Supplementary Information

Can we detect oceanic biodiversity hotspots from space?

Silvia De Monte1,2,3, Alice Soccodato4, Severine Alvain5 and Francesco d'Ovidio4

1Ecole Normale Supérieure, Ecologie et Evolution, Paris, France; 2Université Pierre et MarieCurie-Paris 6, Ecologie et Evolution, Paris, France; 3 CNRS, Ecologie et Evolution, Paris, France; 4 Laboratoired'Océanographie et du Climat Expérimentation et Approches Numériques, CNRS/MNHN/IRD/UPMC,PARIS, France and 5 Laboratoire d’Océanologie et de Géosciences, CNRS-ULCO-Université Lille Nordde France, Wimereux, France

The ISME Journal (2013) 1-3

1. Data accessibility

The data used in this publication (average of the index τ computed at 1 degree resolution from the PHYSAT reanalysis of SeaWiFS ocean color measures), as well as the routines for generating the maps, are available on request by writing to demonte@bioloie.ens.fr.

2. The PHYSAT algorithm

Several algorithms aim at extracting biologically relevant information from spectral anomalies in remotely sensed optical signals (Brewin et al., 2010, Ciotti and Bricaud 2006, Hirata et al., 2011, Kostadinov et al., 2009, Raitsos et al., 2008, Uitz et al., 2006) or their heterogeneity (Rocchini et al. 2010).

The PHYSAT algorithm (Alvain et al., 2005) provides a classification of bio-optical anomalies of Ocean color (http://oceancolor.gsfc.nasa.gov/) based on the identification of specific signatures of normalized water leaving radiances (nLw) measured by ocean color satellites (http://log.univ-littoral.fr/Physat). This is an empirical method based on simultaneous in-situ pigments inventories (mainly from the GeP&Co campaigns: Dandonneau et al., 2004) and remote sensing measurements collected by the SeaWiFS ocean color sensor. Specific anomalies (nLw*) in terms of magnitude and -to a lesser extent- of shape can be empirically associated with the following dominant groups: nanoeucaryotes, Prochlorococcus, Synechoccocus, diatoms, coccolitophorids and Phaeocystis-like (Alvain et al, 2005; Alvain et al., 2008). The nLw* are defined as the second order variability of the satellite signal, obtained by dividing the actual nLw by a mean nLwref model which depends only on the standard Chlorophyll-a concentration. A set of criteria has been defined to characterize each group sampled in-situ, in function of nLw* spectrum. These criteria can be applied to global daily remote sensing observations to get a global map of the most frequent group of dominant phytoplankton. When no dominant groups are detected, pixels are classified as 'unidentified’. The geographical distribution and seasonal succession of PHYSAT groups have been studied using the global Sea-viewing Wide Field-of-view Sensor (SeaWiFS) ten year archive. PHYSAT observations are in good agreement with previous studies based on in-situ observations (Alvain et al., 2008) and a recent validation study based on independent in-situ measurements has permitted to partly unravel the underlying theoretical explanation of these criteria using a realistic model of radiative transfer for surface open ocean waters (Alvain et al., 2012).

Important shortcomings certainly exist, common to all the approaches -PHYSAT among them- that try to estimate phytoplanktonic community composition from remote sensing. Firstly, only a limited number of distinct bio-optical classes can be identified and related to properties of the ecosystem. Moreover, the categorization is subject to errors. We expect such uncertainty not to compromise our analysis, since our approach does not rely on the identification of a specific group, but only requires regions where different communities are stirred to appear as a mosaic in terms of the different PHYSAT bio-optical anomalies. The misclassification or undetectability of some patches would affect the quantitative value of τ at a given point, but not necessarily in a systematic way. Therefore, considering the temporal and spatial scale used in our analysis, we expect the position of the maxima and minima of our index to be generally unaffected.

