Mesopelagic fish biomass and trophic efficiency of the open ocean

ARTICLE Recei ed 28 May 2013 |Accep ed 16 Jan 2014 |Published 7 Feb 2014 La ge mesopelagic fishes biomass and ophic e ficiency in he open ocean Xabie I igoien1, T.A. Kle je 1, A. Røs ad1, U. Ma inez2,G.Boy a 2, J.L. Acun ˜a3, A. Bode4, F. Eche a ia5, J.I. Gonzalez-Go dillo5, S. He nandez-Leon6, S. Agus i7,8, D.L. Aksnes9, C.M. Dua e7,8 & S. Kaa ed 1 Wi h a cu en es ima e o B1,000 million ons, mesopelagic fishes likely domina e he wo ld o al fishes biomass. Howe e , ecen acous ic obse a ions show ha mesopelagic fishes biomass could be significan ly la ge han he cu en es ima e. He e we combine modelling and a sensi i i y analysis o he acous ic obse a ions om he Malaspina 2010 Ci cumna iga ion Expedi ion o show ha he p e ious es ima e needs o be e ised o a leas one o de o magni ude highe . We show ha he e is a close ela ionship be ween he open ocean fishes biomass and p ima y p oduc ion, and ha he ene gy ans e e ficiency om phy oplank on o mesopelagic fishes in he open ocean is highe han wha is ypically assumed. Ou esul s indica e ha he ole o mesopelagic fishes in oceanic ecosys ems and global ocean biogeochemical cycles needs o be e ised as hey may be espi ing B10% o he p ima y p oduc ion in deep wa e s. DOI: 10.1038/ncomms4271 OPEN 1King Abdullah Uni e si y o Science and Technology (KAUST), Red Sea Resea ch Cen e , Thuwal 23955-6900, Saudi A abia. 2AZTI, A an za e a Elikaigin za ako Ins i u u Teknologikoa, He e a Kaia Po ualdea, 20110 Pasaia, Spain. 3Depa amen o de Biologı ´a de O ganismos y Sis emas, Uni e sidad de O iedo, Calle Ca ed a ´ ico Rod igo U ı ´a, Sin Nu ´me o, 33071 O iedo, Spain. 4Ins i u o Espan ˜ol de Oceanog a ı ´a (IEO), Cen o Oceanog a ´fico de A Co un ˜a, Apdo 130, E15080 A Co un ˜a, Spain. 5Depa amen o de Biologı ´a, Facul ad de Ciencias del Ma y Ambien ales, Uni e sidad de Ca ´diz, Campus de Excelencia In e nacional del Ma (CEI MAR), E-11510 Pue o Real, Spain. 6Ins i u e o Oceanog aphy and Global Change, Uni e sidad de Las Palmas de G an Cana ia, Campus Uni e si a io de Tafi a, Las Palmas de G an Cana ia, 35017 Cana y Islands, Spain. 7The UWA Oceans Ins i u e and School o Plan Biology, Uni e si y o Wes e n Aus alia, 35 S i ling Highway, C awley, Wes e n Aus alia 6009, Aus alia. 8Depa men o Global Change Resea ch, IMEDEA (UIB-CSIC), Ins i u o Medi e a ´neo de Es udios A anzados, Espo les 07190, Spain. 9Depa men o Biology, Uni e si y o Be gen, Be gen N-5020, No way. Co espondence and eques s o ma e ials should be add essed o X.I. (email: Xabie .i igoy[email p o ec ed]). NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions 1 &2014 Macmillan Publishe s Limi ed. All igh s ese ed. Mesopelagic fishes— he small fishes li ing in he ocean’s wiligh zone— o m one o he mos cha ac e is ic ea u es o he open ocean: he deep sca e ing laye a dep hs be ween 200 and 1,000 m, isible in he echosounde display o essels sailing all oceans1. Whe eas he mesopelagic fish genus Cyclo hone sp. is likely he mos abundan e eb a e on ea h2, mesopelagic fishes emain one o he leas in es iga ed componen s o he open-ocean ecosys em, wi h majo gaps in ou knowledge o hei biology and adap a ions, and e en majo unce ain ies abou hei global biomass. T awling es ima es sugges ha he biomass o mesopelagic fishes is B1,000 million ons3,4, a numbe commonly used in assessmen s o ecosys em unc ion and he biogeochemis y o he global ocean5,6. Howe e , e en o he o iginal es ima e i was s a ed ha ‘mos o he gea used o ob ain he a ailable in o ma ion ob iously unde es ima e he biomass p esen ’3, and he e ficiency o di e en ypes o ne s o cap u e mesopelagic o ganisms has been u he ques ioned ecen ly, wi h in e -calib a ion exe cises showing o de -o -magni ude di e ences in he cap u ed biomass depending on he ype o gea 7. Mo eo e , awling-based biomass es ima es a e sys ema ically below acous ic es ima es7–9, as mesopelagic fishes ha e been shown o exhibi escape eac ions o ne s, ende ing awling da a suspec o g oss unde es ima ion10. He e we combine a sensi i i y analysis o acous ic da a collec ed du ing Malaspina 2010, he Spanish Ci cumna iga ion Expedi ion (Decembe 2010–July 2011, Fig. 1a), and modelling, o show ha mesopelagic fishes biomass in he open ocean is abou one o de o magni ude highe han p e ious es ima es. We u he mo e examine he mesopelagic fishes biomass ela i e o p ima y p oduc ion (PP) and conside he implica ions o hese es ima es o he unc ioning o he open-ocean ecosys em and biogeochemical cycles. Resul s Acous ic biomass es ima es. A Sim ad EK60 echosounde ope a ing a 38 kHz equency was used o ob ain da a h ough- ou he 32,000-mile oyage (Fig. 1a). We used da a ob ained du ing he day ime om 200–1,000 m dep hs, comp ising he main diu nal habi a o mesopelagic fishes, o calcula e fish biomass and conside ing he di e en sou ces o unce ain ies in ol ed (Me hods). The a e age (±s.d.) nau ical a ea sca e ing coe ficien (s A ,m 2nmi2) in he 200–1,000 m laye o he whole c uise was 1,864±1,341, wi h indi idual es ima es anging om 158–7,617 m2nmi2(N¼209, Fig. 1a,b). The s A was sig- nifican ly co ela ed wi h he 2010 sa elli e-de i ed a e age daily PP (mg C m2d1)11, (Fig. 2, Pea son’s ¼0.77, F co ec ed o spa ial au oco ela ion ¼26, D.F. co ec ed o spa ial au o- co ela ion ¼18, Po0.01). This ela ion shows he e o- scedas ici y (Supplemen a y Fig. 1) and because o he c uise design ( om coas o coas a simila dis ances) i also shows spa ial au oco ela ion a di e en spa ial scales (Mo an I es , Supplemen a y Fig. 2)12. The e o e, we used a geog aphically weigh ed eg ession (GWR)13 on ln- ans o med alues o pa ame e ize he ela ionships be ween PP and s A h ough eg ession analysis (Me hods). The GWR was significan in 96% o he sampling poin s, anging om oceanic gy es o nea shel a eas (Supplemen a y Fig. 3). The only a ea whe e he GWR was no significan was in he icini y o he Humbold upwelling sys em whe e he wa e column was se e ely hypoxic below 100 m (Supplemen a y Fig. 4). We used he equa ions ob ained h ough he GWR in wo ways o de i e es ima es o he s A om sa elli e-es ima ed PP: fi s , he median alues o all he GWR pa ame e s and second, as he eg ession pa ame e s abo e 400 mg C m2d1o PP show less a iabili y (Supplemen a y Fig. 5), we also conside ed di e en equa ions wi h he median alues o da a below and abo e 400 mg C m2d1(Supplemen a y Table 1). Use o he ela ion be ween PP and s A (Fig. 2), oge he wi h he dis ibu ion o PP, o in eg a e mesopelagic fishes biomass o a eas wi h bo om dep hs deepe han 1,000 m be ween 40°N and 40°S ( he la i udinal ange co e ed by he Malaspina expedi ion, Fig. 1a) yielded a o al s A o 5.6 1017 (Table 1). We ans o med he backsca e ing s eng h in eg a ed o e he 200–1,000 m laye , s A , in o mesopelagic fishes biomass using db/weigh a ios de i ed om he li e a u e (Table 2). To accoun o he a iabili y in hese a ios we conside ed he minimum, maximum, a e age, median, 25 and 75% qua iles o he li e a u e alues. Table 1 p esen s he ange o mesopelagic fishes biomass ob ained using di e en app oaches o es ima e he o al s A be ween 40°N and 40°S (o dina y leas squa es eg ession, GWR, a e age s A om he c uise) and db/weigh a