Ecology

Relatively few studies try to understand the value of cultural heritage sites such as Nachi Falls, which is also a pilgrimage route in Japan’s Kii mountain range.Credit: James Fichera/Getty

There are more than 50 ways to value the environment, but most research and policymaking focuses on just a handful of methods. These include counting species and evaluating the cost of replacing a service provided by nature. Yet assessing nature in purely monetary terms can also be harmful to people and the environment, according to the world’s largest assessment of environmental valuation.“Policymaking largely disregards the multiple ways in which nature matters to people,” especially Indigenous people and low-income communities, says the report from the Intergovernmental Science-Policy Panel on Biodiversity and Ecosystem Services (IPBES).For example, in proposals for hydroelectric dams, the needs of affected communities are often seen as secondary to those of urban consumers — especially if communities are required to be displaced, resulting in people losing livelihoods and being compelled to change their way of life, the report finds.The world’s failure to properly value biodiversity has caused a long-term decline in a variety of services that the environment provides, said Anne Larigauderie, an ecologist who leads the IPBES secretariat, at the report’s launch on 11 July. “The capacity to pollinate crops, or regulate water, has been in decline for 50 years,” she said.There is strong evidence that valuing nature on the basis of market prices is contributing to the present biodiversity crisis, said Unai Pascual, an economist at the Basque Centre for Climate Change in Leioa, Spain, at the launch in Bonn, Germany. “Many other values are ignored in favour of short-term profit and economic growth,” added Pascual, who co-chaired the assessment.A summary for policymakers was approved by 139 governments on 8 July. The full assessment report is expected to be released ahead of the Conference of the Parties to the UN Convention on Biological Diversity, which takes place in Montreal in December. This meeting is expected to agree a new set of targets and indicators for biodiversity conservation.Studies of natureEighty-two researchers from around the world, with areas of expertise spanning the sciences, social sciences and humanities, identified 79,000 studies in environmental valuation, and found that their number has been increasing by 10% a year for four decades. But these studies also rarely lead to policy changes. The researchers selected 1,163 of the studies for in-depth review, and found that only for 5% of these cases were recommendations adopted by decision makers.Half of the studies selected for in-depth review used biophysical indicators, such as numbers of species, or quantity of forest biomass. Another 26% used monetary indicators, such as how much it would cost if pollination needed to be carried out by humans, or the amounts that governments pay farmers to conserve biodiversity on agricultural land.Only one-fifth of the studies valued biodiversity according to sociocultural criteria. Those that did included studies on the importance to people of a sacred site; and research on the value that someone attaches to the place where they grew up. Sociocultural values do not necessarily have a numerical quantity, or price tag. The value of sacred sites does not need to be turned into dollars, or euros, Sander Jacobs, one of the IPBES authors and an ecologist at the Research Institute of Nature and Forests in Brussels, said at the report’s launch.The report’s authors found that most studies do not consider multiple values, even when the evidence shows that doing so leads to better outcomes for the environment. The team found that few scientists consult or involve the people who live and work in regions of high biodiversity. Only 2% of the studies reviewed in depth reported having done so. And just 1% involved people in all the steps from designing a study to publishing it.“We need to build coalitions of scientists from different disciplines. But science needs allies too,” Pascual says. “Scientists need to be humble and invite those who represent other ways of knowing. Such a coalition could provide a solutions-oriented approach to the biodiversity and climate crises.” More

