Niche overlap with an exotic competitor mediates the abundant niche‐centre relationship for a native lady beetle

The abundant niche‐centre hypothesis has been used to describe patterns of species abundance relative to position in ecological niche space. Such relationships, however, are inconsistently recovered and may be obscured due to non‐equilibrium distributions, such as those caused novel interactions with exotic species. Here, we explore patterns of fitness for the nine‐spotted lady (Coccinella novemnotata) following the introduction of an exotic competitor, the seven‐spotted lady beetle (C. septempunctata). We examine how niche overlap between species may be causing a range reduction for C. novemnotata through a modified abundant niche‐centre relationship.


| INTRODUC TI ON
Human-induced pressures including changing climates, altered landscapes or exotic species have all resulted in reduced local abundance and altered distributions for many taxa (McKinney & Lockwood, 1999;Neff et al., 2022).These ecological changes associated with the Anthropocene highlight a need for understanding how species will respond to changing environmental conditions.
Importantly, these factors are likely to interact.As an example, shifting species' distributions tracking changing climates may lead to new or modified biotic lateral interactions (Urban et al., 2012) or bottom-up interactions (Damien & Tougeron, 2019).Predicting how species fitness and abundance patterns respond to novel and interacting pressures will help ensure the potential for long-term species persistence in the face of a changing environment.
Past introductions of exotic species may offer insight into how species will react to changing environments.Novel interactions that occur when species are moved beyond their endemic range can result in altered native species distributions and fitness reductions caused by a competitive advantage held by the exotic species (Callaway & Ridenour, 2004;Sexton et al., 2009).Such modifications to geographical distributions reflect changing occupancy within a species' ecological niche (Broennimann et al., 2007;Martínez-Meyer et al., 2013;Pineda-Munoz et al., 2021).Beyond occurrence, changing occupancy in an ecological niche space has been viewed in a less binary fashion (i.e.predicting presence/absence) and instead viewed as a determinant of species' performance.Recent studies have suggested that a species' niche space is not uniform, with increased abundance and fitness found in conditions nearer to the centre of a species ecological niche (Osorio-Olvera et al., 2019;Osorio-Olvera, Yañez-Arenas, et al., 2020).The abundant niche-centre hypothesis, alternatively called distance-abundance relationships (Sagarin & Gaines, 2002), centre-periphery hypothesis (Pironon et al., 2017) or central-marginal hypothesis (Eckert et al., 2008), predicts a dynamic pattern of species abundance across a geographic range as predicted by the position in ecological niche space.
Studies examining abundant niche-centre relationships, however, have recovered inconsistent responses due to methodologies that fail to reflect dispersal limitations or account for Allee effects or unfilled fundamental niche space (Dallalio et al., 2017;Osorio-Olvera et al., 2019;Santini et al., 2019;Yañez et al., 2020).Dallas and Santini (2020) showed in simulations that even when niche centroid relationships occur, they may be masked by demographic stochasticity, environmental heterogeneity or landscape structure.These findings suggest that if we can account for variance occurring at local scales, then niche centroid relationships occurring at regional scales may become more apparent.Local biotic interactions may play an important role in refining these relationships if the species involved occupy increasingly similar niche space and compete for shared resources (Espindola et al., 2022).Novel biotic interactions due to exotic species introductions or intersecting ranges due to climate change may thus alter measured historic abundance niche-centre relationships.
Here, we examine the role of ecological niche overlap and the abundant niche-centre hypothesis in explaining the North American range reduction of the nine-spotted lady beetle (Coleoptera: Coccinella novemnotata Herbst).Previously among the most encountered lady beetle species in North America, it now occupies a fraction of its previous range.The timing of this range reduction is consistent with the introduction of the seven-spotted lady beetle (Coleoptera: Coccinella septempunctata Linnaeus), with community replacement of C. novemnotata by C. septempunctata common (Wheeler & Hoebeke, 1995).Both species are holometabolous, multivoltine species that overlap broadly in terms of ecology and behaviour.Both are considered habitat generalist, aphidophagous species that feed on the same small, soft-bodied insect prey (Wheeler & Hoebeke, 1995).
