University of Groningen
Species selection and the spatial distribution of diversity
Herrera Alsina, Leonel
DOI:
10.33612/diss.99272986
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Herrera Alsina, L. (2019). Species selection and the spatial distribution of diversity. University of Groningen. https://doi.org/10.33612/diss.99272986
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
CHAPTER 4
Dispersal rather than differential speciation rates explains the elevational diversity gradient in songbirds
126 127
4
ABSTRACT
Low elevation regions contain a disproportionately large part of the world’s biodiversity compared to high elevation areas even though the latter areas seem to be nurseries for species as they contain many (young) endemics. This altitudinal species richness gradient ultimately results from the interplay of speciation and immigration, but the relative contribution of these processes is still unclear. Here, we use the phylogenetic tree of all songbird’s (Passeriformes) as well as distributional data to assess a comprehensive set of contrasting hypotheses on homogeneity or heterogeneity of diversification rates along the elevational gradient jointly with differential rates of altitudinal dispersal. We employed a maximum likelihood approach to compare these hypotheses, while accounting for potential differences across biomes and geographic realms. We found no support for differential diversification rates being responsible for the species richness gradient. Instead, this gradient seems primarily due to rates of dispersal from high to low elevations being higher than the rates in the reverse direction. Our findings, highlighting a prominent role for dispersal, contrast with recent studies that instead identified variation in diversification rates along elevational gradients in diversity. We argue that this is due to the fact that we look at diversification at all times (deep time and recent time) and thereby also allow the diversity at one elevation to have arisen at another elevation. Our results thus show how full phylogenetic information can help describe spatial patterns of diversity.
Keywords: lineage dispersal; diversification rates; species distributions
INTRODUCTION
High elevation areas are hotspots of endemism (Körner and Spehn 2002; Merckx et al. 2015) which suggests that a suite of conditions spurs speciation. However, the vast majority of species are not found at highlands but in lowlands. There are two potential mechanisms responsible for the low number of species at high elevations. First, net diversification rates in high elevation areas could be lower than in lowlands. For instance, hummingbird species show relatively low rates of single-copy genome evolution at high elevations (Bleiweiss 1998) while mammals have lower substitution rates at higher elevations (Gillman et al. 2009). Even if rates of molecular evolution are constant along the elevation gradient, ecological limits to diversity could be higher in lowlands. Second, lineages might arise in high elevations and switch to lowlands frequently, which is suggested by various studies (Bates and Zink 1994; Roy 1997; Garcia-moreno et al. 1998; Hall 2005; Ribas et al. 2007). However, these mechanisms are not mutually exclusive and the methods to explore their relative contributions should treat them as interdependent.
For certain taxonomic and geographic scales there are examples of highlands being richer in species than lowlands. In the case of Ericacea species, the rate of lineages dispersing from high areas to lowlands is high, but the rate of montane speciation is even higher which maintains the gradient of richness increasing with elevation (Schwery et al. 2015). Drummond et al. (2012) and Hughes and Eastwood (2006) find a strong association between high rates of speciation and mountain habitats in Lupinus in North America and the Andes. Recently, Quintero and Jetz (2018) reported that within mountain systems bird species show the highest richness and (recent) speciation rates at intermediate elevations. However, what is needed to fully understand the elevational diversity gradient is a global phylogenetic diversification analysis based on a dynamic model of dispersal and elevation-dependent speciation and extinction on the entire time scale of the avian diversification.
Here, we therefore study diversification across the elevational gradient of passerine birds, a group that occupies all habitats from lowland rainforests to areas above the snowline and may persist at higher elevations than other vertebrate groups. We investigate whether diversification rates in montane birds are different from those of lowland birds, and/or whether downward dispersal rates are higher than upward dispersal rates, using a dynamic model of dispersal and elevation-dependent diversification.
METHODS
4
We used the Bayesian pseudo-posterior distribution of time-calibrated phylogenies provided by Jetz et al. (2012; www.birdtree.org), which includes 9993 species of the world’s birds out of a total of 10473 species. Using the program TreeAnnotator from the BEAST package v2.4.2 (Bouckaert et al. 2014) we produced a maximum-clade credibility tree from all available stage 2 trees with the Hackett backbone. We pruned this tree to the level of Passeriformes (5966 species).
Elevation data and large-scale realms
We compiled elevation data for passerines, recording lower and upper elevation bounds of their distribution, based on descriptions in the Handbook of the Birds of the World (del Hoyo et al. 2016). We did not include occasional records at extreme elevation, and elevational distributions are based on breeding range, thus excluding wintering and transient elevation records. Species without known elevational distribution were assigned NA in the data set (n = 260) rather than removed, because our analysis can still make use of their phylogenetic position (see below).
We defined 3 elevational bands: lowlands (from sea level up to 1500 meter above sea level), mid-montane areas (1500-3000 m), and high montane areas (> 3000 m). These categories broadly agree with those established by Chapman (1926) for Neotropical montane birds based on dominant vegetation zones associated with the tropics, subtropics, and alpine zones, respectively. Species could also inhabit more than one elevational band which adds three more categories: low-mid, mid-high and full range (i.e., species living from lowlands to high montane areas). The number of species for each category is shown in Table 1. From hereon we refer to a category as the state of the species.
