New Distributional Records and Characterization of the Climatic Niche of Lepturges (Lepturges) limpidus Bates, 1872 (Coleoptera, Cerambycidae): Sink or Source Population?

Simple Summary Longhorned beetles (Coleoptera: Cerambycidae) are one of the most diverse, ecologically and economically important groups of beetles in the world. In the Neotropical Region, more than 8000 species are known and new records are added each year, linked to increased international trade, reduced transport times, and an increase in the number of potential vectors. In this study, we report Lepturges (Lepturges) limpidus as an exotic species in Argentina and discuss the potential drivers of its expansion through modeling to help understand and predict the possible geographic distributions of the species as a function of environmental factors. In addition, in view of the results obtained, we discuss the relevance of biomonitoring programs and field studies to detect the arrival of potential invasive alien species in the region. Abstract A growing number of cases of the spread and establishment of non-native species outside their previously known ranges has been reported in recent years. Here we report new distributional records of Lepturges (Lepturges) limpidus Bates, 1872 (Cerambycidae) from Argentina and investigate whether these records could represent established populations. We constructed ellipsoid envelope models to characterize climatic niches of L. limpidus, identified areas of climatic suitability, investigated the status of new records as climatic outliers, and evaluated its dependency on its known hostplant as a limiting factor for the beetle distribution. Results indicate widespread climatic suitability in the Neotropical Region, and new records are not outliers with regard to the climatic profile of L. limpidus. Association with its known hostplant is non-dependent, indicating that the species might utilize different hosts plants. New records likely represent established populations, but targeted surveys should be carried out to detect new arrivals and enable the installation of mitigation and control measures.


Introduction
Longhorn beetles (Coleoptera: Cerambycidae) are one of the largest, most diverse, ecologically and economically important groups of beetles [1]. The family comprises more than 38,000 described species worldwide, which are all phytophagous, xylophagous or saproxylophagous [2]. Their larvae consume a great variety of plant parts and tissues, in diverse conditions, from alive and healthy to dead and decaying [3]. Adaptation to this large spectrum of host plant conditions has resulted in great variation in the behavior and ecology of these beetles, whereby many species are considered important pests of agricultural crops, ornamental trees, forestry, and timber products [3][4][5].
New records were obtained during field work performed by the first author (NV), in November 2016 and March 2021. These specimens were deposited at Universidad Nacional del Nordeste, Corrientes, Argentina (CARTROUNNE). Species identification was based on comparison of collected specimens with images in the photographic catalog of the Cerambycidae of the New World [54] and following descriptions provided by [42]. Finally, the identifications were confirmed by the specialist Dra. Marcela Monné (Museo Nacional/UFRJ, Rio de Janeiro, Brazil). The new records from Argentina were not included in the modeling analysis. Environmental data. Environmental data for this study were obtained at 2.5 arc-minute (~4.6 km at the equator) spatial resolution from WorldClim (version 1.4, http://www.worldclim.org (accessed on 1 April 2022) [55]). WorldClim is based on interpolations of weather station data (i.e., monthly precipitation and minimum and maximum temperatures) over the period 1950-2000. From the 19 variables available, we excluded four (mean temperature of wettest quarter, mean temperature of driest quarter, precipitation of warmest quarter, precipitation of coldest quarter) a priori due to known spatial artifacts between adjacent grid cells [56,57]. To avoid overfitting, overly dimensional environmental space, and collinearity among variables, we cropped the variables to the extent of the Neotropical Region and performed principal component analysis (PCA) using the function kuenm_rpca in the package 'kuenm' [58] in R, version 4.0.3 (R Core Team, Vienna, Austria) [59]. To build the ellipsoid niche model of L. limpidus we retained the first three components for model calibration. These explained cumulatively > 88% of the total variance in the dataset (see Supplementary Material Table S1). Following [60], to calculate the ellipsoid niche overlap between L. limpidus and C. fragrans, we tested three distinct environmental sets seeking to avoid bias regarding the combination of variables used to characterize the species niche centrality: 'set 1' included all 15 variables; 'set 2' included only temperature variables (i.e., annual mean temperature; mean diurnal range; isothermality; temperature seasonality; max. temperature of warmest month; min. temperature of coldest month; temperature annual range; mean temperature of warmest quarter and mean temperature of coldest quarter); and 'set 3' included only precipitation variables (i.e., annual precipitation; precipitation of wettest month; precipitation of driest