Coccinella septempunctata also develops at a higher rate creating a relative size disparity between larvae of the two species when raised separately under the same conditions (Ugine & Losey, 2014).
This size disparity could lead to competitive asymmetry even if interactions are infrequent.
Importantly, the outcomes of these competitive interactions are not consistent across the distribution of C. novemnotata.Eastern North American populations are greatly reduced and populated with significantly smaller individuals (Losey et al., 2012), while Western populations appear less effected in terms of beetle size (Evans, 2017).Although these species overlap in terms of resource and habitat use, incomplete overlap in the range of physiological tolerances to environmental factors in the absence of biotic interactions (i.e.Grinnellian ecological niche; Espindola et al., 2019;Soberón & Arroyo-Peña, 2017) may help explain the current differential success of C. novemnotata in North America.
In this study, we examine the relationship between beetle size, here used as a proxy for species' fitness and potential species abundance, and ecological niche overlap.We conducted an extensive review of natural history collections to recover the distribution of C. novemnotata prior to the influence of C. septempunctata.We then used these data to approximate the species' realized Grinnellian ecological niche based on abiotic factors.A time-oriented view of the C. novemnotata niche was constructed that would allow us to view potential niche overlap with C. septempunctata over time.We asked whether beetle size declines were a function of niche overlap between the species, and whether C. novemnotata body size, both before and after the introduction of C. septempunctata, was influenced by the distance from its niche centroid in ecological space.
with a stated level of uncertainty of less than 100 m, were retained.
Finally, all location data were georeferenced using GEOLocate v.

| Data analysis
Prior to analysis, bias in each occurrence data set was examined and corrected.Potential sampling or geographic bias leading to spatial autocorrelation was examined with the mantel r value implemented in ecospat (Di Cola et al., 2017) in R (R Core Team, 2017).Spatial thinning was then carried out in SDMtoobox 2.0 (Brown et al., 2017) at the distance indicated in the correlogram, and then rechecked for spatial autocorrelation until no significant autocorrelation was detected.
We utilized a model selection approach using minimum volume ellipsoids to select a final set of environmental variables to best characterize the ecological niche of C. novemnotata.Analyses were conducted using ntbox in R (Osorio-Olvera, Lira-Noriega, et al., 2020).
The historic data set was chosen for variable selection because the current distribution of C. novemnotata in North America is likely in a non-equilibrium state due to the influence of C. septempunctata.
Our goal was to capture the best estimate of the fundamental niche, including current records in this process would likely introduce bias into this process.Multicollinearity issues were first examined by estimating Spearman correlations between all environmental variables.We retained only variables with correlations below .85.We then randomized the historic C. novemnotata occurrences into a 70% set for training and a 30% set for testing.The 'ellipsoid selection tool' was used under default parameters to calibrate models with all combinations of environmental variables with 0.90, 0.95, and 0.99 niche points included inside the ellipsoid.A single best model representing a reduced set of variables was retained based on (i) a significant value of partial ROC test (Peterson et al., 2008), (ii) having a mean omission rate for the training and tested data that ≥5%, and (iii) ranked by maximum environmental background area under the receiver operating characteristic (AUC).These final set of variables were used in all subsequent analyses.Niche similarity was assessed using humboldt in R (Brown & Carnaval, 2019).Schoener's D niche similarity index (Warren et al., 2008) was used to assess differences in niche overlap and niche divergence for C. novemnotata (historic vs. current ranges) and between C. novemnotata and C. septempunctata (historic vs. exotic; current vs. exotic).The values of Shoener's D range from 0 indicating completely dissimilar niches, to 1 indicating completely similar niches.Significance was evaluated in terms of niche equivalence, using a one-tailed test that evaluates the null hypothesis that the two measured niches are equivalent.Significance was evaluated by resampling from the respective backgrounds and computing Schoener's D between these randomly derived locations.These processes were repeated 1000 times to develop a random distribution by which the actual overlap was compared.We examined niche similarity in terms of both a niche overlap test (NOT), which asks how equivalent the two species' occupied niches are based on total accessible environmental space, and a niche divergence test (NDT) which asks if the species' niches are equivalent given a common background.A background test was then used in the NOT and NDT analyses to examine if the distributions of the two test groups were more or less different than would be expected given the environmental differences present in the conditions available to them (i.e., background).(DNC) and an interaction term describing the impact of C. septempunctata presence on the C. novemnotata niche centroid relationship.Bestfit models were selected for and evaluated using Akaike's information criterion adjusted for small sample sizes (AICc).All statistical analyses were conducted using MuMIn in R (Barton, 2009).Final models were checked via a Q-Q plot for normality, by examining a normality plot of residuals, and assessed for homoscedasticity via the non-constant error variance test using car in R (Fox & Weisberg, 2019).