Palaeogeographical modelWe use palaeogeographical reconstructions describing Earth’s palaeotopography and palaeobathymetry for a series of time slices from 541 Ma to the present day. The reconstructions merge existing models from two published global reconstruction datasets—those of ref. 32 and ref. 33 (https://doi.org/10.5281/zenodo.5348492), which themselves are syntheses of a wealth of previous work.For continental regions, estimates of palaeoelevation and continental flooding rely on a diverse range of geological evidence, such as sedimentary depositional environments and the spatiotemporal distribution of volcanic activity. For a full description, see a recent review34. Together, these data can be used to define the past locations of mountain ranges and palaeoshorelines34. For this part of our reconstruction, we used the compilation of ref. 33 with updated palaeoshorelines based on depositional environment information in current fossil databases35. This compilation comprises 82 palaeotopography maps covering the entire Phanerozoic. Note that each palaeogeographical map is a time slice representing the concatenation of geological data over several million years36.We quantified the impact of using the original compilation of ref. 33 on our model results and found only small changes with respect to using the reconstructions with updated palaeoshorelines (Extended Data Fig. 3a–c). Similarly, eustatic sea level is thought to have varied by around 100 m at timescales much shorter than the duration of the time-slices throughout the Phanerozoic37, such that the extent of continental flooding could have varied within each time slice by an amount significant for our analysis. For this reason, and to assess the uncertainty of our results to continental palaeogeography in general, we computed additional maps of continental flooding in which the sea level is raised or lowered by 100 m compared with the original palaeo–digital elevation model grids of ref. 33 (Extended Data Fig. 3d–f).For deep-ocean regions, the primary control on seafloor depth is the age of the seafloor, so reconstructing palaeobathymetry relies on constructing maps of seafloor age back in time38. As a consequence, we rely on reconstruction models that incorporate a continuous network of plate boundaries. For this part, we used the reconstruction of ref. 32 and derived maps of seafloor age from the plate tectonic model using the method of ref. 39 for which source code is available at GitHub (https://github.com/siwill22/agegrid-0.1). Palaeobathymetry is derived from the seafloor age maps following the steps outlined in ref. 38. It is important to note that seafloor age maps for most of the Phanerozoic (that is, pre-Pangaea times) are not directly constrained by data due to recycling of oceanic crust at subduction zones. Rather, they are model predictions generated by constructing plate motions and plate boundary configurations from the geological and palaeomagnetic record of the continents. Nonetheless, the first-order trends in ocean-basin volume and mean seafloor age are consistent with independent estimates for at least the last 410 million years (Myr)39.The reconstructions of refs. 32,33 differ in the precise locations of the continents through time. To resolve this discrepancy, we reverse reconstructed the continental palaeoelevation model of ref. 33 to present-day coordinates using their rotation parameters, then reconstructed them back in time using the rotations of ref. 32. Owing to the differences in how the continents are divided into different tectonic units, this process leads to some gaps and overlaps in the results40, which we resolved primarily through a combination of data interpolation and averaging. Manual adjustments were made to ensure that the flooding history remained consistent with the original palaeotopography in areas in which interpolation gives a noticeably different history of seafloor ages. The resulting palaeotopography maps are therefore defined in palaeomagnetic reference frame32 appropriate for use in Earth system models.For the biodiversity modelling, we generate estimates of the age of the seafloor for discrete points within the oceans and flooded continents, and track these ages through the lifetime of each point (Supplementary Video 5). For the oceans, this is achieved using the method described in ref. 39 in which the seafloor is represented by points that are incrementally generated at the mid-ocean ridges for a series of time steps 1 Myr apart, with each point tracked through subsequent time steps based on Euler poles of rotation until either the present-day is reached, or they arrive at a subduction zone and are considered to be destroyed.For the continents, tracking the location of discrete points is generally simpler as most crust is conserved throughout the timespan of the reconstruction. In contrast to the deep oceans (where we assume that crust is at all times submerged), we model the ‘age’ of the seafloor from the history of continental flooding and emergence within the palaeogeographical interpretation33. The continents are seeded with uniformly distributed points at the oldest timeslice (541 Ma) at which they are assigned an age of zero. These points are tracked to subsequent time slices of which the palaeogeography is used to determine whether the point lies within a flooded or emergent region. Points within flooded regions of continents are considered to be seafloor, and the age of this seafloor is accumulated across consecutive time slices where a given point lies within a flooded region. When a point is within an emergent region, the seafloor age is reset to zero. Following this approach, individual points within stable continents may undergo several cycles of seafloor age increasing from zero before being reset. At the continental margins formed during the Pangaea breakup, the age of the seafloor continuously grows from the onset of rifting. Intraoceanic island arcs represent an additional case, which can appear as new tectonic units with the reconstructions at various times. In these cases, we assume that the seafloor has a zero-age at the time at which the intraoceanic arc first develops, then remains predominantly underwater for the rest of its lifetime.Thus, for each of the 82 palaeogeographical reconstructions, we annotate 0.5° by 0.5° grids as continental, flooded continental shelf or oceanic for later use in model coupling and production of regional diversity maps.Palaeoenvironmental conditions under the cGENIE Earth system modelWe use cGENIE41, an Earth system model of intermediate complexity, to simulate palaeoenvironmental conditions of seawater temperature and organic carbon export production (as a surrogate for food supply) throughout the Phanerozoic (from 541 Ma to the present day).cGENIE is based on a three-dimensional (3D) ocean circulation model coupled to a 2D energy–moisture balance atmospheric component and a sea-ice module. We configured the model on a 36 × 36 (latitude, longitude) equal area grid with 17 unevenly spaced vertical levels in depth, down to a maximum ocean depth of 5,900 m. The cycling of carbon and associated tracers in the ocean is based on a size-structured plankton ecosystem model with a single (phosphate) nutrient42,43, and adopts an Arrhenius-type temperature-dependent scheme for the remineralization of organic matter exported to the ocean interior44.cGENIE provides a spatially resolved representation of ocean physics and biogeochemistry, which is a prerequisite for the present study to be able to reconstruct the spatial patterns of biodiversity in deep time. However, owing to the computational impracticality of generating a single transient simulation of physics (that is, temperature) and biogeochemistry (that is, export production) over the entire Phanerozoic, we therefore generate 30 model equilibria at regular time intervals throughout the Phanerozoic that are subsequently used as inputs for the regional diversification model (see the ‘Model coupling’ section).We used 30 Phanerozoic palaeogeographical reconstructions through time (~20 Myr evenly spaced time intervals) to represent key time periods. For each continental configuration corresponding to a given age in Earth history, we generate idealized 2D (but zonally averaged) wind speed and wind stress, and 1D zonally averaged albedo forcing fields45 required by the cGENIE model using the ‘muffingen’ open-source software (see the ‘Code availability’ section). For each palaeogeographical reconstruction, the climatic forcing (that is, solar irradiance and carbon dioxide concentration) is adapted to match the corresponding geological time interval. The partial pressure of CO2 is taken from the recent update of the GEOCARB model46. Solar luminosity is calculated using the model of stellar physics of ref. 47. We impose modern-day orbital parameters (obliquity, eccentricity and precession). The simulations are initialized with a sea-ice-free ocean, homogeneous oceanic temperature (5 °C) and salinity (34.9‰). As variations in the oceanic concentration of bio-available phosphate remain challenging to reconstruct in the geological past48,49, we impose a present-day mean ocean phosphate concentration (2.159 μmol kg−1) in our baseline simulations. We quantify the impact of this uncertainty on our model results by conducting additional simulations using half and twice the present-day ocean phosphate concentration (Extended Data Fig. 3g–i). For each ocean phosphate scenario (that is, 0.5×, 1× and 2× the present-day value), each of the 30 model simulations is then run for 20,000 years, a duration ensuring that deep-ocean temperature and geochemistry reach equilibrium. For each model simulation, the results of the mean annual values of the last simulated year are used for the analysis. Note that, although cGENIE makes projections of the distribution of dissolved oxygen ([O2]) in the ocean, our diversification model does not currently consider oxygenation to be a limit on diversity. Thus, we assumed a modern atmospheric partial pressure of O2 in all 30 palaeo simulations and did not use the resulting projected [O2] fields.Regional diversification modelWe tested two models of diversification—the logistic model and the exponential model—describing the dynamics of regional diversity over time. In both models, the net diversification rate (ρ), with units of inverse time (Myr−1), varies within a pre-fixed range of values as a function of seawater temperature and food availability. The net diversification rate is then calculated for a given location and time according to the following equation:$$rho ={rho }_{max }-({rho }_{max }-{rho }_{min })(1-{Q}_{{rm{temp}}}{Q}_{{rm{food}}})$$
(1)
where ρmin and ρmax set the lower and upper net diversification rate limits within which ρ is allowed to vary, and Qtemp and Qfood are non-dimensional limitation terms with values between 0 and 1 that define the dependence of ρ on temperature and food, respectively (Extended Data Table 1).The model considers a direct relationship between seawater temperature, food supply and the rate of net diversification on the basis of the theoretical control that temperature and food supply exert on the rates of origination and extinction (Supplementary Fig. 1). Temperature rise is expected to accelerate the biochemical kinetics of metabolism50 and shorten the development times of individuals51, leading to higher rates of mutation and origination. Greater food availability increases population sizes, which increases the rates of mutation and reduces the probability of extinction52. Furthermore, a large body of observations shows the existence of a positive relationship between resource availability (that is, food supply) and the standing stock of species in marine and terrestrial communities53,54. A larger food supply would support a greater number of individuals. A greater diversity of food resources could also lead to a finer partitioning of available resources55.The temperature dependence of ρ is calculated using the following equation:$${Q}_{{rm{temp}}}=frac{{Q}_{10}^{frac{T-Tmin }{10}}}{{Q}_{10}^{frac{Tmax -Tmin }{10}}}$$
(2)
where the Q10 coefficient measures the temperature sensitivity of the origination rate. In equation (2) above, T is the seawater temperature (in °C) at a given location and time, and Tmin and Tmax are the 0.01 percentile and the 0.99 percentile, respectively, of the temperature frequency distribution in each time interval. In the model, the values of Tmin and Tmax used to calculate Qtemp are therefore recomputed for every time interval (~5 Myr) according to the temperature frequency distribution of the corresponding time interval. This enables us to use updated Tmin and Tmax values in each Phanerozoic time interval and to account for the thermal adaptation of organisms to ever changing climate conditions.The food limitation term is parameterized using a Michaelis–Menten formulation as follows:$${Q}_{{rm{food}}}=frac{text{POC flux}}{left({K}_{{rm{food}}},+,text{POC flux}right)}$$
(3)
where POC flux (mol m−2 yr−1) is the particulate organic carbon export flux, which is used as a surrogate for food availability, at a given location and time of the simulated seafloor. The parameter Kfood (mol m−2 yr−1) in equation (3) is the half-saturation constant, that is, the POC flux at which the diversification rate is half its maximum value, provided that other factors were not limiting. These temperature and food supply limitation terms vary in space and time as a result of changes in seawater temperature and particulate organic carbon export rate, respectively, thereby controlling the spatial and temporal variability of ρ (Supplementary Video 6).The net diversification rate becomes negative (1) in the event of mass extinctions or (2) in response to regional-scale processes, such as sea-level fall and/or seafloor deformation along convergent plate boundaries. Mass extinction events are imposed as external perturbations to the diversification model by imputing negative net diversification rates to all active seafloor points (ocean points and flooded continental points) and assuming non-selective extinction. The percentage of diversity loss as well as the starting time and duration of mass extinctions are extracted from three fossil diversity curves of reference20,21,22 (Source Data for Fig. 1). Each of these fossil diversity curves provides different insights into the Phanerozoic history of marine animal diversity based on uncorrected range-through genus richness estimates20,22 and sampling standardized estimates21. Regional-scale processes—such as sea level fall during marine regressions and/or seafloor destruction at plate boundaries, either by subduction or uplift—are simulated by the combined plate tectonic–palaeoelevation model, and constrain the time that seafloor habitats have to accumulate diversity.The model assumes non-selective extinction during mass extinction events (that is, the field of bullets model of extinction; everything is equally likely to die, no matter the age of the clade and regardless of adaptation)56. However, there is much fossil evidence supporting extinction selectivity57,58. It could be argued that higher extinction rates at diversity hotspots would have delayed their subsequent recovery, flattening global diversity trends. This argument is difficult to reconcile with Sepkoski’s genus-level global diversity curve but could be consistent with the standardized diversity curve of ref. 21. Similarly, the model is also not suitable for reproducing the explosive radiations of certain taxonomic groups after mass extinctions, which could explain the offset between the model and fossil observations in the early Mesozoic (Fig. 1).Letting D represent regional diversity (number of genera within a given seafloor point) and t represent time, the logistic model is formalized by the following differential equation:$$frac{partial Dleft(tright)}{partial t}=rho Dleft[1-frac{D}{{K}_{{rm{eff}}}}right]$$
(4)
where D(t) is the number of genera at time t and Keff is the effective carrying capacity or maximum number of genera that a given seafloor point (that is, grid cell area after gridding) can carry at that time, t. In our logistic model, Keff is allowed to vary within a fixed range of values (from Kmin to Kmax) as a positive linear function of the POC flux at a given location and time as follows:$${K}_{{rm{eff}}}={K}_{max }-left({K}_{max }-{K}_{min }right)frac{{text{POC flux}}_{max }-text{POC flux}}{{text{POC flux}}_{max }-{text{POC flux}}_{min }}$$
(5)
where POC fluxmin and POC fluxmax correspond to the 0.01 and 0.99 quantiles of the POC flux range in the whole Phanerozoic dataset.In the logistic model, the net diversification rate decreases as regional diversity approaches its Keff. The exponential model is a particular case of the logistic model when Keff approaches infinity and, therefore, neither the origination rate nor the extinction rate depend on the standing diversities. In this scenario, diversity grows in an unlimited manner over time only truncated by the effect of mass extinctions and/or by the dynamics of the seafloor (creation versus destruction). Thus, the exponential model is as follows:$$frac{partial Dleft(tright)}{partial t}=rho D$$
(6)
where the rate of change of diversity (the time derivative) is proportional to the standing diversity D such that the regional diversity will follow an exponential increase in time at a speed controlled by the temperature- and food-dependent net diversification rate. Even if analytical solutions exist for the steady-state equilibrium of the logistic and exponential functions, we solved the ordinary differential equations (4) and (6) using numerical methods with a time lag of 1 Myr to account for the spatially and temporally varying environmental constraints, seafloor dynamics and mass extinction events.As the analysis of global fossil diversity curves is unable to discern the causes of diversity loss during mass extinctions, our imputation of negative diversification rates could have overestimated the loss of diversity in those cases in which sea level fall, a factor already accounted for by our model, contributed to mass extinction. This effect was particularly recognizable across the Permian–Triassic mass extinction (Extended Data Fig. 6d–f), and supports previous claims that the decline in the global area of the shallow water shelf exacerbated the severity of the end-Permian mass extinction34.Model couplingAs stated above, the coupled plate tectonic–palaeoelevation (palaeogeographical) model corresponds to a tracer-based model (a Lagrangian-based approach) that simulates and tracks the spatiotemporal dynamics of ocean and flooded continental points. The diversification models start at time 541 Ma with all active points having a D0 = 1 (one single genus everywhere) and we let points accumulate diversity heterogeneously with time according to seafloor age distributions (for ocean points) and the time that continents have been underwater (for flooded continental points). The ocean points are created at mid-ocean ridges and disappear primarily at subduction zones. Between their origin and demise, the points move following plate tectonic motions and we trace their positions while accumulating diversity. The flooded continental points begin to accumulate diversity from the moment that they are submerged, starting with a D value equal to the nearest neighbour flooded continental point with D  > 1, thereby simulating a process of coastal recolonization (or immigration). The diversification process remains active while the seafloor points remain underwater, but it is interrupted, and D set to 0, in those continental points that emerge above sea level. Similarly, seafloor points corresponding to ocean domains disappear in subduction zones, and their diversity is lost. We track the geographical position of the ocean and flooded continental points approximately every 5 Myr, from 541 Ma to the present. Each and every one of the tracked points accumulates diversity over time at a different rate, which is modulated by the environmental history (seawater temperature and food availability) of each point, as described in equations (1)–(3). When a point arrives in an environment with a carrying capacity lower than the diversity it has accumulated through time, we reset the diversity of the point to the value of the carrying capacity, thereby simulating local extinction.Seawater temperature (T) and food availability (POC flux) are provided by the cGENIE model, which has a spatial and temporal resolution coarser than the palaeogeographical model. The cGENIE model provides average seawater T and POC flux values in a 36 × 36 equal area grid (grid cell area equivalent to 2° latitude by 10° longitude at the equator) and 30 time slices or snapshots (from 541 Ma to the present: each ~20 Myr time intervals). To have environmental inputs for the 82 time slices of the plate tectonic–palaeoelevation model, we first interpolate the cGENIE original model output data on a 0.5° by 0.5° grid to match the annotated grids provided by the plate tectonic–palaeoelevation model. As the relatively coarse spatial resolution of the cGENIE model prevents rendering the coast–ocean gradients, we assign surface T and POC flux at the base of the euphotic zone to the flooded continental shelf grid cells, and deep ocean T and POC flux at the bottom of the ocean to the ocean grid cells. As there are time slices without input data of seawater T and POC flux, we interpolate/extrapolate seawater T and POC flux values into the 0.5° by 0.5° flooded continental shelf and ocean grids independently. Finally, we interpolate values from these 0.5° by 0.5° flooded continental shelf and ocean grids into the exact point locations in each time frame. Thus, each active point is tracked with its associated time-varying T and POC flux values throughout its lifetime. On average, 6,000 flooded continental points and 44,000 oceanic points were actively accumulating diversity in each time frame. The model cannot simulate the singularities of relatively small enclosed seas for which the spatial resolution of the palaeogeographical and Earth system models is insufficient to capture relevant features (such as palaeobathymetry, seawater temperature) in detail. The method is also likely to underestimate the diversity of epeiric (inland) seas due to the difficulty of simulating immigration, a process that is strongly influenced by the effect of surface ocean currents and is not considered here. However, as stated above, the model considers recolonization of recently submerged areas by marine biota from nearby coastal environments, which partially explains coastal immigration.Estimation of global diversity from regional diversityOur regional diversity maps are generated by separately interpolating ocean point diversity and flooded continental point diversity into the 0.5° by 0.5° annotated grids provided by the palaeogeographical model. We calculate global diversity at each time step from each of the regional diversity maps following a series of steps to integrate diversity along line transects from diversity peaks (maxima) to diversity troughs (minima) (Extended Data Fig. 1). To select the transects, first, we identify on each of the regional diversity maps the geographical position of the diversity peaks. We identify local maxima (that is, grid cells with diversity greater than their neighbour cells), and define the peaks as those local maxima with diversity greater than the 0.75 quantile of diversity values in all local maxima in the map. In the case of grid cells with equal neighbour diversity, the peak is assigned to the grid cell in the middle. We subsequently identify the geographical position of the diversity troughs, which are defined as newly formed ocean grid cells (age = 0 Myr) and, therefore, with diversities equal to one. The troughs are mostly located at mid-ocean ridges.On each of the 82 spatial diversity maps, we trace a line transect from each diversity peak to its closest trough, provided that the transect does not cross land in more than 20% of the grid cells along the linear path (Supplementary Video 7). On average, for each spatial diversity map, we trace 400 (σ = ±75) linear transects. This sampling design gives rise to transects of different lengths, which may bias the estimates of global diversity. To minimize this bias, we cut the tail of the transects to have a length of 555 km equivalent to 5° at the equator. We tested an alternative cut-off threshold, 1,110 km, and the results do not alter the study’s conclusions.We apply Bresenham’s line algorithm59 to detect the grid cells crossed by the transects and annotate their diversity. To integrate regional diversity along the transects, we developed a method to simplify the scenario of peaks and troughs heterogeneously distributed on the 2D diversity maps. The method requires (1) a vector (the transect) of genus richness (αn) at n different locations (grids) arranged in a line (1D) of L grids, and (2) a coefficient of similarity (Vn,n + 1) between each two neighbouring locations, n and n + 1. Vn,n + 1, the coefficient of similarity, follows a decreasing exponential function with distance between locations. The number of shared genera between n and n + 1 is Vn,n + 1 × min(αn; αn + 1). We integrate diversity from peaks to troughs and assume that, along the transect, αn + 1 is lower than αn. We further assume that the genera present in n and n + 2 cannot be absent from n + 1. Using this method, we integrate the transect’s diversity (γi) using the following equation:$${gamma }_{i}={ {mathbf{upalpha}}}_{1}+{sum }_{n=1}^{L-1}left(1-{V}_{n,n+1}right){ {mathbf{upalpha}}}_{n+1}$$
(7)
To integrate the diversity of all transects (γi) on each 2D diversity map (or time slice), we apply the same procedure as described above (Extended Data Fig. 1). We first sort the transects in descending order from the highest to the lowest diversity. We then assume that the number of shared genera between transect i and the rest of the transects with greater diversity {1, 2, …, i − 1} is given by the distance of its peak to the nearest neighbour peak (NN(i)) of those already integrated {1, 2, …, i − 1}. Thus, we perform a zigzag integration of transects’ diversities down gradient, from the greatest to the poorest, weighted by the nearest neighbour distance among the peaks already integrated. As a result, the contribution of each transect to global diversity will depend on its diversity and its distance to the closest transect out of all those transects already integrated. Using this method, we linearize the problem to simplify the cumbersome procedure of passing from a 2D regional diversity map to a global diversity estimate without knowing the identity (taxonomic affiliation) of the genera. If γtotal is the global diversity at time t:$${gamma }_{{rm{total}}}={gamma }_{1}+{sum }_{i=2}^{j}left(1-{V}_{{rm{NN}}left(iright),i}right){gamma }_{i}$$
(8)
Finally, the resulting global estimates are plotted against the midpoint value of the corresponding time interval to generate a synthetic global diversity curve. To compare the global diversity curves produced by the diversification models with those composed from the fossil record, Lin’s CCC60 is applied to the data normalized to the min–max values of each time series (that is, rescaled within the range 0–1). Lin’s CCC combines measures of both precision and accuracy to determine how far the observed data deviate from the line of perfect concordance (that is, the 1:1 line). Lin’s CCC increases in value as a function of the nearness of the data’s reduced major axis to the line of perfect concordance (the accuracy of the data) and of the tightness of the data around its reduced major axis (the precision of the data).The time series of global diversity generated from the fossil record and from the diversification model exhibit serial correlation and the resulting CCCs are therefore inflated. The use of methods for analysing non-zero autocorrelation time series data, such as first differencing or generalised least squares regression, enables high-frequency variations along the time series to be taken into account. However, the relative simplicity of our model, which was designed to reproduce the main Phanerozoic trends in global diversity, coupled with the fact that biases in the fossil data would introduce uncertainty into the analysis, leads us to focus our analysis on the long-term trends, obviating the effect of autocorrelation.Model parameterization and calibrationThe diversification models are parameterized assuming a range of values that constrain the lower and upper limits of the genus-level net diversification rate (ρmin and ρmax, respectively) (Extended Data Table 1) according to previously reported estimates from fossil records (figures 8 and 11 of ref. 5). A range of realistic values is assigned for the parameters Q10 and Kfood, determining, respectively, the thermal sensitivity and food dependence of the net diversification rate. We test a total of 40 different parameter combinations (Extended Data Table 2). The resulting estimates of diversity are then compared against the fossil diversity curves of ref. 20, ref. 21 or ref. 22, and the 15 parameter combinations providing the highest CCCs are selected.The results of the logistic diversification model rely on the values of the minimum and maximum carrying capacities (Kmin and Kmax, respectively) within which the spatially resolved effective carrying capacities (Keff) are allowed to vary. The values of Kmin and Kmax are therefore calibrated by running 28 simulations of pair-wise Kmin and Kmax combinations increasing in a geometric sequence of base 2, from 2 to 256 genera (Extended Data Fig. 4). We perform these simulations independently for each of the 15 parameter settings selected previously (Extended Data Fig. 4 and Extended Data Table 2). Each combination of Kmin and Kmax produces a global diversity curve, which is evaluated as described above using Lin’s CCC.Calculating estimates of global diversity from regional diversity maps in the absence of information on genus-level taxonomic identities requires that we assume a spatial turnover of taxa with geographical distance (Extended Data Fig. 1). Distance-decay curves are routinely fitted by calculating the ecological similarity (for example, the Jaccard similarity index) between each pair of sampling sites, and fitting an exponential decay function to the points on a scatter plot of similarity (y axis) versus distance (x axis). Following this method, we fit an exponential decay function to the distance–decay curves reported in ref. 61, depicting the decrease in the Jaccard similarity index (J) of fossil genera with geographical distance (great circle distance) at different Phanerozoic time intervals:$$J={J}_{{rm{o}}{rm{f}}{rm{f}}}+(,{J}_{max}text{-}{J}_{{rm{o}}{rm{f}}{rm{f}}}){{rm{e}}}^{-lambda times {rm{d}}{rm{i}}{rm{s}}{rm{t}}{rm{a}}{rm{n}}{rm{c}}{rm{e}}}$$
(9)
where Joff = 0.06 (n.d.) is a small offset, Jmax = 1.0 (n.d.) is the maximum value of the genus-based Jaccard similarity index and λ = 0.0024 (km−1) is the distance-decay rate.The Jaccard similarity index (J) between consecutive points n and n + 1 is bounded between 0 and min(αn; αn + 1)/max(αn; αn + 1). A larger value for J would mean that there are more shared genera between the two communities than there are genera within the least diverse community, which is ecologically absurd. However, using a single similarity decay function can lead the computed value of J to be locally larger than min(αn; αn + 1)/max(αn; αn + 1). To prevent this artefact, we use the Simpson similarity index or ‘overlap coefficient’ (V) instead of J. V corresponds to the percentage of shared genera with respect to the least diverse community (min(αn ; αn + 1)). V is bounded between 0 and 1, whatever the ratio of diversities. As the pre-existing estimates of similarity are expressed using J (ref. 61), we perform the conversion from J to V using the algebraic expression V = (1 + R) × J/(1 + J) where R = max(αn; αn + 1)/min(αn; αn + 1) (Supplementary Note 1). In the cases in which J exceeds the min(αn; αn + 1)/max(αn; αn + 1), V becomes >1 and, in those cases, we force V to be More

