PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of scirepAboutEditorial BoardFor AuthorsScientific Reports
 
Sci Rep. 2017; 7: 43988.
Published online 2017 March 8. doi:  10.1038/srep43988
PMCID: PMC5341063

Geographical variation in morphology of Chaetosiphella stipae stipae Hille Ris Lambers, 1947 (Hemiptera: Aphididae: Chaitophorinae)

Abstract

Chaetosiphella stipae stipae is a xerothermophilous aphid, associated with Palaearctic temperate steppe zones or dry mountain valleys, where there are grasses from the genus Stipa. Its geographical distribution shows several populations that are spread from Spain, across Europe and Asia Minor, to Mongolia and China. Geographical variation in chaetotaxy and other morphological features were the basis to consider whether individuals from different populations are still the same species. Moreover, using Ch. stipae stipae and Stipa species occurrences, as well as climatic variables, we predict potential geographical distributions of the aphid and its steppe habitat. Additionally, for Stipa species we projected current climatic conditions under four climate change scenarios for 2050 and 2070. While highly variable, our results of morphometric analysis demonstrates that all Ch. stipae stipae populations are one very variable subspecies. And in view of predicted climate change, we expect reduction of Stipa grasslands. The disappearance of these ecosystems could result in stronger separation of the East-European and Asian steppes as well as European ‘warm-stage’ refuges. Therefore, the geographic morphological variability that we see today in the aphid subspecies Ch. stipae stipae may in the future lead to speciation and creation of separate subspecies or species.

Genus Chaetosiphella Hille Ris Lambers, 1939 belongs to the tribe Siphini (Aphididae: Chaitophorinae) and includes six taxa (species and subspecies): Ch. berlesei (Del Guercio, 1905), Ch. tshernavini (Mordvilko, 1921), Ch. stipae subsp. stipae Hille Ris Lambers, 1947, Ch. massagetica Kadyrbekov, 2005, Ch. longirostris Wieczorek, 2008 and Ch. stipae subsp. setosa Wieczorek, 2008. Previous studies on this genus related mainly to descriptions of new morphs, distribution and host plants1,2,3,4,5,6,7,8,9. Here, we focus on Ch. stipae stipae which is a xerothermophilous aphid associated with Palaearctic temperate steppe zones or dry mountain valleys. Its distribution ranges from the Central Asia (including the basins and plateaus) and Ukrainian steppes to central and western Europe (Pannonian Basin and other isolated localities – including the dry inner Alpine valleys associated with xerothermic Stipa grasslands)9. Ch. stipae stipae live on the stems and upper side of leaves and are distinguished by their long apical rostral segment, long hind legs, and chaetotaxy. However, variations in chaetotaxy9 suggest that populations may have different taxonomies.

Ch. stipae stipae is a narrow oligophagous subspecies which requires very tight synchrony with its host plant phenology. It is mainly related to plants from the genus Stipa L. (Poaceae) – commonly called feather grass, needle grass, and spear grass. Depending on systematic treatment10, Stipa is comprised of 100 (www.efloras.org) to nearly 400 annual and perennial species (www.theplantlist.org). These grasses grow in dense, shorted clumps, and have no stolons. Leaves are stubbly with open leaf sheaths, and the inflorescence is a panicle with unifloral spikelets. Stipa spp. are mainly pollinated by wind and are facultatively cleistogamous, producing chasmogamous as well as cleistogamous flowers11. Many species of Stipa are protected, particularly in Europe12,13.

The Eurasian Steppe or Great Steppe, is the world’s largest steppe ecoregion where one of the dominant plant communities is S. capillata group14. Generally, the ecoregion extends from Eastern Europe (South Ukraine) to North China15 and is characterized by grassland plains without trees. However, grass and shrub steppes also occur as a narrow transition zone from Morocco in the West to Iran in the East15,16. In the north, the steppe borders the boreal forest (taiga) of central Russia and Siberia; in the south it transitions into desert17,18,19.

Grasslands are one of the most species-rich and diverse plant communities, and especially in Central and Western Europe, are very important refuges for xerothermic species of plants and invertebrates, including endemics16,20,21. Xerothermic grasslands, which are extra-zonal analogues of steppes, are among Europe’s most endangered natural environments12,13. Climatic and environmental conditions led to historical range changes in the distribution of vegetation and animals in Eurasia. During the Pleistocene, mid-latitudes of Eurasia were dominated by steppe-tundra habitat. These habitats have been shifting during interglacial intervals (mostly in the Holocene)22. Because of the regression of steppe-like habitats, steppes are now restricted to their refugial areas, particularly in Europe. So called ‘warm-stage’ refugium of steppe-like habitats23,24 are now restricted to Pannonian Basin25 and the Iberian Peninsula26. Isolated xerothermic grasslands located in the Balkans and central and western Europe could also be considered as cryptic ‘warm-stage’ refugias. These habitats are closely related to the continental steppes of Eurasia and are considered threatened due to anthropogenic transformation of the environment and the fact that they are isolated and limited (mainly to the areas unfavorable for forest plantations and agriculture).

As the first goal of this study we consider morphological variation of Ch. stipae stipae across geographical and environmental space via multivariate morphometric techniques. Aphid multivariate morphometric analyses (i.e. multiple discriminant analysis and the use of canonical varieties) have demonstrated the differences between closely related taxa or samples from clearly defined populations27,28,29,30. We also wanted to investigate how the habitat of this aphid might be affected by climate change and if geographic morphological variability of this subspecies may lead to environmentally-associated speciation in those populations. Therefore, the second goal of this study is use species distribution modeling of Stipa species to predict current Ch. stipae stipae distribution and predict how this distribution might be altered due to climate change.

Results

Statistical analysis of Taxon

Among the 16 morphometric variables and 7 morphometric ratios analyzed, 9 were selected based on the analysis of the correlation: ANT IV, LSH, HL, LS ANT III, LS TIBIA, S7, ARS:III, ANT:BL and IV:III. The first two axes of the CA represent 88.1% of the total variance (the first three axes represent 94.2%). The first component (axis 1) is characterized by the length of the longest seta on the seventh abdominal tergite (S7) and the ratio of ARS:III. The second component (axis 2) reflected the length of longest seta of head (LSH), length of longest seta of third antennal segment (LS ANT III) and the ratio of antennal lengths to body length (ANT:BL). Plotting CV1 and CV2 demonstrated no clear between-sample differentiation according to locality for Ch. stipae stipae (Figs 1 and and2).2). Therefore these results indicated that all studied individuals are most likely one morphologically variable subspecies. These morphological variations were reflected by the inconsistency of the number and the shape of abdominal setae for the individuals analyzed.