ios. We ound ha o al mesopelagic fishes biomass was obus agains di e en app oaches o es ima e s A wi h a di e ence o 25% be ween he maximum (GWR) and he minimum (c uise a e age). I we conside he ange be ween 25 and 75% qua iles o he li e a u e db/weigh a ios as a easonable ep esen a ion o he mesopelagic fishes a ound he wo ld, ou es ima es o mesopelagic fishes biomass would ange om 6,000–200,000 million ons, wi h median alues be ween 11,000 and 15,000 million ons (Table 1). These es ima es o mesopelagic fishes biomass limi ed o 40°N and 40°S a e one o de o magni ude highe han he p e ious global es ima e o 1,000 million ons3,4. As mesopelagic fishes a e no he only sou ce o backsca e we conduc ed a sensi i i y analysis o unde s and which combina- ions o db/weigh a ios and ac ion o he backsca e coming om fishes would esul in he p esen day es ima e o 1,000 million ons. Figu e 3 shows ha he p esen day es ima e can only be a ained i all mesopelagic fishes popula ions had a db/weigh a io close o he maximum obse ed in he li e a u e and mesopelagic fishes ep esen ed o20% o backsca e o i hey we e o10% o he backsca e o a wide dis ibu ion o db/weigh alues (Fig. 3). Any con ibu ion o he backsca e highe han 20%, as gene ally epo ed9,14, esul s in a biomass se e al imes highe han he one accep ed (Fig. 3). Modelling biomass es ima es. Global fishes biomass es ima es de i ed om ood web models a e ypically be ween 900 and 2,000 million ons15–17, abou 10- old below ou di ec , acous ic measu emen . In some o hose models, he mesopelagic biomass is an inpu a ec ing o al fishes biomass es ima es15; in o he s he final alue is e y sensi i e o he ans e e ficiency used16, and finally, he mos ecen wo k by T emblay-Boye e al.17 using ECOTROPH18 de i ed i s es ima e o fishes biomass assuming ha 10% o he PP was ans e ed om PP o he bi o es (Gascuel pe s. comm.). This alue o 10% co esponds o he consump ion o PP by mesozooplank on in p oduc i e a eas (mainly copepods)19. Ye , ample e idence shows ha mic o- zooplank on, no mesozooplank on, a e he majo consume s o PP, consuming 70–80% o he PP on a e age20. The pe cen age o he PP consumed by mesozooplank on in he less p oduc i e oceanic wa e s is also mo e likely o be B20% a he han 10% ( e . 19). The e o e, i we conside mic o and mesozooplank on oge he he pe cen age o PP en e ing he ood web should be close o 90%. Hence, we ha e un he ECOTROPH model wi h a gene ic open-ocean ecosys em model using he a e age PP o he a ea be ween 40°N and 40°S (344 mg C m2d1) and conside ing a flux om PP o he ophic web o 70, 80 and 90%, wi h ans e e ficiencies be ween ophic le els anging om 0.05 o 0.2 ( e . 21) and an a e age empe a u e o 9 °C. Resul s, shown in Table 3, ange om 2,000–70,000 million ones. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 2NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions &2014 Macmillan Publishe s Limi ed. All igh s ese ed. Bo h along he Malaspina ansec (Fig. 4) and in he global es ima e (Table 3), he modelled biomasses using ECOTROPH wi h a ans e e ficiency o 10% all wi hin he ange o acous ic es ima es using he median db/weigh a io (10–15,000 million ons). P oduc ion and ans e e ficiency. We es ima ed mesopelagic fishes p oduc ion (MFP) conside ing equa ions ela ing fishes p oduc ion/biomass (P/B) a ios o he ophic le el, assumed o be 3.2 o mesopelagic fishes (www.Fishbase.o g) and Tempe a u e (°C) Dep h (m) 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 × 104 100 a b c 200 300 400 500 600 700 800 900 1,000 S (dB) −85 −80 −75 −70 −65 −60 −55 −50 AT IO WP EP AT 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 × 104 0 100 200 300 400 500 600 700 800 900 1,000 Dis ance a elled (km) Dep h (m) 0 5 10 15 20 25 30 1,000 P ima y p oduc ion (mg C m–2 d–1)gWW m–2 800 600 400 200 0 351 264 176 89 18 Figu e 1 | The Malaspina c uise. (a) The su ace-in eg a ed es ima ed mesopelagic fishes biomass (g we weigh pe m2) o he 200–1,000 m dep h ange along he Malaspina 2010 Expedi ion c uise ansec (black ci cles) supe imposed on a sa elli e-de i ed global map o PP (a e age in mg C m2d1 o 2010–colou ba ); (b) a day ime echog am om 0–1,000 m along he c uise ack (measu ed in dB—colou ba ), and (c) in e pola ed empe a u e p ofiles along he c uise ack (measu ed in °C—colou ba ). The black iangles in band cindica e he bo de be ween oceanic basins. AT o A lan ic Ocean, IO o Indian Ocean, WP o Wes e n Pacific and EP o Eas e n Pacific. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 ARTICLE NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions 3 &2014 Macmillan Publishe s Limi ed. All igh s ese ed. empe a u e22, which was de i ed om empe a u e p ofiles weigh ed by he dep h dis ibu ion o he fishes biomass (Fig. 1b,c). This esul ed in an a e age P/B o mesopelagic fishes o 0.5 (a e age±s.d. 0.51±0.05). Biomass was es ima ed in a conse a i e way using he median and 75% qua ile o he li e a u e db/weigh a ios. As he mean-weigh ed empe a u e in he deep laye s was ela i ely uni o m (a e age 9 °C, ange 6 °C o 13 °C, Fig. 1c), we ound ha MFP was also significan ly co ela ed o PP (Pea son’s ¼0.68, F co ec ed o spa ial au oco ela ion ¼21, D.F. co ec ed o spa ial au o- co ela ion ¼26, Po0.01). The a e age MFP alues calcula ed along he Malaspina 2010 ansec anged be ween 16 and 27 g m2y1(a e age±s.d. MFP ¼27.2±20.1 and 15.6±11.6 g m2y1 o median and 75% qua ile db/weigh a ios). The co esponding a e age ans e e ficiency om PP (TL1) o MFP (TL3) was 0.02 (a e age±s.d. MFP/PP a io ¼0.022±0.013) when using he median db/weigh a io and 0.01 when using he 75% qua ile (a e age±s.d. MFP/PP a io ¼0.013±0.008) (Fig. 5). Discussion Th ough he use o acous ic da a deepe han 200 m and conside ing he open-ocean a eas be ween 40°N and 40°S wi h bo om dep h deepe han 1,000 m and ophic le els 3–3.5 his analysis ocuses on mesopelagic fishes, excluding fishes li ing in shel a eas, epipelagic fishes and highe -le el p eda o fishes in he open ocean. Ou esul s indica e ha he mesopelagic fishes biomass in he global ocean is much highe han he p e ious es ima e o 1,000 million ons3,4. The sensi i i y analysis (Fig. 3) indica es ha a be e knowledge o he composi ion and acous ic p ope ies o he mesopelagic communi y is needed o ob ain accu a e es ima es o he biomass. Howe e , conside ing he ag eemen be ween he acous ic es ima e using he median db/weigh alue and he model using 10% e ficiency be ween ophic le els we sugges ha he mos likely es ima e o mesopelagic fishes biomass be ween 40°N and 40°S is a leas an o de o magni ude highe han he p e ious es ima e o 1,000 million ons. Ou analysis is limi ed o he la i udinal ange co e ed by he Malaspina expedi ion (40°N–40°S). Ye , mesopelagic fish a e also abundan in highe la i udes, al hough hei abundance s ongly dec eases in pola wa e s. Conside ing only he su ace o he a ea he es ima e could be B30% highe i expanded o he deep ocean be ween 70°N and 70°S. Ac ually, he biomass le els we find using acous ics and he median db/ weigh , as well as he ela ion be ween PP and mesopelagic fishes biomass, ag ee well wi h ecen obse a ions in he no heas Pacific Ocean23. Bo h ou es ima e using acous ics and he local es ima es in he no heas Pacific Ocean coincide