| DISCUSS ION
Coexistence between species with strongly overlapping ecological niches can be accomplished through one, or both, species shifting   Avoiding niche overlap with C. septempunctata appears key in explaining the realized distribution of C. novemnotata.Following the Eltonian Noise Hypothesis (Soberón & Nakamura, 2009), biotic interactions are thought to shape demographics at fine spatial scales, while abiotic variables should primarily influence coarse-grained broad-scale geographical distribution.We find the latter to not only be true but to also highlight the impact of niche overlap in affecting geographical distributions.These results support a growing body of evidence illustrating the importance of local biotic interactions in shaping geographic distributions (Atauchi et al., 2018;Jenkins et al., 2020;Simões & Peterson, 2018).While the direct interactions between these beetle species have a clear negative impact for C. novemnotata at local scales, the strength of these interactions may, however, be context-dependent and mediated by niche overlap.
Ecological niche overlap-mediated competitive exclusion is strongly suggested for the non-equilibrium distribution for C. novemnotata in North America.Importantly, the outcomes of cooccurrence observed locally in geographical space appear to be also displayed in ecological niche space.In both cases co-occurrence leads to decreased C. novemnotata fitness and exclusion.Locally, when introduced into a common environment, C. septempunctata has a documented competitive advantage over C. novemnotata through both interference and exploitation competition (Hoki et al., 2014;Tumminello et al., 2015;Turnipseed et al., 2014Turnipseed et al., , 2015;;Yasuda et al., 2004), resulting in exclusion (Wheeler & Hoebeke, 1995) and reduced fitness (Losey et al., 2012).As co-occurrence decreases in local geographical space due to refugia (Alyokhin & Sewell, 2004;Bahlai et al., 2015;Evans, 2017)  Competitive outcomes between species are known to be mediated by a variety of extrinsic factors, including parasites (Price et al., 1988), herbivore presence (Sakata & Craig, 2021;White et al., 2006), chemicals (Burkepile et al., 2006) and microbe presence (Hodge & Fitter, 2013).Position in ecological niche space, and in particular the degree of overlap present may similarly impact competitive interactions.Temperature-dependent competitive outcomes have been recovered for both Plethodon salamanders (Dallalio et al., 2017) and Drosophila fruit flies (Comeault & Matute, 2021).
Classical theory suggests that species fitness, and subsequent population density, should be greatest when ecological conditions that define the fundamental niche are optimized (Brown, 1984).Not accounting for these factors effectively removes both biotic and movement criteria necessary when considering abundant nichecentre relationships (Yañez et al., 2020).
Spatial distributions of newly interacting species may be mediated by niche overlap, while niche centroids of these species may play a critical role in predicting species displacement.Changes associated with the Anthropocene are expected to introduce new competitive interactions as previously allopatric species enter novel species' competitive environments (Alexander et al., 2015).While local incidence and abundance will be determined by these local interactions, geographical distributions may also be impacted through mediation by niche overlap.Understanding the eventual impact to vulnerable species may be better predicted with an increased understanding of ecological niche mediated abundant niche-centre relationships.