Brandes, N. & Linial, M. Giant viruses—big surprises. Viruses 11, 404 (2019).CAS 
Article 

Google Scholar 
Jover, L. F., Effler, T. C., Buchan, A., Wilhelm, S. W. & Weitz, J. S. The elemental composition of virus particles: implications for marine biogeochemical cycles. Nat. Rev. Microbiol. 12, 519–528 (2014).CAS 
Article 

Google Scholar 
Madsen, E. L. Microorganisms and their roles in fundamental biogeochemical cycles. Curr. opinion biotechnology 22, 456–464 (2011).CAS 
Article 

Google Scholar 
Gummow, B. Challenges posed by new and re-emerging infectious diseases in livestock production, wildlife and humans. Livest. Sci. 130, 41–46 (2010).CAS 
Article 

Google Scholar 
Becker, D. J., Streicker, D. G. & Altizer, S. Linking anthropogenic resources to wildlife–pathogen dynamics: a review and meta-analysis. Ecol. letters 18, 483–495 (2015).Article 

Google Scholar 
Woolhouse, M. E. & Gowtage-Sequeria, S. Host range and emerging and reemerging pathogens. Emerg. infectious diseases 11, 1842 (2005).Article 

Google Scholar 
Foster, R. et al. Pathogens co-transported with invasive non-native aquatic species: implications for risk analysis and legislation. NeoBiota 69, 79–102 (2021).Article 

Google Scholar 
Brasier, C. The biosecurity threat to the uk and global environment from international trade in plants. Plant Pathol. 57, 792–808 (2008).Article 

Google Scholar 
Ruiz, G. M. et al. Global spread of microorganisms by ships. Nature 408, 49–50 (2000).ADS 
CAS 
Article 

Google Scholar 
Essl, F. et al. Which taxa are alien? criteria, applications, and uncertainties. BioScience 68, 496–509 (2018).Article 

Google Scholar 
Blackburn, T. M., Bellard, C. & Ricciardi, A. Alien versus native species as drivers of recent extinctions. Front. Ecol. Environ. 17, 203–207 (2019).Article 

Google Scholar 
Hawkins, C. L. et al. Framework and guidelines for implementing the proposed iucn environmental impact classification for alien taxa (eicat). Divers. Distributions 21, 1360–1363 (2015).Article 

Google Scholar 
Corrales, X. et al. Advances and challenges in modelling the impacts of invasive alien species on aquatic ecosystems. Biol. Invasions 22, 907–934 (2020).Article 

Google Scholar 
Regulation, E. Regulation (eu) no 1143/2014 of the European Parliament and of the Council of 22 October 2014 on the prevention and management of the introduction and spread of invasive alien species. Off. J. Eur. Union 57, 35–55 (2014).
Google Scholar 
EU. Regulation (eu) 2016/2031 of the European Parliament of the Council of 26 October 2016 on protective measures against pests of plants, amending regulations (eu) 228/2013,(eu) 652/2014 and (eu) 1143/2014 and repealing council directives 69/464/eec, 74/647/eec, 93/85/eec, 98/57/ec, 2000/29/ec, 2006/91/ec and 2007/33/ec. Off. J. 317, 4–104 (2016).
Google Scholar 
Murtaugh, M. P. et al. The science behind one health: at the interface of humans, animals, and the environment. Tech. Rep. (2017).Ogden, N. H. et al. Emerging infectious diseases and biological invasions: a call for a one health collaboration in science and management. Royal Soc. open science 6, 181577 (2019).ADS 
Article 