Table 1. Elevational distribution of passerine diversity by geographic and biome realm. We defined 3
elevational bands: lowlands (from sea level up to 1500 meter above sea level), mid-montane areas (1500-3000 m), and high montane areas (> (1500-3000 m). Species could also inhabit more than one elevational band which adds three more categories: low-mid, mid-high and full range (i.e., species living from lowlands to high montane areas).
Elevation Total Temperate Tropical Old World New World
High 79 27 52 35 44 Medium-High 406 90 316 179 227 Medium 126 13 113 63 63 Low-Medium 2140 314 1826 1367 773 Low 2561 306 2255 1392 1169 Full-range 394 133 261 240 154 Total 883 4823 3276 2430
We accounted for the fact that drivers of diversification may differ across latitude/biome and longitude/continents in the following way. First, we performed a null analysis with no effect of latitudinal/longitudinal or realm on diversification rates and thus only considering the effects of differences in elevation, resulting in 6 states (global analysis). Second, we distinguished species occurring the tropics from species from those in temperate areas (latitudinal analysis), and both latitude and elevation can differentially affect diversification. This means that the number of states increased to 12 (e.g., a species could be a tropical lowland species, or in a temperate mid-high state). We classified species as tropical when most of their geographic distribution (del Hoyo et al. 2016) lies between tropics of Cancer and Capricorn. Similarly, we ran a third analysis where we separated Old World from New World species (longitudinal analysis), again resulting in 12 possible states (e.g. Old World low-mid state) in which both longitude and elevation can differentially affect diversification. We did not perform an analysis where diversification rates depend on elevation, latitude and longitude, because the large state space required for such an analysis was computationally unfeasible and numerically unstable given the size of our phylogeny.
State-dependent diversification analysis and parametrization
We used the SSE framework (State-dependent Speciation and Extinction) which allows determining whether diversification rates of a clade are associated with an evolving trait (Maddison et al. 2007). In this model, speciation rate (λi) or extinction rate (μi) of a lineage depends on its trait state i (here elevation, or a combination of elevation and latitude/biome or longitude/continent). However, this state is not static but the species can switch to a different state at a given rate qij, where i and j represent the state of origin and the state of destination, respectively. This allows us to use trait and branching patterns simultaneously to study macroevolutionary dynamics. In other words, when lineages living at different elevations experience different speciation/extinction regimes, the shift from one elevational band to another will have an effect on diversification rates. Statistical support for elevation affecting diversification rates is found when the likelihood of a model where speciation (or extinction) differs across elevation categories (Examined Trait-Dependent model, ETD) is higher – after correcting for differences in numbers of parameters - than a model where rates are associated to an unknown (hidden or concealed) trait (Concealed Trait-Dependent, CTD) and a model with constant rates (CR) (Beaulieu and O’Meara 2016; Herrera-Alsina et al. 2018). We used the R package secsse (Several Examined and Concealed States-Dependent Speciation and Extinction, Herrera-Alsina et al. 2018a) which computes and optimizes the likelihood of the model with 2 or more states.
The transition between elevations can take place according to two scenarios: expansion-contraction and up-down. In both types of transition, species can only transit to an adjacent band (i.e., from low to low-mid or from high to mid-high). Under the
expansion-130 131
4
contraction model, all expansion rates (i.e., change from one single band to two, or from two to three) are equal but different from contraction rates (i.e., change from two bands to a single band, or from three to two bands). In the up-down scenario, “uphill” transitions have a different rate than “downhill” transitions. Note that under the up-down scenario, changing from low-mid elevation to low elevation means going downhill. For the concealed trait we assumed a similar structure. We also included models that allow the transition rates in the concealed trait to be different from the rates of the examined trait, in order to relax the assumption that both traits evolve at the same pace. Because differences in diversification across states could be due to either differential speciation or extinction rates, we tested both cases for all the models.
For the longitudinal and latitudinal analyses, we assumed that the exchange of species between realms (realm exchange) happens to and from the same elevational band (low, mid and high only). When estimating these rates, they are assumed to be all the same (one single rate for dispersal to other latitudes/longitudes) but different from the rates of transition across elevations within latitudinal or longitudinal zones.
In order to keep the number of estimated parameters as low as possible during the likelihood optimization, we only optimized the rates of low, mid and high elevation and used these values to average the states that are a combination of those elevations (e.g. the low and mid elevation rates of speciation are averaged to yield the rate of low-mid elevation). In the case of longitudinal/latitudinal analyses, rates of diversification at different latitude/longitude differed by the same factor, which was optimized as well (e.g. the rates at latitude or longitude are this factor times the rates of the other latitude/longitude).
