  47. Osorio-Olvera L., Yañez-Arenas C., Martínez-Meyer E., et al. Relationships between population densities and niche-centroid distances in North American birds. Ecol. Lett. 2020;23:555–564. doi: 10.1111/ele.13453. [DOI] [PubMed] [Google Scholar]
  48. Ovaskainen O., Abrego N. Cambridge University Press; 2020. Joint Species Distribution Modelling: with Applications in R. [Google Scholar]
  49. Pearson R.G., Dawson T.P. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecol. Biogeogr. 2003;12:361–371. [Google Scholar]
  50. Peterson A.T., Soberón J., Pearson R.G., et al. Princeton University Press; Princeton: 2011. Ecological Niches and Geographic Distributions. [Google Scholar]
  51. Pironon S., Papuga G., Villellas J., et al. Geographic variation in genetic and demographic performance: new insights from an old biogeographical paradigm: the centre-periphery hypothesis. Biol. Rev. 2017;92:1877–1909. doi: 10.1111/brv.12313. [DOI] [PubMed] [Google Scholar]
  52. Qian H., Kissling W.D. Spatial scale and cross-taxon congruence of terrestrial vertebrate and vascular plant species richness in China. Ecology. 2010;91:1172–1183. doi: 10.1890/09-0620.1. [DOI] [PubMed] [Google Scholar]
  53. R Core Team . R Foundation for Statistical Computing; 2023. R: a Language and Environment for Statistical Computing. [Google Scholar]
  54. Ren H., Condit R., Chen B., et al. Geographical range and local abundance of tree species in China. PLoS One. 2013;8 doi: 10.1371/journal.pone.0076374. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Sagarin R.D., Gaines S.D. The “abundant centre” distribution: to what extent is it a biogeographical rule? Ecol. Evol. 2002;5:137–147. [Google Scholar]
  56. Sagarin R.D., Gaines S.D., Gaylord B. Moving beyond assumptions to understand abundance distributions across the ranges of species. Trends Ecol. Evol. 2006;21:524–530. doi: 10.1016/j.tree.2006.06.008. [DOI] [PubMed] [Google Scholar]
  57. Santini L., Pironon S., Maiorano L., et al. Addressing common pitfalls does not provide more support to geographical and ecological abundant-centre hypotheses. Ecography. 2019;42:696–705. [Google Scholar]
  58. Schoener T.W. Nonsynchronous spatial overlap of lizards in patchy habitats. Ecology. 1970;51:408–418. [Google Scholar]
  59. Sexton J.P., McIntyre P.J., Angert A.L., et al. Evolution and ecology of species range limits. Annu. Rev. Ecol. Evol. Syst. 2009;40:415–436. [Google Scholar]
  60. Smith A.B., Santos M.J. Testing the ability of species distribution models to infer variable importance. Ecography. 2020;43:1801–1813. [Google Scholar]
  61. Soberón J. Grinnellian and Eltonian niches and geographic distributions of species. Ecol. Lett. 2007;10:1115–1123. doi: 10.1111/j.1461-0248.2007.01107.x. [DOI] [PubMed] [Google Scholar]
  62. Soberón J., Nakamura M. Niches and distributional areas: concepts, methods, and assumptions. Proc. Natl. Acad. Sci. U.S.A. 2009;106:19644–19650. doi: 10.1073/pnas.0901637106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Sporbert M., Keil P., Seidler G., et al. Testing macroecological abundance patterns: the relationship between local abundance and range size, range position and climatic suitability among European vascular plants. J. Biogeogr. 2020;47:2210–2222. [Google Scholar]
  64. Stein A., Kreft H. Terminology and quantification of environmental heterogeneity in species-richness research. Biol. Rev. 2015;90:815–836. doi: 10.1111/brv.12135. [DOI] [PubMed] [Google Scholar]
  65. VanDerWal J., Shoo L.P., Johnson C.N., et al. Abundance and the environmental niche: environmental suitability estimated from niche models predicts the upper limit of local abundance. Am. Nat. 2009;174:282–291. doi: 10.1086/600087. [DOI] [PubMed] [Google Scholar]
  66. Vellend M. Conceptual synthesis in community ecology. Q. Rev. Biol. 2010;85:183–206. doi: 10.1086/652373. [DOI] [PubMed] [Google Scholar]
  67. Waldock C., Stuart-Smith R.D., Edgar G.J., et al. The shape of abundance distributions across temperature gradients in reef fishes. Ecol. Lett. 2019;22:685–696. doi: 10.1111/ele.13222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Wang Z., Fang J., Tang Z., et al. Patterns, determinants and models of woody plant diversity in China. Proc. R. Soc. Lond. B-Biol. Sci. 2011;278:2122–2132. doi: 10.1098/rspb.2010.1897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Weber M.M., Stevens R.D., Diniz-Filho J.A.F., et al. Is there a correlation between abundance and environmental suitability derived from ecological niche modelling? A meta-analysis. Ecography. 2017;40:817–828. [Google Scholar]
  70. Williams H.M., Siegrist J., Wilson A.M. Support for a relationship between demography and modeled habitat suitability is scale dependent for the purple martin Progne subis. J. Anim. Ecol. 2021;90:356–366. doi: 10.1111/1365-2656.13369. [DOI] [PubMed] [Google Scholar]
  71. Wiens J.A. Spatial scaling in ecology. Funct. Ecol. 1989;3:385–397. [Google Scholar]
  72. Williams J.W., Shuman B.N., Webb III T., et al. Late-quaternary vegetation dynamics in North America: scaling from taxa to biomes. Ecol. Monogr. 2004;74:309–334. [Google Scholar]
  73. Woodward F.I. Cambridge University Press; Cambridge: 1987. Climate and Plant Distribution. [Google Scholar]
  74. Woudenberg S.W., Conkling B.L., O'Connell B.M., et al. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station; Fort Collins, CO: 2010. The Forest Inventory and Analysis Database: Database Description and Users Manual Version 4.0 for Phase 2 (General Technical Reports No. RMRS-GTR-245) [Google Scholar]
  75. Xing D., Swenson N.G., Weiser M.D., et al. Determinants of species abundance for eastern North American trees. Global Ecol. Biogeogr. 2014;23:903–911. [Google Scholar]
  76. Yancovitch Shalom H., Granot I., Blowes S.A., et al. A closer examination of the ‘abundant centre’ hypothesis for reef fishes. J. Biogeogr. 2020;47:2194–2209. [Google Scholar]
  77. Yañez-Arenas C., Martínez-Meyer E., Mandujano S., et al. Modelling geographic patterns of population density of the white-tailed deer in central Mexico by implementing ecological niche theory. Oikos. 2012;121:2081–2089. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1
mmc1.docx (7.6MB, docx)

Data Availability Statement

Data and code are available on Zenodo at https://zenodo.org/records/17188563.

Articles from Plant Diversity are provided here courtesy of Kunming Institute of Botany, Chinese Academy of Sciences

RESOURCES