Niche conservatism of Eulophia alta , a trans-Atlantic orchid species

The trans-Atlantic range disjunction is observed within about 110 angiosperm genera [1]. Despite over 40 years of biogeographical studies [2–4] the understanding of timing, direction and pattern of long-distance dispersal across the Atlantic is still very limited. Little is known also about the niche variation of the taxa characterized by remarkable distribution gap. The niche conservatism of numerous organisms was recently intensively studied using ecological niche modeling (ENM) methods [5–7], but so far this analysis was not apply to evaluate niche variation of species exhibiting trans-Atlantic disjunction. Based on the incoming research results it is difficult to formulate any explicit principle about the tendency of such species to retain characteristics of their fundamental niche over time. The degree of niche conservatism varies among groups of species (small-ranged and specialist [6]) and it is related to the evolutionary history of each taxon [8]. While some authors [9] considered evolutionary and biogeographic patterns of species diversity as completely separated processes, the concept of niche conservatism offers a bridge between them [10–13]. Studies on ecological niche variation are particularly important to recognize fundamental distribution factors and evolutionary models of species characterized by the disjunctive range. While in some taxa the geographical discontinuity resulted in adaptive [14,15], other adjusted evolutionary stable strategy [16]. The object of our study was Eulophia alta (L.) Fawc. & Rendle, a sole representative of the genus in Neotropics where its range extends from USA (Florida) south to Argentina. Except Americas E. alta occurs in the Africa, from Senegal to Zimbabwe and it is assumed that it originates in the Old World where high specific diversity of the genus is observed. However, no clear evidence of this hypothesis was presented so far. Eulophia alta is terrestrial, large plant producing a subglobose, subterranean corm and loosely many-flowered, racemose inflorescence. The species is found in open areas, along roadsides and on cleared land reverting to bush lower montane forest, usually in heavy, moisture-retentive soils. Populations of E. alta were found in lowland and premontane areas, at the altitudes up to 1500 m. The main flower visitors and most effective pollinators of this species are anthophorid bees. Six Hymenopteran and two Lepidopteran families were observed visiting the flowers [17]. The aim of this study was to evaluate the similarity of the niches occupied by African and Neotropical populations of E. alta using ENM tools. Because adaptation to local climatic conditions is a significant force driving morphological evolution and speciation, the availability of the suitable niches of the studied species during last glacial maximum (LGM; 26 500–19 000 years ago) was evaluated to estimate the possible postglacial niche shift and geographical range changes of E. alta. In the studies on Orchidaceae the ENM was applied so far mainly in the research on invasive species [18,19] and Abstract


Introduction
The trans-Atlantic range disjunction is observed within about 110 angiosperm genera [1].Despite over 40 years of biogeographical studies [2][3][4] the understanding of timing, direction and pattern of long-distance dispersal across the Atlantic is still very limited.Little is known also about the niche variation of the taxa characterized by remarkable distribution gap.
The niche conservatism of numerous organisms was recently intensively studied using ecological niche modeling (ENM) methods [5][6][7], but so far this analysis was not apply to evaluate niche variation of species exhibiting trans-Atlantic disjunction.Based on the incoming research results it is difficult to formulate any explicit principle about the tendency of such species to retain characteristics of their fundamental niche over time.The degree of niche conservatism varies among groups of species (small-ranged and specialist [6]) and it is related to the evolutionary history of each taxon [8].While some authors [9] considered evolutionary and biogeographic patterns of species diversity as completely separated processes, the concept of niche conservatism offers a bridge between them [10][11][12][13].Studies on ecological niche variation are particularly important to recognize fundamental distribution factors and evolutionary models of species characterized by the disjunctive range.While in some taxa the geographical discontinuity resulted in adaptive [14,15], other adjusted evolutionary stable strategy [16].
The object of our study was Eulophia alta (L.) Fawc.& Rendle, a sole representative of the genus in Neotropics where its range extends from USA (Florida) south to Argentina.Except Americas E. alta occurs in the Africa, from Senegal to Zimbabwe and it is assumed that it originates in the Old World where high specific diversity of the genus is observed.However, no clear evidence of this hypothesis was presented so far.Eulophia alta is terrestrial, large plant producing a subglobose, subterranean corm and loosely many-flowered, racemose inflorescence.The species is found in open areas, along roadsides and on cleared land reverting to bush lower montane forest, usually in heavy, moisture-retentive soils.Populations of E. alta were found in lowland and premontane areas, at the altitudes up to 1500 m.The main flower visitors and most effective pollinators of this species are anthophorid bees.Six Hymenopteran and two Lepidopteran families were observed visiting the flowers [17].
The aim of this study was to evaluate the similarity of the niches occupied by African and Neotropical populations of E. alta using ENM tools.Because adaptation to local climatic conditions is a significant force driving morphological evolution and speciation, the availability of the suitable niches of the studied species during last glacial maximum (LGM; 26 500-19 000 years ago) was evaluated to estimate the possible postglacial niche shift and geographical range changes of E. alta.In the studies on Orchidaceae the ENM was applied so far mainly in the research on invasive species [18,19] and a sole analysis [20] concerned the potential distribution of orchid during last glacial maximum.