month; precipitation seasonality; precipitation of wettest quarter; precipitation of driest quarter). As a result, we retained the first three components of each set, which explained cumulatively > 90% of the total variance in the dataset for model calibration (see Supplementary  Material Table S2). Environmental data. Environmental data for this study were obtained at 2.5 arc-minute (~4.6 km at the equator) spatial resolution from WorldClim (version 1.4, http://www. worldclim.org (accessed on 1 April 2022) [55]). WorldClim is based on interpolations of weather station data (i.e., monthly precipitation and minimum and maximum temperatures) over the period 1950-2000. From the 19 variables available, we excluded four (mean temperature of wettest quarter, mean temperature of driest quarter, precipitation of warmest quarter, precipitation of coldest quarter) a priori due to known spatial artifacts between adjacent grid cells [56,57]. To avoid overfitting, overly dimensional environmental space, and collinearity among variables, we cropped the variables to the extent of the Neotropical Region and performed principal component analysis (PCA) using the function kuenm_rpca in the package 'kuenm' [58] in R, version 4.0.3 (R Core Team, Vienna, Austria) [59]. To build the ellipsoid niche model of L. limpidus we retained the first three components for model calibration. These explained cumulatively > 88% of the total variance in the dataset (see Supplementary Material Table S1). Following [60], to calculate the ellipsoid niche overlap between L. limpidus and C. fragrans, we tested three distinct environmental sets seeking to avoid bias regarding the combination of variables used to characterize the species niche centrality: 'set 1' included all 15 variables; 'set 2' included only temperature variables (i.e., annual mean temperature; mean diurnal range; isothermality; temperature seasonality; max. temperature of warmest month; min. temperature of coldest month; temperature annual range; mean temperature of warmest quarter and mean temperature of coldest quarter); and 'set 3' included only precipitation variables (i.e., annual precipitation; precipitation of wettest month; precipitation of driest month; precipitation seasonality; precipitation of wettest quarter; precipitation of driest quarter). As a result, we retained the first three components of each set, which explained cumulatively > 90% of the total variance in the dataset for model calibration (see Supplementary Material Table S2).
Ellipsoid niche model. To characterize the environmental niches of L. limpidus and C. fragans we used ellipsoids as climatic niche models. Ellipsoid niches have been hypothesized to provide a close approximation to a fundamental niche [61][62][63] and assume that the ecological niche of a species has only one optimum (i.e., niche centroid; see [62,64,65]). Therefore, this is an optimal method for interpreting the multidimensional parameter space of habitat variation and roughly categorizing records as belonging to source or sink populations [60].
Models were built using the 'ellipsenm' package [66], calibrated using the 95% pairwise confidence region for the ellipsoid, and evaluated using the function 'ellipsoid_calibration' [66]. Two different methods were employed to construct ellipsoid models: (1) 'covmat', which creates ellipsoids based on the centroid and a matrix of co-variances of the variables and (2) 'mve1', which generates an ellipsoid that reduces the volume contained in it without losing the data contained (i.e., minimum volume ellipsoid; [67]). Best-model selection was based on statistical significance (partial ROC; [68]); the proportion of testing data known to be in suitable areas and prediction of unsuitable areas was based on omission rates (E = 5%; [69]) and prevalence (i.e., proportion of space identified as suitable for the species; [66]). To calculate the partial ROC metric, we used 500 bootstrap iterations with 50% of testing data to be used in each bootstrapped process with 5% of testing data error due to uncertainty. Prevalence was calculated in geographical and environmental space. In geographic space, proportions were calculated using all pixels from the raster, while in environmental space only pixels that have distinct combinations of all variable values were considered [63,66]. The calibration area (i.e., region accessible to the species; [70]) included a buffer of 50 km from the occurrence records included in our models. The buffer size was defined based on the known dispersal ability of species of longhorn beetle subfamily Lamiinae [71] and on previous studies on the distribution of the hostplant (i.e., [72,73]). Final parameters were selected based on the best evaluated models and used to create the final models using 10 replicates with bootstrapped subsamples of 75% of the data using the function 'ellipsoid_model' available in the 'ellipsenm' R package [66]. The replicates were produced by excluding one occurrence record at a time. We represented ecological niche and suitability levels of L. limpidus in geographic space, which were binarized using a threshold for suitability aiming to exclude the 5% of the data with the most extreme values. Visualization of results was performed in QGIS (version 3.10, QGIS Development Team, http://www.qgis.org (accessed on 22 July 2022) [74]).