Google Scholar 
Roy, H. E. et al. Alien pathogens on the horizon: Opportunities for predicting their threat to wildlife. Conserv. Lett. 10, 477–484 (2017).Article 

Google Scholar 
Ikner, L. A., Gerba, C. P. & Bright, K. R. Concentration and recovery of viruses from water: a comprehensive review. Food Environ. Virol. 4, 41–67 (2012).
Google Scholar 
Taylor, M. W. Introduction: A short history of virology. In Viruses and Man: A History of Interactions, 1–22 (Springer, 2014).Thakur, M. P., Van der Putten, W. H., Cobben, M. M., van Kleunen, M. & Geisen, S. Microbial invasions in terrestrial ecosystems. Nat. Rev. Microbiol. 17, 621–631 (2019).CAS 
Article 

Google Scholar 
Desprez-Loustau, M.-L. et al. The fungal dimension of biological invasions. Trends ecology & evolution 22, 472–480 (2007).Article 

Google Scholar 
Rivett, D. W. et al. Elevated success of multispecies bacterial invasions impacts community composition during ecological succession. Ecol. Lett. 21, 516–524 (2018).Article 

Google Scholar 
Dunn, A. M. & Hatcher, M. J. Parasites and biological invasions: parallels, interactions, and control. TRENDS Parasitol. 31, 189–199 (2015).Article 

Google Scholar 
Pyšek, P. et al. Macroecological framework for invasive aliens (mafia): disentangling large-scale context dependence in biological invasions. (2020).Hulme, P. E. et al. Blurring alien introduction pathways risks losing the focus on invasive species policy. Conserv. Lett. 10, 265–266 (2017).Article 

Google Scholar 
Gilroy, J. J., Avery, J. D. & Lockwood, J. L. Seeking international agreement on what it means to be “native”. Conserv. Lett. 10, 238–247 (2017).Article 

Google Scholar 
Webber, B. L. & Scott, J. K. Rapid global change: implications for defining natives and aliens. Glob. Ecol. Biogeogr. 21, 305–311 (2012).Article 

Google Scholar 
CBD Secretariat. Decision VI/23: Alien species that threaten ecosystems, habitats and species. Document UNEP/CBD/COP/6/23 (2002).World Health Organization. A brief guide to emerging infectious diseases and zoonoses. Tech. Rep. https://apps.who.int/iris/handle/10665/204722 (2014).Firrao, G. et al. Candidatus phytoplasma’, a taxon for the wall-less, non-helical prokaryotes that colonize plant phloem and insects. Int. J. Syst. Evol. Microbiol. 54, 1243–1255 (2004).CAS 
Article 

Google Scholar 
CBD. Pathways of introduction of invasive species, their prioritization and management (Secretariat of the Convention on Biological Diversity Montreal, 2014).OIE. Terrestrial Animal Health Code 2021 (OIE, 2021).Magliozzi, C. et al. bacteria and viruses traits and species-related factors. figshare https://doi.org/10.6084/m9.figshare.18550907.v2 (2022).Katsanevakis, S. et al. Implementing the European policies for alien species: networking, science, and partnership in a complex environment. Manag. Biol. Invasions 4, 3–6 (2013).Article 

Google Scholar 
Tsiamis, K. et al. The EASIN Editorial Board: quality assurance, exchange and sharing of alien species information in europe. Manag. Biol. invasions 7, 321–328 (2016).Article 

Google Scholar 
Wieczorek, J. et al. Darwin core: an evolving community-developed biodiversity data standard. PloS one 7, e29715 (2012).ADS 
CAS 
Article 

Google Scholar 
Darwin Core. Darwin Core quick reference guide. https://dwc.tdwg.org/terms/ (2018).R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, https://www.R-project.org/ (2020).Wickham, H. ggplot2: Elegant Graphics for Data Analysis, https://ggplot2.tidyverse.org (Springer-Verlag New York, 2016).Schwarzl, T. ggBubbles: Mini Bubble Plots for Comparison of Discrete Data with ‘ggplot2’ R package version 0.1.4 (2019).Moon, K. R statistics and graphs for medical papers (Hannarae Seoul, 2015).Current, C. Invasive species compendium. Wallingford, UK: CAB Int. Available online: www.cabi.org/isc (accessed on 19 August 2020) (2011).Adams, M. J. & Antoniw, J. F. Dpvweb: An open access internet resource on plant viruses and virus diseases. Outlooks on Pest Manag. 16, 268 (2005).Article 

Google Scholar 
Adams, M. J. & Antoniw, J. F. Dpvweb: a comprehensive database of plant and fungal virus genes and genomes. Nucleic acids research 34, D382–D385 (2006).CAS 
Article 

Google Scholar 
Benson, D. A. et al. Genbank. Nucleic acids research 41, D36–D42 (2012).Article 

Google Scholar  More

Sakamoto, M. Primary production by phytoplankton community in some Japanese lakes and its dependence on lake depth. Archiv für Hydrobilogie. 62, 1–28 (1966).
Google Scholar 
Vollenweider, R. A. Scientific fundamentals of the eutrophication of lakes and flowing waters, with particular reference to nitrogen and phosphorus as factors in eutrophication (Organisation for Economic Co-operation and Development, 1968).
Google Scholar 
Edmondson, W. T. Phosphorus, nitrogen, and algae in Lake Washington after diversion of sewage. Science 169, 690–691 (1970).ADS 
CAS 
Article 

Google Scholar 
Dillon, P. J. & Rigler, F. H. The phosphorus-chlorophyll relationship in lakes. Limnol. Oceanogr. 19, 767–773 (1974).ADS 
CAS 
Article 

Google Scholar 
Jones, J. R. & Bachmann, R. W. Prediction of phosphorus and chlorophyll levels in lakes. J. Water Pollut. Control Feder. 48, 2176–2182 (1976).CAS 

Google Scholar 
Schindler, D. W. Evolution of phosphorus limitation in lakes. Science 195, 260–262 (1977).ADS 
CAS 
Article 

Google Scholar 
Filstrup, C. T. & Downing, J. A. Relationship of chlorophyll to phosphorus and nitrogen in nutrient-rich lakes. Inland Waters. 7, 385–400 (2017).CAS 
Article 

Google Scholar 
Schindler, D. W. Recent advances in the understanding and management of eutrophication. Limnol. Oceanogr. 51, 356–363 (2006).ADS 
Article 

Google Scholar 
Quinlan, R. et al. Relationships of total phosphorus and chlorophyll in lakes worldwide. Limnol. Oceanogr. 66, 392–404 (2020).ADS 
Article 

Google Scholar 
Yuan, L. L. & Jones, J. R. Rethinking phosphorus–chlorophyll relationships in lakes. Limnol. Oceanogr. 65, 1847–1857 (2020).ADS 
CAS 
Article 

Google Scholar 
Carlson, R. E. A trophic state index for lakes. Limnol. Oceanogr. 11, 361–369 (1977).ADS 
Article 

Google Scholar 
Neveux, J. et al. Comparison of chlorophyll and phaeopigment determinations by spectrophotometric, fluorometric, spectrofluorometric and HPLC methods. Mar. Microb. Food Webs 4, 217–238 (1990).
Google Scholar 
Lampert, W. & Sommer, U. Limnoecology (Oxford University, 2007).
Google Scholar 
Kovalevskaya, R. Z., Zhukava, H. A. & Adamovich, B. V. Modification of the method of spectrophotometric determination of chlorophyll a in the suspended matter of water bodies. J. Appl. Spectrosc. 87, 72–78 (2020).ADS 
CAS 
Article 

Google Scholar 
Søndergaard, M., Lauridsen, T. L., Johansson, L. S. & Jeppesen, E. Nitrogen or phosphorus limitation in lakes and its impact on phytoplankton biomass and submerged macrophyte cover. Hydrobiologia 795, 35–48 (2017).Article 

Google Scholar 
Søndergaard, M., Jensen, J. P., Jeppesen, E. & Møller. P. H. Seasonal dynamics in the concentrations and retention of phosphorus in shallow Danish lakes after reduced loading. Aquat. Ecosyst. Health Manag. 5(1), 19–29 (2002).Magumba, D., Atsushi, M., Michiko, T., Akira, K. & Masao, K. Relationships between Chlorophyll-a, phosphorus and nitrogen as fundamentals for controlling phytoplankton biomass in lakes. Environ. Control. Biol. 51(4), 179–185 (2013).CAS 
Article 

Google Scholar 
Smith, V. H. & Shapiro, J. Chlorophyll-phosphorus relations in individual lakes. Their importance to lake restoration strategies. Environ. Sci. Technol. 15(4), 444–451 (1981).Pothoven, S. A. & Vanderploeg, H. A. Seasonal patterns for Secchi depth, chlorophyll a, total phosphorus, and nutrient limitation differ between nearshore and offshore in Lake Michigan. J. Great Lakes Res. 46, 519–527 (2020).CAS 
Article 

Google Scholar 
Søndergaard, M. & Jeppesen, E. Lake Søbygaard, Denmark: phosphorus dynamics during the first 35 years after an external loading reduction. In: Internal Phosphorus Loading: Causes, Case Studies, and Management (ed. Steinman, A.D. & Spears, B. M.) 285–299 (J. Ross, Plantation, 2020).Guildford, S. J. & Hecky, R. E. Total nitrogen, total phosphorus, and nutrient limitation in lakes and oceans: Is there a common relationship?. Limnol. Oceanogr. 45, 1213–1223 (2000).ADS 
CAS 
Article 

Google Scholar 
Jones, J.R. et al. Nutrients, seston, and transparency of Missouri reservoirs and oxbow lakes. An analysis of regional limnology. Lake Reser. Manag. 24, 155–180 (2008).Pikovsky, A., Rosenblum, M. & Kurths, J. Synchronization. A universal concept in nonlinear sciences (Cambridge University, 2001).Book 

Google Scholar 
Kuramoto, Y. Chemical Oscillations, Waves, and Turbulence (Springer, 1984).Book 

Google Scholar 
Sazonov, A. V. et al. An investigation of the phase locking index for measuring of interdependency of cortical source signals recorded in the EEG. Biol. Cybern. 100, 129–146 (2009).Article 

Google Scholar 
Medvinsky, A. B. et al. Temperature as a factor affecting fluctuations and predictability of the abundance of lake bacterioplankton. Ecol. Complex. 32, 90–98 (2017).Article 

Google Scholar 
Zhukova, T. V. & Ostapenya, A. P. Estimation of efficiency of nature protection measures in water catchment area of the Naroch lakes. Natural Resources. 3, 68–73 (2000) ((in Russian)).
Google Scholar 
Burlakova, L. E., Karatayev, A. Y. & Padilla, D. K. Changes in the distribution and abundance of Dreissena polymorpha within lakes through time. Hydrobiologia 571, 133–146 (2006).Article 