3. Area-based Shannon entropy as a proxy of stirring-induced tracer diversity

The index τ is a local measure of the heterogeneity in the ocean surface bio-optical properties at a certain scale, that is the radius of a disk over which the occurrence frequencies of remote-sensed bio-optical frequencies are estimated.The rationale behind considering τ a possible proxy of biodiversity is that at one point in the ocean we expect to observe a higher diversity, the more heterogeneous the distribution of planktonic communities - in this case, emerging as bio-optical anomalies - around that point. Conversely, if the conditions at a given location are such to support high diversity, we can expect that this will reflect on the ability of that local community to prosper in nearby niches shaped by transport, thus creating high spatial heterogeneity in the nearby environment. These two cases correspond to two kinds of hotspots in Fig. 1: hotspots located at confluence regions, where nearby communities are mixed by horizontal stirring, and those located around specific geographical features, for instance islands or reefs. In our analysis we computed τ within disks of 100 km radius. Quantitatively, water masses that are 100 kms apart are mixed on a time scale of the order of few weeks, comparable to that of a planktonic bloom (d'Ovidio et al., 2010), so that we expect that each of the communities present in the area will also be locally represented. On the other hand, communities that are locally present will be given the chance to colonize niches in that radius before the bloom season is over.We stress here that this measure exploits a specific feature of oceanic ecosystems at the submesoscale, where horizontal transport and mixing occur on the same time scale of planktonic blooms (d'Ovidio et al. 2010). The concept of niches that are fixed in time, typical of terrestrial landscapes, does not apply to the open ocean environment. This is especially true at the scale ~10s of km (meso- and submesoscale), where microbial communities are disturbed (Connell 1978) by stirring on the same time scale of their demography. See also Paragraph 4.3 about the sensibility of τ on the disk radius.

This stirring-mediated diversity, which we try to capture, is only one component of oceanic biodiversity at a given location, since other processes may occur, that do not appear to the level of the dominant type or cannot be captured by our area-based index. One example of physical mechanisms other than horizontal stirring which is likely to provide a strong structuring effect to marine communities is the vertical dynamics occurring in upwelling regions (see Paragraph 4.1).

4. Comparison with biodiversity observations

In order to test whether a measure of biodiversity based on the τ index is consistent with other characterizations of biodiversity variability, we discuss here a comparison with previous studies of different aspects of marine ecosystem diversity:1. In-situ phytoplanktonic biodiversity measures of the AMT-2 cruise;2. Hotspots for higher trophic levels, retrieved mainly by telemetry tracking of top predators;3. Diversity-temperature relation for oceanic biota from a compilation of OBIS data, and from models for the evolution of trophic webs, based on metabolic theory.A detailed comparison with the ECCO2-Darwin Model, a global circulation model which provides a time-evolving physical and biological environment of a virtual phytoplanktonic ecosystem, is in progress. Even if such ecosystem is way simpler than any real biological community, there is still an overall agreement between the map of the τ index and Fig. 1 of Barton et al., 2010.

4.1 In-situ measures of planktonic biodiversity

We compare the τ index with in-situ oceanic biodiversity estimates, obtained by the Atlantic Meridional Transect programme (http://amt-uk.org/) that samples the Atlantic Ocean twice a year, measuring biological, chemical and physical parameters. We consider here the AMT-2 transect, that was run from April to May 1996. In Fig. S1 a, the trajectory is overlaid to the τ index global map averaged on the same two months. The AMT-2 cruise measured nano- and micro-planktonic (cyanobacteria, phytoplankton, heterotrophic plankton, ciliates) community composition in both cellular (n cells m-³) and carbon biomass (mg C m-³) by inverted microscopy and such data have already been analyzed to investigate planktonic community structure (Cermeño et al., 2008, Irigoien et al., 2004, Marañón et al., 2000). We compute the Shannon index HAMT by considering the abundance (integrated over the mixed layer) of the organisms classified in functional groups, and weighting organisms belonging to different size fractions by their carbon biomass, following Cermeño et al., 2008. The index is normalized with respect to the maximal value that would be attained if all genera observed in the cruise were equally represented at a single location, and is reported in table ST1. While considering depths separately, or using higher levels of taxonomic classification, results in slightly different values of the Shannon index, the overall pattern is the same irrespective of the particular method of computation of the Shannon index.