Localities and georeferencing
The database of E. alta localities was prepared based on the examination of herbarium specimens deposited in herbaria AMES, BM, BR, COL, HUA, K, MO, NY, P, UGDA, VALLE, W, and WAG.Herbaria acronyms are cited according to "Index Herbariorum" [21].To enlarge the dataset, the information obtained from the electronic database of the Missouri Botanical Garden (available at http://www.tropicos.org)was included in the analysis.
The georeferencing process followed Hijmans et al. [22] and only the localities, which could be precisely placed on the map were used in the study.The information about the latitude and longitude provided on the herbarium sheet labels were verified.If no geographic coordinates were indicated by the plant collector, they were assigned based on the data about the collection place.The Google Earth (v.6.1.0.5001,Google Inc.) application was used to validate all gathered information.
In total 54 localities were included in the database (Fig. 1, Tab. 1), 8 African and 46 Neotropical, which is more than the minimum number of localities (>5) required by Maxent to obtain reliable predictions [23].

Maximum entropy analysis
The maximum entropy method implemented in Maxent version 3.3.2[24][25][26] was used to create models of the distribution of the ecological niches of E. alta.Because Maxent is relatively robust against collinear variables [27,28], all available climatic factors (Tab.2) in 2.5 arc-minutes developed by Hijmans et al. [29] as well as the altitudinal data were used as an input data.This was also justified by the insufficient data on habitat requirements of the studied species that could lead to premature exclusion of the correlated variables.The analogical bioclimatic data for the last glacial maximum period was mapped by Paleoclimate Modelling Intercomparison Project Phase II [30].
To assess high specificity of the modeling, the maximum iterations was set to 10 000 and convergence threshold to 0.00001.For each run 20% of the data were used to be set aside as test points [31].The "random seed" option which provided random test partition and background subset for each run was applied.The run was performed as a bootstrap with 100 replicates, and the output was set to logistic.All operations on GIS data were carried out on ArcGis 9.3 (ESRI).

Niche similarity
The geographical overlap of the niches calculated based on two different datasets (African and Neotropical locations only) as well as the similarity of niches occupied by the populations from the New World and Africa were defined based on the test implemented in ENMTools application.The Schoener's D [32] and I [33] statistics were calculated.In Schoener's D statistic the local species density measures are compared with each other."I" statistic is based on Hellinger distance and measures the ability of the model to estimate the true suitability of the habitat.Both metrics range from 0 (no similarity) to 1 (overlapping).

Models evaluation
All repeated ecological niche models for the present time received high area under the curve (AUC) scores of 0.972-0.984(Tab.3) that indicate very high reliability of the analysis.

Bioclimatic limiting factors
The results of the niche modeling indicated three main bioclimatic variables limiting the distribution of suitable habitats for E. alta.The crucial factor is the temperature seasonality (bio4), which influenced significantly models of both African and Neotropical populations of the studies species.While the range of latter geographical group is also related with the amount of precipitation in the coldest quarter (bio19), the African plants are more dependent on the mean temperature of the coldest quarter (bio11).The estimates of relative contributions of the environmental variables to the Maxent model for both present time and LGM are presented in Tab. 4.