Figure 1
Canonical analysis of 61 specimens of Chaetosiphella stipae stipae (individuals divided according to the place of occurrence) and 3 specimens of Ch. tshernavini based on the analysis of 9 morphological variables and ratios; specimens projected onto the ...
Figure 2
Canonical analysis of specimens of Chaetosiphella stipae stipae (individuals divided according to the place of occurrence) with the appointment of the impact of 9 morphometric variables and ratios; specimens projected onto the first and second principal ...

Individuals from central part of Europe (Austria, Czech Republic, Hungary, Poland (Fig. 3a–d) are characterized by having the typical set of abdominal chaetotaxy: marginal setae on abdominal tergites I-VI are 0.10–0.12 mm long and have forked and jagged apices; abdominal tergites VII and VIII have pointed setae 0.10–0.15 mm long. Pleural and spinal setae are 0.05–0.075 mm long, have forked and jagged apices interspaced with numerous, short, fan-shaped setae 0.025–0.03 mm long; the abdominal tergite VII has the longest, forked setae 0.10–0.12 mm long. Moreover, specimens from Turkey share some morphological features with specimens from central part of Europe (i.e. ARS, S7 or ARS:III) and Kazakhstan (i.e. LSH). Although some specimens from Swiss and Spanish populations are apparently different (Fig. 4a,b), morphometric analysis, shows they are within the range of variation. Specimens from Kazakh and Spanish populations share characters different than other populations studied: longer antennae, shorter apical segment of the rostrum, longer jagged marginal seta on the first abdominal segment, longer setae of hind tibia and the lowest ratio of the apical segment of the rostrum to the third antennal segment. Abdominal chaetotaxy is also similar (Fig. 5a,b), however some specimens from Spain are more setosae, with additional numerous long and pointed marginal setae on the abdominal tergites IV–VII (Fig. 4a), (these individuals are the most distant on the graph in the third quadrant (−, −)). Specimens from Mongolia (Fig. 5c) and Switzerland (Fig. 5d) populations share the following characters: shorter third antennal segment, shorter hind legs, similar length of the longest seta on the first abdominal tergite (S1), as well as the highest ratios of ANT V:III, ARS:ANT III and ARS:HT II. Similarly, some individuals from Switzerland are more setosae with additional numerous long and pointed marginal setae on abdominal tergites IV-VII (Fig. 4b), (these specimens are the most distant on the graph in the first quadrant (+, +)).

Figure 3
Abdominal chaetotaxy of representatives of Chaetosiphella stipae stipae from: (a) Austria, (b) Czech Republic, (c) Hungary, (d) Poland (typical set of abdominal chaetotaxy).
Figure 4
Abdominal chaetotaxy of representatives of Chaetosiphella stipae stipae from: (a) Spain, (b) Switzerland (individuals much “hairy”, with additional numerous long and pointed marginal setae on the tergites IV-VII).
Figure 5
Abdominal chaetotaxy of representatives of Chaetosiphella stipae stipae from: (a) Kazakhstan, (b) Spain, (c) Mongolia, (d) Switzerland (typical set of abdominal chaetotaxy).

Evaluation of the model and the importance of environmental predictors

The selection of settings for each model was chosen based on the results of pAUC, AICc and ΔAICc. In both models for Ch. stipae stipae, we observed better values of pAUC, AICc and ΔAICc were achieved where we used linear, quadratic and product (LQP) features together, regardless of the number of iterations. Otherwise it was with model for Stipa species – ΔAICc had the best value for models with auto features, but pAUC values were the best for models with LQP features. Most likely, this is due to the amount of the occurrence points used in the models. In both cases, we decided that we will use models with LQP features because models with auto features produced biologically nonsensical curves (jagged and not smooth).

Moreover, for Ch. stipae stipae models, the regularization multiplier performed better at a lower level than the default (0.5 and 0.75). Therefore, the areas predicted for the aphid generally corresponded to vegetation types where it is known to occur. For Stipa species model, the regularization multiplier performed the best result at value 1.25.

Two Maxent models were initiated for aphids and one for the Stipa species (one for present presentclimateclimate projected to sixty-four for future climate scenarios). The average AUC values for the test data for Ch. stipae stipae are 0.863 (standard deviation = ±0.110 which is 11.0%) for the model based only on climate variables and 0.868 (SD =  ± 0.091, i.e. 9.1%) for the model based on climate variables plus the outcome model of potential distribution of host plants. For the Stipa species (present climate scenario), AUC was 0.878 (SD = ±0.03, i.e. 3.0%).

Table 1
Total area predicted to have probability of suitable habitat conditions for steppe habitat under climate change scenarios.
Table 2
Bioclimatic variables chosen for modeling and their association with the habitat preferences of Ch. stipae stipae and Stipa species.

Because only four climatic variables were used for Ch. stipae stipae (see Table 3 for explanation which ones and why), the jackknife test (for details and other Maxent model outputs see Supplementary File S1) shows all of them were rather important (precipitation of warmest quarter (Bio18) was the least important). When the projected distribution of suitable habitat for Stipa species was included in model calibration, Maxent identified this variable as the most important, and affecting the reduction of areas potentially favorable for Ch. stipae stipae. In the case of Stipa species, the most important environmental variable was the mean temperature of warmest quarter (Bio10).

Table 3
Settings selected for the Maxent models based on the results of evaluation methods.

Ecological niches and potential distribution

For both the aphid (Fig. 6a,b) and Stipa species (Fig. 7), maps of the median of the 10 model replicates were derived. The arithmetic mean was not used because it is not as resistant to outliers as the median (see Supplementary File S1 for plots showing distribution of occurrence records of Ch. stipae stipae and Stipa species in reference to used predictors). The logistic output was used. Therefore, the results range from 0 to 1.

Figure 6
Maps of potentially suitable niches for Chaetosiphella stipae stipae.
Figure 7
Maps of potentially suitable niches for representatives of the genus Stipa.