Google Scholar 
Ostapenia, A. P. et al. Bentification of lake ecosystem: causes, mechanisms, possible consequences, prospect for future research. Trudy BGU. 7, 135–148 (2012) ((in Russian)).
Google Scholar 
Karatayev, A.Y., Burlakova, L.E. & Padilla, D.K. Impacts of Zebra Mussels on aquatic communities and their role as ecosystem engineers. In: Leppäkoski, E., Gollasch, S., Olenin, S. (eds) Invasive Aquatic Species of Europe. Distribution, Impacts and Management (Springer, Dordrecht, 2002).Adamovich, B. V. et al. The divergence of chlorophyll dynamics in the Naroch Lakes. Biophysics 60, 632–638 (2015).CAS 
Article 

Google Scholar 
Zhukova, T. V. et al. Long-term dynamics of suspended matter in Naroch Lakes: Trend or intervation. Inland Water Biol. 10, 250–257 (2017).Article 

Google Scholar 
Adamovich, B. V. et al. Eutrophication, oligotrophication, and benthiphication in Naroch Lakes: 40 years of monitoring. J. Siber. Federal Univ. Biol. 10, 379–394 (2017).Article 

Google Scholar 
Ostapenya A.P. et al. Ecological passport of Lake Myastro (EcoMir, Minsk, 1994) (in Russian).Kantz, H. & Schreiber, T. Nonlinear time series analysis (Cambridge University, 1997).MATH 

Google Scholar 
Kot, M. Elements of mathematical ecology (Cambridge University, 2001).Book 

Google Scholar 
Turchin, P. Complex population dynamics. A Theoretical/Empirical Synthesis (Princeton University, Princeton, 2003).MATH 

Google Scholar 
Cazelles, B. & Stone, L. Detection of imperfect population synchrony in an uncertain world. J. Anim. Ecol. 72, 953–968 (2003).Article 

Google Scholar 
Karatayev, A. Y., Burlakova, L. & Padilla, D. K. The effects of Dreissena polymorpha (Pallas) invasion on aquatic communities in Eastern Europe. J. Shellfish Res. 16, 187–203 (1997).
Google Scholar 
Lia, J. et al. Benthic invaders control the phosphorus cycle in the world’s largest freshwater ecosystem. PNAS 118(6), e2008223118. https://doi.org/10.1073/pnas.2008223118 (2021).CAS 
Article 

Google Scholar 
Mikheyeva, T. M. et al. The dynamics of freshwater phytoplankton stability in the Naroch Lakes (Belarus). Ecol. Ind. 81, 481–490 (2017).Article 

Google Scholar 
Harris, P. H. Phytoplankton ecology. Structure, functioning and flucttuation (Chapman & Hall, London, New York, 1986).Jeppesen, E., Jensen, J. P., Søndergaard, M. & Lauridsen, T. L. Response of fish and plankton to nutrient loading reduction in eight shallow Danish lakes with special emphasis on seasonal dynamics. Freshw. Biol. 50, 1616–1627 (2005).CAS 
Article 

Google Scholar 
Nezlin, N.P. & Li, B-L. Time-series analysis of remote-sensed chlorophyll and environmental factors in the Santa Monica–San Pedro Basin off Southern California. J. Mar. Syst. 39, 185–202 (2003).French, T. D. & Petticrew, E. L. Chlorophyll a seasonality in four shallow eutrophic lakes (northern British Columbia, Canada) and the critical roles of internal phosphorus loading and temperature. Hydrobiologia 575, 285–299 (2007).CAS 
Article 

Google Scholar 
SCOR-UNESCO Working Group no. 17. Determination of photosynthetic pigments in sea-water. Monographs on Oceanologic Methodology 9–18 (UNESSCO, Paris, 1966).Semenov, A. D. Guide on the chemical analysis of continental surface waters (Gidrometeoizdat, 1977) ((in Russian)).
Google Scholar 
Wetzel, R. G. & Likens, G. E. Limnological analysis (Springer, 2000).Book 

Google Scholar 
Steffen, M. & Bartz-Beielstein, T. imputeTS: time series missing value imputation in R. R J. 9(1), 207–218 (2017).Article 

Google Scholar 
R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, 2020). More

Giorgi, A., Pentimalli, D., Giupponi, L. & Panseri, S. Quality traits of saffron (Crocus sativus L.) produced in the Italian Alps. Open Agric. 2(1), 52–57 (2017).Article 

Google Scholar 
Winterhalter, P. & Straubinger, M. Saffron—Renewed interest in an ancient spice. Food Rev. Intl. 16(1), 39–59 (2000).CAS 
Article 

Google Scholar 
Schmidt, M., Betti, G. & Hensel, A. Saffron in phytotherapy: Pharmacology and clinical uses. Wien Med. Wochenschr. 157, 315–319 (2007).PubMed 
Article 

Google Scholar 
Siddique, H. R., Fatma, H. & Khan, M. A. Medicinal properties of saffron with special reference to cancer—A review of preclinical studies. in Saffron: The Age-Old Panacea in a New Light (ed. Sarwat,
M. & Sumaiya, S.) 233–244 (Academic Press, 2020).Chapter 

Google Scholar 
Abdullaev, F. I. Cancer chemopreventive and tumoricidal properties of saffron (Crocus sativus L.). Exp. Biol. Med. 227(1), 20–25 (2002).CAS 
Article 

Google Scholar 
Kafi, M., Koocheki, A. & Rashed, M. H. Saffron (Crocus sativus): Production and Processing (Science Publishers, 2006).Book 

Google Scholar 
Mir, G.M. Saffron Agronomy in Kashmir (1992).Melnyk, J. P., Wang, S. & Marcone, M. F. Chemical and biological properties of the world’s most expensive spice: Saffron. Food Res. Int. 43(8), 1981–1989 (2010).CAS 
Article 

Google Scholar 
Menia, M. et al. Production technology of saffron for enhancing productivity. J. Pharmacognosy Phytochem. 7(1), 1033–1039 (2018).
Google Scholar 
Tanra, M. A., Dar, B. A. & Sing, S. Economic analysis of Production and Marketing of saffron in Jammu and Kashmir. J. Social Relevance Concern 5(10), 12–19 (2017).
Google Scholar 
Husaini, A. M., Hassan, B., Ghani, M. Y., Teixeira da Silva, J. A. & Kirmani, N. A. saffron (Crocus sativus Kashmirianus) cultivation in Kashmir: Practices and problems. Functional Plant Sci. Biotechnol. 4(2), 108–115 (2010).
Google Scholar 
Amirnia, R., Bayat, M. & Tajbakhsh, M. Effects of nano fertilizer application and maternal corm weight on flowering at some saffron (Crocus sativus L.) ecotypes. Turkish J. Field Crops. 19(2), 158–168 (2014).Article 

Google Scholar 
Kumar, R. et al. State of art of saffron (Crocus sativus L.) agronomy: A comprehensive review. Food Rev. Int. 25(1), 44–85 (2009).Article 

Google Scholar 
Dhar, A. K. Saffron breeding and agrotechnology. Status Rep. PAFAI J. 12, 18–22 (1990).
Google Scholar 
Ehsanzadeh, P., Yadollahi, A. A. & Maibodi, A. M. Productivity, growth and quality attributes of 10 Iranian saffron accessions under climatic conditions of Chahar-Mahal Bakhtiari, Central Iran. Int. Symp. Saffron Biol. Biotechnol. 650, 183–188 (2003).
Google Scholar 
Duke, J. A. Ecosystematic data on economic plants. Quart. J. Crude Drug Res. 17(3–4), 91–109 (1979).Article 

Google Scholar 
Kanth, R.H., Khanday, B.A. & Tabassum, S. Crop weather relationship for saffron production. Saffron Production in Jammu and Kashmir, Directorate of Extension Education. SKUAST-K, India 170–188 (2008).Shinde, D. A., Talib, A. R. & Gorantiwar, S. M. Composition and classification of some typical soils of saffron growing areas of Jammu and Kashmir. J. Indian Soc. Soil Sci. 32(3), 473–477 (1984).CAS 

Google Scholar 
Nazir, N. A., Khitrov, N. B. & Chizhikova, N. P. Statistical evaluation of soil properties which influence saffron growth in Kashmir. Eurasian Soil Sci. 28(4), 120–138 (1996).
Google Scholar 
Ganai, M. R., Wani, M. A. & Zargar, G. H. Characterization of saffron growing soils of Kashmir. Appl. Biol. Res. 2(1/2), 27–30 (2000).
Google Scholar 
Ganai, M.R.D. Nutrient status of saffron soils and their management. in Proceedings of Seminar-cum-Workshop on saffron (Crocus sativus) 51–54 (2001).Molina, R. V., Valero, M., Navarro, Y., Guardiola, J. L. & Garcia-Luis, A. Temperature effects on flower formation in saffron (Crocus sativus L.). ScientiaHorticulturae 103(3), 361–379 (2005).
Google Scholar 
Galavi, M., Soloki, M., Mousavi, S. R. & Ziyaie, M. Effect of planting depth and soil summer temperature control on growth and yield of saffron (Crocus sativus L.). Asian J. Plant Sci. 7(8), 747 (2008).Article 

Google Scholar 
Kamyabi, S., Habibi Nokhandan, M. & Rouhi, A. Effect of climatic factors affecting saffron using analytic hierarchy process (AHP); Case Study Roshtkhar Region, Iran. (2014).Gupta, R. K. Saffron status and cultivation in northwestern Himalayas. Vegetos 20(1), 1–7 (2007).
Google Scholar 
Qin, A. et al. Maxent modelling for predicting impacts of climate change on the potential distribution of Thuja sutchuenensis Franch., an extremely endangered conifer from southwestern China. Glob. Ecol. Conserv. 10, 139–146 (2017).Article 

Google Scholar 
Fielding, A. H. & Bell, J. F. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ. Conserv. 24(1), 38–49 (1997).Article 

Google Scholar 
Swets, J. A. Measuring the accuracy of diagnostic systems. Science 240(4857), 1285–1293 (1988).ADS 
MathSciNet 
CAS 
PubMed 
MATH 
Article 

Google Scholar 
Muscarella, R. et al. ENMeval: An R package for conducting spatially independent evaluations and estimating optimal model complexity for Maxent ecological niche models. Methods Ecol. Evaluat. 5(11), 1198–1205 (2014).Article 

Google Scholar 
Hao, T., Elith, J., Arroita, G. G. & Monfort, J. J. L. A review of evidence about use and performance of species distribution modelling ensembles like BIOMOD. Divers. Distrib. 25(5), 839–852 (2019).Article 

Google Scholar 
Thuiller, W. BIOMOD—Optimizing predictions of species distributions and projecting potential future shifts under global change. Glob. Change Biol. 9, 1353–1362 (2003).ADS 
Article 

Google Scholar 
Mykhailenko, O., Desenko, V., Ivanauskas, L. & Georgiyants, V. Standard operating procedure of Ukrainian saffron cultivation according to with good agricultural and collection practices to assure quality and traceability. Ind. Crops Prod. 151, 112376. https://doi.org/10.1016/j.indcrop.2020.112376 (2020).CAS 
Article 

