Performance of Coffea arabica F1 hybrids in agroforestry and full-sun cropping systems in comparison with American pure line cultivars

Transcription

1 Euphytic (2011) 181: DOI /s Performnce of Coffe ric F1 hyrids in groforestry nd full-sun cropping systems in comprison with Americn pure line cultivrs B. Bertrnd E. Alpizr L. Lr R. SntCreo M. Hidlgo J. M. Quijno C. Montgnon F. Georget H. Etienne Received: 22 Octoer 2010 / Accepted: 25 Jnury 2011 / Pulished online: 16 Ferury 2011 Ó The Author(s) This rticle is pulished with open ccess t Springerlink.com Astrct Coffe ric F1 hyrids derived from crosses etween wild Sudn-Ethiopin nd Americn cultivrs nd propgted y somtic emryogenesis hve een otined in Centrl Americ. These new hyrids considerly enhnced the genetic diversity of coffee in the region. We conducted 15 trils to ssess whether using hyrids represents sustntil genetic progress in terms of productivity in groforestry nd full-sun cropping systems. The new germplsm ws grown in the sme conditions s the est Americn cultivr (homozygous pure lines). The results showed tht yields of hyrids were erlier nd superior to those of Americn cultivrs. The hyrids were lso more stle thn the Americn cultivrs in ll environments. In the groforestry system, the men yield of hyrids ws 58% higher thn tht of the Americn cultivrs, while the men yield of hyrids in the full-sun system ws 34% higher. Coffee-sed groforestry systems (AS) re considered effective in protecting the environment in the volcnic cordillers of Centrl Americ. We found tht introducing hyrids in coffee-sed AS cn considerly increse productivity. This finding could e convincing rgument to encourge coffee growers who hve dopted the full-sun cropping system to return to groforestry cropping systems. Finlly, the conditions for lrge-scle dissemintion of those new hyrids which represent mjor innovtion for C. ric cropping ws nlysed. B. Bertrnd C. Montgnon F. Georget H. Etienne Déprtement des Systèmes Biologiques, UMR-RPB, CIRAD-IRD-UM2, Centre de Coopértion Interntionle en Recherche Agronomique pour le Développement (CIRAD), 911 Avenue Agropolis, BP 64501, Montpellier, Frnce E. Alpizr Cfé Totl, AP Aljuel, Cost Ric L. Lr ECOM Coffee Group Nicrgu. Rotond Jen Pul Geni, 300 m jo. Edificio Sn Mrino, Mngu, Nicrgu M. Hidlgo Instituto del Cfé de Cost Ric (ICAFE), AP Sn José, Cost Ric J. M. Quijno Fundción Slvdoreñ pr l investigción del Cfé (PROCAFE), AP 23 Snt Tecl, El Slvdor B. Bertrnd (&) CIRAD, UMR RPB, Montpellier, Frnce e-mil: enoit.ertrnd@cird.fr R. SntCreo Instituto Hondureño del Cfé (IHCAFE), AP 3174 Teguciglp, Hondurs