Station n Lat Long HAMT Station n Lat Long HAMT

1 48.7466 -8.0992 0.53 14 -7.4763 -29.7628 0.51

2 46.4879 -14.062 0.71 15 -11.2464 -31.4648 0.51

3 44.1269 -20.0239 0.83 16 -15.1714 -33.3054 0.46

4 39.8961 -19.9964 0.76 17 -19.9067 -35.5242 0.55

5 35.6675 -20.4248 0.74 18 -24.3494 -37.8635 0.49

6 30.9106 -21.2587 0.71 19 -27.6185 -40.7125 0.57

7 26.5409 -21.7865 0.56 20 -30.31 -43.4269 0.58

8 21.7646 -21.4735 0.60 21 -33.3946 -46.6163 0.72

9 11.7404 -21.0588 0.70 22 -36.0588 -48.8114 0.64

10 7.5785 -22.9468 0.64 23 -39.3514 -53.3012 0.70

11 3.4015 -24.8052 0.53 24 -43.4254 -54.7414 0.60

12 -0.1951 -26.3851 0.63 25 -47.5886 -55.597 0.68

13 -3.9165 -28.2124 0.68

Table ST1: AMT-2 cruise data: station number, position and Shannon index HAMT computed as indicated in the text.

The AMT measures of biodiversity are compared to the τ index at their resolution, which is lower than that obtainable through the satellite measures (Fig. S1 b). We consider the τ index at the same geographical position of the AMT-2 stations. We compute the mean of the τ index as in Fig. 1, both on all days where the PHYSAT reanalysis was available (2003-2010), and restricted to the months in which the cruise took place (Fig S1 b). The comparison between the means performed over a yearly or bimonthly time frame (black and blue lines of Fig. S1) shows that, although some regions may display a rather strong seasonal variability, the latitudinal overall pattern with bands of biodiversity appears to be preserved.

Figure S1: (a) Sampling stations for the AMT-2 transect overlayed to the global map of the τ index averaged over the cruise monthly time lag (April-May, from 2003 to 2010); (b) Diversity as a function of latitude: blue, normalized τ index calculated at the AMT transect positions and averaged over the months of April and May from 2003 to 2010; black, normalized τ index calculated at the AMT transect positions over the years 2003 to 2010; red, nano- and micro-plankton normalized Shannon index from the AMT-2 in-situ measures.

Qualitatively, the in situ and satellite-estimated measures of biodiversity show a comparable latitudinal alternation of maxima and minima: from north to south, a shallow minimum centered in 15:30N, then a peak in the subtropics (15:20N); an equatorial peak around 10S followed by a deep minimum; and a recovering trend for both curves when continuing from 20S further poleward. However, whereas the correlation is significant in the Southern hemisphere (Pearson's r=0.7, p=4 10-3, Fig. S2 a), in the Northern hemisphere the quantitative agreement is lost. We note that the localization of Northern hemisphere AMT stations includes upwelling systems where vertical processes are more relevant than in the open ocean, and where possibly stirring may not be the dominant structuring process shaping biodiversity hotspots. The importance of the Mauritanian upwelling, that is regularly crossed by the AMT transect, in affecting picophytoplankton community composition has already been reported (Zubkov et al., 1998, pag. 1346). Note moreover that the in situ biodiversity measure in stations 1 and 3 appear to be extreme with respect to the general trend of the temperature dependence of biodiversity (Fig. S2b).

Figure S2: (a) τ index versus AMT-2 cruise Shannon index HAMT (filled circles, Southern hemisphere; empty circles, Northern hemisphere). The black line is the best linear fit for the Southern hemisphere data, that shows a statistically significant correlation (r=0.7, p=4 10-3). (b) Shannon index HAMT (the number of the station is indicated next to the point) versus average Sea Surface Temperature. The best quadratic fit of the points relative to the in situ measures (red line, R2=0.26, p=0.038) is displayed together with the fit of the τ index and the confidence interval to one standard deviation (light blue lines, same as in Fig. 2).

Altogether, the AMT data provide a picture of the variation of diversity with temperature that is consistent to what displayed in Fig. 2 for the τ index, although the best quadratic fit is only significant to a 0.05 level. Both regression lines show a very similar trend with an maximum at intermediate temperatures.

4.2 Marine predators hotspots

In contrast to the observations of phytoplanktonic biodiversity, which provide a sparse coverage for many remote regions of the global oceans, a relatively larger amount of data has been collected in recent years by tracking the movement of marine predators. Regions with increased probability of localization of different species of marine predators have been identified (Block et al., 2011, Kaschner et al., 2011), and are considered as hotspots of biodiversity for higher levels of the trophic chain.

Interestingly, the hotspots in the distribution of the τ index seem to match several independent observations relative to marine top predators. The τ index is particularly high in the North Pacific Transition Zone (Fig. S3), identified by Block et al., 2011 as a region where a high number of predator species concentrate, and attains high values more to the West, where we encounter the well-known biodiversity hotspot of the California Current large marine ecosystem. Moreover, the same paper reports a minimum in predator density between the California Coast and the Hawaii. This is also found in the τ index map.