Google Scholar 
Kothari, D., Thakur, M., Joshi, R., Kumar, A. & Kumar, R. Agro-climatic suitability evaluation for saffron production in areas of western Himalaya. Front. Plant Sci. 12, 657819. https://doi.org/10.3389/fpls.2021.657819 (2021).Article 
PubMed 
PubMed Central 

Google Scholar 
Mir, J. I., Ahmed, N., Wafai, A. H. & Qadri, R. A. Variability in stigma length and apocarotenoid content in Crocus sativus L. selections of Kashmir. J. Spices Aromatic Crops. 21(2), 169–171 (2012).
Google Scholar 
Nehvi, F. A. et al. New emerging trends on production technology of saffron. II Int. Symp. Saffron Biol. Technol. 739, 375–381 (2006).
Google Scholar 
Golmohammadi, F. Sustainable agriculture and rural development in Iran, Some modern issues in sustainable agriculture and rural development in Iran Germany, LAP LAMBERT Academic Publishing GmbH & Co. LAP Lambert Academic Publishing. Germany. ISBN-13, 978-3 (2012).Golmohammadi, F. Saffron and its importance, medical uses and economical export situation in Iran. in Oral Article Presented in: International Conference on Advances in Plant Sciences 14–18 (2012).Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maximum entropy modelling of species geographic distributions. Ecol. Model. 190(3–4), 231–259 (2006).Article 

Google Scholar 
Elith, J. et al. Novel methods improve prediction of species distributions from occurrence data. Ecography 29(2), 129–151 (2006).Article 

Google Scholar 
Pearson, R. G., Raxworthy, C. J., Nakamura, M. & Townsend Peterson, A. Predicting species distributions from small numbers of occurrence records: A test case using cryptic geckos in Madagascar. J. Biogeogr. 34(1), 102–117 (2007).Article 

Google Scholar 
Wisz, M. S. et al. NCEAS Predicting species distributions working group. Effects of sample size on the performance of species distribution models. Diversity Distributions. 14(5), 763–773 (2008).Article 

Google Scholar 
Rebelo, H. & Jones, G. Ground validation of presence only modelling with rare species: A case study on Barbastella barbastellus (Chiroptera: Vespertilionidae). J. Appl. Ecol. 47(2), 410–420 (2010).Article 

Google Scholar 
Elith, J. & Leathwick, J. R. Species distribution models: Ecological explanation and prediction across space and time. Annu. Rev. Ecol. Evol. Syst. 40, 677–697 (2009).Article 

Google Scholar 
Palomera, S. et al. Mapping from heterogeneous biodiversity monitoring data sources. Biodiversity Conservation 21(11), 2927–2948 (2012).Article 

Google Scholar 
Garcia, K., Lasco, R., Ines, A., Lyon, B. & Pulhin, F. Predicting geographic distribution and habitat suitability due to climate change of selected threatened forest tree species in the Philippines. Appl. Geogr. 44, 12–22 (2013).Article 

Google Scholar 
Marcer, A., Sáez, L., Molowny-Horas, R., Pons, X. & Pino, J. Using species distribution modelling to disentangle realised versus potential distributions for rare species conservation. Biol. Cons. 166, 221–230 (2013).Article 

Google Scholar 
Phillips, S.J., Dudík, M. & Schapire, R.E. A maximum entropy approach to species distribution modelling. in Proceedings of the Twenty-First International Conference on Machine Learning 83 (2004).Baldwin, R. A. Use of maximum entropy modelling in wildlife research. Entropy 11(4), 854–866 (2009).ADS 
Article 

Google Scholar 
Izadpanah, F., Kalantari, S., Hassani, M. E., Naghavi, M. R. & Shokrpour, M. Variation in Saffron (Crocus sativus L.) accessions and Crocus wild species by RAPD analysis. Plant Syst. Evolut. 300, 1941–1944 (2014).Article 

Google Scholar 
Nemati, Z., Harpke, D., Gemicioglu, A., Kerndorff, H. & Blattner, F. R. Saffron (Crocus sativus) is an autotriploid that evolved in Attica (Greece) from wild Crocus cartwrightianus. Mol. Phylogenet. Evol. 136, 14–20 (2019).PubMed 
Article 

Google Scholar 
Proosdij, A. S. J. V., Sosef, M. S. M., Wieringa, J. J. & Raes, N. Minimum required number of specimen records to develop accurate species distribution models. Ecography 39, 542–552 (2016).Article 

Google Scholar  More

Rafael Almeida and his colleagues estimate that floating solar panels on 5–10% of the area of large reservoirs could help the world to reach electricity decarbonization targets by 2050 (R. M. Almeida et al. Nature 606, 246–249; 2022). On small lakes in Europe and Asia, however, the existing coverage is significantly higher (averaging 50%, according to our unpublished data), with potentially greater ecological impact (G. Exley et al. Solar Energy 219, 24–33; 2021).
Competing Interests
The authors declare no competing interests. More