in biomasses one o de o magni ude highe han p e ious es ima es. As mesopelagic fishes likely domina e he global fishes biomass e en wi h he o me es ima e, ou esul s indica e ha he cu en global fishes biomass needs o be upg aded by one o de o magni ude. P ima y p oduc ion in he oligo ophic ocean is domina ed by picoplank on24, which a e no e ficien ly cap u ed by copepods. I is he e o e gene ally implied ha he ‘mic obial loop’ domina es he ophic web in oligo ophic a eas, which he e o e should suppo a lowe ans e e ficiency om PP o mesozooplank on, and hence o fishes, han mo e p oduc i e ma ine sys ems25. Howe e , he e is no di ec e idence o a lowe e ficiency in less p oduc i e a eas26. On he con a y some da a sugges a igh e coupling be ween PP and g azing in oligo ophic seas19. Al hough a ans e e ficiency o 0.1 be ween consecu i e ophic le els is commonly assumed, alues o 0.2 and highe a e also used in ophic models o aqua ic ecosys ems21. Recen s udies ha e shown ha ood- ans e e ficiencies can a y se e al o de s o magni ude, om less han 0.001 o significan ly mo e han 0.1 ( e . 27). The ans e e ficiency be ween consecu i e ophic le els implici in ou ans e e ficiency es ima es be ween TL1 and TL3 appea s o be in conflic wi h he common wisdom ha ophic chains in oligo ophic sys ems a e less e ficien han in p oduc i e shel wa e s28 (gene ally conside ed o be 0.1). Howe e , he ans e e ficiency om PP o mic ozooplank on is p obably 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 0 200 400 600 800 1,000 s A (m 2 nmi –2 ) P ima y p oduc ion (mg C m–2 d–1) Figu e 2 | Rela ionship be ween PP and acous ic backsca e . The ela ionship be ween PP (2010 annual a e age mg C m2d1) and he nau ical a ea sca e ing coe ficien (s A ,m 2nmi2) om 200–1,000 m du ing day ime. Table 1 | Acous ic fishes biomass es ima es. s A es ima e Acous ic fishes biomass es ima es To al s A A e age Median 75% 25% Max Min 34.6 db kg130.8 db kg 128.4 db kg142.2 db kg126.8 db kg 146.8 db kg1 OLS: s A ¼2384.4* ln(PP)—11678 4.24E þ17 28,363 11,824 6,804 163,215 4,707 470,717 OLS: ln (s A )¼1.52* ln (PP)—1.36 4.70E þ17 31,449 13,110 7,544 180,972 5,219 521,930 GWR: ln (s A )¼1.36* ln (PP)—0.2 5.57E þ17 37,264 15,534 8,939 214,433 6,184 618,432 GWR di e en equa ions o PP abo e and below 400* 4.38E þ17 29,321 12,223 7,034 168,725 4,866 486,607 C uise a e age s A xocean su ace deepe 1,000 m 4.14E þ17 27,427 11,433 6,579 157,826 4,552 455,176 GWR, geog aphically weigh ed eg ession; OLS, o dina y leas squa es eg ession. To al backsca e be ween 40°N and 40°S es ima ed om PP ( o al s A ) and di e en acous ic o weigh (db kg1) a ios (see Table 2). *See Supplemen a y Table 1 o de ails on he GWR equa ion pa ame e s abo e and below 400 mg C m2d1. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 4NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions &2014 Macmillan Publishe s Limi ed. All igh s ese ed. highe in wa m a eas o he ocean (oligo ophic zones) han in eu ophic a eas, as g azing by he e o ophs is mo e igh ly coupled o PP in wa me seas whe e he e o ophic me abolic a es inc ease as e han pho o ophic a es ela i e o inc eases in empe a u e29. Fu he , he dominance o pico-sized p ima y p oduce s in he oligo ophic ocean24 implies small losses due o sinking pa icles. Mesopelagic fishes a e isual p eda o s eeding on mesozoo- plank on (TL3–4, see Me hods). The ophic ans e e ficiency om mesozooplank on o fishes is also p obably highe in less p oduc i e and clea oceanic wa e s han in mo e p oduc i e and u bid shel and coas al wa e s (Supplemen a y Table 2): clea wa e has a low beam a enua ion coe ficien (c), which a o ds isual p eda o s long sigh ing dis ances ( ), acco ding o ¼k/c, whe e kis de e mined by he con as and he con as sensi i i y o he p ey and he p eda o , espec i ely30. The olume sea ched pe uni ime (V) scales wi h 2 o a c uising p eda o 31 and is he e o e e ec i ely enhanced by dec eased c( ha is, inc eased wa e cla i y) acco ding o Vpc2(Me hods). In addi ion, clea wa e also has a low a enua ion coe ficien o downwelling i adiance (K), which means ha ligh pene a es deepe . Wi hin he ange allowed by deep hypoxic laye s32, e idence sugges s ha e ical ex ension o fishes habi a (H) scales acco ding o Hp1/K( e . 33). Thus, inc eased wa e cla i y ends o inc ease bo h he sho ( ) and he long (H) ange o isual o aging, he eby enhancing he ans e e ficiency om mesozooplank on o mesopelagic fishes h ough mo e e ec i e isual p eda ion. The quan i y c2K1, which is p opo ional o he po en ial olume sea ched, is abou one o de o magni ude highe o clea oceanic wa e (K¼0.044 m1, he a e age in ou s udy) han o u bid coas al wa e s wi h K40.1 m1(Supplemen a y Table 2). Thus, al hough inc eased PP in p oduc i e a eas is likely o inc ease p oduc ion a TL2 and TL3, i is also likely, beyond a ce ain poin whe e high plank on biomass leads o high ligh a enua ion and isual cons ain s become se e e, o dec ease he ans e e ficiency be ween mesozooplank on and isual p eda o s34 ela i e o he ans e e ficiency in clea wa e s a he oligo ophic ocean. The e idence ha mesopelagic fishes biomass, and conse- quen ly he o al fishes biomass, is 10- old highe , o e en mo e, han p e iously assumed has impo an implica ions o ou unde s anding o he ca bon fluxes in he ocean. A disc epancy in he es ima es o expo p oduc ion using 234Th:238U disequilib ia and shallow sedimen aps has been sys ema ically epo ed35. This disc epancy has gene ally been a ibu ed o a i ac s in he aps36 o o he episodic na u e o he sinking e en s37. A compa ison be ween e e se modelling es ima ed C expo fluxes and sedimen aps sugges ed ha shallow aps (o1,000 m) unde es ima e he flux, whe eas deep aps (41,000 m) measu ed fluxes highe han he modelled ones38. An o de o magni ude highe biomass o mesopelagic fishes migh explain he di e ence. Mesopelagic fishes pe o m diel e ical mig a ion, eeding a nigh in he uppe laye s (eupho ic) and exc e ing and espi ing a dep h a day. This implies ha Table 2 | Ta ge s eng h o mesopelagic fishes a su ey equencies. Species/ ype Taxonomy/ g oup F equency (kHz) Swimbladde Leng h (cm) Weigh (g) A g TS (dB) dB kg1Acous ic model Acous ic model sou ce Rema k L/W sou ce Ce a oscopelus wa mingii Myc ophidae 38 No 6 2.0 73.8 46.8 49.4*log10(L) 112.2 56 Ce a oscopelus made ensis 59 Diaphus ch yso hyncus Myc ophidae 38 No 6 1.4 72.6 43.9 30.5*log10(L) 96.3 56 Diaphus ga mani 60 Diaphus ga mani Myc ophidae 38 No 6 1.4 71.5 42.9 54*log10(L) 113.5 56 —60 Myc ophidae 38 No 6 2.2 67.3 40.8 52.7*log10(L) 108.3 56 Myc ophum punc a um 59 No oscopelus japonicus Myc ophidae — No 6 1.8 71.1 43.7 20*log10(L) 86.7 61 No oscopelus elonga us 59 S enob achius leucopsau us Myc ophidae 38 No 6 0.8 72.2 41.1 32.1*ln(log10(L)) 64.1 62 —63 Symbolopho us cali o niensis Myc ophidae — No 6 2.6 70.1 44.4 20*log10(L)) 85.7 62 Symbolopho us e anyi 59 Ben hosema glaciale Myc ophidae 38 Yes 6 3.0 58.0 32.8 — 64 —— Ce a oscopelus wa mingii Myc ophidae 38 Yes 6 2.0 57.6 30.6 26.3*log10(L) 78.1 56 Ce a oscopelus made ensis 59 Diaphus ga mani Myc ophidae 38 Yes 6 1.4 56.7 28.0 34.5*log10(L) 83.5 56 —60 Diaphus he a Myc ophidae 70 Yes 5.55 2.6 55.7 29.8 20*log10(F23)) 70.6 65 —66 Diaphus he a Myc ophidae