Supplementary Text This document is a supplementary methods section for “Coupled energy pathways and the resilience of size-structured ecosystems.” by Julia L. Blanchard, Richard Law, Matthew D. Castle & Simon Jennings Dynamics of coupled size-structured communities The analyses of food web resilience at steady state and responses to perturbations were contingent on the development of a size-based model that allowed us to couple the dynamics of the "pelagic predators" and "benthic detritivores". The underlying size-based model was developed from the coupled dynamic size spectra model proposed by Blanchard et al. (2009) but incorporated a dynamic reproductive process, which is a consequence of the energy density and energy budgets of individual organisms within the size spectra. All model equations and parameter values are shown in Tables S1 and S2. Model structure The coupling in the model consists of predation and production linkages between two size-structured communities: “pelagic predators” and “benthic detritivores”. In both communities, we are concerned with the continuous function N (m,t) (m-3 g-1) which gives the density per unit mass per unit volume for organisms of mass m at time t. The continuous processes of growth G and mortality D that arise from organisms encountering and eating available and suitable food govern the temporal dynamics and lead to a partial differential equation for each size spectrum i {P=pelagic predators, B =benthic detritivores} (Table S1). In the pelagic community there is a background “primary producer” spectrum that spans the size range 10-12 to 10-3 g which forms part of the food for “pelagic predators”. For simplicity we do not model the dynamics of the primary producer size spectrum and hold the numerical density as a constant function of body mass through time Npp=c·m-1. Feeding The feeding rate FPi (m,t) of a given size pelagic predator is a function of the preference for prey i in spectrum i, the volume of water searched per unit time AP m P (m-3 yr-1), and the amount of suitably sized food available in spectrum i, m m' N (m' , t ) m' dm' (Table S1) . The probability of a predator of size m eating an i encountered prey of size m' is given by the lognormal probability density function: ln m m' 2 1 2 2 2 m, m' exp (S.1) when m > m' and m, m' =0 otherwise, where is the logarithm of the preferred predator: prey mass ratio and the width. The assumption of a linear predator functional response is relaxed later. Benthic consumers compete for an unstructured shared pool of food. For simplicity, we call the latter group detritivores since in most benthic invertebrate communities detritus forms the bulk of their food (but it could also be supplemented by living phytoplankton). Detritus (BD) is produced by sinking phyto- and zooplankton, faeces from the pelagic predator size spectrum, and other dead organic matter from both communities. The feeding rate of a given size detritivore FB (m,t) depends on the volume of water either searched or filtered per unit time AB m B (m-3 yr-1) and the available biomass density of detritus BD (t ) (g m-3). Assimilation, growth and reproduction Of the food consumed only a fraction is assimilated. Assimilated food is allocated to growth K and reproduction R after accounting for losses associated with meeting metabolic requirements and activity (SM). We assume the fraction of each type of food that is assimilated, 1-Ei, reflects food quality (Table S2). Ki is the net growth conversion efficiency, the fraction of each type of assimilated food that is converted to production of new somatic tissue with the surplus fraction for reproduction given by Ri = 1 – (SMi + Ki). We assume that lower food quality is associated with lower assimilation efficiency and higher metabolic costs (hence lower conversion efficiencies to growth and reproduction). The combinations of conversion efficiencies for each food type meet the requirements of a simple energy budget (Table S2). Multiplying feeding rates by conversion efficiencies for each food type gives rise to the size-specific growth Gi(m,t) and reproductive ri(m,t) rates in each size spectrum i {P=pelagic predators, B =benthic detritivores} shown in Table S1. The reproductive investment per unit mass of an individual declines with body mass in each community, consistent with the levels of reproductive investment reported for marine teleost fishes and bivalves (Gunderson 1997, Van der Veer et al. 2006). The total biomass allocated to reproduction per unit time by all individuals in the size mi ,max spectrum larger than the egg size r (m, t ) N (m, t ) dm i mi ,min dm i divided by the mass of an egg gives the rate at which new offspring enter each size spectrum. We assume that both females and males invest the same energy to support the production of fertilized eggs. The renewal term or rate of change in the numerical density at the smallest (egg) size, mi ,m in , in each size spectrum is given by the rate that new individuals (eggs) enter the spectrum minus the rates that individuals either grow larger or die (S.2). The renewal function is similar to that given in Maury et al. (2007): dN i (mi ,min , t ) dt Gi (mi ,min , t ) N i (mi ,min , t ) Di (mi ,min , t )mi ,min N i (mi ,min , t ) 1 mi ,min (S.2) mi , max r (m, t ) N (m, t ) dm i mi , min dm i Mortality The overall mortality rate Di in each size spectrum i {P, B} arises from feeding by predators (as described above) and other sources of mortality (disease, senescence). Other mortality DiO includes an intrinsic term that decreases as a function of body size, and senescent mortality that increases with body size at the size ms. Detritus The biomass density of detritus BD changes from the fluxes of biomass being added and removed through