Widespread occurrence and environmental stimuli responsiveness of trypsin in marine phytoplanktonTo assess whether trypsin occurs broadly in marine phytoplankton and what ecological functions phytoplankton trypsin genes may play, we investigated the occurrence of trypsin genes and environmental stimuli regulating their expression based on PhyloDB, Tara Oceans unigenes and metatranscriptomes datasets. From Tara Oceans unigenes and metatranscriptomes, trypsin homologs were found at all the sampling stations worldwide and in all major phytoplankton phyla (Fig. 1a and Supplementary Fig. 1). The broad phylogenetic representation is corroborated by the prevalence of trypsin in the individual species’ transcriptomes in the PhyloDB database (Fig. 1b), most notably in Bacillariophyta, Dinophyta, Chlorophyta, Cryptophyta and Haptophyta, the major eukaryotic groups of phytoplankton in the ocean. These indicate that trypsin is widely distributed in phytoplankton both taxonomically and geographically, a finding that advances our knowledge on the distribution of this ancient enzyme. Moreover, phylogenetic and structure alignment analysis showed that phytoplankton trypsins are more closely related with bacterial trypsins than metazoan and fungal counterparts, but contain the conserved important residues and structure typical of animal trypsins (Supplementary Figs. 2–4). These observations suggest some complex evolutionary trajectory that might result in functional innovation of phytoplankton trypsin.Fig. 1: Widespread occurrence and environmental nutrient responsiveness of trypsin in global marine phytoplankton.a Wide geographic distribution of trypsin in phytoplankton found in Tara Oceans. Color scale depicts trypsin mRNA abundance. b Wide taxonomic distribution of trypsin in algae found in PhyloDB. c Environmental nutrient drivers of phytoplankton trypsin abundance. Pairwise comparisons of environmental nutrient concentrations are shown with a color gradient denoting Pearson’s correlation coefficient. The trypsin abundance and taxonomic distribution based on the 5–180 µm size fraction from SRF layer from Tara Ocean datasets. Taxonomic trypsin abundance was related to each nutrient factor by partial (geographic distance-corrected) Mantel tests. Edge width corresponds to the Mantel’s r statistic for the corresponding distance correlations, and edge color denotes the statistical significance based on 9999 permutations. Baci Bacillariophyta, Dino Dinophyta, Chlo Chlorophyta, Cryp Cryptophyta, Hapt Haptophyta. Source data are provided as a Source Data file.Full size imageWe found a large amount of trypsin gene duplication, 5 copies to 65 copies in each algal genome we examined6. The evolution of the gene family, in gene sequence and organization relative to other functional domain, need to be treated in a separate paper6, but the rampant gene duplication suggests that trypsin may have important roles in phytoplankton. Moreover, our correlation analysis for trypsin gene expression with environmental parameters in the Tara Oceans metatranscriptomic data showed that the phytoplankton trypsin transcript abundance was correlated with environmental conditions in some taxa, size fractions, and water depths, evidence that trypsin may be important in phytoplankton to adapt to dynamical environmental conditions6. To further explore specific environmental drivers modulating the expression of trypsin, we analyzed distance-corrected dissimilarities of phytoplankton trypsin transcript abundance with environmental nutrient factors using the partial Mantel test. Analyses were restricted to the 5–20 and 20–180 µm size fractions from surface layer as their trypsin appeared to be more responsive to environmental stimuli. As shown in Fig. 1c, trypsin expression in Bacillariophyta, Dinophyta, Chlorophyta, Cryptophyta and Haptophyta was differentially correlated with nutrient availability, most notably in Bacillariophyta and Chlorophyta. Moreover, nitrate and nitrite (NO3, NO3_5m*, and NO3_NO2) and phosphate (PO4) were the strongest correlates of both Bacillariophyta and Chlorophyta trypsin transcript abundances (Fig. 1c). Hence, we posit that trypsin have important functions in the response of phytoplankton to N and P nutrient conditions.Involvement of trypsin in nitrogen and phosphorus nutrient responsesTo gain mechanistic insights into the function of trypsin in phytoplankton, we conducted experiments on the model diatom Phaeodactylum tricornutum. We identified ten trypsin genes from its genome (Supplementary Table 1), and based on qRT-PCR, we observed their growth stage- and condition-specific expression variations (Fig. 2a and Supplementary Fig. 5). Interestingly, one of these genes (PtTryp2) exhibited opposite directions of expression dynamic under N- and P-depleted conditions: downregulated under N-depleted but upregulated under P-depleted condition (Fig. 2a). Furthermore, PtTryp2 transcript increased with increasing cellular N content but decreased with increasing cellular P content (Fig. 2b, c). These results suggest that PtTryp2 is involved in an opposite-direction regulation of responses to nitrogen and phosphorus nutrient status.Fig. 2: Involvement of PtTryp2 in nitrogen and phosphorus nutrient responses.a PtTryp2 expression in P. tricornutum under different growth stages and conditions based on qRT-PCR. Nutrient-replete, HNHP; N-depletion, LNHP; P-depletion, HNLP. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. b Time-course expression patterns of PtTryp2 when P. tricornutum was grown with different forms of nitrogen nutrients. Data are presented as mean values ± SD (n = 2 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. c PtTryp2 expression pattern after phosphorus supplement. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. Source data are provided as a Source Data file.Full size imageTo interrogate the function of PtTryp2 in N and P nutrient responses, we analyzed the physiology of homologous overexpression and CRISPR/Cas9 knockout lines we generated. A PtTryp2-overexpression cell line with C-terminal eGFP fusion (named PtTryp2-OE) was generated, and the expression of OE cell line was confirmed at a protein level through Western blot (Fig. 3a). Because the function of a protein corresponds with its subcellular location, we first examined where PtTryp2 is located inside P. tricornutum cells. By computational simulation, we find PtTryp2 is potentially localized in the chloroplast via the secretory pathway (Supplementary Table 2), in accordance with the fact that chloroplasts contain a rather high number of proteases and are the main location of nutrients assimilation and remobilization7. To obtain experimental verification of the chloroplast localization, we carried out subcellular localization analysis in the OE and OEC cell lines using confocal fluorescence microscopy. Interestingly, results show PtTryp2-eGFP are localized in both the chloroplast and cytoplasmic endoplasmic reticulum (ER), to the exclusion of the nucleus and Golgi apparatus, whereas the fluorescence from the eGFP blank vector control is outspread in the cell instead of being co-localized with chloroplast and ER (Fig. 3b and Supplementary Figs. 6–8). Further analyses show that PtTryp2 lacks the C-terminal -(K/H) DEL sequences, a typical ER-retention signal that prevents ER-resident proteins from being transported to downstream locations of the secretory system8,9. Hence, PtTtryp2 is evidently transported via the ER to the chloroplast, as in the case of the previously documented light-harvesting chlorophyll a/b-binding protein in Euglena10.Fig. 3: Subcellular localization of PtTryp2.a Detection of the expression of GFP-PtTryp2 by Western blot using anti-GFP primary antibody. Left panel, GFP-PtTryp2 fusion protein. Middle panel, GFP protein. GAPDH (on the right) was detected using anti-GAPDH as the control to indicate equal protein quantities loaded to each lane. The GFP-PtTryp2 was confirmed expressed successfully at protein level in OE cell line. All experiments were repeated independently three times, and similar results were obtained. b Confocal micrographs showing subcellular localization of GFP-PtTryp2 in chloroplast (PAF, showing red autofluorescence) and endoplasmic reticulum (ER, showing blue fluorescent stain by ER-Tracker) but not in nucleus (Hoechst 33342, showing blue fluorescent stain). TL merge, merger of the fluorescence images with transmission light image. Scale bar, 10 µm, applies to all images. All experiments were repeated independently three times, and similar results were obtained. Source data are provided as a Source Data file.Full size imagePtTryp2 contains one trypsin domain and two internal repeats 1 (RPT) (Fig. 4a), offering one single target for trypsin mutagenesis. Using an optimized efficient CRISPR/Cas9 gene editing system11, we obtained three PtTryp2 mutants with different mutation characteristics in the trypsin domain (named KO1, KO2, and KO3, respectively; Fig. 4b). As shown in Fig. 4c, compared with the knockout control cell line (KOC), all three PtTryp2-KO lines exhibited a significantly diminished PtTryp2 expression under both nutrient depletion and repletion; conversely, the OE cell line displayed markedly elevated PtTryp2 expression in comparison to the overexpression control cell line (OEC). Moreover, the PtTryp2 expression level in KOC cell lines strongly responded to the ambient N and P level, but consistently showed a constant and low expression pattern in KO lines (Fig. 4d). These results verified that KO cell lines with the loss of PtTryp2 function, and OE with enhanced function of PtTryp2, can be used for subsequent functional analyses of PtTryp2.Fig. 4: Mutation generations of PtTryp2 and characters of mutants.a Schematic presentation of PtTryp2 protein. The target site (vertical arrow) for CRISPR/Cas9-based knockout is located within the conserved functional domain (green pentagon), with PAM motif shown in orange font. Red rectangle on the left depicts signal peptide; RPT: internal repeat 1; b Alignment of partial PtTryp2 sequences of the CRISPR/Cas9-generated mutants showing frameshift indels compared to wild type. The frequency by which the sequence was detected within the same colony is indicated in parenthesis. Font color coding: Black, WT sequence; Orange, functional domain containing target for CRISPR/Cas9; Purple, PAM sequence; Blue, Inserted bases; Red dashes, deleted bases. c PtTryp2 expression patterns of knockout and overexpression mutants under different conditions. FC fold change. Data are presented as mean values ± SD (n = 3 biologically independent samples). d PtTryp2 expression of knockout mutants exhibited no response to ambient N and P fluctuation. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. e Growth curves of different PtTryp2 mutants under different N and P conditions. Nutrient conditions in c–e are indicated by HNHP (Nutrient-replete), LNHP (N-depleted, P-replete), HNLP (N-replete, P-depleted), and LNLP (Nutrient-depleted). Data are presented as mean values ± SD (n = 3 biologically independent samples). Source data are provided as a Source Data file.Full size imageMoreover, we observed the growth physiology of different PtTryp2 mutants across different nutrient conditions. As shown in Fig. 4e and Supplementary Fig. 9, both of the knockout and overexpression of PtTryp2 resulted in decreases in the exponential growth rates (days 1–4) and maximum cell density across different N and P culture conditions. Taken together, these results demonstrate that both elevation and reduction of PtTryp2 expression result in cell growth repression, evidence that PtTryp2 has a crucial role in modulating cell growth in response to different N and P conditions.
PtTryp2 represses nitrogen assimilation and metabolismTranscriptomic data show that PtTryp2 knockout led to the upregulation of most of the nitrogen assimilation and metabolism genes under both N-depleted and replete conditions (Fig. 5a). The transcriptomic data are confirmed to be reproducible based on the correlation analysis of housekeeping genes (Supplementary Fig. 10 and Supplementary Table 3). Notably, the expression fold change of most N assimilation and metabolism genes under N-depleted, P-replete (LNHP) versus nutrient repete (HNHP) conditions were moderated in the PtTryp2 knockout mutant compared to that in its control (KOC), with the exception of GOGAT, which exhibited larger response to the nutrient changes in KOC (Fig. 5a). All these indicate that the inactivation of PtTryp2 enhanced N assimilation and metabolism to mitigate cell stress and reduce overall transcriptomic swing from N-depletion. Under replete conditions (HNHP), substantial transcriptional reprogramming and a significant increase in nitrate uptake rate and cellular N content was observed in the knockout mutants (KO1, KO2 and KO3) (Fig. 5b). The physiological changes were reversed in the overexpression cell lines: a decline in nitrate uptake rate and cellular N content was noted in PtTryp2-OE (Fig. 5c). All the results demonstrate that PtTryp2 functions as a repressor of nitrogen assimilation and metabolism.Fig. 5: Transcriptomic and physiological evidence that PtTryp2 directly represses nitrogen assimilation and metabolism.a PtTryp2 knockout resulted in upregulation of major nitrate-uptake and N-metabolism genes in PtTryp2 knockout (KO1) and control (KOC) under N-depleted (LNHP), P-depleted (HNLP), and nutrient-replete conditions (HNHP). NRT nitrate transporter, NR nitrate reductase, NiR nitrite reductase, GS glutamine synthetase, GOGAT glutamate synthase, GDH glutamate dehydrogenase, 2OG 2-Oxoglutarate; b NO3− uptake rate and cellular N content, increasing dramatically in PtTryp2-KO under HNHP, but decreasing remarkably under HNLP. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. c NO3− uptake rate and cellular N content, decreasing remarkably in PtTryp2-overexpressing P. tricornutum under HNHP, but increasing under HNLP. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. d Venn diagram showing the number of N-depletion induced DEGs in PtTryp2-KO1 and KOC. In parentheses, total number of DEGs; red font, upregulated; green font, downregulated. e Log2 fold changes (FC) of N-depletion induced differential gene expression in PtTryp2-KO1 against that in KOC. Most data points (93.37%) are distributed in 1,3 quadrants, indicating the same direction of change. Source data are provided as a Source Data file.Full size imageIn addition, when comparing N-depleted with N-replete conditions, 646 differentially expressed genes (DEGs) were identified in the blank vector control (KOC) but only 187 in PtTryp2-KO1, considerably fewer in the knockout mutant (Fig. 5d). Besides, the magnitude of change was smaller in PtTryp2-KO1 than in KOC for the majority (73%) of the DEGs (Fig. 5e). It is thus evident PtTryp2 in the wild type functions as an amplifier of general metabolic response to N-starvation by repressing nitrogen assimilation and metabolism. Notably, the PtTryp2-KO-promoted and PtTryp2-OE-repressed NO3− uptake patterns observed under nutrient repletion were reversed under P-depletion, indicating that PtTryp2’s roles in N and P signaling are not separated, but rather the protein might mediate the cross-talk between N and P signaling.Besides the direction of action (repression or promotion) shown above, the function of PtTryp2 involves another layer of regulation: the direction of its own expression changes. We find that PtTryp2 expression decreased under N-depletion and increased after N-supplement. Under this two-level regulatory scheme, PtTryp2 is a repressor of N uptake and assimilation genes and a promoter of N starvation-responsiveness in general metabolic pathways per se; yet its own expression decreases under N-limitation to upgrade N-uptake and assimilation under N depletion, and increases under N richness to prevent excessive N-uptake and assimilation; meanwhile, the decreased expression of PtTryp2 actually dampens the dynamic swing in the metabolic landscape in response to N-starvation. This PtTryp2-based regulatory mechanism might enable cells to swiftly respond to fluctuating N availability and cellular demand in order to finetune N responses so that N acquisition is optimized.