To prevent finding only local optima during the likelihood optimization, we used five different initial parameter sets. The first set of parameters were the estimates of speciation and extinction from a birth-death model fit to the branching times, transition rate was a fifth of speciation rate. For the second set we doubled the speciation rates of the first set, and halved the transition rates. In the third, we halved the speciation rates of the first set and doubled the transition rates. Similarly, the fourth had doubled extinction rates and halved transition rates, and the fifth had halved extinction rates and doubled transition rates compared to the first set. The highest likelihood of the five starting points was taken as the global optimum and used to compare across models. We used AIC weights – thus penalizing the number of free parameters - to select the best models per analysis.
Our global, latitudinal and longitudinal analyses differ in their assumptions on what factors diversification rates depend on (elevation only, elevation + latitude, elevation + longitude). Using only the data necessary to study these dependencies would prevent model comparison, because the data sets would differ. Therefore, we made the AIC values
comparable by adding an extra likelihood term to the likelihood computed by secsse that covers the transitions not covered in secsse, using the function fitDiscrete from the R package Geiger (Harmon et al. 2008). That is, for the longitudinal analysis we added the (maximum) loglikelihood of a model of transitions in latitude given the phylogenetic tree, for the latitudinal analysis we added the (maximum) loglikelihood of a model of transitions in longitude given the phylogenetic tree, and for the global analysis we added the (maximum) loglikelihood of a model of transitions in both longitude and latitude – which in fact is the sum of the two previous loglikelihoods.
RESULTS
Speciation, extinction, and dispersal across elevations
We found that a CTD model with concealed-state dependent speciation rates had the highest support (AICw ~ 0.99) indicating that lineages are not homogeneous in speciation rates but this variation cannot be explained by elevation (Table 2). This heterogeneity in
diversification is due to differences in speciation rates rather than in extinction rates as elevation-dependent extinction models performed poorly. The results are robust to latitudinal and longitudinal variation (i.e., differences between tropical and temperate regions and between Old World and New World; Table 2)
Our analyses suggest that up-down models are more likely (Figure 1) than
expansion-contraction models: all expansion-expansion-contraction models ranked low in AIC weight. Furthermore, the best-supported model in each analysis has different transition rates for examined and concealed traits (Table 2).
4
ble 2 . M od els of pa sse rin e d iv ersi ficat io n d ep en ding o n elev atio n w hil e a cco un tin g fo r p ote ntia l la titu din al an d lo ng itu dina l ren ce s (i .e . t em pe ra te - tro pical a nd te m pe rate - tro pi ca l r eg io ns ). Spec ia tion or ex tinc tion c an d ep end on t he el ev at ional ng e (Ex am ine d T rait -Dep en de nt; E TD) , o r o n a n u nkn ow n tra it (C on ce ale d Trait -Dep en de nt; CT D) . A m od el w ith Con sta nt es (C R) a cross li ne ag es is als o incl ud ed . M od els assu m e e xp licit co nstra int s (U p-D ow n an d E xpans io n-Co ntra ctio n) o n w li ne ag es d ispe rse a cross elev atio ns o ve r ev olut ion ary tim e. Th e fo urth c olu m n s ho w s w he th er th e e xa m ine d a nd nce aled tra its are as su m ed to tra nsit at th e sa m e r ate s. W he n la tit ud ina l a na ly sis is ind icat ed in th e fif th c olu m n, lat itu di nal te w as inco rpo ra te d in th e se cs se an al ys es (P ar tial Log lik el ihood ) w her ea s l on gi tud e w as a nal yz ed s epar at el y under a ple m od el o f tra it e vo lut ion a nd a dd ed to o bta in th e To ta l L og like lih oo d. L ik ew ise, w he n L on gitu dina l a na ly sis is ind icat ed , itu de w as a na ly ze d se pa rat ely (see M et ho ds). W he n Gl ob al is in dicat ed , a m od el of tra it e vo lut io n t ha t co nsi