2 148 Euphytic (2011) 181: Keywords Coffee Hyrids Productivity Ecologicl innovtion Agroforestry system Introduction Aric coffee is the min source of export ernings for numerous countries. Ltin Americ is the min re of production of this commodity, which ccounts for 80% of the world export volume. In Centrl Americ, coffee ws trditionlly grown in groforestry systems (AS) under forest or plnted trees. The cropping prctices evolved through series technicl dvnces tht were implemented in the 1950s nd 1960s, thus creting idel conditions for veritle Green Revolution. A switch ws mde from n extensive AS to n unshded intensive system with undnt fertilizer (notly units of N/h) nd pesticide inputs. This oosted yields from 150 to 1,000 kg/h of green coffee to more thn 1,000 1,500 kg, notly in Cost Ric nd Colomi. However, in mny countries (Anden Americ, Hondurs, El Slvdor, Nicrgu, Guteml nd the Chips region of Mexico), most coffee growers hve not ndoned the AS (Fig. 1). Coffe ric is n utogmous species. Growers in Ltin Americ currently hve the choice etween two types of homozygous lines propgted y seed: Americn trditionl cultivrs (Bouron, Typic, Cturr, Ctui) nd Ctimor cultivrs. These recent ltter cultivrs re derived from cv. Hirido de Timor, which is nturl cross etween C. ric (4n = 2x = 44) nd C. cnephor (2n = 2x = 22). C. ric (4n = 2x = 44) nd C. cnephor (2n = 2x = 22). The mjority of cultivrs Ctimors hve shown complete resistnce to ll physiologicl rces of the coffee lef rust (Hemilei vsttrix). He we refer to this seed-propgted mteril s Americn cultivrs or lines. Since 1997, coffee growers throughout Centrl Americ hve hd ccess to some F1 hyrid clones micropropgted y somtic emryogenesis (Etienne- Brry et l. 1999; Etienne 2005). These new hyrids, derived from crosses etween wild Sudn-Ethiopin nd Americn trditionl cultivrs or Ctimor cultivrs, hve considerly incresed the nrrow genetic se of coffee trees cultivted in Ltin Americ. We refer to this mteril s hyrids, while keeping their inter-origin genetic ckground in mind. In controlled trils conducted in three environments in Cost Ric nd Nicrgu, the heterosis found in hyrid popultions reltive to the prentl mteril ws 20 50% (Bertrnd et l. 2005). In sensory evlutions in which the hyrids were compred with trditionl lines under vrious soil-climte conditions, the hyrids performed s well s or etter thn the trditionl lines (Bertrnd et l. 2006). As coffee rust resistnce genes re dominnt, when hyrids re derived from cross etween rust-resistnt Ctimor prent nd n Ethiopin ccession, they will hve the sme level of resistnce s the Ctimor prent. Coffee growers who hve dopted Green Revolution principles re wre of the need to crop their coffee in AS s they did in the pst, ut this cn result in n estimted 20 40% drop in productivity (Vst et l. 2006). Otherwise, growers who hve not ndoned AS, strive to increse the profitility of AS y generting income for ecologicl service provision, dopting commercil strtegies sed on etter everge qulity, etc. (Vst nd Hrmnd 2002). Strngely enough, reeding hs een overlooked in this wve of innovtions for improving AS, lthough the Green Fig. 1 Illustrtion of the C. ric network trils. Full-sun system nd groforestry system

3 Euphytic (2011) 181: Revolution hs een lrgely sed on reeding. The present study ws crried out to ssess whether introducing new hyrid coffee cultivrs in coffeesed AS in Centrl Americ would increse cropping system productivity nd therefore enhnce the profitility of AS. We nlyzed the yield performnce of these new hyrids in groforestry (shde) nd full-sun (unshded) cropping systems in comprison with the highest yielding Americn pure line cultivrs distriuted in Centrl Americ. Mterils nd methods Plnt mteril Two types of coffee cultivr were tested in the network: Thirteen lines (i.e. Americn cultivrs): nmely Cturr, Pcs, Ctui nd Bouron s trditionl cultivrs, nd CR95 (lso clled Lempir), Ctisic, Ipr59, T5296, T17931, T17933, T18121, T18138 nd T18141 s Ctimor cultivrs. Twenty-one C. ric F1 hyrids: crosses of Americn cultivrs with Ethiopin ccessions (ET6, ET15, ET25, E41, E416, E531, Anfilo nd Rume Sudn). The hyrids were micropropgted vi somtic emryogenesis (see Fig. 2), s previously descried (Etienne 2005). Field trils The tests were crried out from 2000 to 2006 in network of 15 trils (i.e. loctions) set up in 1999, 2000, nd 2001 in three Centrl Americn countries. A rndomized lock design ws used in ech tril. Inside ech lock, cultivrs were grown in unit plots of 10 trees. The min tril chrcteristics nd the genotype replictions in the 15 trils re summrized in Tles 1 nd 2, respectively. The trils were set up on frms of vrious sizes (5 150 h). The elevtion of the frms rnged from 750 to 1,580 m.s.l. Unlike mny multi-site tril networks, we considered tht comprisons of new germplsm with trditionl cultivrs should not involve chnge in growers cropping prctices. We thus decided to llow plot mngers to mke decisions ccording to their experience nd resources. In Cost Ric, the trils were mostly mnged in n unshded intensive system, prt from two lowltitude trils mnged with coffee trees growing under slight shde from Erythrin poeppigin t density of trees/h. In El Slvdor nd Hondurs, the trils were mnged in AS with shde Fig. 2 Illustrtion of the coffee somtic emryogenesis micropropgtion process under industril conditions (Nicrgu, CIRAD-ECOM project)., Multipliction of emryogenic suspensions in Erlenmeyer flsks; c mss regenertion in temporry immersion iorectors of directly cclimtizle pregerminted somtic emryos; d plnt hrdening in the nursery efore field trnsfer