Spatial error analysis. Suitability values in ellipsoid envelope models represent the Mahalanobis distance from the optimum (i.e., ellipsoid centroid), while maximum values will be close to the centroid and minimum values will be close to the periphery of the ellipsoid [66]. Thus, to verify whether the new records in Northern Argentina represent outliers in the environmental space, we extracted values for all known occurrence points of L. limpidus from its predicted environmental suitability, using the mean of replicates. Then we visually inspected the position of the new records within the distribution of Mahalanobis distance values of records of the known distribution of L. limpidus.
Ellipsoid niche overlap. To test whether the hostplant might be constraining L. limpidus distribution, ellipsoid niche overlap was calculated considering the union of the environmental conditions relevant for both species, L. limpidus and C. fragans. The process was repeated 1000 times, and the observed overlap values were compared to the values found for pairs of random ellipsoids. The null hypothesis is that the niche of L. limpidus is contained within the C. fragans ellipsoid, and if the observed values are as extreme or more extreme than the lower limit of the values found for the random ellipsoids, the null hypothesis is rejected. That is, if the observed overlap is lower than the 95% of the random-derived values of overlap, the niches are considered non-overlapped. If the observed values cannot be distinguished from the random, the null hypothesis cannot be rejected. A p-value and the predefined confidence limit is added to the overlap matrix when the test is performed.
Examined material. Ellipsoid niche model of L. limpidus. The geographic projections of the ecological niche of L. limpidus showed widespread suitable areas across the Neotropics, including northern and central Argentina. Specifically, climatic suitability was widespread within the Chaco biogeographic province and concentrated mainly on the Eastern Chaco district, where new records for L. limpidus are located ( Figure 1A). The best fitting method to construct the climatic ellipsoids was 'covmat'; mean AUC, p-value of partial ROC and omission rates were significantly better than random expectations (p-value < 0.05; Table 2). Prevalence of mean ellipsoidal models in geographical (G-space) and environmental (E-space) space were relatively high (0.712; Table 2). The complete specifications of ellipsoid characteristics (e.g., centroid, covariance matrix, semi-axes length, etc.) can be found in Supplementary Material Table S3.    Ellipsoid niche overlap of L. limpidus and C. fragans. We rejected the hypothesis that L. limpidus is contained within the niche of C. fragans (see p-values in Table 3, Figure 3). All combinations of bioclimatic variables revealed that the longhorn beetle has ca. three times the hypervolume of the hostplant C. fragans ellipsoid. Niche volumes were similar between these species when only precipitation or temperature variables were considered, but when all variables were included, niche volume of L. limpidus increased substantially, while the volume of C. fragans reduced in size (Table 3).

Discussion
Species distributions are conceived as the intersection of three limiting factors: movement capacities, abiotic conditions, and biotic interactions [23]. The relative significance of these components is influenced by spatial scale and resolution [24,25,75]. Recent fieldwork revealed new distributional records for the longhorn beetle Lepturges (L.) limpidus Bates, 1872 (Cerambycidae) in northern Argentina. The beetle species shows a wide distribution throughout the Neotropical Region, but this is the first record for the Chaco biogeographic province. Our results indicate widespread climatic suitability within the Neotropical Region, but new records in northeastern Argentina are near the periphery of the L. limpidus ellipsoid envelope. Further, based on the ellipsoid envelopes, no dependent association to its known hostplant C. fragans was recovered.
A central postulate in biogeography is that climate exerts a dominant influence on the distribution of species (e.g., [76][77][78][79]). For L. limpidus, climatic niche projections revealed widespread suitable areas within the Neotropics, with high geographic (i.e., proportion of total suitable area) and environmental space prevalence (i.e., proportion of the multidimensional space identified to be suitable considering only distinct combinations of environmental conditions). This was particularly observed within the eastern part of the Chaco biogeographic province (Eastern Chaco district), characterized by a humid subtropical climate without a marked dry season and by a hot summer season [37], favorable for the development of different cerambycid species and the growth of their potential hosts [80][81][82]. However, Mahalanobis distance values associated with occurrences in northeast Argentina were relatively low (MD = 0.043 and 0.046; Figure 2), suggesting that these records are probably attributable to climatically peripheral populations.