de rs th e nsition s a cross L atit ud e a nd L on gitu de w ere an alyz ed se pa rat ely. Fo r e ach m od el, t he n um be r o f f re e p ara m ete rs k is icat ed , A IC w eig hts a nd Δ AIC v alue s a re co m pu te d u sing T ota l L og like lih oo ds a nd u se d to co m pa re m od els. B est rfo rm in g la titu din al, l on gitu dina l a nd g lob al m od els a re hig hli gh te d in g ray . ait -El ev atio n t ran si tion m ode Varia bi lit y across state Q ex amin ed = Qcon cea led ? Ty pe of an al ysi s Part ial Log ik el iho od Tot al Lo gl ik el iho od k AI Cw Δ AIC D Up -D own Spe ci atio n No Lat itu di nal -26 73 0.47 -26 78 5.54 12 ~ 0 .99 0 D Up -D own Spe ci atio n Ye s Lat itu di nal -26 85 6.93 -26 91 2.00 9 < 0 .00 01 247 D Up -D own Spe ci atio n Ye s Lat itu di nal -27 06 9.99 -27 12 5.06 9 < 0. 00 01 673 D Up -D own Spe ci atio n No Lat itu di nal -27 07 0.06 -27 12 5.13 12 < 0 .00 01 679 Up -D own Spe ci atio n Ye s Lat itu di nal -27 28 7.98 -27 34 3.05 6 < 0 .00 01 1103 D Up -D own Ex tinctio n Ye s Lat itu di nal -27 28 7.94 -27 34 3.01 9 < 0 .00 01 1109 Up -D own Spe ci atio n No Lat itu di nal -27 30 7.17 -27 36 2.24 9 < 0 .00 01 1147 D Up -D own Ex tinctio n No Lat itu di nal -27 31 0.58 -27 36 5.65 12 < 0 .00 01 1160 D Up -D own Ex tinctio n No Lat itu di nal -27 32 9.65 -27 38 4.72 12 < 0 .00 01 1198 D Up -D own Spe ci atio n No G lob al -24 92 7.06 -27 41 8.86 10 < 0 .00 01 1263 D Up -D own Ex tinctio n Ye s Lat itu di nal -27 40 9.29 -27 46 4.36 9 < 0 .00 01 1352 D Up -D own Spe ci atio n Ye s G lob al -24 98 0.61 -27 47 2.41 8 < 0 .00 01 1366 D Ex pa nsi on -C on tractio n Spe ci atio n No Lat itu di nal -27 73 5.41 -27 79 0.48 12 < 0 .00 01 2010 D Ex pa nsi on -C on tractio n Spe ci atio n No G lob al -25 32 2.63 -27 81 4.43 10 < 0 .00 01 2054 D Ex pa nsi on -C on tractio n Spe ci atio n Ye s G lob al -25 33 5.06 -27 82 6.86 8 < 0 .00 01 2075 D Up -D own Spe ci atio n Ye s G lob al -25 33 6.69 -27 82 8.49 8 < 0 .00 01 2078 D Up -D own Spe ci atio n No G lob al -25 33 6.94 -27 82 8.74 10 < 0 .00 01 2082 D Up -D own Spe ci atio n No Lo ng itud ina l -25 49 1.37 -27 92 8.10 12 < 0 .00 01 2285 D Ex pa nsi on -C on tractio n Spe ci atio n Ye s Lat itu di nal -27 87 7.70 -27 93 2.77 9 < 0. 00 01 2288 Up -D own Spe ci atio n Ye s G lob al -25 47 2.91 -27 96 4.72 6 < 0 .00 01 2346 D Up -D own Ex tinctio n Ye s G lob al -25 47 2.88 -27 96 4.68 8 < 0 .00 01 2350 Up -D own Spe ci atio n No G lob al -25 47 2.91 -27 96 4.72 8 < 0 .00 01 2350 D Up -D own Ex tinctio n Ye s G lob al -25 47 2.92 -27 96 4.72 8 < 0 .00 01 2350 D Up -D own Ex tinctio n No G lob al -25 47 2.88 -27 96 4.68 10 < 0 .00 01 2354 D Up -D own Ex tinctio n No G lob al -25 47 6.31 -27 96 8.11 10 < 0 .00 01 2361 D Up -D own Spe ci atio n Ye s Lo ng itud ina l -25 54 8.50 -27 98 5.23 9 < 0. 00 01 2393 D Ex pa nsi on -C on tractio n Spe ci atio n Ye s Lat itu di nal -28 19 1.26 -28 24 6.33 9 < 0 .00 01 2916 D Ex pa nsi on -C on tractio n Spe ci atio n Ye s G lob al -25 75 6.98 -28 24 8.78 8 < 0 .00 01 2918 D Ex pa nsi on -C on tractio n Spe ci atio n No G lob al -25 76 6.41 -28 25 8.21 10 < 0 .00 01 2941 D Ex pa nsi on -C on tractio n Spe ci atio n No Lat itu di nal -28 21 8.90 -28 27 3.97 12 < 0 .00 01 2977 D Up -D own Spe ci atio n Ye s Lo ng itud ina l -25 84 8.43 -28 28 5.16 9 < 0 .00 01 2993 D Up -D own Spe ci atio n No Lo ng itud ina l -25 87 0.24 -28 30 6.97 12 < 0 .00 01 3043 D Ex pa nsi on -C on tractio n Ex tinctio n No G lob al -25 83 0.22 -28 32 2.02 10 < 0 .00 01 3069 Ex pa nsi on -C on tractio n Spe ci atio n Ye s G lob al -25 84 1.22 -28 33 3.02 6 < 0 .00 01 3083 D Ex pa nsi on -C on tractio n Ex tinctio n Ye s G lob al -25 84 0.72 -28 33 2.52 8 < 0 .00 01 3086 Ex pa nsi on -C on tractio n Spe ci atio n No G lob al -25 84 1.22 -28 33 3.02 8 < 0 .00 01 3087 D Ex pa nsi on -C on tractio n Ex tinctio n Ye s G lob al -25 84 1.23 -28 33 3.03 8 < 0 .00 01 3087 D Ex pa nsi on -C on tractio n Ex tinctio n No G lob al -25 84 1.24 -28 33 3.04 10 < 0 .00 01 3091 D Ex pa nsi on -C on tractio n Spe ci atio n Ye s Lo ng itud ina l -25 91 4.45 -28 35 1.18 9 < 0 .00 01 3125 D Ex pa nsi on -C on tractio n Spe ci atio n No Lo ng itud ina l -25 92 3.45 -28 36 0.18 12 < 0 .00 01 3149 D Ex pa nsi on -C on tractio n Ex tinctio n Ye s Lat itu di nal -28 30 9.92 -28 36 4.99 9 < 0 .00 01 3153 Ex pa nsi on -C on tractio n Spe ci atio n Ye s Lat itu di nal -28 42 4.00 -28 47 9.07 6 < 0 .00 01 3375 D Ex pa nsi on -C on tractio n Ex tinctio n No Lat itu di nal -28 42 0.98 -28 47 6.05 12 < 0. 00 01 3381 Ex pa nsi on -C on tractio n Spe ci atio n No Lat itu di nal -28 42 4.00 -28 47 9.07 9 < 0 .00 01 3381 D Ex pa nsi on -C on tractio n Ex tinctio n Ye s Lat itu di nal -28 42 4.01 -28 47 9.08 9 < 0 .00 01 3381134 135
