Almost 50 years ago MacArthur & Wilson (1963) provided a concept linking ecological pro-cesses with observable species diversity patterns in their equilibrium theory on island biogeography. The idea of a constant species richness that is defined by island specific geographical properties (but with changing species composition) was ground-breaking and has influenced research and discussion in all fields of ecology (Vitousek 2002; Whittaker &
Fernández-Palacios 2007). Still, island biogeography is stimulating the development of ecological theory in general. With an increasing understanding of island-like biological systems, strong calls for an incorporation of further aspects (especially temporal and evolutionary dynamics) into a general theory have been addressed (e.g. Brown & Lomolino 2000; Heaney 2000; Lomolino 2000a; Whittaker 2000; Heaney 2007). Meanwhile strong contributions (Whittaker et al. 2007, 2008; Rosindell & Phillimore 2011 with remarks in Manuscript 4) head towards a new synthesis for a general theory of island biogeography.
Additional theoretical contribution to island biogeography will probably be enhanced by the public awareness of the 50 years anniversary of the MacArthur & Wilson (1963) paper in 2013.
The hump shaped relation between the percentage of endemic species (pSIE) with island age and a positive relation with species richness has initiated an intense debate on the drivers of evolutionary processes on oceanic islands (Chapter 1.4.7.). While investigating the geo-ecologi-cal factors driving species diversity on the Canary Islands (Manuscript 2) I could identify first hints that other factors than a direct effect of age or species richness is responsible for the observed relations of pSIE with age and species richness. Overall species richness is on oceanic islands in general, but on the Canary Islands in particular, positively related to an island´s maxi-mum elevation (Manuscript 2). However, the differentiation between the effect of elevation and age is limited by the strong negative relation between the two variables within the characteristic ontogeny of volcanic islands.
I suggest that the positive relation between pSIE and elevation identified when using entire islands as units comes from an increasing pSIE with elevation within these islands. The under-ling ecological explanation is an increasing ecological isolation with elevation on islands lead-ing to a higher isolation and thus speciation in higher elevated parts of an island in comparison to the low elevation ecosystems (Figure 1 in Manuscript 2 and Figure 1 in Manuscript 5). On the Canary Islands the closest comparable elevations to the Pico de Teide (3718 m, Tenerife)
Introduction occur in the High Atlas Mountains of Morocco (Djebel Toubkal, 4167 m) at a distance of ca.
900 km. This distance is by far larger than the direct distance from coast to coast (ca. 100 km).
Thus diaspores or potential colonising individuals from ecosystems with similar environmental conditions have a much stronger dispersal filter for high-elevation ecosystems in comparison to low elevation ecosystem (Manuscript 2 & 5). This directional ecological filtering along the ele-vational gradient is crucial for the colonisation of non-native species into high-elevation regions worldwide (Alexander et al. 2011) and leads to an increasing isolation with elevation. Species exchange among elevational bands is limited due to niche conservatism and the associated climatic barriers (see Wiens et al. 2007 for further references). The result of this elevation-driven ecological isolation hypothesis (Manuscript 2 & 5) is an increasing speciation with elevation resulting in a higher pSIE in higher elevated insular ecosystems. The hypothesis is supported by data from Mediterranean islands (Manuscript 5) as well as the Canary Island (Manuscript 2) and the application of mixed effects models to account for differences among islands (Hortal 2012).
Whittaker et al. (2008) interpreted the influence of age primarily via its effect on topographic complexity and carrying capacity on islands. Interestingly, not only an increase of pSIE with elevation but also the hump shaped relation with age within similar elevational bands on different islands was confirmed. This stresses time as an independent driver for evolutionary dynamics.
There are only very few general patterns in ecology. However, there are first hints that the elevation-diversification pattern identified within this thesis may be one of them (Manuscript 2
& 5; Jump et al. 2012). Even within mountain chains such as the European Alps, dispersal limitation (or recolonisation limitation) is a likely cause of the high rate of endemic species around areas suspected to be climatic refugia during the ice age (Dullinger et al. 2012). It has indeed been shown that species distribution of endemic species is strongly influenced by limited post-glacial migration (Essl et al. 2011). Thus environmental filtering along elevational gradients (Alexander et al. 2011) together with the isolating effect induced by dispersal limitation of species and the geographical isolation in high-elevation ecosystems may cause similar changes in speciation with elevation on most mountain systems that are old and large enough to document this process. Especially “sky islands“ - i.e. mountains that differ remarka-bly in their environmental setting from the surrounding matrix - pose optimal study objects but are surprisingly seldom studied (Gehrke & Linder 2011; Nogué et al. 2012). Körner (2007) introduced the idea of an “archipelago of climatic mountain islands“ to highlight the distinct elevation-driven ecological isolation (Manuscript 2 & 5). As there is a general lack of studies investigating changes in speciation, extinction and colonisation along elevational gradients (Wiens et al. 2007) future studies addressing these issues (see Chapter 1.7.) may not only pose valuable results, but especially contribute to the theoretical understanding of biotic systems.