The model suggested favorable climatic conditions for Ch. stipae stipae mainly in Europe. As potentially the most suitable areas in terms of climate, the model has appointed northern Spain, France, Belgium, the Netherlands, Germany, Denmark, the United Kingdom and Ireland. Other regions are Central Europe (excluding areas of the Alps and part of the Carpathians), Italy, Eastern Europe (excluding the areas of Russia), Turkey, the Balkans (excluding Dinaric Alps). In the Mediterranean region, such conditions occur in Morocco and the northern part of Algeria. In Asia, the model suggests suitable regions in the Caucasus, in certain regions of Iran, Afghanistan, Pakistan, Kyrgyzstan, Kazakhstan, and in foothill areas of north-western China (excluding the Tibetan Plateau and Quiling Mountain, and the Tarim Basin, Dzungarian Basin and Turfan Depression) (Fig. 6a).

When the projected distribution of Stipa species was included as a biotic variable in the model for Ch. stipae stipae, the distribution of suitable habitat for the aphid was projected as more constrained. Primarily, range has been limited in the north and south of Europe as well as in Asia. Mountainous areas (e.g., the Alps, Carpathian, Pyrenees and Caucasus Mountains) have also been eliminated (Fig. 6b). Supplementary Table S1 details ecoregions which are occupied by Ch. stipae stipae. These are mainly regions within the temperate broadleaf and mixed forests, temperate conifer forests, temperate grasslands, savannas and shrublands, Mediterranean forests, woodlands and scrub, and montane grasslands and shrublands.

In addition, according to the model, the species of grasses of the genus Stipa will find suitable conditions in Europe similar to those of the aphid (albeit slightly broader), but in Asia the area is definitely broader. The model suggested areas typical for these grasses in the Pontic-Caspian steppe, Central and Eastern Anatolian steppe in Turkey, Kazakh steppe and forest steppe, Emin Valley steppe, Mongolian-Manchurian grassland and Daurian forest steppe. These areas extend from Moldova, Romania and Ukraine, by Russia and Kazakhstan, to Mongolia and China.

Climatic preferences

Possible climatic preferences of aphid as well as Stipa species were inferred by comparing potential ecological niches to the Köppen-Geiger climate classification. The analysis of climate data from known localities of Ch. stipae stipae shows that it prefers places with an average monthly temperature above 10 °C (50 °F) in warmest months (April to September). Such conditions are characteristic for continental climates like warm summer continental climates (Dfb, Dwb) in central and eastern Europe and Russia, as well as Mediterranean (Csb) (Spain) and oceanic (Cfb) climates in Spain and Germany. Some individuals were located in the steppe type of climate associated with cold semi-arid climate (BSk) in Spain, or even cold desert climate (BWk) on the border of Russia and Mongolia. However, individuals from the population in Switzerland and Italy, which was located in the Alps, apparently have other preferences. A mountain or highland climate in this area is characterized by absence of a month with a mean temperature higher than 10 °C (50 °F). However, the locations of aphids are in the lower parts of the mountains – in the valleys, where the temperature is slightly higher.

Grasses of the genus Stipa have a similar climate preference to Ch. stipae stipae which feed on them. However, besides the aforementioned preference on warm summer subtype of a continental climate, Stipa species preferred additional areas include interior Eurasia, east-central Asia, and parts of India.

Comparison of current and future suitable habitat for Stipa species habitat

Suitable climatic conditions for Stipa steppes habitat were predicted to decline under all four RCP scenarios by the 2050 s (Supplementary File S1 Fig. 1.16a–d and 1.18a–d) and the 2070 s (Supplementary File S1 Fig. 1.17a–d and 1.19a–d). At present time, 32,471,127 km2 were designated as suitable in Europe, Asia and North Africa (with 10th percentile training presence threshold). For the 2050 climate scenarios, the amount of suitable habitat could decrease to 81.9–87.9% of the area in common with the current model (it decrease to 63.3–75.2% for the higher threshold) (Table 1). For the 2070 climate scenarios, those amounts could decrease to 75.6–88.1% of the area in common with the current model (46.1–73.5% for the higher threshold). In both versions of the future climate scenarios, the higher the concentration of carbon dioxide, the more mean annual temperature increased (Supplementary File S1 Fig. 1.20a,b). Therefore, generally our results indicated that the potential distribution range of steppes habitat with Stipa species will be reduced over time.

Discussion

Ch. stipae stipae is an example of a strictly xerothermophilous subspecies originating in the East-European and West-Asian steppe. It is also encountered in European ‘warm-stage’ refuges of steppe-like habitats. Similar to its host plants (Stipa spp.), the existing populations of Ch. stipae stipae are highly isolated from each other. However, based on the results of our morphometric analysis, these populations all represent one highly variable aphid subspecies, Ch. stipae stipae. Stipa grasslands developing on limestone bedrock are exposed to high-amplitude variation in temperature, both on a daily and annual basis. Those local conditions (e.g., in the dry inner Alpine valleys), possibly influence morphological characters and lead to multiplication of abdominal chaetotaxy in some individuals of Ch. stipae stipae. This confirms the presence of closely related Ch. stipae setosa presently known only from a few localities in Alpes-Côte d’Azur Provence, France. Ch. stipae setosa can be distinguished from nominal taxon by the length and form of apical segment of rostrum (much shorter) and numerous, pointed marginal setae. The marginal setae with forked or jagged apices, which are diagnostic character of Ch. stipae stipae, are absent8.

Models confirm favorable climatic conditions in most of areas where the presence of individuals representing the studied subspecies has already been recorded9 (see Supplementary File S1 for detailed maps). In Asia, suitable regions have been suggested by the model in foothill areas of the Caucasus, Altay Mts., Pamir Mts., Turbagatay submountain region and in north-western China (Mt. Xiaowutaishan). In Europe, this subspecies also prefers dry valleys at the foothills of the Alps and Pyrenees. When projected habitat suitability of Stipa was included as a predictor variable, suitable habitat for Ch. stipae stipae did not encompass the Alps and the Carpathians. However, on the enlarged map (Supplementary File S1 Fig. 1.9) we can see that suitable habitat was projected in mountain valleys. It should be noted that scale plays an important role in the modeling31,32,33. Therefore, we modeled the Eurasian area using a resolution of 2 km. But modeling on such a scale will negatively affect areas of refugia. In our example, it affected on warm mountain valleys that create the appropriate microclimate conditions for Stipa species and associated with them Ch. stipae stipae. By using a larger scale in the modeling, the result probably would be even more accurate.