4
CR Up -D own Spe ci atio n Ye s Lo ng itud ina l -26 04 7.86 -28 48 4.59 6 < 0 .00 01 3386 CT D Ex pa nsi on -C on tractio n Ex tinctio n No Lat itu di nal -28 42 4.00 -28 47 9.07 12 < 0 .00 01 3387 CR Up -D own Spe ci atio n No Lo ng itud ina l -26 04 7.87 -28 48 4.60 9 < 0 .00 01 3392 ET D Up -D own Ex tinctio n No Lo ng itud ina l -26 04 7.69 -28 48 4.42 12 < 0 .00 01 3398 CT D Up -D own Ex tinctio n No Lo ng itud ina l -26 04 7.91 -28 48 4.64 12 < 0 .00 01 3398 ET D Up -D own Ex tinctio n Ye s Lo ng itud ina l -26 07 5.04 -28 51 1.77 9 < 0 .00 01 3446 CT D Up -D own Ex tinctio n Ye s Lo ng itud ina l -26 11 3.74 -28 55 0.47 9 < 0. 00 01 3524 ET D Ex pa nsi on -C on tractio n Spe ci atio n Ye s Lo ng itud ina l -26 40 3.33 -28 84 0.06 9 < 0 .00 01 4103 ET D Ex pa nsi on -C on tractio n Spe ci atio n No Lo ng itud ina l -26 40 9.87 -28 84 6.60 12 < 0 .00 01 4122 ET D Ex pa nsi on -C on tractio n Ex tinctio n Ye s Lo ng itud ina l -26 51 5.80 -28 95 2.53 9 < 0 .00 01 4328 CR Ex pa nsi on -C on tractio n Spe ci atio n Ye s Lo ng itud ina l -26 57 4.03 -29 01 0.76 6 < 0 .00 01 4438 CR Ex pa nsi on -C on tractio n Spe ci atio n No Lo ng itud ina l -26 57 4.03 -29 01 0.76 9 < 0 .00 01 4444 CT D Ex pa nsi on -C on tractio n Ex tinctio n Ye s Lo ng itud ina l -26 57 4.03 -29 01 0.76 9 < 0 .00 01 4444 ET D Ex pa nsi on -C on tractio n Ex tinctio n No Lo ng itud ina l -26 57 2.24 -29 00 8.97 12 < 0 .00 01 4447 CT D Ex pa nsi on -C on tractio n Ex tinctio n No Lo ng itud ina l -26 57 4.03 -29 01 0.76 12 < 0 .00 01 4450Figure 1. Rates of passerine dispersal across elevation, latitude and longitude over evolutionary time. A)
Best supported model of transitions across elevational bands. Three elevational bands were defined: lowlands, mid-montane areas, and high montane areas. Species could also inhabit more than one elevational band which adds three more categories: low-mid, mid-high and full range (i.e., species living from lowlands to high montane areas). Lineages disperse uphill (rate indicated with a gray arrow) at a lower rate than downhill (indicated with a black arrow). Note that lineages could not disperse directly to certain elevational bands (e.g., from lowlands to high montane areas) without transiting through intermediate elevational categories. B) Estimated rates of latitudinal (tropics and temperate regions) and longitudinal (Old and New World) exchange of lineages.
According to the best model across latitudinal, longitudinal and global analyses, the estimated rate of dispersal uphill is 0.072 and the rate of downhill dispersal is 0.189 (Figure 1). This pattern of higher downhill than uphill dispersal holds true in all three
analyses. In all cases, extinction was estimated to be low: 0.0001.
Geographic origin of passerines and accumulation of lineages over time
Figure 1. Rates of passerine dispersal across elevation, latitude and longitude over evolutionary time. A)
Best supported model of transitions across elevational bands. Three elevational bands were defined: lowlands, mid-montane areas, and high montane areas. Species could also inhabit more than one elevational band which adds three more categories: low-mid, mid-high and full range (i.e., species living from lowlands to high montane areas). Lineages disperse uphill (rate indicated with a gray arrow) at a lower rate than downhill (indicated with a black arrow). Note that lineages could not disperse directly to certain elevational bands (e.g., from lowlands to high montane areas) without transiting through intermediate elevational categories. B) Estimated rates of latitudinal (tropics and temperate regions) and longitudinal (Old and New World) exchange of lineages.
