Biogeography as a synthetic discipline emerged because of the need to understand spatial patterns of biological diversity (Lomolino et al. 2006). From Darwin and Wallace through most of the 20th century, form and space were at the epicenter of the field. As phylogenetic studies became pivotal for inferring biogeographical hypotheses, the connection between patterns and their processes was still concealed by the lack of temporal information (Donoghue and Moore 2003). More recently, we have been able to include time in bio-geographic analyses due to the development of modern methods that better integrate fossil and molecular data (Sanderson 1997, 2002; Rambaut and Bromham 1998; Thorne et al. 1998; Cutler 2000; Huelsenbeck et al. 2000; Kishino et al. 2001; Thorne and Kishino 2002). Additionally, new methods for estimating ancestral area reconstruction in a temporal framework allow us to answer questions about the historical biogeography of individual lineages (Ree et al. 2005; Ree and Smith 2008) by integrating information about time of lineage divergences, geological history of the areas, and other relevant biological data (e.g., dispersal mechanisms).
Given all of the above, in many cases phylogenetic data are still very incomplete or completely lacking, although knowledge of the organisms living in an area and the history of the area is available. Therefore, in the absence of molecular phylogenies and temporal information, can we still generate testable hypotheses on the historical biogeography of an area and its organisms using modern approaches? We propose a study where species occurrence data in conjunction with well-known abiotic parameters may still inform the biogeographical history of an area and the individual taxa that inhabit it. Our approach is based on distribution data alone when this is the only available information that can be used in a biogeographic analysis, when the only alternative would be to perform no analyses. The information inherent to floristic data is clearly limited, but may still allow a first attempt to generate viable hypotheses on species history that can be further tested with more appropriate methods or when more data become available.
The Volga is Europe's longest and largest river and an important feature in Russia's landscape, culture, and bio-diversity. It rises in the Valdai Hills, northwest of Moscow, and runs 3530 km across southeastern Russia to finally empty in the Caspian Sea, making it the largest river that does not drain into an ocean (Kroonenberg et al. 1997). The Caspian Sea, in turn, is the largest continental water body on Earth and has over 80% of water inflow from the Volga River (Dumont 1998). Currently at –26.3 m below global sea level, this large basin has a complex history characterized by tectonic movements caused by the closing of the Thethys Sea, and several transgression and regression events that are deeply associated with the overall formation of the Low Volga region and its biodiversity (Kroonenber et al. 1997; Mamedov 1997; Dumont 1998).
The genesis of the modern Volga River and its Delta dates back from the Pliocene when glaciation cycles have largely contributed to a series of dramatic transgression events that culminated with the Late Khvalyn phase (the last major transgression in the Pleistocene) and were intercalated by regression events that include the current Novocaspian phase (Kroonenberg et al. 1997; Mamedov 1997; Dumont 1998). Although the sequence of events is not controversial, the actual chronology is still unclear and debated in the literature despite the numerous 14C dates generated by various studies (Kaplin et al. 1993; Kroonenberg et al. 1997; Mamedov 1997; Dumont 1998). Overall, the last transgression event, with all of its phases, may have occurred approximately within 16–7 ka ago (Svitoch 2008), although older dates are provided by Kaplin et al. (1993), Kroonenberg et al. (1997), Mamedov (1997), and Dumont (1998).
The Astrakhan region comprises the Volga River's southern most part (Fig. 1) with 13 distinct local floristic areas that include Mountain Bolshoe Bogdo and Lake Baskunchak in the northeast (reviewed in Popov 2004; A. V. Popov, unpubl. data) and 11 additional areas forming the Lower Volga Valley (LVV) (Golub et al. 2002; Laktionov 2003; Laktionov and Afanasiev 2007; Losev et al. 2008). The LVV is about 500 km long and nearly 1.5 million sq. km and comprises two parts, namely the Volga-Akhtuba floodplain and the Volga Delta (Golub and Mirkin 1986). As broadly defined (Golub et al. 2002; Laktionov 2003), the LVV includes adjacent steppe and semidesert areas. Along the arid Caspian lowlands, the valley is remarkable for its vegetation diversity and includes swamps, meadows, fens and forests, whose overall biological productivity is considerably superior to that of zonal xerophytic communities (Golub and Mirkin 1986). Generally, the floodplain is characterized by high concentration of salts that reaches its maximum levels in salt marshes where only halophytes can grow. However, when covered with fresh water, salts are carried away giving rise to less hostile ecosystems and a variety of vegetation structures (Shein et al. 2011).