3.22(Rios & Bart, 2010) with accuracy metrics established at less than or equal to 100 m.This work required a time-oriented view of C. novemnotata occurrence in North America.Occurrence records were classified as occurring before or after the introduction of C. septempunctata.To do this we used the first detection of C. septempunctata within a particular state or province in the United States and Canada, respectively (Appendix), to delineate the C. novemnotata occurrences into pre-introduction (historic) data set and a post-introduction (current) dataset.Coccinella septempunctata was widely introduced into North America, and first detection in a state or province does not necessarily indicate species establishment.Identifying the point of establishment, however, is not possible across the entire exotic range.First detection is used here as it offers a uniform delineation between no possible interaction (historic) and potential interaction (current) between species.The C. septempunctata data set was represented by occurrences consisting of all North America records (exotic).We expressed the ecological niche of the two Coccinella species based on 19 climate variables constructed using the BIOCLIM WorldClim Global Climate Dataset (www.world clim.org) representing averages, extremes and seasonality in temperature and precipitation.Climate variables represent averages of temperature and precipitation for 1970-2000 with a spatial resolution of 5 km(2.5 arc-min)(Fick & Hijmans, 2018).
Adult beetle size is a strong approximation of potential reproductive output and potential population size (i.e.fitness;Kajita & Evans, 2010).Hemptinne et al. (2022) recovered a strong correlation (r = .7520;p > .01) between coccinellid beetle body mass and potential reproductive rate.Losey et al. (2012) found dorsal body area (l × w) to be strongly correlated with beetle volume (r 2 = .988)and was a strong predictor of prey consumption (r 2 = .81;p > .01).The size of C. novemnotata individuals was, therefore, used as a measure of potential fitness by measuring the dorsal area of collected specimens.Beetle area was measured for available C. novemnotata specimens by viewing the dorsum of preserved collected specimens and calculating its dorsal area (mm 2 , length × width).Length was measured from the anterior tip of the head to the posterior tip of the elytra.Width was measured as the maximum width across the elytra.Measurements were taken using a digital camera or microscope micrometre, both with an accuracy to 0.01 mm and measured using Motic Images Plus 2.0 (Motic China Group Co., Ltd.).The effect of exotic species introduction on the C. novemnotata abundant niche-centre relationship was then examined by relating beetle size to the distance from its ecological niche centroid.We calculated Mahalanobis distance(Mahalanobis, 1936) through convex hull modelling to establish minimum volume ellipsoids using ntbox in R based on occurrences in the historic C. novemnotata data set.Generalized linear models were used to test relationships between size and distance to the niche centroid.Beetle size (mm 2 ) was used as the response variable.Predictor variables included temporal C. novemnotata groupings (current or historic) separated by the present of C. septempunctata (C7), the distance to the C. novemnotata niche centroid After spatial rarefaction, the final data sets contained 505 historic and 66 current occurrences of C. novemnotata, and 1703 exotic occurrences of C. septempunctata.Final environmental variable selection through the minimum volume ellipsoid selection process resulted in 741 total models tested and 183 significant candidate models.A single best model was selected with a significant partial ROC test (p < .01),low omission rate of testing and training data (.04), and having the highest environmental background AUC (.86).The final environmental set included five variables, including annual mean temperature (bio1), maximum temperature of the warmest month (bio5), temperature annual range (bio7), mean temperature of the driest quarter (bio9) and annual precipitation (bio12).Niche similarity tests provide strong evidence that the ecological niche of Coccinella novemnotata has diverged since the introduction of C. septempunctata into North America (Figure 1, Table 1).Niche similarity, in terms of both overlap (D = 0.46, p = .02)and divergence (D = 0.42, p = .01),suggests C. novemnotata is currently occupying non-equivalent niche space compared to its historic distribution.Background tests indicate that the niche space measured from the F I G U R E 1 Intraspecific histogram density plots for the current versus historic range of Coccinella novemnotata (a), exotic range of C. septempunctata versus historic range of C. novemnotata (b) and exotic range of C. septempunctata versus the current range of C. novemnotata (c).The density of each species data set in environmental space is displayed along with lines representing the kernel density isopleths from 1% to 100% kernel densities (red = sp1 and blue = sp2).