According to the best model across latitudinal, longitudinal and global analyses, the estimated rate of dispersal uphill is 0.072 and the rate of downhill dispersal is 0.189 (Figure 1). This pattern of higher downhill than uphill dispersal holds true in all three
analyses. In all cases, extinction was estimated to be low: 0.0001.
Geographic origin of passerines and accumulation of lineages over time
Figure 1. Rates of passerine dispersal across elevation, latitude and longitude over evolutionary time. A)
Best supported model of transitions across elevational bands. Three elevational bands were defined: lowlands, mid-montane areas, and high montane areas. Species could also inhabit more than one elevational band which adds three more categories: low-mid, mid-high and full range (i.e., species living from lowlands to high montane areas). Lineages disperse uphill (rate indicated with a gray arrow) at a lower rate than downhill (indicated with a black arrow). Note that lineages could not disperse directly to certain elevational bands (e.g., from lowlands to high montane areas) without transiting through intermediate elevational categories. B) Estimated rates of latitudinal (tropics and temperate regions) and longitudinal (Old and New World) exchange of lineages.
According to the best model across latitudinal, longitudinal and global analyses, the estimated rate of dispersal uphill is 0.072 and the rate of downhill dispersal is 0.189 (Figure 1). This pattern of higher downhill than uphill dispersal holds true in all three
analyses. In all cases, extinction was estimated to be low: 0.0001.
4
We found that lineages disperse latitudinally and longitudinally at rates that are smaller than dispersal across elevational bands. The best model (Table 2) shows that the rate of
lineage exchange between tropics and temperate zones is 0.0187 and events of lineage dispersal across Old Word to New World happen rarely with a rate of 0.00016.
The best supported models all pointed to an Old World state at the root of the phylogeny. They did not agree on the latitudinal root: the latitudinal analysis (i.e. latitudinal transitions modelled by secsse) suggested a temperate origin, whereas the longitudinal analysis (i.e. latitudinal transitions modelled by fitDiscrete model) suggested a tropical lineage as the most likely state at the root.
DISCUSSION
In spite of clear differences in passerine diversity at different elevations, we find no differences in speciation or extinction rates across elevational zones. Instead, a Concealed-Trait Dependent model is identified as the most likely which suggests that the altitudinal gradient of diversity depends on downhill dispersal. The CTD model being the best suggests that there are differences in speciation rates but they are not related to elevation.
There are two caveats in our analysis. First, many of the world’s high-elevation habitats did not develop until relatively recently in geological history which gives lowlands areas a head start in species accumulation. The Andes did not exist until ~15 mya (Graham 2009), the Himalayan-Tibetan orogenic system’s higher elevations are of similar age (Yin and Harrison 2000), and the New Guinea Highlands built up to current-elevation levels at ~5 mya (van Ufford and Cloos 2005). Many low-elevation regions are much older and more climatically stable, including those associated with cratonic systems like the Guiana Shield (Lujan and Armbruster 2011) and the large pan-African craton (Rino et al. 2008). Second, the lack of a signal of differential net diversification rates may be due to the broad taxonomic scale at which the analysis was performed, as evolutionary drivers could differ across passerine families obscuring the overall importance of elevation (see below). Moreover, the drivers of evolution can be acting at a smaller geographic scale than the broad-scale latitudinal and longitudinal variation we considered, which might hamper the detectability of an association of speciation rates and elevation. As shown by Quintero and Jetz (2018), patterns of diversification and species accumulation can be different from one mountain system to another where particular circumstances could have shaped diversity across elevation. For instance Fjeldså (1994) showed that African montane regions hold a larger number of young lineages than surrounding lowlands which they explained by the presence of forest refuges that housed avian species during the Pleistocene. For other highland systems, Pleistocene refugia might not have played an important role.
Our finding of uniform diversification rates along the elevational gradient contrasts with that of a recent study showed that under different scenarios, diversification rates always increase with altitude regardless of a strong or mild gradient in species richness (Quintero and Jetz 2018). It is important to note that the methods used by Quintero and Jetz are different from ours. Quintero and Jetz calculated diversification rate for each extant passerine bird species (tip DR) within their geographic scope, and then averaged this across all the species inhabiting each location (locations were distributed along altitudinal gradient). However, this approach only considers recent diversification and hence ignores deep-time diversification events, and links the tip diversification rates to the current location, thereby ignoring the possibility that the recent diversification events occurred at a different location (elevation) followed by dispersal into the current location. Our model accounts for speciation events occurring throughout the entire history of the passerine radiation and explicitly models the elevation where the speciation occurs, by accounting for the dispersal of lineages across altitudinal bands.