1.5.2. The effect of isolation on the speciation-elevation interaction
The elevation-driven ecological isolation hypothesis (Manuscript 2) indicates above average speciation in high elevations (Manuscript 2 & 5). This is exactly what is found on the Canary Islands (Manuscript 2) and in the Mediterranean (Manuscript 5). On the contrary, knowledge gained from investigations of the latitudinal diversity gradient (Chapter 1.4.4.) and species-area relations (Chapter 1.4.3.) suggests the opposite, namely enhanced speciation in low elevations (Figure 3). First, the metabolic theory of ecology strongly favours an increase of speciation rate with temperature or more general with available energy. The above average speciation rate in warm areas originates either form a direct effect via fast rates of speciation or due to a larger number of individuals in comparison to cold areas (Rohde 1992). Second, the larger number of species found in low to mid elevations (Alexander et al. 2011) is claimed to enhance speciation via interspecific interactions and pressure for adaptive radiation (diversity begets diversity;
Emerson & Kolm 2005). Third, no matter whether speciation per species increases with area linearly or via thresholds (Lomolino 2000b; Losos & Schluter 2000; Kisel & Barraclough 2010)
Introduction it would be above average in low elevations where the area of altitudinal belts tends to be larger (Körner 2007). Forth, extinction rates are suspected to be higher in temperate than in tropical climates (Weir & Schluter 2007), which could be reflected in an increase in extinction rate with elevations (McCain 2009). MacArthur (1972) already denoted enhanced extinction rates in mountain environments due to the small area and high isolation. High extinction risk of species results in a fast temporal species turnover. The resulting low time available for each species likely additionally decreases the chance of one species for adaptive radiation (time for speciation hypothesis; Stephens & Wiens 2003).
Time for speciation has been suggested as a causal explanation for richness patterns along elevational gradients with clades in low to mid elevations being older then in high elevation (Wiens et al. 2007). Wiens et al. (2007) suspects an above average diversification rate at mid elevations and highlight that it may be caused either by an increased speciation rate at mid elevations or by a increased extinction rate at high and low elevations. However, Wiens et al.
(2007) do not consider any elevational movement of species, a fact that may indeed become more and more relevant in face of climatic changes (Jentsch & Beierkuhnlein 2003). Species diversity and hence intensified species interactions have in addition been suggested to enhance extinction risk of single species (diversity begets diversity; Emerson & Kolm 2005). In summary, taking away the effect of isolation, which is one prerequisite for speciation (Heaney 2000), theory predicts a higher rate of speciation and especially diversification in low, not in high elevations (Figure 3).
Figure 3: The elevational dependence of different drivers of speciation (grey) and extinction (black) on diversification of species along an elevational gradient of a high oceanic island.
Only elevation-driven ecological isolation supports higher speciation and thus diversification rates in high-elevation ecosystems on islands (and mountains). The opposite effect is triggered by temperature, area and diversity dependent interactions among a smaller number of spe-cies. *)Asterisks indicate mechanisms that are under debate.
The effect of elevation-driven ecological isolation diminishes if the difference in isolation between high and low elevation ecosystems declines. On very remote oceanic islands like Hawaii high and low elevation ecosystems are both equally isolated from a continental source region, while on Crete, high and low elevation ecosystems differ considerable in distance to comparable ecosystems on the mainland (Figure 1 in Manuscript 5). As a consequence, I expect a counterintuitive interaction between the relation of diversification with elevation and the degree of isolation of islands. For continental mountains the isolation contrast between low and high elevations is very strong as low elevations directly merge into the surrounding matrix.
Here and for less isolated islands, I expect a significant increase in diversification processes with elevation. This would be reflected by an above average pSIE in high elevations if the
Introduction mountains are large enough to support speciation and old enough to document the pattern. I expect that the more isolated an island is, the less pronounced is the positive relation of diversification (and pSIE) with elevation. Ultimately, if the contrast in isolation becomes negligi-ble, the diversification-elevation relation should be reversed with higher or at least equal diversification rates in low elevation (Figure 4). The theory – yet to be tested – does however not question the comparably high rates of diversification (and thus endemism) on isolated islands.
Figure 4: The increase in diversification with elevation should be reduced with increasing isola-tion of the island (a & b). The more isolated an island is the less pronounced is the difference in isolation of its low and high elevation ecosystems. Whether diversification per species on very remote islands more prominent in low elevation ecosystems, as predicted by theory (continuous line), or not (dashed line) is yet to be tested.