Furthermore, according to the model, France, Belgium, the Netherlands and the United Kingdom should be suitable areas for Ch. stipae stipae, but this subspecies has not been recorded in those regions of Europe. Milder and warmer oceanic climate, conditions in these localities, may be appropriate for this aphid subspecies if the host plant condition is suitable. Unlike other aphids, Ch. stipae stipae do not form large and numerous colonies. Instead, they are cryptic and feed individually on stems and the upper side of the leaves. Because of this, their observation and collection, as with most of the tribe Siphini, is difficult and infrequent. There may be more species awaiting discovery in this tribe which would expand what is currently defined as their geographical range. For example, the recently described species Atheroides vallescaldera Miller and Jensen, 2014 (Siphini) from the USA, belongs to a genus previously considered exclusively Palaearctic in origin34. The use of ecological-niche models could be used as a convenient tool to predict both potential new species habitats and localities.

It is known that climate change affects various organisms. Both temperature and precipitation are important abiotic factors, which specifically affect plant life cycles, growth and their ranges of occurrence35. Such factors consequentially change interspecific relationships and plant community structure36,37. There are numerous studies regarding the influence of climate change on different types of vegetation in different regions of the world (e.g. European flora in general38) or just European forest ecosystems39, Aloe tree, Namib Desert40, temperate steppe in Inner Mongolia, China41 or agriculture in Africa and Latin America42. Climate change and its effects on terrestrial insects and herbivory patterns was reviewed by Cornelissen43. Predicting the potential impact of climate change on species distributions via ecological niche modeling represents a useful tool. And as noted by Sanchez et al.44, such models could be used by conservation practitioners, order to help them in the management of vegetation under the climate change. However, we should remember that apart from climatic changes, other abiotic parameters (e.g., land transformations or soil) also affect the habitat preferences and adaptation possibilities45. Unfortunately, due to lack of such data for the present study area, we were not able to include those parameters in the current analysis.

Each of the conducted scenarios in our study suggest that steppe habitat with representatives of the genus Stipa will shrink rather than expand into new areas. Stipa spp. are a C3 grasses, and similar like other C3 plants33, are characterized by the tendency to reduce abundance with increasing mean annual temperature46. Part of the areas that are now climatically suitable, according to data from raster used in the modeling, over time will become too warm and too dry. This applies in Europe to areas of Spain, but in particular to areas of Asia: southern Russia, Kazakhstan, Turkey, Azerbaijan, Iran, as well as China and Mongolia. Among the areas that will remain stable is Europe (excluding Spain) as well as a central part of Mongolia. Areas most at risk are the eastern part of the Pontic–Caspian steppe, Kazakh steppe and Emin Valley steppe – one of the most common Eurasian steppe ecoregions. The disappearance of these ecosystems could result in stronger separation of the East-European and Asian steppes as well as European ‘warm-stage’ refuges. There is the possibility of greater speciation events among both plants as well as insects associated with this environment. Therefore, the geographic morphological variability that we see today in the aphid subspecies Ch. stipae stipae may in the future lead to speciation and creation of separate subspecies or species47,48,49,50,51. The isolation of habitats may also promote this mechanism52,53.

Methods

Taxon sampling and statistical analysis

Sixty-one microscope slide-mounted specimens of Ch. stipae stipae were examined (three slide-mounted specimens of Ch. tshernavini Mordv. were also examined as an out group) from the local populations distributed in Spain (UL coll.), Switzerland (BMNH, RMNH coll.), Austria (BMNH coll.), Czech Republic (BMNH coll.), Poland (UŚ coll.), Hungary (ZMPA coll.), Turkey (MNHN coll.), Kazakhstan (USNM coll.) and Mongolia (ZMPA coll.). The slides were examined using a Nikon® Ni-U light microscope and photographed with a Nikon® DS-Fi2 camera. Measurements are given in mm (see Supplementary Table S2). The following abbreviations were used (partly following54): BL – body length (from anterior border of the head to the end of cauda); ANT – antennae or their lengths; ANT I, II, III, IV, V, VI – lengths of antennal segments I, II, III, IV, V, VI (ratios between antennal segments are simply given as e.g. ‘VI:III’); BASE – basal part of last antennal segment or its length; PT – processus terminalis of last antennal segment or its length; ARS – apical segment of rostrum or its length; HL – length of hind leg; HT II – second segment of hind tarsus or its length; LSH – length of longest seta of head; LS ANT III – length of longest seta of third antennal segment; BD ANT III – basal articular diameter of ANT III; S1 – the length of the longest seta on the first tergite; S7 – the length of the longest seta on the seventh tergite; LS TIBIA – length of longest seta of hind tibia.

Specimens were borrowed from the following scientific collections (preceded by acronyms used in this paper): MNHN – Muséum national d’Histoire naturelle, Paris, France; BMNH – the Natural History Museum, London, UK; UL – University of Leon, Leon, Spain; RMNH – Nationaal Natuurhistorisch Museum, Leiden, The Netherlands; USNM – U.S. National Museum of Natural History Aphid collection, located at the Henry A. Wallace Beltsville Agricultural Research Center, Beltsville, Maryland, USA; UŚ – entomology collection of the Department of Zoology, University of Silesia, Katowice, Poland; ZMPA – Museum and Institute of Zoology of the Polish Academy of Sciences, Warsaw, Poland.

Twenty-three morphometric variables and ratios were analyzed with the use of STATISTICA (StatSoft Inc, Tulsa, Oklahoma, USA) (Supplementary Table S2). The discriminant analysis module, applying stepwise discriminant function analysis (DFA), followed by canonical analysis (CA) was performed. Canonical analysis was used to determine variables which contributed most to separation of the locality-based groups.

Occurrence data

Thirty-two unique occurrence localities were compiled for the Ch. stipae stipae; 188 occurrence localities were selected within the steppe ecoregions to include known host plants of the described aphid – species of the genus Stipa (S. capillata, S. dasyphylla, S. joannis, S. kirghisorum, S. pennata subsp. eriocaulis, S. sibirica, S. splendens). All occurrence data for the aphid were based on the detailed review of specimens studied in the museum collections and scientific literature (see abbreviations for depositories and literature references in Supplementary Table S1). Occurrence data for Stipa species were obtained from the GBIF database (www.gbif.org). Records with unknown or unspecified localities were not used. A Geographic Distance Matrix Generator 1.2.3 was used to calculate the geographic distance between each pair of localities55,56. Points close to each other less than 20 km have been removed – this distance was chosen to reduce the inherent geographic biases (effect of spatial autocorrelation) which are associated with methods of sample collecting. All localities were geo-referenced using Google Earth 7.1.2.204157 (coordinates for localities were collected in decimal degrees, datum: WGS84).