Yes 6 3.3 54.3 29.5 11.8*log10(L) 63.5 61 —66 Fish sb — 38 Yes o353.0 27.8 — 67 W¼3g — Fish sb — 38 Yes o352.0 26.8 — 67 67W¼3g — Myc ophid Myc ophidae 38 Yes 5.8 — 55.0 — * epo ed modal alue 68 —— Myc ophum aspe um Myc ophidae 38 Yes 6 2.2 53.3 26.8 45.4*log10(L) 88.6 56 Myc ophum punc a um 59 Myc ophid Myc ophidae 38 ? NA 3 56.2 31.0 31 dB kg 169 —— Myc ophid Myc ophidae 38 Yes 6–10 — 53.0 — — 70 —— Myc ophid Myc ophidae 38 Yes 9 50 29.5 * epo ed modal alue 68 —— Fish sb 38 Yes — 3–10 49.0 27.1 — 67 W¼6.5 g — ——————— — — —— Median wi hou swimbladde —————71.5 43.7 — — — — Median wi h swimbladde —————54.7 29.5 — — — — Median all — — — — — 56.7 30.8 — — — — A e age all — — — — — 60.6 30.7* — — — — 75% qua ile all — — — — — 70.6 42.4 — — — — 25% qua ile all — — — — — 53.8 28.4 — — — — This able summa izes some ecen a ge s eng h (TS) esul s o mesopelagic fishes. In cases whe e acous ic models exis , we ha e calcula ed TS a leng hs close o 6 cm, and we ha e used published eg essions o leng h and weigh (L/W) ela ionships o he same o closely ela ed species o es ima e sca e ing om 1 kg biomass (dB kg 1). *A e age alue compu ed on he linea domain. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 ARTICLE NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions 5 &2014 Macmillan Publishe s Limi ed. All igh s ese ed. mesopelagic fishes d i e a e ical flux om he su ace o he mesopelagic laye , bypassing he de ec ion capaci y o sedimen aps. A he same ime, by de aeca ing in deep wa e s, aeces p oduced wi h su ace o ganic ma e s a sinking a dep hs o B500–700 m, bypassing consump ion in a la ge ac ion o he wa e column and inc easing he obse ed flux in deep aps. Basically, mesopelagic fishes accele a e he flux by ac i ely anspo ing o ganic ma e in he op laye o he wa e column, whe e mos o ganic ca bon is los om he sedimen a y pa icle flux. Del Gio gio and Dua e39, using fishes p oduc ion es ima es de i ed om fishe ies landings (mesopelagic fishes a e no comme cially fished), conside ed fishes espi a ion o li le significance o he global ocean. Howe e , a subs an ial e ision o he mesopelagic fishes biomass upwa ds ende s fishes espi a ion ele an . Assuming fishes espi a ion o be nine imes p oduc ion39 and hal o i o happen in deep wa e s, up o 10% o he PP could be espi ed by mesopelagic fishes in deep wa e s (Table 4). This es ima e is again in ag eemen wi h he local es ima e in he no heas Pacific23. The es ima e needs efinemen in e ms o ime occupied and he espi a ion a es in deep laye s (usually mo e han hal o he day, bu also a low empe a u es), bu indica es ha , in deep laye s, he sum o MFP and espi a ion is in he o de o magni ude needed o explain he disc epancies be ween 234Th:238U disequilib ia and shallow sedimen aps. Mo eo e , he exc e ion in deep laye s o ma e ials inges ed by mesopelagic fishes in he su ace migh pa ly explain he unexpec edly la ge mic obial espi a ion in he deep ocean40. Ou esul s s eng hen he p e ious claim ha mesopelagic fishes a e he mos abundan fishes and, indeed, he mos abundan e eb a es in he biosphe e2. The esul s om he su ey p esen ed he e sugges ha ophic ans e e ficiency om p ima y p oduce s o fishes has been unde es ima ed in he oligo ophic ocean, wi h he high ans e e ficiency om p ima y p oduce s o fishes associa ed wi h wa m wa e empe a u es and ex eme wa e anspa ency, maximizing p ey cap u e by isual p eda o s. As many mesopelagic fishes, dominan in oceanic a eas, a e s ong e ical mig a o s, eeding in he uppe wa e column and exc e ing a dep h, hese esul s ha e impo an implica ions o he biogeochemical cycles o he ocean, as hese animals p o ide ophic connec i i y and anspo o ganic ca bon be ween he su ace and he mesopelagic ocean, and could help explain exis ing disc epancies be ween flux es ima es ob ained by he 234Th:238U me hod and sedimen aps35, as well as he unexpec edly la ge mic obial espi a ion in deep wa e 40. E en wi h he cu en 109 ons es ima e mesopelagic fishes a e conside ed o play a key ole in he wo ld’s oceans as a link be ween plank on and op p eda o s41 and in he oxygen 1 10 10 10 100 100 100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 –50 –45 –40 –35 – 30 –25 Fish ac ion o he acous ic biomass db kg–1 Max db kg–1 75% qua ile db kg–1 25% qua ile db kg–1 Min db kg–1 Figu e 3 | Sensi i i y analysis o he assump ions used o es ima e biomass. Compu a ions a e based on he o al backsca e es ima ed om he GWR eg ession (Table 1, 5.57E þ17), con ou s a e he a io o ou es ima e o 1,000 million ons. Qua iles, maximum and minimum alues o he db/weigh a ios ound in li e a u e a e indica ed by he do ed lines. Table 3 | Fishes biomass Eco oph es ima es. T ophic e ficiency be ween TL % PP o ood chain 70% 80% 90% 5% 2,322 2,655 2,985 10% 10,691 12,224 13,744 20% 57,054 64,148 71,299 Eco oph fishes biomass es ima es (millions o ons) be ween 40°N and 40°S as a unc ion o he e ficiency be ween ophic le els and he flux om PP o he fi s ophic le el (TL). 0 50 100 150 200 250 0 200 400 600 800 1,000 Biomass (g m–2) P ima y p oduc ion (mg C m–2 d–1) dB/weigh –30.8 ECOTROPH Figu e 4 | Rela ion be ween PP and fish biomass. The ela ionship be ween PP (2010 annual a e age mg C m2d1) and he es ima ed mesopelagic fishes biomass using he median db/weigh a io (black ci cles) and he ECOTROPH model wi h a ans e e ficiency o 0.1 and conside ing ha 90% o he PP en e s he ood web ( ed line). The eg ession equa ion o he acous ic biomass es ima e agains PP is mesopelagic fishes biomass (g m 2)¼0.185 PP (mg C m2d1)—6.66, 2¼0.46, Po0,001, n¼209). 0.00 0.02 0.04 0.06 0.08 0.10 0.12 100 1,000 MFP/PP P ima y p oduc ion (mg C m–2 d–1) Figu e 5 | T ophic e ficiency. The es ima ed ophic e ficiency om PP o MFP ( he a io o fishes p oduc ion o PP) as a unc ion o PP. The line is a 50 pe iod unning a e age o 50 con iguous es ima es. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 6NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions &2014 Macmillan Publishe s Limi ed. All igh s ese ed. deple ion o he open ocean deep laye s32. Wi h he 10- old highe biomass ound in his s udy and in ecen local s udies23 he conclusion abou po en ial impac s o ha es ing mesopelagic fishes ex ends o he global biogeochemical cycles. This finding calls o an e o o imp o e he accu acy o he es ima es o he biomass and composi ion o he mesopelagic communi y. A mo e accu a e es ima e will equi e echnological de elopmen s o inc ease he cap u abili y o mesopelagic fishes and ob ain de ailed a ge s eng hs, as well as coo dina ed c uises ac oss ep esen a i e a eas o he wo ld ocean wi h su ficien esolu ion o add ess mesoscale s uc u es. Me hods Echosounde .Con inuous acous ic measu emen s we e made wi h a calib a ed42 Sim ad EK60 echosounde (7°beam wid h), ope a ing a a equency o 38 kHz and wi h a ping a e o 1 ansmi ed pulse pe 2 s. The da a we e s o ed o la e analysis, ca ied ou using he LSSS so wa e43. The echosounde da a we e episodically a ec ed by noise om a ious sou ces; consequen ly, p io o impo in o he LSSS so wa e o pos -p ocessing, he da a we e subjec ed o a se ies o fil e s o emo e bad da a. These fil e s in oduced a bias by emo ing he highes in ensi y da a, howe e . The backsca e es ima es, which a