Fursajew (1937, 1940) indicated that the level of endemism and general diversity of the LVV is essentially underestimated. He found a significant number of undescribed taxa in several traditionally circumscribed genera including Alopecurus, Althaea, Astragalus, Bromopsis, Carex, Corispermum, Elytrigia, Glyceria, Hierochloe, Lotus, Phalaris, Plantago, Poa, Polygonum, Rorippa, Salix, Senecio, Setaria, Simphytum, Thalictrum, Trapa, and Veronica among many others (see also the numerous Fursajew collections deposited in the herbarium of Saratov State University [SARAT]). This large number of taxa unknown to science was probably due to recent and rapid evolution in the LVV area, most likely caused by selective responses to the numerous Volga floods and more recent transgression processes that took place since the Pleistocene (Fursajew 1937, 1940). Some of these taxa were later formally described by other botanists. According to Fursajew (1937, 1940), the LVV is a highly dynamic area that contains many more unknown endemics than recognized thus far, and may actually represent a biodiversity hot spot in southeast Eurasia that clearly requires more investigation. More evidence of the LVV high diversity is provided by the Herbarium of Otto Kuntze who traveled from Sarepta (modern Volgograd) to Astrakhan in 1886 during his famous Caucasian trip, and collected and described many new taxa (Zanoni 1980; Zanoni and Schofield 1981).
Over the past 12 years, we gathered a large amount of detailed floristic data covering 13 local floristic regions that characterize the LVV (Laktionov 2003; Laktionov and Afanasiev 2007; Losev et al. 2008; Laktionov 2009; A. P. Laktionov, unpubl. data). Specifically, we recorded the occurrence of 1018 species representing nearly all of the known LVV local flora, and including 23 endemic species restricted to one or few of the 13 floristic regions. The goal of this study is to infer the putative history of the areas comprising the LVV and reconstruct the individual histories of each species in order to generate hypotheses on the general processes affecting the LVV in view of its known complex geological history.
Taxon-area analyses have dominated primarily the field of cladistic biogeography and have been, and still are, extensively used in approaches such as parsimony analysis of endemicity (PAE) and its several methodological variations (Morrone and Crisci 1995). Cladistic biogeography and PAE have been criticized in the literature, particularly due to lack of integrated temporal information leading to significant misinterpretation of the connection between patterns and their causal processes (Donoghue and Moore 2003). However, general PAE approaches are still used (Porzecanski and Cracraft 2005; Contreras-Medina et al. 2007; Escalante et al. 2009, among many others), regardless of old and new critiques (Garzon-Orduna et al. 2008). We are fully aware and acknowledge the limitations of PAE for not laying its methodological foundations on phylogenetic or temporal information, and de-emphasizing dispersal and extinction as likely causal processes. In this study, we adopt the basic assumption that through modern taxon distribution alone we may still recover a strong historical signal about area relationships (Cracraft 1991; Porzecanski and Cracraft 2005); however, rather than using maximum parsimony (MP), we choose to analyze our data in a maximum likelihood (ML) framework. In PAE, synapomorphies are usually interpreted as vicariance events, parallelism and convergence as dispersal events, and reversals as extinctions (e.g., Morrone and Crisci 1995). ML provides a model-based approach in which area relationships are explained based on the degree of likelihood that species are present on each node of the taxon-area phylogeny. No preferred causal process is evoked and vicariance, dispersal, and extinction become equally possible. Moreover, in contrast to PAE, all taxa, including widespread and narrow endemics, are considered informative and the history of each species can be individually reconstructed on the ML topology, allowing us to generate testable hypotheses based on their putative biogeography.