Distribution data for individuals of Ch. stipae stipae subspecies will be published in the Global Biodiversity Information Facility (GBIF) database. For details of all occurrence localities for Ch. stipae stipae and Stipa species used in the MaxEnt model refer to Supplementary Table S1 and S3.

Environmental predictors and climate classification

The standard nineteen bioclimatic variables, as well as future climate data (downscaled CMIP5 data) were downloaded from WorldClim 1.4 dataset (http://www.worldclim.org)58. Coupled Model Intercomparison Project Phase 5 (CMIP5) evaluates global warming based on global climate models, and one of the variables studied by these models is the climate sensitivity to changes in the concentration of carbon dioxide in various scenarios change59. To estimate the influence of global climate change on the potential distribution of Stipa species, the species distribution for three different time periods (present, 2050, and 2070) and for four future representative concentration pathways (RCPs) (+2.6, +4.5, +6.0 and +8.5 W/m2) were modeled. For future climate scenarios, the map of the average values were presented from eight CMIP5 model outputs: BCC-CSM1-1, CCSM4, GISS-E2-R, HadGEM2-AO, HadGEM2-ES, IPSL-CM5A-LR, MRI-CGCM3 and NorESM1-M. A spatial resolution of 60 arc-seconds (approximately 2 km2) for models was chosen (30 arc-seconds spatial resolution grids downloaded from WorldClim were interpolated to 60 arc-seconds spatial resolution). All maps were prepared in SAGA GIS 3.0.0 (http://www.saga-gis.org)60 using WGS84 datum and EPSG: 3395 (World Mercator). To calculate the area occupied by Stipa species, Lambert azimuthal equal-area projection was used61.

The Köppen-Geiger climate classification system62 was used to define climate preferences of Ch. stipae stipae and Stipa species. Raw climate classification data were extracted from raster layer at Ch. stipae stipae and Stipa occurrence records. Additionally, the resulting raster from Maxent were imposed on raster of Köppen-Geiger climate classification using SAGA GIS.

Ecological niche modeling

In our study we present two models for aphids – one based only on climate variables, and the second additionally contains the result of modeling for Stipa species, included as a biotic variable. Models of the current distribution of Ch. stipae stipae as well as current and future (under climate change scenarios for 2050 and 2070) potential distribution of Stipa species were made using the Maxent software (version 3.3.3k; http://www.cs.princeton.edu/~schapire/maxent), which is based on a maximum entropy algorithm63. Ecological niche modeling was used to discern ecological aspects of Ch. stipae stipae and its habitat represented by Stipa species. This modeling tool has been widely used in faunal and floral studies and in many aspects of biology including ecology, evolutionary biology and biogeography64,65,66. The primary task of modeling is to predict the range of distribution of a species or plant community. This is based, in simple approach, on identifying areas with environmental conditions that will allow species to survive67,68,69.

By using SAGA GIS 3.0.0, raw environmental data were extracted from bioclimatic raster layers at Ch. stipae stipae and Stipa occurrence records, as well as from 10 000 background points from the entire study area (Europe and Asia). Through the use of Spearman rank correlation, performed in the R software (version 3.3.1)70 using Rattle package (version 3.5.0)71, the number of variables was minimized by discarding those which were highly correlated (r > 0.75 or r < −0.75). Variables that did not have any significant contribution to the model (did not match the habitat preferences of surveyed species) and were highly correlated, were removed. Table 2 presents selected variables and their association with the habitat preferences of Ch. stipae stipae and Stipa species.

As species occurrence points came from museum data and were not collected randomly, bias files were provided during Maxent modeling. Both bias grid files were generated in SAGA GIS, weighted by a Gaussian kernel with a standard deviation (SD) of 200 km (for instruction see ref. 72).

Many authors points out that the default settings in Maxent do not always produce the best predictions73,74,75,76. Therefore we used different regularization multiplier values (ranging from 0.5 to 2.5) and different combinations of feature types (auto features or linear, quadratic and product features together (LQP)). A 10-fold cross-validation was ran in both models – for aphid and for Stipa species63,77,78. A jackknife test was selected to show relative importance of each predictor by comparing models with all environmental variable combinations with individual variable importance.

As a threshold-independent assessment of overall model performance we used area under the ROC (receiver‐operating characteristic) curve (AUC), partial AUC (pAUC) (calculated using NicheA79) and the sample-size corrected Akaike’s information criterion (AICc and ΔAICc) (calculated using ENMTools 1.4.480) (see Supplementary Table S4 for results). Calculation of AICc requires both the ‘.asc’ file and the .‘lambdas’ file associated with each model, and it is crucial for output to be in “raw” format.

We also employed threshold-dependent measure (omission rate based on threshold rule) –10th percentile training presence threshold. It is set at a value that excludes the 10% of calibration localities with lowest prediction, so it has an expected omission rate of 0.1081. As there is still no consensus as to which threshold is the best, the 10th percentile threshold has been more commonly used82,83,84. It is considered to provide a more ecologically significant result85,86. Moreover, other thresholds often require both presence and absence data, while 10th percentile threshold is widely used when true absence data is not available87. Therefore, in order to convert from the continuous to binary maps and to define habitat and non-habitat areas, we used 10th percentile training presence threshold for Ch. stipae stipae and Stipa species77,88. We also checked what changes will occur in the study area under the climate changes at the 50th percentile training presence threshold. We did it because we wanted to know how much vulnerable to climate change are areas with a higher probability of suitable conditions.

The logistic output of Maxent with prediction values from 0 (unsuitable habitat) to 1 (optimal habitat) was used for final models. Table 3 presents settings that were finally chosen for all models.