e he basis o ou biomass es ima es, a e he e o e conse a i e. The fil e s wo ked by compa ing he in eg a ed backsca e o e a dep h ange wi h he backg ound backsca e o e he same dep h ange. Backg ound in ensi ies we e de ec ed using a median fil e ha was 400 pings wide, upda ed e e y 100 pings. Pings a ec ed by a enua ion we e defined as pings wi h backsca e 46dB below he median in ei he o he dep h anges o 50–600 m o 600–1,000 m. Pe iods wi h backsca e 44 dB abo e backg ound le els in he dep h ange 800–1,000 m we e also ma ked. Pings agged by hese fil e s we e excluded in u he analyses. Las ly, a simple 9-poin unning median (ho izon al) emo ed sho e i egula spikes. A e manual sc u iny o he emaining da a, he da a we e in eg a ed in 2-minu e-by-2-me e bins a a h eshold o 90 dB. A e in eg a ion in LSSS, da a we e impo ed in o R44 o u he analysis. Acous ic esul s we e spli in o day, nigh and c epuscula da a, using he unc ion ‘sun ise ’ om he ‘map ools’ package, wi h c epuscula pe iods defined as sunse /sun ise±1h. To plo he echog am (Fig. 1b), he p ocessed LSSS da a we e expo ed in 10-minu e-by-1-me e bins o Ma lab, whe e he nau ical a ea sca e ing coe ficien (s A ,m 2nmi2) was con e ed o olume backsca e ing s eng h (S ,dB e1m1). The day ime da a we e ex ac ed (2 h a e sun ise and 2 h be o e sunse ) and in e pola ed o emo e he nigh - ime gaps. Longe dis ances wi h missing o emo ed da a (be ween c uise legs) we e plo ed as whi e. Sa elli e da a.The sa elli e da a we used we e all annual a e ages o he yea 2010. Annual a e ages o PP o 2010 we e gene a ed by a e aging mon hly da a o PP downloaded om he Ocean P oduc i i y websi e (h p://www.science. o egons a e.edu/ocean.p oduc i i y/index.php)11. C uise segmen s we e hen gene a ed by combining all posi ion fixes wi hin a s a poin ±(8 4.6) km in no h, sou h, eas and wes di ec ions. The s a poin o he nex segmen was he fi s posi ion egis e ed ou side his box. Alignmen o in si u and sa elli e da a was done by selec ing he 64 (8 8, size pe bin B4.6 4.6 km) chlo ophyll-abins along he segmen s ha we e closes o he midpoin (median posi ion) wi hin a c uise segmen , wi h he added es ic ion ha no chlo ophyll-abin could be used wice. Values o hese 64 bins, co esponding o an a ea o B37 37 km we e a e aged. Fo he o he sa elli e-de i ed measu emen s, he maximum and minimum posi ions o he chlo ophyll-abins we e used as bounda ies o selec ion p io o calcula ion o he a e ages. Da a om he conduc i i y empe a u e and dep h (CTD) p obe we e aligned o c uise segmen s, and only CTD cas s wi hin a gi en c uise segmen we e used o a gi en segmen . A eas o biomass es ima ion.We used he PP–backsca e ela ion o es ima e he mesopelagic biomass om sa elli e-de i ed PP da a. We de e mined he bio- mass om he sum o he biomasses es ima ed om sa elli e-de i ed PP es ima es using only a eas whe e he bo om dep h was 41,000 m. We used he ETOPO1 da a se (h p://www.ngdc.noaa.go /mgg/global/ global.h ml) o es ima e he a ea. The ba hyme y da a se was ansla ed down o a 100a c g id, and o e e y cell in he PP da a se g id (Bsame spa ial esolu ion, no iden ical g ids), we assigned he dep h om he closes g id-poin in he ba hyme y da a se . P ima y p oduc ion g id-poin s/cells wi h bo om dep hs shallowe han 1,000 m we e hen excluded om ou biomass es ima ion, as we e a eas no h and sou h o 40 deg ees no h and sou h. This esul ed in an a ea o 222.3 million km2. Tempe a u e da a.The empe a u e da a o es ima e he MFP/biomass a io (P/B) we e ob ained om he Malaspina Expedi ion CTD p ofiles. P ofiles inside he PP boxes, o hose closes o he boxes we e used. As mo e da a-poin s we e a ailable o day ime, we used he empe a u e a he day ime weigh ed mean dep h (WMD) o he acous ic da a. In a eas whe e day- and nigh - ime acous ic WMDs we e a ailable, he a e age di e ence be ween he day ime WMD empe a u e and he a e age empe a u e be ween day- and nigh - ime WMD was o0.1 °C (see Supplemen a y Table 3). Modelling.The model ECOTROPH18 was un using he plugin inco po a ed o Ecopa h wi h Ecosim45, (EwE, www.Ecopa h.o g) and he gene ic model Ocean Ecos as a basis. We used he a e age PP o he oceanic a ea deepe han 1,000 m be ween 40°N and 40°S (344 mg C m 2d1, h p://www.science.o egons a e.edu/ ocean.p oduc i i y/index.php), conside ing a flux om PP o he ophic web o 70, 80 and 90% ( e . 20), ans e e ficiencies be ween ophic le els anging om 0.05 o 0.2 ( e . 21) and a empe a u e o 9 °C (see abo e). ECOTROPH es ima es biomass and p oduc ion pe ophic le el (TL) in s eps o 0.1 and we conside ed mesopelagic fishes as he main communi y be ween TL 3 and 3.5 in he open sea and biomass es ima es a e p o ided o ha ange. Mesopelagic fishes gene ally eed on zooplank on o ganisms such as copepods and euphausiids (www.fishbase.o g). To de e mine he TL o mesopelagic fishes we used fishbase (www.fishbase.o g) wi h sea ches o ‘lan e nfish’ and ‘b is lemou hs’ (myc ophids and Cyclo hone sp.). This yielded 96 and 15 TL alues o lan e nfish and b is elmou hs espec i ely. Fo lan e nfish he TL anged om 3 o 4.6, wi h an a e age o 3.2, mode o 3.1 and median o 3.2. Fo b is elmou hs he TL anged om 3 o 3.6, wi h an a e age o 3.3, mode o 3.5 and median o 3.3. Biomass ans o ma ions.Fo compa ison, fishes p oduc ion and PP we e ans o med om ca bon in o we weigh using a ac o o 10:1 ( e . 21). S a is ics.A ea sca e ing coe ficien (s A ) was significan ly co ela ed o PP. Howe e , s aigh eg ession be ween he wo ac o s (s A ¼2374 ln (PP) þ11624) was limi ed by he he e oscedas ici y o he da a (Supplemen a y Fig. 1A). In p inciple he e oscedas ici y does no a ec o dina y leas squa es eg ession coe ficien es ima es, bu can bias he significance es ima es. Loga i hmic ans- o ma ion o he da a (ln (s A )¼1.52 ln (PP)–1.36) elimina es he e oscedas ici y (Supplemen a y Fig. 1B) bu s ill p esen s spa ial au oco ela ion (Supplemen a y Fig. 2), ha can inc ease ype I e o s12. GWR is a me hod o ob ain eg ession pa ame e s o each o he poin s in a spa ial g id using he su ounding poin s weigh ed by dis ance. In spa ial analysis GWR o e s he ad an age o ex ac ing addi ional in o ma ion a each loca ion, as well as usually no being a ec ed by spa ial au oco ela ion13. He e we ca ied a GWR eg ession on ln- ans o med da a using a bi-squa e spa ial weigh ing unc ion and a bandwid h o 4,130 uni s. The eg ession was pe o med using he Spa ial Analysis in Mac oecology so wa e (SAM),46. Supplemen a y Table 1 p esen s he compa ison o he h ee eg essions in e ms o pa ame e s es ima es, 2and Akaike coe ficien . Supplemen a y Fig. 3 p esen s he spa ial a ia ions o he significance o he GWR slopes. The slopes we e gene ally highly significan excep in he Eas e n Pacific, nea he Humbold cu en , whe e he laye s below 100 m showed se e e hypoxia (Supplemen a y Fig. 4). This sugges s ha he s A –PP ela ion is a ec ed in ha specific a ea ei he because se e e hypoxic condi ions influence he niche space, o because he e is a s ong ex e nal inpu o o ganic ma e and local PP does no eflec he ood condi ions. Rega dless, he ype o eg ession used o es ima e s A om sa elli e PP has a limi ed e ec on he o e all biomass es ima e (see Table 1). The K- and c-e ec s on ision-based eeding habi a s.Bo