We found that lineages are more likely to move to a lower altitude over evolutionary time and this seems to contrast with what has been reported for some taxa. For Leptopogon flycatchers (Bates and Zink 1994) a repeated upward movement into highlands has been suggested. Chat-tyrants (Ochthoeca, Silvicultrix; Garcia-moreno et al. 1998) and Andropadus greenbuls (Roy 1997) similarly show upward dispersal, followed by in situ montane diversification. Ribas et al. (2007) found that most transitions in Pionus parrots are from lowlands to highlands, but the timing of these transitions is before major Andean orogenic events, and thus highland taxa may have been passively transported to higher elevations with their preferred habitats. McGuire et al. (2007) showed that within Trochilidae there were cases of colonization of the lowlands by Andean lineages, but also the reverse. Our findings across passerines, do not necessarily contradict these earlier results because uphill dispersal followed by a radiation is still possible in our model. Nevertheless, our study shows that such upward movement is the exception rather than the rule and that dispersal from high to low elevations is the dominant biogeographic trend. Furthermore, in these studies, the age of closely related lineages inhabiting different elevations is often used to infer what elevation the common ancestor lived at and to describe an uphill or downhill change. Our approach is different because we explicitly modelled the switch of lineages from one elevation to another.
Our results suggest that passerine birds had an Old World origin. The Old World origin we report is in line with Oliveros et al. (2019) who recently found that passerine species are likely to have originated in the Australo-Pacific region. Moreover, we do not find clear evidence on where this clade arose. The temperate origin seems at odds with the fact that the vast majority of passerines (84%) inhabit tropical areas in the present. However, the estimated rate of dispersal from temperate to tropical regions suggests that tropical areas were colonized not long after the clade first appeared. Neither the latitudinal or
138 139
4
longitudinal analysis indicates lowlands as the most likely elevation at which passerines initially developed, but perhaps the high rates of overall altitudinal dispersal do not allow an accurate inference of elevational origin. Moreover our reported origin is actually the state of the ancestral species just before its split at the crown of the phylogeny, so the true origin of the species (at the stem age) may have been different.
Our findings suggest that even though there is an important heterogeneity of both richness along elevation and diversification rates across the passerine clade, the relationship between these patterns is not causal, but they are connected via lineage dispersal. Phylogenies contain information on how diversity gradients and other spatial patterns were formed, even when speciation and extinction do not show a gradient themselves.
ACKNOWLEDGMENTS
PVE thanks the Stichting P.A. Hens Memorial Fund, an American Ornithologists’ Union Alexander Wetmore Memorial Research Award, the American Museum of Natural History Frank M. Chapman Memorial Fund, the Adaptive Life Program of the Groningen Institute for Evolutionary Life Sciences, Faculty of Science and Engineering at the University of Groningen for financial support. LHA thanks the Consejo Nacional de Ciencia y Tecnologia of Mexico for funding (CVU 385304 L). RSE thanks the Netherlands Organization for Scientific Research (NWO) for financial support through a VICI grant.
REFERENCES
Bates J.M., Zink R.M. 1994. Evolution into the Andes : Molecular Evidence for Species Relationships in the Genus Leptopogon Linked references are available on JSTOR for this article : Auk. 111:507–515.
Beaulieu J.M., O’Meara B.C. 2016. Detecting hidden diversification shifts in models of trait-dependent speciation and extinction. Syst. Biol. 65:583–601.
Bleiweiss R. 1998. Slow rate of molecular evolution in high-elevation hummingbirds. Proc. Natl. Acad. Sci. 95:612–616.
Bouckaert R.R., Heled J., Kuhnert D., Vaughan T., Wu C., Xie D., Suchard M.A., Rambaut A., Drummond A.J. 2014. BEAST 2 : A Software Platform for Bayesian Evolutionary Analysis. PloS Comput. Biol. 10:1–6.
Drummond C., Eastwood R.J., Miotto S.T.S., Hughes C.E. 2012. Multiple Continental Radiations and Correlates of Diversification in Lupinus ( Leguminosae ): Testing for Key Innovation with Incomplete Taxon Sampling. Syst. Biol. 61:443–460.
Fjeldså J. 1994. Geographical patterns for relict and young species of birds in Africa and South America and implications for conservation priorities. Biodivers. Conserv. 226:207– 226.
Garcia-moreno J., Arctander P., Fjeldså J. 1998. Pre-Pleistocene Differentiation among Chat-Tyrants. Condor. 100:629–640.
Gillman L.N., Keeling D.J., Ross H.A., Wright S.D. 2009. Latitude , elevation and the tempo of molecular evolution in mammals. Proc. R. Soc. B Biol. Sci. 276:3353–3359. Graham A. 2009. The andes: a geological overview from a biological perspective. Ann. Missouri Bot. Gard. 96:371–385.
Hall J.P.W. 2005. Montane speciation patterns in Ithomiola butterflies (Lepidoptera : Riodinidae ): are they consistently moving up in the world ? Proc. R. Soc. B Biol. Sci. 272:2457–2466.
Harmon L.J., Weir J.T., Brock C.D., Glor R.E., Challenger W. 2008. GEIGER : investigating evolutionary radiations. Bioinformatics. 24:129–131.