4 150 Euphytic (2011) 181: Tle 1 Tril chrcteristics Elevtion (m.s.l.) Plnting yer Country Region T (trees/h) Numer of cultivrs tested per type Cropping system shding Soils R (mm yer -1 ) Tver ( C) Nprod Hondurs L Fé 4,000 TC = 1, F1 = 12, Ctimors = 1 AS, 40% Luvisols 3, Cost Ric Plmir 5,850 TC = 1, F1 = 10, Ctimors = 1 AS, 20% Inceptisol 4, El Slvdor Sn Antonio 4,000 TC = 1, F1 = 6, Ctimors = 1 AS, 40% Luvisols 1, El Slvdor Sn José 4,000 TC = 2, F1 = 8 AS, 30% Andisols 2, Cost Ric Coto Brus 5,000 TC = 1, F1 = 8 AS, 20% Andisols 3, El Slvdor Sn Jorge 4,000 TC = 1, F1 = 8 AS, 40% Nitisols 1, Hondurs Linderos 4,000 TC = 1, F1 = 7, Ctimors = 1 AS, 50% Luvisols 2, Cost Ric Brv de Heredi 5,000 TC = 2, F1 = 5, Ctimors = 7 FS, 0% Andisol 2, Cost Ric Brv de Heredi 5,000 TC = 2, F1 = 13 FS, 0% Andisol 2, El Slvdor Usulután, Los Pirineos 4,000 TC = 1, F1 = 6, Ctimors = 1 AS, 50% Nitisols 2, El Slvdor Ahuchpán, El Milenio 4,000 TC = 2, F1 = 4 AS, 30% Luvisols 1, Cost Ric Trrzu 5,000 TC = 2, F1 = 12 FS, 0% Ultisol 2, Hondurs Mrcl 4,000 TC = 1, F1 = 7, Ctimors = 1 AS, 30% Luvisols 1, Cost Ric Snill de Aljuel 5,000 TC = 2, F1 = 1, Ctimors = 7 FS, 0% Andisols 2, Cost Ric Snt Mrí de Dot 5,000 TC = 2, F1 = 2, Ctimors = 7 FS, 0% Ultisol 2, Elevtion (m.s.l.), plnting yer; country; region; T plnting density (numer of trees/h); numer of cultivrs tested per type, with TC trditionl cultivrs, Ctimors lines derived from Hirido de Timor, F1 clones of F1 hyrids; cropping system nd shding (%) with FS full-sun, AS groforestry system; type of soils (Bornemisz et l. 1999); Tver, men dily temperture ( C); R nnul rinfll (mm yer -1 ); Nprod Numer of production yers monitored

5 Euphytic (2011) 181: Tle 2 Distriution of cultivrs (hyrids nd lines) in the C. ric multisite tril Type of cultivr Cultivrs Tril numer nd elevtion in m.s.l (750) (800) (820) (880) (1,000) (1,060) (1,100) (1,180) (1,185) (1,260) (1,340) (1,400) (1,420) (1,425) (1,580) Hyrid Centroméric X X X X X X X X X Hyrid L02A11 X X X X X Hyrid L02A30 X X X X X X X X Hyrid L03A07 X X X X X Hyrid L03A15 X X X X X Hyrid L03A17 X X X X X X X Hyrid L04A05 X X X Hyrid L04A20 X X X X X X Hyrid L04A34 X X X X X Hyrid L04A42 X X X X X Hyrid L05A26 X X X X X X X X X X Hyrid L05A27 X X Hyrid L09A22 X X Hyrid L10A25 X X Hyrid L11A26 X X X X X X Hyrid L12A05 X Hyrid L13A12 X X X X Hyrid L13A22 X X X X X X Hyrid L14A08 X X Hyrid L22A08 X X X X X Line Ctisic X Line Ctui X X X X X X X X Line Cturr X X X X X X X X X X Line CR95 X X X X X X X Line IAPAR59 X X X Line T17931 X X X Line T17933 X X X Line T18121 X X X Line T18138 X X X Line T18141 X X X Line T5296 X Line Tekisic X X X X X Type of cultivr, cultivrs, tril numer (1 15) nd elevtion ( ) in m.s.l. X = presence in the tril. Rust-resistnt hyrids nd lines re indicted in old chrcter