PtTryp2 promotes P starvation-induced genes and Pi uptakeAs shown above, PtTryp2 expression is downregulated under N-deficiency to release PtTryp2’s repressing effects on N-starvation response and to promote N uptake, thereby the cells achieve N homeostasis, and an opposite expression pattern of PtTryp2 was observed under P-deficiency, suggesting a N-P coregulation. However, the role of PtTryp2 in P-starvation responses and P homeostasis still needs to be unraveled. Toward that goal, we examined the effects of PtTryp2 inactivation on the expression changes of P starvation-induced genes and the inhibitory regulator of P signaling (SPX), which in plants is a typical P starvation response mechanism12. Consistently, most of Pi transporters (PTs) and alkaline phosphatase (APs) exhibited upregulation to P starvation response in KOC, but most of SPX genes showed downregulation (Fig. 6a).Fig. 6: Transcriptomic and physiological evidence that PtTryp2 positively modulates P starvation-induced genes during Pi starvation.a PtTryp2 knockout resulted in a reverse regulation of most P starvation-induced genes relative to that in control (KOC) under N-depleted (LNHP), P-depleted (HNLP), and nutrient-replete (HNHP) conditions. b PtTryp2 knockout caused decreases in Pi uptake and cellular P content under nutrient-replete condition (HNHP) but caused increases under N-depleted condition (LNHP). Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. c PtTryp2 knockout caused increases in Pi uptake rate and cellular P content under HNHP and LNHP. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. d Venn diagram showing the number of P-depletion induced DEGs in PtTryp2-KO1 and KOC. In parentheses, total number of DEGs; red font, upregulated; green font, downregulated. e Log2 fold changes (FC) of P-depletion induced differential gene expression in PtTryp2-KO1 against that in KOC. Most data points (95.69%) are distributed in 1,3 quadrants, indicating the same direction of change. Source data are provided as a Source Data file.Full size imageInterestingly, under P-depletion, PtTryp2 knockout downregulated the expression of most of PTs and APs, but upregulated most of the SPX genes (Fig. 6a), revealing PtTryp2’s role in WT to promote P-starvation responses. Consistent with gene transcription, PtTryp2 knockout lowered Pi uptake rate and cellular P content under the nutrient-replete condition (Fig. 6b), whereas an increase was noted in the overexpression cell line PtTryp2-OE (Fig. 6c). Based on RNA-seq, remarkably more DEGs were found for the P-depleted versus nutrient-replete comparison in PtTryp2-KO1 (1501) than that in KOC (277) (Fig. 6d). Besides, in PtTryp2-KO1, the majority of these DEGs (77.25%) exhibited greater fold changes than that in KOC (Fig. 6e). These results indicate that PtTryp2 upregulation in the wild type would dampen metabolic reprogramming in responses to P-limitation, and PtTryp2 downregulation would prevent cells from over P accumulation after P supplement, as opposed to the response to N-depletion. All these findings are indicative that PtTryp2 in the WT functions to upregulate the P starvation-induced genes and restrict general metabolic reconfiguration in response to P-limitation, a mechanism to maintain P homeostasis. Similar to that the PtTryp2-KO-promoted and PtTryp2-OE-repressed NO3− uptake patterns were reversed under P-depletion, the PtTryp2-KO-repressed Pi uptake pattern was reversed under N-depletion (Fig. 6b), implying that PtTryp2 might mediate the cross-talk between N and P signaling. The PtTryp2-OE-promoted Pi uptake pattern was not reversed under N-depletion, however, because N-depletion downregulated the expression of PtTryp2, resulting in the PtTryp2 expression pattern between OEC and OE similar to that under nutrient repletion.
PtTryp2 coordinately regulate N and P uptake and mediates N-P cross-talkGiven the PtTryp2-mediated cross-talk between N and P signaling in P. tricornutum implied in the results presented above, we were tempted to investigate the nature and the mechanism the cross-talk. Here, we uncover Pi and NO3− antagonistic interactions in P. tricornutum, which resemble that in land plants to achieve an optimal N-P nutrient balance13,14. In wild-type (WT) P. tricornutum, we observed a significant repression of NO3− uptake under P starvation and a significant repression of Pi uptake rate under N starvation. Consequently, cellular N content decreased under the P-depleted condition, and cellular P content decreased under the N-depleted condition, relative to nutrient-replete conditions (Fig. 7a, b). In accordance, the transcription of N assimilation and metabolism genes was repressed by P deficiency, and that of P starvation-induced genes was repressed by N limitation (Supplementary Fig. 11). Moreover, transcriptomic results demonstrated that PtTryp2 knockout led to the magnification of Pi and NO3− antagonistic interaction (Supplementary Fig. 11), linking PtTryp2 inactivation to the disruption of the N-P homeostasis. Taken together, our data reveal that PtTryp2’s function operates in opposite directions for N and P responses, but in a coordinated manner, consistent with a role to coregulate N and P signaling.Fig. 7: Illustration that PtTryp2 coordinately regulates N and P acquisition under fluctuating nutritional conditions.a NO3− uptake and cellular N content repressed under HNLP in wild-type cells (WT). Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. b Pi uptake and cellular P content repressed under LNHP in wild-type cells (WT). Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. c Time-course expression of PtTryp2 showed co-varied with the N/P nutrient ratio. Moreover, PtTryp2 expression fluctuated less at the N/P ratio of 16:1 compared to other N/P ratios. The 4 h after nutrient addition represents nutrient-repletion and 72 h nutrient-depletion. Data are presented as mean values ± SD (n = 3 biologically independent samples). d The cellular N/P ratio was significantly elevated by the inactivation of PtTryp2. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. e The cellular N/P ratio was significantly decreased by the overexpression of PtTryp2. Data are presented as mean values ± SD (n = 3 biologically independent samples). The comparisons between the averages of the two groups were evaluated using the one-tailed Student’s t test. The p values with significance (p ≤ 0.05) are shown. f Hypothetical model depicting the role of PtTryp2 in balancing N and P acquisition. Under N-depletion, PtTryp2 expression is downregulated to promote N-starvation responses and repress P-starvation responses. In contrast, under P-depletion, PtTryp2 expression is upregulated to reinforce P-starvation responses and lessen N-starvation response. By this feedback loop, optimal N-P uptake is achieved to maintain stoichiometric homeostasis. Upregulated genes and enhancement processes are shown in red, downregulated genes and weakened processes colored green. The black arrows depict transcriptional activation. Black bar at line’s end depicts inhibitory regulation. The gray arrows depict possible but unverified interaction between PtTryp2 and the existing P regulating cascade SPX-PHR or an equivalent of the N regulating cascade known in plants (SPX-NLP where NLP stands for NIN-like protein, a transcription factor). Source data are provided as a Source Data file.Full size imageTo further illustrate this, we have carried out PtTryp2 expression pattern analysis across different N/P nutrient stoichiometric ratio conditions, and found that PtTryp2 expression co-varied with the N/P nutrient ratio (Fig. 7c). The time-course analysis showed that PtTryp2 expression fluctuated less under different N or P conditions at the N/P ratio of 16:1 compared to other N/P ratios. The N/P nutrient ratio of 16:1 is considered balanced stoichiometry (Redfield ratio) and appears to be optimal for P. tricornutum growth (Supplementary Fig. 12), as previously documented15, suggesting that at this nutrient stoichiometry there is no need for a significant change in PtTryp2 expression to maintain N/P balance, but other N:P nutrient ratios deviating from 16:1 caused changes in PtTryp2 expression to maintain N/P balance. Moreover, the extent of change in PtTryp2 expression varied between cultures with different levels of N:P nutrient ratios, and between 4 and 72 h after culture inoculation from N- and P-depletion-acclimated parent culture into the experimental nutrient conditions. At 72 h PtTryp2 expression level increased with the degree of P stress (the higher the N:P ratio, the more P stressed the cultures were), except for the N:P = 1:1 condition, an extreme N-limited condition that seemed to cause PtTryp2 expression not to respond according to the general trend. Overall, all these data indicate that PtTryp2 responds strongly to the variability of the N:P ratio. Correspondingly, the cellular N/P ratio under nutrient-repletion also seems to be influenced by PtTryp2 expression level: the cellular N/P ratio was significantly elevated by PtTryp2 knockout, but conversely, was significantly decreased by the overexpression of PtTryp2 (Fig. 7d). Evidently, PtTryp2 serves to coordinate N and P uptake and metabolism to dampen the amplitude of N:P ratio changes that occur when the P. tricornutum cells experience fluctuations in nutrient conditions16,17. That is, PtTryp2 in P. tricornutum acts like an amplitude reducer of the N-P seesaw to achieve the N and P stoichiometric homeostasis (Fig. 7f).As critical nutrients for phytoplankton and plants, the balance and homeostasis of N and P are crucial to the growth of the organisms. For plants, nutrient supply in the soil is highly variable; therefore, to achieve optimal and coordinated utilization of N and P, integration of N and P signaling into an integrated network is required18. Recent studies have revealed the critical components of the network in the model plants Arabidopsis thaliana and Oryza sativa12,19,20,21. Similarly, phytoplankton in the ocean face remarkable environmental nutrient variations, and N and P nutrients are often limited22,23. Although the respective responses to N and P deficiencies have been extensively studied in phytoplankton24,25, an integrative signaling pathway of N-P nutrition cross-talk has remained unknown until now. It is striking to find that trypsin, rather than homologs of plant NRT1.1 and NIGT114,19, mediates and regulates the nitrate-phosphate signaling cross-talk.The two-level model of PtTryp2 function (Fig. 7f), including the direction of PtTryp2 action and the direction of PtTryp2 expression changes, demonstrate that PtTryp2 functions by shifting the setpoints, by tuning its own expression level, at which N signaling or P signaling is triggered in response to environmental nutrient fluctuations so that cells commit to appropriate responses. However, much of the mechanics in the regulatory cascade, from environmental nutrient sensing, PtTryp2-mediated signaling, to the regulation of the effectors such as N- and Pi-transporters and assimilatory genes, remains to be elucidated. Although the interplay between N and P nutrition based on SPX-NLP-NIGT1 and SPX-PHR-NIGT1 cascades, respectively have been uncovered in plants12,19, how PtTryp2 interacts with the SPX-PHR cascade26 and whether a SPX-NLP cascade or other regulatory cascades exist and interact with PtTryp2 for P and N nutrient regulation in phytoplankton remain to be addressed.As an initial attempt, we have performed transcriptional regulatory interaction analysis based on the Inferelator algorithm27 to predict the potential co-regulated genes in the PtTryp2-dependent regulatory cascade. Consequently, a set of 1034 genes co-regulated with PtTryp2 were identified, including 10 transcription factors (Supplementary Table 4), 10 N metabolism and assimilation genes, and a P responsive gene (Supplementary Fig. 13). Moreover, the functional enrichment of the gene set showed that PtTryp2 is possibly involved in post-transcriptional regulation, intracellular signal transduction pathway and kinase-based phosphorus metabolism and recycle pathway (Supplementary Fig. 14). The results hint on a potentially complex regulatory network that requires much more transcriptomes derived from more growth conditions than just the N and P conditions used in this study and other experimental approaches to unravel.We used the potential co-regulated gene list identified in this study in a comparative analysis with the published co-regulatory analysis datasets that contained hundreds of public RNA-seq datasets: DiatomPortal28 and PhaeoNet29. Interestingly, based on the DiatomPortal dataset, the PtTryp2 was found in the Phatr_hclust_0381 hierarchical cluster that consists of 10 genes, which has been identified as the GO term of ubiquitin-dependent protein catabolism. In terrestrial plants, the ubiquitination and degradation of SPX4 was found to mediate the nitrate-phosphate interaction signaling pathway by enabling the release of PHR2 and NLP3 into the nucleus to activate the expression of both phosphate- and nitrate-responsive genes12,19. In addition, we found 120 genes that were common in our gene list and PhaeoNet, some of which are transcription factors.Taken together, our analyses showed that the deletion and overexpression of PtTryp2 simultaneously impacted nitrogen and phosphorus uptake, nitrogen and phosphorus contents of the cell, and the N:P ratio. The simultaneous impact on N and P in opposite directions suggests that this protein either directly regulates the N and P uptake machinery or is close to the direct regulator, e.g., functioning through the ubiquitination and degradation of the direct regulators as in terrestrial plants. Furthermore, it is conceivable that one or more intermediate relays between PtTryp2 and the direct regulator would make it extremely challenging, if not impossible, to exert such precise and coordinated bidirectional regulation on N and P. To understand the mechanics of the regulatory mechanism, co-immunoprecipitation and Chromation immunoprecipitation sequencing are underway in our laboratory to experimentally identify the potential proteins and DNAs interacting with PtTryp2. Further studies on multiple fronts surrounding trypsin and its regulatory pathway are required for gaining an in-depth understanding of the interplay between N and P nutrition in phytoplankton and how phytoplankton will adapt to the potentially more variable and skewed N-P environment in the Anthropocene oceans. More

Brazil’s government risks fuelling the climate and biodiversity crisis by offsetting the fertilizer shortage resulting from Russia’s invasion of Ukraine this year (J. Liu et al. Nature 604, 425 (2022); S. Osendarp et al. Nature 604, 620–624; 2022). To produce an alternative fertilizer, it plans to mine up to 12 million tonnes annually of rhodoliths taken from an area in the South Atlantic that is roughly the size of the United Kingdom (see go.nature.com/3yhiyio).A full list of co-signatories to this letter appears in Supplementary Information.
Competing Interests
The author declares no competing interests. More