Figure S3: τ index map, centered in the North Pacific, where the North Pacific Transition Zone, the region East of Hawaii and the California current stand out as hotspots in agreement with biodiversity hotspots found by Block et al., 2011 in the Pacific.

On a more global scale, the τ index map shows similarities with the maps of marine mammal occurrence and species richness derived by predictions of global distributional ranges and empirical observations (Fig. 4 of Kaschner et al., 2011). Similarities are also evident considering global patterns of species richness encompassing several phyla, from zooplankton to marine mammals (see Tittensor et al., 2010, Fig. 2d).The matching of areas considered hotspots for predator diversity with transport-induced hotspots is a feature that supports the bottom-up propagation of biodiversity across the trophic chain, and thus the possible use of the τ index for detecting habitat shift and degradation for higher tropic levels.

4.3 τ index over global gradients

The relationship between biodiversity and latitude and temperature gradients has been highlighted in many studies on both land and marine ecosystems (Hillebrandt 2004): diversity appears to decrease at higher latitudes and for lower temperatures.For instance, in the ocean, latitudinal gradients have been shown to exist for species richness of planktonic marine bacteria (Fuhrman et al., 2008), and for functional diversity (Raes et al., 2011). However, quantitative differences in the latitudinal profiles, related to the relatively small number of observations and the heterogeneity of the conditions in which these have been performed, do not allow to go much beyond the qualitative conclusion that diversity decreases at high latitudes and that a mid-latitude maximum may exist. This conclusion is anyway consistent with the zonal mean of the τ index (Fig. 1 a).

Other studies analyzed also the relationship between variation of diversity with temperature. We focus on three of these studies, which compile a large number of observations.Figure 2 of our paper can be compared to Fig. 3 c of Tittensor et al., 2010 , to Fig. 2 of Worm et al., 2005, and to Fig. 2 b of Rutherford et al., 1999, respectively reporting the variation of species richness in the Census of Marine Life, for tuna and billfish and for zooplankton. All curves display an increase

of diversity with temperature, corresponding to the latitudinal trend of increase in biodiversity when moving from the poles towards the tropics. The curves for fish and zooplankton peak between 20 and 25° C, whereas the relationship appears to saturate at high temperature in Fig. 3 c of Tittensor et al., 2010. However, mid-latitude biodiversity maxima are present in most oceanic taxa-specific global diversity maps (Fig. 1 h-m, Tittensor et al., 2010) across all oceans.The quadratic relationship between SST and the τ index is highly significant (R2=0.26, p<10-4) and is consistent with the best fit of AMTin situ data (Fig. S2 b).

The existence of a unimodal diversity-temperature relationship is also supported by theoretical studies of food chain evolution, based on the hypothesis that metabolic rates universally depend on temperature: see Stegen et al, 2012, in particular their Fig. 1g, to be compared to Fig. S4, where we plotted the same data we have presented in Fig. 2 of the main text, but as a function of the inverse of temperature.

Figure S4: τ index versus inverse average sea surface temperature. The axis have been chosen as in Fig. 1g of ref. Stegen et al., 2012, for an easier comparison. The light blue line is the best quadratic fit of Fig. 2.

With current ocean color products, it is not possible to compute τ over disks with radius less than 100 km, since this is the typical scale of patches observed in PHYSAT data (d'Ovidio et al., 2010). In order to test the sensitivity of our results to larger radii, we have calculated the mean global map of τ500, the area-based Shannon index for disks of 500 km radius. The relationship between such an index and the SST is still highly significant (R2=0.08, p<10-4) but the correlation is inferior to that for the index computed at 100 km . The fact that the index computed at 100 km provides a higher correlation to the SST supports the appropriateness of the higher resolution for studying the biodiversity patterns generated by transport at the (sub)mesoscale.

References

Alvain S, Moulin C, Dandonneau Y, Bréon FM (2005). Remote sensing of phytoplankton groups in case 1 waters from global SeaWiFS imagery. Deep-Sea Res Pt 1 52: 1989-2004.

Alvain S, Moulin C, Dandonneau Y, Loisel H (2008). Seasonal distribution and succession of dominant phytoplankton groups in the global ocean: A satellite view. Global Biogeochem Cy 22: GB3001.