6 152 Euphytic (2011) 181: provided y Ing edulis ( trees/h), sometimes with few tll ntive trees such s Cordi lliodor, Juglns steyermrkii nd Quercus costricensis. The shde levels vried etween trils nd within ech plot depending on the seson. They were estimted visully t etween 20 nd 50%. The coffee tree plnting density rnged from 4000 to 5,800 trees/ h -1. In Cost Ric, the plnts received [ ], [ ], [20 50] kg/h -1 /yer -1 of N, K 2 O nd P 2 O 5, respectively, long with 2 4 yerly pplictions of copper hydroxide or trizole to control lef diseses (coffee lef rust nd rown eye spot). In Hondurs nd El Slvdor, the plnts received [50 100], [50 100], [0 20] kg/h -1 /yer -1 of N, K 2 O nd P 2 O 5, respectively, long with zero or two yerly pplictions of copper hydroxide to control coffee lef rust. Trits oserved The trits monitored were verge production during the first production cycle efore pruning nd coppicing. The numer of hrvests during the first production cycle vried depending on the elevtion, degree of intensifiction, nd soil-climte conditions. Tle 1 shows the numer of production yers monitored per tril. These hrvests were mesured in grms of fresh erries nd then expressed in grms of green coffee per tree sed on the ssumption tht the weight of green coffee mounted to 20% of the fresh erry weight. To fcilitte comprisons etween trils, we clculted Y (yield), i.e. the rtio of the sum of hrvests (CUM) to the numer of hrvests. Y is expressed in grms of green coffee per tree nd per yer. Two other vriles were clculted: erliness (E), which represented the percentge of the first hrvest (Y1) over CUM, nd the coefficient of vrition for the hrvests (cvy%), i.e. the rtio etween the stndrd devition of CUM nd Y. Dt nlysis The SAS System for Windows V9.1 ws used for ll sttisticl nlyses. Within-loction nd multi-loction nlyses of vrince were performed on dt gthered in the 15 trils. For ech tril, we compred lines to hyrids y ANOVA followed y Newmn nd Keul s test t P B 0.05 for Y, E, nd cvy%. A two-wy ANOVA (referred s multi-loction nlysis), where trils (loctions) nd types of cultivr (i.e. hyrids vs. lines) were considered s fixed effects, ws performed to study interctions etween the type of cultivr nd the loctions. For this two-wy ANOVA, yield (Y), erliness (E) nd production stility (cvy%) were studied. Hyrid stility ws compred to line stility y joint regression nlysis on loctions (Eerhrt nd Russel 1966). In second nlysis, prticulrly productive hyrid (cv. Centromeric) with interesting sensory chrcteristics (dt not shown) ws compred with the most productive line in ech tril for Y, E, nd cvy%. This hyrid ws present in nine trils. As ove, multi-loction nlysis ws crried out for Y on these nine loctions (i.e. trils), using two-wy ANOVA (loctions nd genotypes considered s fixed effects) followed y joint regression nlysis on loctions. Finlly, in third nlysis, we estimted Y ccording to the cropping system. We used nested model s follows: Y ijk ¼ l þ v i þ S k þ B jjk þ l þ v i l þ E ijkl where Y is the yield, l is the overll men, v i represents the fixed effect of the type of cultivr i (i.e. hyrid men vs. line men), S k represents the rndom effect of loction S k,b j k is the rndom effect due to the jth lock within loction k, l represents the fixed effect of the cropping system (i.e. groforestry system vs full-sunlight system), v * il represents the rndom interction etween the type of cultivr i nd cropping system l, nd E ijkl represents the rndom error ssocited with prticulr oservtion. Results Hyrids compred to lines For ech tril, mens of lines nd hyrids were compred through n nlysis of vrince, focusing on yield, erliness, nd production stility. Hyrids hd significntly greter yields thn lines t 14 of the 15 tril sites (Tle 3). The differences in yield rnged from 8 to 127% (Tle 3). At low elevtions ( m.s.l.), the differences were not uniform, with extremes rnging from 16 to 127% nd medin of 52% for this group of four trils. At moderte elevtion (1,000 1,340 m.s.l.), which ws the cse for most of the coffee res in Centrl Americ, the