Additional Information

How to cite this article: Wieczorek, K. et al. Geographical variation in morphology of Chaetosiphella stipae stipae Hille Ris Lambers, 1947 (Hemiptera: Aphididae: Chaitophorinae). Sci. Rep. 7, 43988; doi: 10.1038/srep43988 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Material

Supplementary Information:
Supplementary Table S1:
Supplementary Table S2:
Supplementary Table S3:
Supplementary Table S4:

Acknowledgments

We would like to express our thanks to the late Georges Remaudière, Danièle Matile-Ferrero and Thierry Bourgoin, Museum national d’Histoire naturelle, Paris, France, Diana Percy and Paul Brown, Natural History Museum, London, UK, as well as Nicolás Pérez Hidalgo, Department of Biodiversity and Environmental Management, University of Leon, León, Spain for the loan of slides and for all their help during the visits in the collections of the MNHN and BMNH. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA; USDA is an equal opportunity provider and employer.

Footnotes

The authors declare no competing financial interests.

Author Contributions K.W. and A.B.-N. conceived the ideas, collected the data. K.W., A.B.-N. and M.K. analysed the data. A.B.-N. conducted modeling and described the results. K.W., A.B.-N. and G.M. led the writing.

References

  • Richards W. R. The Chaitophorinae of Canada (Homoptera: Aphididae). The Entomological Society of Canada, Ottawa (1972).
  • Ivanovskaja O. I. Tli zapadnoj Sibiri. 1. Adelgidae-Chaitophoridae. [Aphids of the Western Siberia. Family Adelgidae]. Nauka, Nowosibirsk (1977).
  • Szelegiewicz H. Leveltetvek I. – Aphidinea I. Chaitophorinae, Callaphidinae. Fauna Hung. 17(18), 1–175 (1977).
  • Szelegiewicz H. Klucze do oznaczania owadów Polski (Homoptera, Aphidodea), Chaitophoridae. PWN, Warszawa (1985).
  • Heie O. E. The Aphidoidea (Hemiptera) of Fennoscandia and Denmark. 2. The family Drepanosiphidae. Fauna entomologica scandinavica, Scandinavian Science Press LTD, Klampenborg (1982).
  • Nieto Nafria J. M. & Mier Durante M. P. Hemiptera, Aphididae I. [In:] Fauna Ibérica. (ed. Ramos M. A. et al. .). Museo Nacional De Ciencias Naturales, Madrid (1998).
  • Qiao G. X. & Zhang G. X. Taxonomic study of the Chinese subfamily Atheroidinae (Homoptera: Aphidoidea). Acta Zootaxonomica Sinica 27(4), 756–767 (2002).
  • Wieczorek K. Review of the genus Chaetosiphella Hille Ris Lambers, 1939 (Hemiptera, Aphididae: Chaitophorinae). Insect Syst. Evol. 39(3), 327–340 (2008).
  • Wieczorek K. A monograph of Siphini Mordvilko, 1928 (Hemiptera, Aphidoidea: Chaitophorinae). Wydawnictwo Uniwersytetu Śląskiego, Katowice (2010).
  • Vazquez Pardo F. M. & Gutierrez Esteban M. Classification of species of Stipa with awns having plumose distal segments. Telopea 13(1–2), 155–176 (2011).
  • Ponomarev A. N. Klejstogomiya u Kovylej. Bot Z. 9, 1229–1235 (1961).
  • Ludwig G. & Schnittler M. Rote Liste gefährdeter Pflanzen Deutschlands. Bundesamt für Naturschutz, Bonn (1996).
  • Mirek, Z., Zarzycki, K., Wojewoda & W., Szeląg, Z. Red list of plants and fungi in Poland. W. Szafer Institute of Botany, Polish Academy of Sciences, Kraków (2006).
  • Suzuki K. et al. . What kind of vegetation exists in the Central Eurasian steppe? J. Arid Land Stud. 22, 239–241 (2012).
  • Schultz J. The ecozones of the world. The ecological division of the geosphere, 2nd ed. Springer, New York (2002).
  • Dengler J., Janišová M., Török P. & Wellstein C. Biodiversity of Palaearctic grasslands: a synthesis. Agric. Ecosyst. Environ. 182, 1–14 (2014).
  • Coupland R. T. Ecosystems of the world, Vol. 8B. Natural grasslands: eastern hemisphere and résumé. Elsevier, Amsterdam (1993).
  • Rafferty J. P. Deserts and steppes (Living Earth). The Rosen Publishing Group, New York (2011).
  • Shahgedanova M. The physical geography of Northern Eurasia (Oxford Regional Environments). Oxford University Press, New York (2003).
  • Cremene C. et al. . Alterations of steppe-like grasslands in Eastern Europe: a threat to regional biodiversity hotspots. Conserv. Biol. 19(5), 1606–1618 (2005).
  • Chmura D., Adamski P. & Denisiuk Z. How do plant communities and flower visitors relate? A case study of semi-natural xerothermic grasslands. Acta Soc. Bot. Pol. 82(2), 99–105 (2013).
  • Zimov S. A. et al. . Steppe-tundra transition: a herbivore-driven biome shift at the end of the Pleistocene. Am. Nat. 146(5), 765–794 (1995).
  • Schmitt T. Molecular biogeography of Europe: Pleistocene cycles and postglacial trends. Front. Zool. 4, 1–13 (2007). [PMC free article] [PubMed]
  • Ashcroft M. B. Identifying refugia from climate change. J. Biogeogr. 37, 1407–1413 (2010).
  • Stewart J. R., Lister A. M., Barnes I. & Dalén L. Refugia revisited: individualistic responses of species in space and time. P. Roy. Soc. B-Biol. Sci. 277(1682), 661–671 (2010). [PMC free article] [PubMed]
  • Ribera I. & Blasco-Zumeta J. Biogeographical links between steppe insects in the Monegros region (Aragon, NE Spain), the eastern Mediterranean, and central Asia. J. Biogeogr. 25(5), 969–986 (1998).
  • Blackman R. L. & De Boise E. Morphometric correlates of karyotype and host plant in genus Euceraphis (Hemiptera: Aphididae). Syst. Entomol. 27(3), 323–335 (2002).
  • Lozier J. D., Foottit R. G., Miller G. L., Mills N. J. & Roderick G. K. Molecular and morphological evaluation of the aphid genus Hyalopterus Koch (Insecta: Hemiptera: Aphididae), with a description of a new species. Zootaxa 1688, 1–19 (2008).
  • Mier Durante M. P. et al. . A new genus and species of Asteraceae-inhabiting aphid (Hemiptera: Aphididae) from Costa Rica and Mexico. Fla. Entomol. 96(2), 323–331 (2013).