Herrera-Alsina L., van Els P., Etienne R.S. 2018. Detecting the Dependence of Diversification on Multiple Traits from Phylogenetic Trees and Trait Data. Syst. Biol. 0:1– 12.
del Hoyo J., Elliott A., Sargatal J., Christie D.A., Kirwan G., editors. 2016. Handbook of the Birds of the World Alive. Barcelona: Lynx Edicions.
Hughes C., Eastwood R. 2006. Island Radiation on a Continental Scale : Exceptional Rates of Plant Diversification after Uplift of the Andes Island radiation on a continental scale : Exceptional rates of plant diversification after uplift of the Andes. Proc. Natl. Acad. Sci. 103:10334–10339.
Jetz W., Thomas G.H., Joy J.., Hartmann K., Mooers A.O. 2012. The global diversity of birds in space and time. Nature. 491:1–5.
Körner C., Spehn E.M. 2002. Mountain Biodiversity: a Global Assessment. Boca Raton: Parthenon.
Kozak K.H., Wiens J.J. 2010. Niche Conservatism Drives Elevational Diversity Patterns in Appalachian Salamanders. Am. Nat. 176:DOI: 10.1086/653031.
Lujan N.K., Armbruster J.W. 2011. The Guiana Shield. In: Albert J.S., Reis R.E., editors. Historical Biogeography of Neotropical Freshwater Fishes. Berkeley: University of California Press. p. 211–224.
Maddison W.P., Midford P.E., Otto S.P. 2007. Estimating a binary character’s effect on speciation and extinction. Syst. Biol. 56:701–710.
4
McGuire J.A., Witt C.C., Altshuler D.L., Remsen J. V. 2007. Phylogenetic Systematics and Biogeography of Hummingbirds: Bayesian and Maximum Likelihood Analyses of Partitioned Data and Selection of an Appropriate Partitioning Strategy. Syst. Biol. 56:837– 856.
Merckx V.S.F.T., Hendriks K.P., Beentjes K.K., Mennes C.B., Becking L.E., Peijnenburg K.T.C.A., Afendy A., Arumugam N., de Boer H., Biun A., Buang M.M., Chen P.-P., Chung A.Y.C., Dow R., Feijen F.A.A., Feijen H., Soest C.F., Geml J., Geurts R., Gravendeel B., Hovenkamp P., Imbun P., Ipor I., Janssens S.B., Jocqué M., Kappes H., Khoo E., Koomen P., Lens F., Majapun R.J., Morgado L.N., Neupane S., Nieser N., Pereira J.T., Rahman H., Sabran S., Sawang A., Schwallier R.M., Shim P.-S., Smit H., Sol N., Spait M., Stech M., Stokvis F., Sug J.B., Suleiman M., Sumail S., Thomas D.C., van Tol J., Tuh F.Y.Y., Yahya B.E., Nais J., Repin R., Lakim M., Schilthuizen M. 2015. Evolution of endemism on a young tropical mountain. Nature. 524:347–350.
Oliveros C.H., Field D.J., Ksepka D.T., Barker F.K., Aleixo A., Andersen M.J., Alström P., Benz B.W., Braun E.L., Braun M.J., Bravo G.A., Brumfield R.T., Chesser R.T., Claramunt S., Cracraft J., Cuervo A.M., Derryberry E.P., Glenn T.C., Harvey M.G., Hosner P.A., Joseph L., Kimball R.T., Mack A.L., Miskelly C.M., Peterson A.T., Robbins M.B., Sheldon F.H., Silveira L.F., Smith B.T., White N.D., Moyle R.G., Faircloth B.C. 2019. Earth history and the passerine superradiation. Proc. Natl. Acad. Sci. U. S. A. 116:7916–7925. Quintero I., Jetz W. 2018. Global elevational diversity and diversification of birds. Nature. 555:246–250.
Ribas C.C., Moyle R.G., Miyaki C.Y., Cracraft J. 2007. The assembly of montane biotas : linking Andean tectonics and climatic oscillations to independent regimes of diversification in Pionus parrots. Proc. R. Soc. B Biol. Sci. 274:2399–2408.
Rino S., Kon Y., Sato W., Maruyama S., Santosh M., Zhao D. 2008. The Grenvillian and Pan-African orogens : World ’ s largest orogenies through geologic time , and their implications on the origin of superplume. Gondwana Res. 14. 14:51–72.
Roy M. 1997. Recent diversification in African greenbuls (Pycnonotidae : Andropadus) supports a montane speciation model. Proc. R. Soc. London B Biol. Sci. 264:1337–1344. Schwery O., Onstein R.E., Bouchenak-khelladi Y., Xing Y., Carter R.J., Linder H.P. 2015. As old as the mountains : the radiations of the Ericaceae. New Phytol. 207:355–367. van Ufford A.Q., Cloos M. 2005. Cenozoic tectonics of New Guinea. Am. Assoc. Pet. Geol. Bull. 89:119–140.
Weir J.T. 2006. Divergent timing and patterns of species accumulation in lowland and highland neotropical birds. Evolution (N. Y). 60:842–855.
Yin A., Harrison T.M. 2000. Geologic Evolution of the Himalayan-Tibetan Orogen. Annu. Rev. Earth Planet. Sci. 28:211–280.