Potential glacial refugia
The model of habitats available in Neotropics during the LGM indicate few possible glacial refugia of E. alta (Fig. 2).In Mesoamerica the suitable niches were located in the Gulf of Honduras, Mosquitia lowlands, Mosquito Coast, Cordillera de Talamanca, the Darién Gap as well as in Jamaica.In South America the possible glacial refugia were located in the northern Andes, eastern pre-Andean highland and partially Amazon basin, Guiana Highlands as well as western Brazilian coast.The African refugia were much more limited and there were distributed in the northern Grain Coast and around the Gulf of Guinea.Less suitable habitats were located in the Zanzibar Archipelago and northern Madagascar.The areas determined for both, African and Neotropical populations,  were covered during LGM with seasonal tropical forest as well as with tropical savanna and woodland [34].

Current potential range
The combined model (95% confidence level summary grids, Fig. 3a-c) of suitable habitats created using all available location data shows the general decreasing of the available niches in Neotropics and increasing of the African ones in comparison with the analogical models for LGM.In South America the highest concentration of proper niches is observed near the estuaries of Orinoco and Amazon rivers as well as in Cachimbo and dos Gradaús mountains.Less suitable areas are located in the lowland between Andes and Amazon basin.The African niches seem extended within the tropical Africa to east, but their general distribution did not change.

Niches overlap and identity
The calculated statistics confirm the geographical differences in the distribution of the suitable niches for African and Neotropical populations of E. alta (D = 0.319; I = 0.587), however the niche identity test indicate moderate similarity between the preferred habitats [D = 0.657 (SD = 0.0609); I = 0.883 (SD = 0.036)].The overlap of models created for LGM and present time created based all known locations of E. alta gave results of D = 0.800 and I = 0.947.

Ecological niches distribution vs. current range
The known geographical range of E. alta corresponds to the distribution of its suitable niches estimated in ENM analysis.The only areas where there are no proper niches for the studied species, according to the ecological model, and where the specimens of the studied orchid were found, are southern Florida and northern Cuba.The first region is significantly affected by non-native plants, which are naturalized from the horticultures, and it can be hypothesized that the ornamental usage of E. alta is the source of its North American populations [35].The most probable reason of the occurrence of this orchid in Cuba is the migration of E. alta from other Caribbean islands.As it was shown in the previous studies the invasive populations may shift their niches to invade new areas [18,36].As the Neotropical populations of E. alta often grow in disturbed, open areas, such as roadsides, their ecological amplitude is rather wide and it was confirmed in the ENM analysis.Hereby the chances of the North American populations to survive in habitats Tab. 4 Relative contributions of the environmental variables to the Maxent models.
less suitable for the studied orchid are high.Unfortunately, the number of the known locations of E. alta in this region is not sufficient to conduct reliable analysis of their ecology and to verify the possible niche shift of those populations.

African vs. Neotropical niches
The differences between niches occupied by African and Neotropical populations of E. alta suggest preglacial disjunction of its range and gradual, independent adaptation of both groups to slightly different climatic conditions.This timing is indicated by the consistency of the glacial refugia of this orchid in Africa and Neotropics with its current distribution that confirm the postglacial migration of this orchids on different continents from the refugia defined in the ENM analysis.Nevertheless, the occurrence of adaptation processes requires confirmation in the genetic studies.Apparently the climate changes during LGM were more harmful for African populations and while E. alta is now common in Neotropics, its African potential range declined.

Niche conservatism
While numerous species shift their niches in response to warming climate after the last glacial maximum [37][38][39], the habitats suitable for E. alta seem to be rather stable as indicated by the niche overlap test for LGM and present time.Considering also the relative similarity of the suitable niches and limiting factors for E. alta populations from different continents the phylogenetic niche conservatism theory seems to be applicable to the evolution history or this species.Despite the significant geographical disjunction of E. alta a high degree of its fundamental niche conservatism was found.This is also implied by the morphological uniformity of species representatives that suggest lack of substantial, abrupt adaptation to distinct habitat conditions.

Fig. 1
Fig. 1 Locations of E. alta used in the ecological niche modeling.

Fig. 2
Fig. 2 The Maxent projections of suitable habitats of E. alta onto climatic conditions of LGM derived from Paleoclimate Modelling Intercomparison Project Phase II.

Fig. 3
Fig. 3 The Maxent projections of suitable habitats of E. alta based on all localities (a), African localities only (b), Neotropical localities only (c).
Localities of Eulophia alta used in ENM analysis.
Variables used in the modeling.The AUC scores for each model created during the study with information on standard deviation values (SD).