Alvain S, Loisel H, Dessailly D (2012). Theoretical analysis of ocean color radiances anomalies and implications for

phytoplankton groups detection in case 1 waters. Opt Express 20: 1070-1083.

Barton AD, Dutkiewicz S, Flierl G, Bragg J, Follows MJ (2010). Patterns of Diversity in Marine Phytoplankton. Science 327: 1509-1511.

Block BA et al (2011). Tracking apex marine predator movements in a dynamic ocean. Nature 475: 86-90.

Brewin RJW et al (2010). A three-component model of phytoplankton size class for the Atlantic Ocean. Ecol Model 221: 1472-1483.

Cermeño P et al (2008). Resource levels, allometric scaling of population abundance, and marine phytoplankton diversity. Limnol Oceanogr 53: 312-318.

Ciotti A, Bricaud A (2006). Retrievals of a size parameter for phytoplankton and spectral light absorption by Colored Detrital Matter from water-leaving radiances at SeaWiFS channels in a continental shelf region off Brazil. Limnol Oceanogr-Meth 4: 237-253.

Connell JH (1978). Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310.

d’Ovidio F, De Monte S, Alvain S, Dandonneau Y, Lévy M (2010). Fluid dynamical niches of phytoplankton types. PNAS 107: 18366-18370.

Dandonneau Y et al (2004). Seasonal and interannual variability of ocean color and composition of phytoplankton communities in the North Atlantic, equatorial Pacific and South Pacific. Deep-Sea Res Pt 2 51: 303-318.

Fuhrman JA et al (2008). A latitudinal diversity gradient in planktonic marine bacteria. PNAS 105: 7774 –7778.

Hillebrandt H (2004). On the generality of the latitudinal diversity gradient. Am Nat 163: 192-211.

Hirata T. et al (2011) Synoptic relationships between surface Chlorophyll-a and diagnostic pigments specific to phytoplankton functional types Biogeosciences, 8: 311–327

Irigoien X, Huisman J, Harris RP (2004). Global biodiversity patterns of marine phytoplankton and zooplankton. Nature 429: 863-867.

Kaschner K, Tittensor DP, Ready J, Gerrodette T, Worm B (2011). Current and Future Patterns of Global Marine Mammal Biodiversity. PLoS ONE 6: e19653.

Kostadinov TS, Siegel DA, Maritorena S (2009). Retrieval of the particle size distribution from satellite ocean color observations. J Geophys Res 114: C09015.

Marañón E, Holligan PM, Varela M, Mouriño B, Bale AJ (2000). Basin-scale variability of phytoplankton biomass, production and growth in the Atlantic Ocean. Deep-Sea Res Pt 1 47: 825-857.

Raes J, Letunic I, Yamada T, Jensen LJ, Bork P (2011). Toward molecular trait-based ecology through integration of biogeochemical, geographical and metagenomic data. Mol Syst Biol 7: 473

Raitsos ED et al (2008). Identifying Four Puhytoplankton Functional Types from Space: An Ecological Approach. Limnol Oceanogr 53: 605-613.

Rocchini D et al (2010) Remotely sensed spectral heterogeneity as a proxy of species diversity:

Recent advances and open challenges. Ecological Informatics 5: 318–329

Rutherford S, D'Hondt S, Prell W (1999). Environmental controls on the geographic distribution of zooplankton diversity. Nature 400: 749-753.

Stegen JC, Ferriere R, Enquist BJ (2012). Evolving ecological networks and the emergence of biodiversity patterns across temperature gradients. Proc R Soc B 279: 1051-1060.

Tittensor DP et al (2010). Global patterns and predictors of marine biodiversity across taxa. Nature 466: 1098-1101.

Uitz J, Claustre H, Morel A, Hooker SB (2006). Vertical distribution of phytoplankton communities in open ocean: An assessment based on surface chlorophyll. J Geophys Res 111: C08005.

Worm B, Sandow M, Oschlies A, Lotze HK, Myers RA (2005). Global Patterns of Predator Diversity in the Open Oceans. Science 309: 1365-1369.

Zubkov MV, Sleigh MA, Tarran GA, Burkill PH, Leakey RJG (1998). Picoplanktonic community structure on an Atlantic transect from 50°N to 50°S. Deep-Sea Res I 45: 1339—1355