time. Flux into the detritus pool, ID, is the total rate at which mass is egested and dead mass that is generated by non-predation mortality; a term for dead plankton is also included (Table S1). The proportion of detritus from the pelagic zone reaching the benthic zone is S. Within the benthic community, detritus is derived from dead biomass but not from egested material (which is assumed instead to become part of the sediment). The flux out of the detritus pool OD, is the biomass density consumed per unit time by all detritivores. Numerical integration method Prior to the stability analyses, numerical integrations were carried out to determine whether the model communities would tend to a steady state. For mathematical convenience and because size spectra usually span many orders of magnitude in mass, we carried out our numerical analyses on a log10 mass scale (see Blanchard et al. 2009). The numerical technique used to solve these equations was implicit time upwind finite differencing (Press et al. 1992). All simulations were run for a period of 50 years (with a daily time step and 0.1 log10 gram body mass intervals). Initial conditions Initial conditions for the predator and benthic detritivore size spectra comprised a low density of individuals in small sizes (< 1 g). The constant intercept for the density of the primary producer spectrum c was 0.006 m-3 g-1 and the initial detritus biomass density was 0.6 g m-3. These values represent average annual levels of productivity for the North Sea (Blanchard et al. 2009). Ecosystem-level properties at steady state 1. Total biomass density for each size spectrum (i) is expressed as: mmax N (m) m dm with units g m-3. i (S.3) mi min The detritus pool, BD, is expressed as a biomass density and has units of g m-3. 2. Total throughput is the sum of all consumption rates (or input rate in the case of the detritus pool) and is expressed as: F PP (m) N P m dm FPB (m) N P m dm FB (m) N B m d m I D (S.4) with units: g m-3 yr-1. 3. Production was assumed to be the total of all energy fluxes out (due to mortality) for each size spectrum (Allen 1971): D (m) N (m)m dm i i with units: g m-3 yr-1 (S.5) and the production:biomass ratio was equal to the production divided by the total biomass of each size spectrum at steady state : D (m) N (m)m dm i i N (m) m dm i with units: yr-1 (S.6) In order to examine production:biomass ratio for each resource (primary producers and detritus) and each consumer (pelagic predators and benthic detritivores) separately, the pelagic size spectrum needed to be separated into primary producer and pelagic predator size ranges. The above integral was therefore evaluated across the range [mmin, mP,min) for primary producers and [mP,min, mmax] for pelagic predators.The equivalent measure of production:biomass ratio for the non-living detritus pool is flux out of the detritus pool, OD (g m-3 yr-1) divided by biomass density of detritus, BD (g m-3). Predator functional response The feeding rates described in Table S2 do not include satiation and individuals simply feed in proportion to the suitably sized encountered prey (equivalent to a linear Lotka-Volterra functional response). We also tested the effects of incorporating a saturating Type II (hyperbolic) functional response on our results. We incorporated a handling time that was constant across predator size where h 5.7 E 7 years per gram (Ware 1978). The realised feeding rate of an individual predator at size m feeding on prey of size m' in each size spectrum i {P, B} then became: i AP m m m' N (m' , t ) m' dm' P i FPi (m, t ) P 1 h i AP m m m' N (m' , t ) m' dm' (S.7) i Similarly, the realised feeding rates of an individual benthic detritivore was changed to: FB (m, t ) AB m B BD (t ) 1 h AB m B BD (t ) (S.8) The predation mortality rate inflicted on an individual prey at size m in each size spectrum i by all size m' predators became: DiP (m, t ) i AP m' m' m N P (m' , t ) P P 1 h i AP m' m' m N (m, t ) m dm dm' (S.9) i The biomass flux per unit time out of the detritus pool from detritivore feeding became: OD (t ) = N B (m) FB (m, t ) dm (S.10) Using a Type II functional response did not affect our overall results. For local stability analyses the max values were marginally closer to zero when a Type II functional response was used, but there were no discernible differences in the patterns of how coupling affected local stability, steady state size spectra or returns following large perturbations. This is likely due to the fact that the levels of food availability in our model simulations were lower that the levels at which feeding rates would saturate. References Allen KR (1971) Relation between production and biomass. J Fish Res Bd Can 28: 15731581. Blanchard JL, Jennings S, Law R, Castle MD, McCloghrie P, Rochet M-J, Benoit E (2009) How does abundance scale with body size in coupled size-structured food webs? J Anim Ecol. 78: 270–280. Gunderson DR (1997) Trade-off between reproductive effort and adult survival in oviparous and viviparous fishes. Can J Fish Aquat Sci 54(5): 990-998. Maury O, Faugeras B, Shin Y-J, Poggiale JC, Ari TB, Marsac F (2007) Modeling environmental effects on the size-structured energy flow through marine ecosystems. Part 1: the model. Progr Oceanogr 74(4): 479-499. Press WH, Teukolsky SA, Vetterling WT, Flannery BP (1992) Numerical Recipes - The art of scientific computing, Cambridge University Press, 40 West 20th Street, New York, NY 10011, USA, second edition. Van der Veer HW, Cardoso JFMF, Van der Meer J (2006) Estimation of DEB parameters for various North Atlantic bivalve species. J Sea Res 56: 107-124. Ware DM (1978) Bioenergetics of pelagic fish: theoretical change in swimming speed and ration with body size. J Fish Res Bd Can 35:220-228.