  • Rakauskas R., Havelka J., Zaremba A. & Bernotienė R. Mitochondrial COI and morphological evidence for host specificity of the black cherry aphids Myzus cerasi (Fabricius, 1775) collected from different cherry tree species in Europe (Hemiptera, Aphididae). ZooKeys 388, 1–16 (2014). [PMC free article] [PubMed]
  • Austin M. P. & Van Niel K. P. Improving species distribution models for climate change studies: variable selection and scale. J. Biogeogr. 38(1), 1–8 (2011).
  • Gillingham P. K. et al. . The relative importance of climate and habitat in determining the distributions of species at different spatial scales: a case study with ground beetles in Great Britain. Ecography 35, 831–838 (2012).
  • West A. M. et al. . Using high-resolution future climate scenarios to forecast Bromus tectorum invasion in Rocky Mountain National Park. PLoS ONE 10(2), e0117893 (2015). [PMC free article] [PubMed]
  • Miller G. L., Jensen A. S., Metz M. A. & Parmenter R. R. A new species of Atheroides Haliday (Hemiptera, Aphididae) native to North America. ZooKeys 452, 35–50 (2014). [PMC free article] [PubMed]
  • Cavaliere C. The effects of climate change on medicinal and aromatic plants. Herbal Gram. 81, 44–57 (2009).
  • Klein J. A., Harte J. & Zhao X. Q. Experimental warming causes large and rapid species loss, dampened by simulated grazing, on the Tibetan Plateau. Ecol. Lett. 7, 1170–1179 (2004).
  • Klanderud K. & Totland O. The relative role of dispersal and local interactions for alpine plant community diversity under simulated climate warming. Oikos 116(8), 1279–1288 (2007).
  • Thuiller W., Lavorel S., Araújo M. B., Sykes M. T. & Prentice I. C. Climate change threats to plant diversity in Europe. Proc. Natl. Acad. Sci. 102(23), 8245–8250 (2005). [PubMed]
  • Lindner M. et al. . Climate change impacts, adaptive capacity, and vulnerability of European forest ecosystems. For. Ecol. Manage. 259(4), 698–709 (2010).
  • Foden W. et al. . A changing climate is eroding the geographical range of the Namib Desert tree Aloe through population declines and dispersal lags. Divers. Distrib. 13(5), 645–653 (2007).
  • Yang H. et al. . Community structure and composition in response to climate change in a temperate steppe. Glob. Chang. Biol. 17(1), 452–465 (2011).
  • Jones P. G. & Thornton P. K. The potential impact of climate change on maize production in Africa and Latin America in 2055. Glob. Chang. Biol. 13(1), 51–59 (2003).
  • Cornelissen T. Climate change and its effects on terrestrial insects and herbivory patterns. Neotrop. Entomol. 20(2), 155–163 (2011). [PubMed]
  • Sanchez A. C., Osborne P. E. & Haq N. Climate change and the African baobab (Adansonia digitata L.): the need for better conservation strategies. Afr. J. Ecol. 49(2), 234–245 (2011).
  • Pearson R. G. & Dawson T. P. Long-distance plant dispersal and habitat fragmentation: identifying conservation targets for spatial landscape planning under climate change. Biol. Conserv. 123(3), 389–401 (2005).
  • Paruelo J. M. & Lauenroth W. K. Relative abundance of plant functional types in grasslands and shrublands of North America. Ecol. Appl. 6(4), 1212–1224 (1996).
  • Hoskin C. J., Higgie M., McDonald K. R. & Moritz C. Reinforcement drives rapid allopatric speciation. Nature 437, 1353–1356 (2005). [PubMed]
  • Lozier J. D., Rodericka G. K. & Mills N. J. Molecular markers reveal strong geographic, but not host associated, genetic differentiation in Aphidius transcaspicus, a parasitoid of the aphid genus Hyalopterus. Bull. Entomol. Res. 99(1), 83–96 (2009). [PubMed]
  • Jean P. & Jean-Christophe S. The pea aphid complex as a model of ecological speciation. Ecol. Entomol. 35(1), 119–130 (2010).
  • Nyman T., Vikberg V., Smith D. R. & Boeve J. L. How common is ecological speciation in plant-feeding insects? A ‘Higher’ Nematinae perspective. BMC Evol. Biol. 10, 266 (2010). [PMC free article] [PubMed]
  • Jousselin E. et al. . Is ecological speciation a major trend in aphids? Insights from a molecular phylogeny of the conifer-feeding genus Cinara. Front. Zool. 10, 56 (2013). [PMC free article] [PubMed]
  • Kajtoch Ł. Conservation genetics of xerothermic beetles in Europe: the case of Centricnemus leucogrammus. J. Insect. Conserv. 15(6), 787–797 (2011).
  • Kubisz D., Kajtoch Ł., Mazur M. A. & Rizun V. Molecular barcoding for central-eastern European Crioceris leaf-beetles (Coleoptera: Chrysomelidae). Cent. Eur. J. Biol. 7(1), 69–76 (2012).
  • Blackman R. L. & Eastop V. F. Aphids on the Worlds Trees. An Identification and Information Guide. CAB International, Wallingford, Oxon (1994).
  • Ersts P. J. Geographic Distance Matrix Generator (Version 1.2.3). American Museum of Natural History. Center for Biodiversity and Conservation. Available from http://biodiversityinformatics.amnh.org/open_source/gdmg (Accessed: 28th September 2016) (2012).
  • Boria R. A., Olson L. E., Goodman S. M. & Anderson R. P. Spatial filtering to reduce sampling bias can improve the performance of ecological niche models. Ecol. Model. 275, 73–77 (2014).
  • Google Inc. Google Earth (Version 7.1.2.2041). Mountain View, CA, USA. Available at: http://www.google.com/earth/index.html. (Accessed: 6th July 2016) (2016).
  • Hijmans R. J., Cameron S. E., Parra J. L., Jones P. G. & Jarvis A. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25(15), 1965–1978 (2005).
  • Taylor K. E., Stouffer R. J. & Meehl G. A. An Overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc. 93, 485–498 (2012).
  • Conrad O. et al. . System for Automated Geoscientific Analyses (SAGA) v. 2.2.7. Geosci. Model Dev. 8, 1991–2007 (2016).
  • Zhu X. GIS for Environmental Applications: A Practical Approach. Routledge, New York (2016).