h he beam a enua ion coe ficien (c,m 1) and he downwelling i adiance a enua ion coe ficien (K,m 1) a e a ec ed by wa e cla i y. We compa ed he quan i y c1K2, which is p opo ional o a heo e ical sea ch olume, o clea oceanic and less clea coas al wa e (Supplemen a y Table 2). The quan i y c1K2has Table 4 | Fish p oduc ion and espi a ion es ima es. PP, MFP and espi a ion es ima es along he Malaspina ansec . P ima y p oduc ion (g C m2y1) Fish p oduc ion median db/ weigh (g C m 2y1) Fish p oduc ion 75% db/ weigh (g C m2y1) Respi a ion median db/ weigh (g C m2y1) Respi a ion 75% db/ weigh (g C m2y1) A e age (s d) 116 (50) 2.7 (2) 1.6 (1.2) 12.2 (9) 7.0 (5.2) Max PP 318.0 8.7 5.0 39.0 23.0 Min PP 52.3 0.6 0.4 2.8 1.6 NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 ARTICLE NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions 7 &2014 Macmillan Publishe s Limi ed. All igh s ese ed. uni m3and combines he e ec s ha wa e cla i y has on he sho ( , he sigh ing dis ance) and he long (H, he e ical ex ension o he habi a ) ange sea ch abili y o a e ically mig a ing isual p eda o . The sigh ing dis ance is30: ¼k=c;ð1Þ whe e k¼ln(C 0 /C min ), C 0 is he inhe en con as o he p ey, and C min is he minimum con as ha he p eda o can de ec 30. The ision-based p ey de ec ion a e (p) o a c uising p eda o ends o be p opo ional o 2( e . 47), which in combina ion wi h Equa ion (1) gi es: p/c2:ð2Þ A fish ha mig a es e ically has a la ge po en ial eeding habi a han a fish ha emains a he same dep h. A g ea e a ea o mig a ion also p o ides a la ge po en ial eeding habi a 48. Wi hin he ange allowed by oxygen le els32, o a fish wi h a ce ain ligh p e e ence (isolume), he mig a ion dis ance (H, m) scales wi h Kacco ding o e s 33,47: H/K1:ð3Þ The p oduc o he igh -hand side o Equa ions (1) and (2), c2K1(m3), combines he op ical cha ac e is ics o he sho - and he long- ange sea ch abili y o a e ically mig a ing p eda o . We app oxima ed his quan i y, which is p opo ional o a heo e ical sea ch olume, o he a e age wa e cla i y o he s a ions o he c uise and o hypo he ical coas al wa e s wi h K- alues o 0.10 and 0.15, which a e in he lowe ange o hose epo ed in Table 6.2 in he s udy by Ki k and Ligh 49. Acco ding o Supplemen a y Table 3, he es ima e o his sea ch olume is abou one o de o magni ude highe o he a e age Malaspina s a ion han o he hypo he ical coas al wa e s. This sugges s ha , in o de o he ision-based p ey de ec ion a es o be equal, he p ey concen a ion mus be 10 imes highe a he coas al loca ion, assuming all o he ac o s o be equal. This sugges s ha he high cla i y o oceanic wa e enables e ficien isual eeding a low p ey concen a ions. Unce ain y analysis.T ans o ming acous ic measu es in o biomass in ol es unce ain y, dependen on he dis ibu ion o fishes sizes ela i e o acous ic wa e leng hs, on co ec ly asc ibing he acous ic backsca e o fishes in sca e ing laye s composed o di e en axonomic g oups as well as on he use o app op ia e con e sion ac o s om backsca e o biomass (TS alues). In o de o assess he bias in oduced h ough ou non-s anda d pos - p ocessing me hods, we compa ed ou esul s wi h esul s om s anda d acous ic pos -p ocessing in selec ed low-noise sec ions. S anda d pos -p ocessing he e e e s o au oma ed emo al o noise spikes and backg ound noise as implemen ed in he so wa e LSSS, in addi ion o manual sc u iny o da a and emo al o pe iods whe e hese au oma ed fil e s did no elimina e all noise, o emo ed subs an ial po ions o he da a. Resul s a e p esen ed in Supplemen a y Figs 5 and 6 as a ac ion o he di e ence be ween s anda d esul s and ou esul s (only 200–800 m da a, s anda d s A –fil e ed s A /fil e ed s A ). F ac ions we e a e aged ei he pe dep h channel (Supplemen a y Fig. 6) o pe ime bin (Supplemen a y Fig. 7). Time-a e aged esul s (Supplemen a y Fig. 7) sugges ha he e is a linea ela ionship be ween ou es ima es and s anda d es ima es, wi h ou es ima es B30–40% lowe han s anda d es ima es. The e ical bias p ofile (Supplemen a y Fig. 6) shows ha he e is a e ical influence on he bias, bu ha he bias appea s almos cons an a 30–40% in he 400–800 me e dep h ange (sugges ing he magni ude o unde es ima ion o his dep h ange, encompassing mos o he mesopelagic backsca e ; c . Supplemen a y Fig. 6). The bias is highe a dep hs shallowe han 400 m; i d ops apidly a dep hs lowe han 800 m, sugges ing ha he e is no bias a B900 m and a nega i e bias deepe han his ( ha is, ou es ima es a e highe han s anda d es ima es a hese dep hs). The axonomic composi ion o he o ganisms esponsible o he mesopelagic backsca e along he pa h o his ci cumna iga ion oyage is no known. Mo eo e , no all o he backsca e o igina es om fish. Ea ly s udies o he deep sca e ing laye concluded ha la ge c us aceans and in pa icula euphausiids we e significan in he deep sca e ing laye 50, bu la e s udies showed ha a he low equencies used in hese ea ly s udies, mesopelagic fish we e he mos significan sca e e s, al hough wi h a possible con ibu ion om gas-bea ing siphonopho es51. La ge c us aceans a e ela i ely weak sca e e s compa ed wi h o ganisms wi h ai -inclusions a he equency used in his s udy52, and he e o e hey p obably made up a negligible p opo ion o he o al backsca e a ou equency. Acco dingly, ecen s udies om oceans a ound he wo ld conclude ha mesopelagic fish make up he majo i y o he backsca e 9,14 ye p ope ‘g ound u hing’ is no possible due o he highly a ying ca ch e ficiency o sampling gea 7, which ende s es ima es de i ed om ne s un eliable. Size o he fish is ano he po en ial issue. A 38 kHz he acous ic wa e leng h is B3.9 cm. Indi iduals much smalle han he wa eleng h can be de ec ed, pa icula ly when occu ing in high concen a ions (al hough mesopelagic fish do no school). The e is, howe e , an exponen ial dec ease in acous ic backsca e wi h dec easing size in his so- called Rayleigh sca e ing egion53. I is likely ha a la ge pa o he mesopelagic fish communi y is smalle han 3.9 cm. This would lead o unde es ima ing mesopelagic fish biomass, as fish in he Rayleigh domain would esul in small db/weigh a ios, excep o indi iduals wi h a esonan swim bladde . Resonance has he po en ial o esul in a la ge bias o he es ima e by di ec ly a ec ing he acous ic esul s by up o 25 imes54. Swimbladde esonance may inc ease acous ic backsca e om small mesopelagic fish and o ganisms wi h ai bubbles55. Howe e , al hough a ce ain le el o esonance canno be excluded, he da a sugges ha is no a majo sou ce o bias because (1) he a io o pai ed day and nigh backsca e alues is gene ally o1, (2) i esonance was high and widesp ead he geog aphically weigh ed ela ion be ween s A and PP should disappea and (3) he eco oph model esul s ag ee well wi h he acous ic es ima e. Modelling and field s udies sugges ha he e ec s o esonance a 38 kHz inc ease wi h dep h9,56. The e o e, i he esonance e ec was a majo bias in ou da a se , we would expec he o al nigh - ime backsca e o be much lowe han co esponding day ime alues, because, on a e age, animals a e dis ibu ed conside ably close o he su ace a nigh , and hus p oducing lowe esonance. We would also expec o obse e high a iabili y in he day/nigh a io along he ansec when going o e di e en communi ies (fish species