Moreover, association with the climatic space of its hostplant C. fragrans was nondependent. As a native species of the Guianas [83], C. fragrans is intensively harvested and traded as a commercial timber tree [84,85], which could have potentially facilitated the translocation (i.e., accidental or intentional human-mediated movement[s] of living organisms from one area, with release in another; IUCN 2012) of L. limpidus to northeastern Argentina. The endophytic lifestyle of cerambycid species allows their easy and widespread movement through the international trade in timber and timber derivatives [86][87][88]. Unintentional transport of species within or on a specific commodity is frequently reported. In Europe, for instance, this accounts for ca. 90% of alien arthropod species invasions [89]. A particular case is the polyphagous invasive beetle Anoplophora glabripennis (Motschulsky, 1854) native to Southeast Asia, which threatens forests in Europe and North America and spreads in wooden packaging materials (boxes/pallets) and their derivatives, managing to adapt to new territories due to its wide range of hostplants [90]. Thus far, evidence suggests that the new records of L. limpidus are a result of translocation, potentially representing sink populations, while biotic and abiotic interactions in the new distributional area seem sub-optimal for the establishment of populations.
Nevertheless, it is hard to ascertain the status of a population as sink or source. Local adaptation of a translocated species to a new geographic area can occur due to its ability to exploit empty niches in the ranges to which it is introduced [91], or the frequency and magnitude of local processes (e.g., disturbance), and the absence of its natural enemies, including competitors, predators or pathogens [92][93][94][95][96]. Likewise, translocations may enhance evolutionary changes partly owing to founder effects and genetic bottlenecks and partly because of the triggering of evolution by new environmental factors [97]. Zygogramma suturalis (Fabricius, 1775) (Coleoptera: Chrysomelidae) for instance, showed rapid evolutionary changes in flight capacity (development of flight ability and morphological changes) within only five generations [98].
Then again, L. limpidus could alternatively be utilizing other hostplant species as resources; however, our knowledge regarding additional biotic interactions is inadequate. Despite great economic relevance and prominence in ecology and evolutionary literature, the biotic interactions of xylophagous insects and the essence of their interaction (e.g., hostplants as food source, oviposition sites, or shelter for hibernation) remain poorly known [29,99]. This deficit hinders the development of targeted monitoring programs for species at high risk of becoming invasive pests and the protection of areas that could represent areas of potential distribution for pest species [29]. Among xylophagous insects, most cerambycid species present different specialization strategies, occupying various host plant species [100]. As expected in such a large insect family, some cerambycids are strictly monophagous, while others develop in multiple plant species of the same family (oligophagous) or multiple families (polyphagous) [71,101]. Also, some cerambycids infest living, healthy plants, while others develop in dying plants; likewise, some species prefer moist wood, while others prefer dry wood [3]. In the Neotropics, [43] determined the pattern of host specificity for cerambycids from the tropical rainforest of French Guiana, where almost 24% of the species had a single recorded host, while 53% were generalists, mostly Acanthoderini or Acanthocinini. Alternatively, a new record based on only a few specimens could represent a recent colonization (e.g., the three males and two females of L. limpidus from northeast Argentina). However, as in many cases involving arthropod species, the introduction of the species cannot be ascertained [102], and the small number of specimens might reflect a lack of target sampling or incongruency between species biology and field surveys. Whatever the case, further monitoring surveys should be implemented to detect, document and investigate potential establishment of L. limpidus in the area.
Finally, it is important to highlight that while our models help to understand and predict potential geographic distributions of species based on environmental factors, they do so without considering important factors of microenvironments and vegetation cover. We cannot overlook that cerambycid species are often closely linked to forest environments [71], so including vegetation cover layers (e.g., normalized difference vegetation index, NDVI) in future studies could contribute to a broader interpretation of the species' distributional model [29,103,104].

Conclusions
In summary, we report new records of L. limpidus in Argentina, and our results show that its new distributional record possibly represents a sink population, potentially resulting from translocation associated with the increased commercialization of its hostplant as timber. Although Chaco is the second-most forested area in South America [105], it has undergone extensive land-use changes that have led to significant deforestation [106,107]. Hence, target surveys are essential for the detection of new arrivals of exotic species, their patterns of invasion, and exotic species impacts to allow installation of mitigation and control measures.  Table S1. Principal component analysis results for model calibration. We retained the first three components, which cumulatively explained >88% of the total variance in the dataset. Supplementary Material Table S2. Principal component analysis results for model calibration. We retained the first three components, which cumulatively explained >90% of the total variance in the data set. Supplementary Material Table S3. Complete report on the characteristics of the ellipsoid. Data Availability Statement: The data that supports the findings of this study are available in the supplementary material of this article.