  • Peel M. C., Finlayson B. L. & McMahon T. A. Updated world map of the Köppen-Geiger climate classification. Hydrol. Earth. Syst. Sc. 11(5), 1633–1644 (2007).
  • Phillips S. J., Anderson R. P. & Schapire R. E. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190(3–4), 231–259 (2006).
  • Kozak K. H., Graham C. H. & Wiens J. J. Integrating GIS-based environmental data into evolutionary biology. Trends. Ecol. Evol. 23(3), 141–148 (2008). [PubMed]
  • Wieczorek K. & Bugaj-Nawrocka A. Invasive aphids of the tribe Siphini: a model of potentially suitable ecological niches. Agric. For. Entomol. 16(4), 434–443 (2014).
  • Chłond D. & Bugaj-Nawrocka A. Distribution Pattern and Climate Preferences of the Representatives of the Cosmopolitan Genus Sirthenea Spinola, 1840 (Heteroptera: Reduviidae: Peiratinae). PLoS ONE 10(10), e0140801 (2015). [PMC free article] [PubMed]
  • Guisan A. & Thuiller W. Predicting species distribution: offering more than simple habitat models. Ecol. Lett. 8(9), 993–1009 (2005).
  • Elith J. et al. . Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29(2), 129–151 (2006).
  • Elith J. & Leathwick J. R. Species distribution models: ecological explanation and prediction across space and time. Annu. Rev. Ecol. Evol. S. 40, 677–697 (2009).
  • R Development Core Team. R: a language and environment for statistical computing (R foundation for Statistical Computing, Vienna, Austria, Available at: http://www.R-project.org. (Accessed: 5th June 2015) (2015).
  • Williams G. J. Data Mining with Rattle and R: The Art of Excavating Data for Knowledge Discovery. Use R! series. Springer, New York (2011).
  • Elith J., Kearney M. & Phillips S. The art of modelling range-shifting species. Methods Ecol. Evol. 1(4), 330–342 (2010).
  • Merow C. et al. . What do we gain from simplicity versus complexity in species distribution models? Ecography 37(2), 1267–1281 (2014).
  • Kumar S., Neven L. G. & Yee W. L. Assessing the potential for establishment of western cherry fruit fly using ecological niche modeling. J. Econ. Entomol. 107(3), 1032–1044 (2014). [PubMed]
  • Kumar S., Neven L. G. & Yee W. L. Evaluating correlative and mechanistic niche models for assessing the risk of pest establishment. Ecosphere 5(7), 1–23 (2014).
  • Samy A. M. et al. . Climate change influences on the global potential distribution of the mosquito Culex quinquefasciatus, vector of West Nile Virus and Lymphatic Filariasis. PLoS ONE 11(10), e0163863 (2016). [PMC free article] [PubMed]
  • Khanum R., Mumtaz A. S. & Kumar S. Predicting impacts of climate change on medicinal asclepiads of Pakistan using Maxent modeling. Acta Oecol. 49, 23–31 (2013).
  • Phillips S. J. & Dudík M. Modelling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31(2), 161–175 (2008).
  • Qiao H. et al. . NicheA: Creating Virtual Species and Ecological Niches in Multivariate Environmental Scenarios. Ecography 39(8), 805–813 (2015).
  • Warren D. L., Glor R. E. & Turelli M. ENMTools: a toolbox for comparative studies of environmental niche models. Ecography 33(3), 607–611 (2010).
  • Anderson R. P. & Gonzalez I. Species-specific tuning increases robustness to sampling bias in models of species distributions: An implementation with Maxent. Ecol. Model. 222(15), 2796–2811 (2011).
  • Ficetola G. F., Thuiller W. & Padoa-Schioppa E. From introduction to the establishment of alien species: bioclimatic differences between presence and reproduction localities in the slider turtle. Diversity Distrib. 15(1), 108–116 (2009).
  • Liu X., Guo Z., Ke Z., Wang S. & Li Y. Increasing potential risk of a global aquatic invader in Europe in contrast to other continents under future climate change. PLoS ONE 6(3), e18429 (2011). [PMC free article] [PubMed]
  • Kramer-Schadt S. et al. . The importance of correcting for sampling bias in MaxEnt species distribution models. Divers. Distrib. 19(11), 1366–1379 (2013).
  • Svenning J. C., Normand S. & Kageyama M. Glacial refugia of temperate trees in Europe: insights from species distribution modelling. Journal of Ecology 96(6), 1117–1127 (2008).
  • Jarnevich C. S. & Reynolds L. V. Challenges of predicting the potential distribution of a slow-spreading invader: a habitat suitability map for an invasive riparian tree. Biol. Invasions 13(1), 153–163 (2011).
  • Brito J. C., Santos X., Pleguezuelos J. M. & Sillero N. Inferring evolutionary scenarios with geostatistics and geographical information systems for the viperid snakes Vipera latastei and Vipera monticola. Biol. J. Linn. Soc. 95, 790–806 (2008).
  • Pearson R. G., Dawson T. P. & Liu C. Modelling species distributions in Britain: a hierarchical integration of climate and land-cover data. Ecography 27(3), 285–298 (2004).
  • O’Donnell M. S. & Ignizio D. A. Bioclimatic predictors for supporting ecological applications in the conterminous United States. U.S. Geological Survey Data Series 691(10) (2012).
  • Jörg E. & Lampel G. Eastern elements in the swiss aphid fauna (xerothermophilous steppe inhabitants). Acta Phytopathol. Hun. 25(1–4), 351–363 (1990).
  • Kawada K. Polymorphism and morph determination. [In:] Aphids, their biology, natural enemies and control (ed. Minks A. K., Harrewijn P.) 255–266. Elsevier, Amsterdam (1987).
  • Ivanov D. & Koleva-Rekalova E. Palynological and sedimentological data about late Sarmatian palaeoclimatic changes in the Forecarpathian and Euxinian Basins (North Bulgaria). Acta Palaeobot. Supl. 2, 307–313 (1999).
  • Schwinning S., Starr B. I. & Ehleringer J. R. Dominant cold desert shrubs do not partition warm season precipitation by event size. Oecologia 136(2), 252–260 (2003). [PubMed]

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group