and sizes) wi h di e en esonance le els. We he e o e compa ed day and nigh column (10–1,000 m) o al backsca e alues o check o he po en ial influence o esonance in ou da a. The a io o pai ed day and nigh backsca e alues (Supplemen a y Fig. 8 day s A /nigh s A ) shows ha mos o hese alues a e o1 wi h low a iabili y (a e age day/nigh a io alue 0.96, s.d. 0.53). The e ical bias would end o d i e his a io in he same di ec ion as esonance, while mig a ions om deepe in he wa e column would oppose he end. Ou low a ios a e inconsis en wi h esonance being a majo ac o . Howe e , in one sec ion along he c uise- ack (Eas Pacific), he a ios appea o be highe (2–3), consis en wi h esonance playing a ole. This is he same zone, wi h hypoxic wa e s a 100 m (see Supplemen a y Fig. 4), whe e he GWR is no significan . Howe e , he biomass es ima es in ha a ea a e no ou lie s in he gene al eg ession (Supplemen a y Fig. 9) o highe han ha p edic ed by Eco oph o he zone (Supplemen a y Fig. 10), which sugges s ha he day/nigh a ios 41 could be explained by ac o s o he han an o e es ima ion o he biomass. In his egion, he nigh - ime e ical p ofiles we e pa icula ly close o he su ace, sugges ing ha a la ge p opo ion o he biomass ac ually mig a es o he nea -su ace dead-zone o he echosounde , leading o a lowe o al nigh - ime backsca e 57. In any case he da a om ha limi ed a ea do no influence he es ima ions o global backsca e using di e en me hods ( eg essions o a e age). The spa ial cohe ence o he GWR (GWR pa ame e s and significance) is also a s ong indica ion ha esonance does no play a majo ole. Unless esonance was he same all along he ansec (same communi y composi ion, sizes, dep h dis ibu ion and mig a ion pa e ns all a ound he wo ld), esonance should eclipse he local ela ion be ween s A and PP in a eas whe e esonance was ele an . The GWR does no show such a iabili y and he ela ion emains significan all along he ansec , excep o he No h Eas Pacific a ea wi h hypoxic deep laye s ha also show highe day/nigh a ios (Supplemen a y Fig. 4). Finally, he gene al es ima e and he es ima ions along he ansec ag ee wi h he Eco oph model esul s using a e age pa ame e s. The model is comple ely independen om acous ics es ima es, based on PP and ans e e ficiency. The ag eemen be ween he wo independen app oaches sugges s ha esonance is no a majo sou ce o bias in he acous ic da a. We use a ange o li e a u e db/weigh a io alues o es ima e biomass. This is a simplifica ion ha p ecludes exac es ima ion a each single poin , bu is la gely o gene a e eliable a e age es ima es, as he inaccu acies go bo h ways. A la ge po ion o he backsca e om an indi idual fish no mally o igina es om i s gas- filled swimbladde 52, bu in mesopelagic fish educed swimbladde s o a -filled swimbladde s a e common58 and ha e a s ong e ec on he TS o he fish. Fo ins ance, adul s o some species o he genus Cyclo hone may ha e gas-filled swimbladde s, whe eas only ju eniles in o he Cyclo hone species ha e gas-filled swimbladde s. The e a e o he species o he same genus ha ne e ha e gas-filled swimbladde s58. G ound u hing in each a ea is no possible, bu he 25–75% qua ile ange used in he es ima es p o ided he e should encompass he a e age TS alue o he oceanic mesopelagic fish popula ions. As ou ocus was mesopelagic fish, we did no include he backsca e om he uppe 200 m laye (apa om he es on esonance). The addi ional fish biomass in he uppe 200 m would con ibu e o a highe biomass es ima e o o al fish. The in eg a ed s A o he uppe laye is on a e age 27% o he in eg a ed alue om 200–1,000 m. Howe e , i fi e a eas wi h excep ionally high alues we e excluded, he a e age alue would d op o 7% (Supplemen a y Fig. 11). Re e ences 1. Ma shall, N. B. Ba hypelagic fishes as sound sca e e s in he ocean. J. Ma . Res. 10, 1–17 (1951). 2. Nelson, J. S. Fishes o he Wo ld (Wiley, 2006). 3. Gjøsae e , J. & Kawaguchi, K. A e iew o he wo ld esou ces o mesopelagic fish Vol. 193 (Be nan P ess, 1980). 4. Lam, V. & Pauly, D. Mapping he global biomass o mesopelagic fishes. Sea A ound Us P ojec Newsle e 30, 4 (2005). 5. T e ´gue , P., Legend e, L., Ri kin, R. T., Ragueneau, O. & N, D. Ocean Biogeochemis y: The Role o Ocean Ca bon Cycle in Global Change 145–156 (Sp inge , 2003). 6. Ch is ensen, V. e al. Da abase-d i en models o he wo ld’s La ge Ma ine Ecosys ems. Ecol. Modell. 220, 1984–1996 (2009). ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 8NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions &2014 Macmillan Publishe s Limi ed. All igh s ese ed. 7. Pakhomo , E. & Yamamu a, O. Repo o he Ad iso y Panel on Mic onek on Sampling In e -calib a ion Expe imen (No h Pacific Ma ine Science O ganiza ion (PICES), 2010). 8. La a-Lopez, A. L., Da ison, P. & Koslow, J. A. Abundance and communi y composi ion o mic onek on ac oss a on o Sou he n Cali o nia. J. Plank. Res. 34, 828–848 (2012). 9. Klose , R. J., Ryan, T. E., Young, J. W. & Lewis, M. E. Acous ic obse a ions o mic onek on fish on he scale o an ocean basin: po en ial and challenges. Ices J. Ma . Sci. 66, 998–1006 (2009). 10. Kaa ed , S., S aby, A. & Aksnes, D. L. E ficien awl a oidance by mesopelagic fishes causes la ge unde es ima ion o hei biomass. Ma . Ecol. P og. Se . 456, 1–6 (2012). 11. Beh en eld, M. J. & Falkowski, P. G. Pho osyn he ic a es de i ed om sa elli e-based chlo ophyll concen a ion. Limnol. Oceanog . 42, 1–20 (1997). 12. Do mann, C. F. e al. Me hods o accoun o spa ial au oco ela ion in he analysis o species dis ibu ional da a: a e iew. Ecog aphy 30, 609–628 (2007). 13. Fo he ingham, A. S., B unsdon, C. & Cha l on, M. Geog aphically weigh ed eg ession (Wiley, 2002). 14. Godo, O. R., Pa el, R. & Pede sen, G. Diel mig a ion and swimbladde esonance o small fish: some implica ions o analyses o mul i equency echo da a. Ices J. Ma . Sci. 66, 1143–1148 (2009). 15. Wilson, R. W. e al. Con ibu ion o Fish o he Ma ine Ino ganic Ca bon Cycle. Science 323, 359–362 (2009). 16. Jennings, S. e al. Global-scale p edic ions o communi y and ecosys em p ope ies om simple ecological heo y. P oc. R. Soc. B Biol. Sci. 275, 1375–1383 (2008). 17. T emblay-Boye , L., Gascuel, D., Wa son, R., Ch is ensen, V. & Pauly, D. Modelling he e ec s o fishing on he biomass o he wo ld’s oceans om 1950–2006. Ma . Ecol. P og. Se . 442, 169–185 (2011). 18. Gascuel, D. & Pauly, D. EcoT oph: modelling ma ine ecosys em unc ioning and impac o fishing. Ecol. Modell. 220, 2885–2898 (2009). 19. Calbe , A. Mesozooplank on g azing e ec on p ima y p oduc ion: a global compa a i e analysis in ma ine ecosys ems. Limnol. Oceanog . 46, 1824–1830 (2001). 20. Calbe , A. & Land y, M. R. Phy oplank on g ow h, mic ozooplank on g azing, and ca bon cycling in ma ine sys ems. Limnol. Oceanog . 49, 51–57 (2004). 21. Pauly, D. & Ch is ensen, V. P ima y p oduc ion equi ed o sus ain global fishe ies. Na u e 374, 255–257 (1995). 22. Gascuel, D., Mo isse e, L., Paloma es, M. L. D. & Ch is ensen, V. T ophic flow kine ics in ma ine ecosys ems: owa d a heo e ical app oach o ecosys em unc ioning. Ecol. Modell. 217, 33–47 (2008). 23. Da ison, P. C., Checkley, D. M., Kolslow, J. A. & Ba low, J. Ca bon expo media ed by mesopelagic fishes in he no heas Pacific Ocean. P og. Oceanog . 