current distribution has significant overlap with the historic niche (p = .07);however, the historic niche space does not have significant overlap with the current niche (p = .04).This background test result suggests niche exclusion from previously occupied habitat (seeBrown & Carnaval, 2019).The exotic range of C. septempunctata was found to be occupying non-equivalent niche space to both the historic (D = 0.28, p = .01)and current (D = 0.15, p = .04)distributions of C. novemnotata.A change in Schoener's D suggests that the current niche space occupied by C. novemnotata has decreased overlap with the ecological niche of C. septempunctata.The changes illustrated here suggest a change in the current realized niche of C. novemnotata that decreased niche overlap with C. septempunctata but does not likely indicate a change to its fundamental niche.Beetle size varied across the species' geographical range, ranging from 16.48 to 76.40 mm 2 (average = 40.41mm 2 ; n = 466).A single best model (p < .01;Adj R 2 = .43)relating beetle size to the distance from the niche centroid was recovered.The non-constant variance test indicated homoscedastic error variance (χ 2 = 3.66, df = 1, p = .06)with normally distributed model residuals.A significant shift in the size-niche-centre relationship based on the introduction of C. septempunctata into North America was recovered (Table 2).Historically, larger beetles were observed in conditions approaching the species' niche centroid and smaller beetles were observed as the Mahalanobis distance increased (β = −3.60,p < .01, Figure 2).After the introduction of C. septempunctata there was a significant shift with the smallest beetles occurring near the niche centroid and largest beetles found at larger Mahalanobis distances (β = 2.58, p < .01, Figure 2).
behaviourally, temporally or geographically to avoid overlap.Our results indicate that niche exclusion due to overlap with C. septempunctata is the likely cause of decline for C. novemnotata in North America.As the geographic range of C. septempunctata expanded across North America it came to occupy similar environmental space to that of C. novemnotata, resulting in a two-fold outcome.First, a significant change to the abundant niche-centre relationship was observed for C. novemnotata.Beetle size and potential fitness decreased across formerly optimal conditions.Next, C. novemnotata experienced a range reduction that, when viewed in environmental TA B L E 1 Niche similarity between Coccinella novemnotata (C9) and C. septempunctata (C7) species pairs.

F I G U R E 2
Interaction plot showing the effect of Coccinella septempunctata occurrence on C. novemnotata body size relative to the distance from its niche centroid (pre-C.septempunctata = orange; post-C.septempunctata = blue).space, resulted in decreased niche overlap with C. septempunctata.This range is not equivalent to its historic range and suggests a nonequilibrium distribution due to exclusion.By observing these species in a time-oriented fashion we were able to relate the decline of a native species across a broad geographical range to niche exclusion by an exotic competitor.Niche overlap with C. septempunctata does not appear to uniformly impact the size and subsequent distribution of C. novemnotata in North America.When viewed in environmental space, C. novemnotata is absent from conditions nearest its niche centroid, displays decreased size at distances approaching its centroid, and is less impacted at the niche periphery.These results support the findings of both Losey et al. (2012) who found significantly smaller C. novemnotata individuals in areas of range restriction, and Evans (2017) who recorded no size reductions from geographic areas with less range restriction.Now excluded from optimal conditions, C. novemnotata now persists at the niche margins and displays an inverse abundant niche-centre relationship.The demarcation presented by the introduction of C. septempunctata appears to present a novel biotic pressure.
e. climate) competition between these two Coccinella species could explain the significant shifts in both the abundant niche centroid relationship and in ecological niche space that reduces overlap.Other niche dimensions detailing biotic and movement factors may additionally play a role.Decreased natural habitat and increased agricultural habitat have been shown to enhance the exploitative competitive advantage of alien species over other native coccinellid species While a positive relationship between fitness and environmental suitability should serve as a null hypothesis (seeOsorio-Olvera et al., 2019), these patterns may be difficult to recover, particularly if the current distribution based on the realized niche is not an accurate approximation of the fundamental niche.The niche overlap informed abundant niche-centre relationship recovered here illustrates how patterns may be obscured if additional niche dimensions are not accounted for.Novel interactions between competing species can lead to fitness reductions and displacement.Both outcomes decrease a species capacity to effectively fill ecological niche space.

Species group (1 vs. 2) Niche overlap (NOT) Niche divergence (NDT)
Note: D refers to Schoener's D niche similarity metric, E is the equivalence statistic and B refers to the background statistic for pairwise comparisons (historic = before C7 introduction; current = before C7 introduction; exotic = North America range).Model selection table describing Coccinella novemnotata body size (Area) as a function of Mahalanobis distance (DNC) occurring post-or pre-introduction of C. septempunctata (C7).
Note: Bolded text represents the best fit model.