116, 14–30 (2013). 24. Agawin, N. S. R., Dua e, C. M. & Agus i, S. Nu ien and empe a u e con ol o he con ibu ion o picoplank on o phy oplank on biomass and p oduc ion. Limnol. Oceanog . 45, 591–600 (2000). 25. Ry he , J. H. Pho osyn hesis and fish p oduc ion in Sea. Science 166, 72–76 (1969). 26. San Ma in, E. e al. Va ia ion in he ans e o ene gy in ma ine plank on along a p oduc i i y g adien in he A lan ic Ocean. Limnol. Oceanog . 51, 2084–2091 (2006). 27. Bonsall, M. B. & Hassell, M. P. In: Theo e ical Ecology P inciples and Applica ions (eds May, R. M. & McLean, A. R.) (Ox o d Uni e si y P ess, 2007). 28. Jennings, S., Wa , K. J. & Mackinson, S. Use o size-based p oduc ion and s able iso ope analyses o p edic ophic ans e e ficiencies and p eda o -p ey body mass a ios in ood webs. Ma . Ecol. P og. Se . 240, 11–20 (2002). 29. Rose, J. M. & Ca on, D. A. Does low empe a u e cons ain he g ow h a es o he e o ophic p o is s? E idence and implica ions o algal blooms in cold wa e s. Limnol. Oceanog . 52, 886–895 (2007). 30. Johnson, S. The Op ics o Li e (P ince on Uni e si y P ess, 2012). 31. Kio boe, T. How zooplank on eed: mechanisms, ai s and ade-o s. Biol. Re . 86, 311–339 (2011). 32. Bianchi, D., Galb ai h, E. D., Ca ozza, D. A., Mislan, K. & S ock, C. A. In ensifica ion o open-ocean oxygen deple ion by e ically mig a ing animals. Na . Geosci. 6, 545–548 (2013). 33. Aksnes, D. L. E idence o isual cons ain s in la ge ma ine fish s ocks. Limnol. Oceanog . 52, 198–203 (2007). 34. Ha aldsson, M., To ¨nnesson, K., Tiselius, P., Things ad, T. F. & Aksnes, D. L. Rela ionship be ween fish and jellyfish as a unc ion o eu ophica ion and wa e cla i y. Ma Ecol. P og. Se . 471, 73–85 (2012). 35. Buessele , K. O. Do uppe -ocean sedimen aps p o ide an accu a e eco d o pa icle-flux? Na u e 353, 420–423 (1991). 36. Buessele , K. O. e al. An assessmen o he use o sedimen aps o es ima ing uppe ocean pa icle fluxes. J. Ma . Res. 65, 345–416 (2007). 37. Ka l, D. M. e al. Building he long- e m pic u e. Oceanog aphy 14, 6–17 (2001). 38. Usbeck, R., Schli ze , R., Fische , G. & We e , G. Pa icle fluxes in he ocean: compa ison o sedimen ap da a wi h esul s om in e se modeling. J. Ma . Sys . 39, 167–183 (2003). 39. del Gio gio, P. A. & Dua e, C. M. Respi a ion in he open ocean. Na u e 420, 379–384 (2002). 40. A is egui, J., Dua e, C. M., Gasol, J. M. & Alonso-Saez, L. Ac i e mesopelagic p oka yo es suppo high espi a ion in he sub opical no heas A lan ic Ocean. Geophys. Res. Le . 32, L03608 (2005). 41. Smi h, A. D. e al. Impac s o fishing low- ophic le el species on ma ine ecosys ems. Science 333, 1147–1150 (2011). 42. Foo e, K., Knudsen, H., Ves nes, G., MacLennan, D. & Simmonds, E. Calib a ion o Acous ic Ins umen s o Fish Densi y Es ima ion: a P ac ical Guide (In e na ional Council o he Explo a ion o he Sea, 1987). 43. Ko neliussen, R. J., Heggelund, Y., Eliassen, I. K. & Johansen, G. O. Acous ic species iden ifica ion o schooling fish. Ices J. Ma . Sci. 66, 1111–1118 (2009). 44. Ihaka, R. & Gen leman, R. R: a language o da a analysis and g aphics. J. Compu . G aph. S a . 5, 299–314 (1996). 45. Ch is ensen, V. & Pauly, D. Ecopa h-II—a so wa e o balancing s eady-s a e ecosys em models and calcula ing ne wo k cha ac e is ics. Ecol. Modell. 61, 169–185 (1992). 46. Rangel, T. F., Diniz, J. A. F. & Bini, L. M. SAM: a comp ehensi e applica ion o spa ial analysis in mac oecology. Ecog aphy 33, 46–50 (2010). 47. Aksnes, D. L., Nejs gaa d, J., Soedbe g, E. & So nes, T. Op ical con ol o fish and zooplank on popula ions. Limnol. Oceanog . 49, 233–238 (2004). 48. Cla k, C. W. & Le y, D. A. Diel Ve ical mig a ions by ju enile sockeye salmon and he an ip eda ion window. Am. Na . 131, 271–290 (1988). 49. Ki k, J. T. O. Ligh and Pho osyn hesis in Aqua ic Ecosys ems (Camb idge Uni e si y P ess, 1994). 50. Moo e, H. B. The ela ion be ween he sca e ing laye and he Euphausiacea. Biol. Bull. 99, 181–212 (1950). 51. Ba ham, E. G. Deep sca e ing laye mig a ion and composi ion—obse a ions om a di ing sauce . Science 151, 1399–1403 (1966). 52. MacLennan, D. N. & Simmonds, E. J. Fishe ies Acous ics (Chapman & Hall, 1992). 53. Ho ne, J. K. & Jech, J. M. in Sounds in he Sea: F om Ocean Acous ics o Acous ical Oceanog aphy (ed Medwin, H.) 374–397 (Camb idge Uni e si y P ess, 2005). 54. Clay, C. S. & Medwin, H. Acous ical Oceanog aphy: P inciples and Applica ions Vol. 4 (Wiley, 1977). 55. Benfield, M. C. e al. Dis ibu ions o physonec siphonulae in he Gul o Maine and hei po en ial as impo an sou ces o acous ic sca e ing. Can. J. Fish. Aqua . Sci. 60, 759–772 (2003). 56. Yasuma, H., Sawada, K., Takao, Y., Miyashi a, K. & Aoki, I. Swimbladde condi ion and a ge s eng h o myc ophid fish in he empe a e zone o he No hwes Pacific. Ices J. Ma . Sci. 67, 135–144 (2010). 57. O’D iscoll, R. L., Gau hie , S. & De ine, J. A. Acous ic es ima es o mesopelagic fish: as clea as day and nigh ? Ices J. Ma . Sci. 66, 1310–1317 (2009). 58. Da ison, P. The specific g a i y o mesopelagic fish om he no heas e n Pacific Ocean and i s implica ions o acous ic backsca e . Ices J. Ma . Sci. 68, 2064–2074 (2011). 59. Fock, H. & Eh ich, S. Deep-sea pelagic nek on biomass es ima es in he No h A lan ic: ho izon al and e ical esolu ion o e ised da a om 1982 and 1983. J. Appl. Ich hyol. 26, 85–101 (2010). 60. Be na des, R. & Rossi-Wong schowski, C. Leng h-weigh ela ionship o small pelagic fish species o he sou heas and Sou h B azilian Exclusi e Economic Zone. Naga, he ICLARM Qua e ly 23, 30–32 (2000). 61. Yasuma, H., Sawada, K., Olishima, T., Miyashi a, K. & Aoki, I. Ta ge s eng h o mesopelagic lan e nfishes ( amily Myc ophidae) based on swimbladde mo phology. Ices J. Ma . Sci. 60, 584–591 (2003). 62. Yasuma, H., Takao, Y., Sawada, K., Miyashi a, K. & Aoki, I. Ta ge s eng h o he lan e nfish, S enob achius leucopsa us ( amily Myc ophidae), a fish wi hou an ai bladde , measu ed in he Be ing Sea. Ices J. Ma . Sci. 63, 683–692 (2006). 63. Smoke , W. & Pea cy, W. G. G ow h and ep oduc ion o he lan e nfish S enob achius leucopsa us. J. Fish. Boa d Can. 27, 1265–1275 (1970). 64. To ge sen, T. & Kaa ed , S. In si u swimming beha iou o indi idual mesopelagic fish s udied by spli -beam echo a ge acking. Ices J. Ma . Sci. 58, 346–354 (2001). 65. Sawada, K. e al. In si u and ex si u a ge s eng h measu emen o mesopelagic lan e nfish, Diaphus The a (Family Myc ophidae). J. Ma . Sci. Tech. Taiw. 19, 302–311 (2011). 66. Acuna, E. Biology o he myc ophid fish, Diaphus he a Eigenmann and Eigenmann 1890, o he O egon coas (O egon S a e Uni e si y, 1983). 67. Koslow, J. A., Klose , R. J. & Williams, A. Pelagic biomass and communi y s uc u e o e he mid-con inen al slope o sou heas e n Aus alia based upon acous ic and midwa e awl sampling. Ma . Ecol. P og. Se . 146, 21–35 (1997). 68. Klose , R. J., Williams, A. & Koslow, J. A. P oblems wi h acous ic a ge s eng h measu emen s o a deepwa e fish, o ange oughy (Hoplos e hus a lan icus, Colle ). Ices J. Ma . Sci. 54, 60–71 (1997). NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4271 ARTICLE NATURE COMMUNICATIONS | 5:3271 | DOI: 10.1038/ncomms4271 | www.na u e.com/na u ecommunica ions 9 &2014 Macmillan Publishe s Limi ed. All igh s ese ed.