7 Euphytic (2011) 181: Tle 3 Comprison of F1 hyrids with lines in the 15 trils over the first production cycle efore coppicing Elevtion (m.s.l.) Yhyrids (g) Ylines (g) D (%) PrY Ehyrids Elines PrE cvy%hyrids cvy%lines PrcvY% NS NS NS NS NS NS NS NS NS NS NS NS NS NS For ech tril, re indicted: the elevtion (m.s.l.), the men nnul production of green coffee (Y in g nd expressed per tree) for the men of F1 hyrids (Yhyrids) nd for the men of lines (Ylines), D (%) tht represents (D ¼ Yhyrids Ylines Ylines 100) nd PrY is the ssocited proility of this difference, Ehyrids nd Elines re the erliness of yields respectively representing the men of F1 hyrids nd the men of lines, nd PrE is the ssocited proility of this difference, cvy%hyrids nd cvy%lines re the coefficients of vrition for hrvests, respectively for the men of F1 hyrids nd for the men of lines nd PrcvY% is the ssocited proility of this difference. NS not significnt t P = 0.05 superiority of the hyrid men ws sustntil, rnging from 11 to 121%, with medin of 48%. Lstly, t high elevtion (C1400 m.s.l.), the differences in yield incresed from 8 to 95%, with medin of 33%. A two-wy ANOVA reveled highly significnt effect of loction nd of type of cultivr, respectively (F = 69.7, P \ nd F = 87.5 nd P \ ) nd low ut significnt interction etween the two (F = 2.65, P \ 0.01). There thus ws n interction etween type of cultivr nd loctions, even though the men squre for the interction (1.33) ws lower thn tht for the loctions nd type of cultivr (34.7 nd 44.1, respectively). This significnt interction ws due to the different rte of response of hyrids to lines t ech site, s shown in Fig. 3. The nlysis of interction y regression slope reveled tht the hyrid men hd non-significnt regression coefficient (R =-0.04 ± 0.02) nd tht the line men hd significnt regression coefficient (R =-1.03 ± 0.09). We mesured erliness s n index representing the percentge of the first hrvest (Y1) over the cumultion of hrvests for the production cycle. The percentges were highly vrile etween trils, rnging from 0.6 to 33.4% for the hyrid mens nd from 0 to 22.6% for the line mens (Tle 3). These differences were sttisticlly significnt in only nine out of the 15 trils in the network. During the first production yer (dt not shown), the hyrid men yield ws significntly higher thn the line men yield t ll loctions. Regrding the production stility over the production cycle, the coefficients of vrition (cvy%) for the hyrid mens rnged from 29 to 70%, wheres the line mens rnged from 46 to 94% (Tle 3). The coefficients of vrition for hyrids were lwys lower thn those for lines. Nevertheless, the differences were significnt t only eight loctions. Overll, it ppered tht the hyrid mens were more stle thn the line mens, i.e. production differences etween yers were less mrked for hyrids. The est hyrid compred to the est line According to dt gthered from this tril network nd other loctions (dt not shown), Centromeric ws

8 154 Euphytic (2011) 181: Fig. 3 Comprison of the men yield of lines nd of hyrids in the 15 trils; (Xxis: elevtions in m.s.l.,y-xis: yield in green coffee). Two different letters indicte significnt differences t P \ 0.05 etween hyrids nd lines in the sme tril (i.e. elevtion) Yield of green coffee (in g per tree nd per yer) Lines Hyrids Elevtion (m.s.l.) Tle 4 Comprison of the est F1 hyrid, cv. Centromeric with the est line of ech tril over the first production cycle Elevtion (m.s.l.) Yhyrid Centroméric (g) Yline (est line) (g) D (%) PrY Nme of the est line CR Ctisic Pcs Pcs CR T Cturr CR NS IAPAR59 For ech tril, re indicted: the elevtion (m.s.l.), the men nnul production of green coffee (Y in g nd per tree) for the the F1 hyrid cv. Centromeric (Yhyrid) nd for the est line of the tril (Ylines), D (%) represents (D ¼ Yhyrid Yline Yline 100) nd PrY is the ssocited proility of this difference, nd the nme of the est line per tril. NS not significnt t P = 0.05 the most productive hyrid. This hyrid, which ws present in nine trils in the network, ws compred to the est line in ech of the nine trils, sed on the sme criteri s ove. The est line in ech tril is indicted in Tle 4. In five trils, Ctimor line ws involved. In four trils, trditionl cultivrs were involved (i.e. Ctui or Cturr). A multi-loction nlysis reveled highly significnt difference etween the est hyrid nd the est line (F = 27.3, Pr \ ) nd nonsignificnt interction (F = 1.3, Pr = 0.22). Depending on the tril, the hyrid Centromeric produced from 14 to 243% more thn the est line (Tle 4), nd the men estimted difference mounted to 37%. When the nlysis ws crried out for ech loction, the differences were significnt for eight trils nd nonsignificnt for one tril. At ll loctions, the hyrid Centromeric displyed etter erliness thn the est line. However, the percentges were highly vrile

9 Euphytic (2011) 181: etween trils, rnging from 0.02 to 22% for the hyrid nd from 0 to 16% for the line (dt not shown). The differences were significnt in five of the nine trils. The multi-loction nlysis reveled significnt difference etween the hyrid nd the est line for ech tril (F = 8.3, Pr = 0.03) nd non-significnt interction (F = 1.5, Pr = 0.07). The first yer production for the hyrid mounted to 10% on verge of the totl for the first production cycle s opposed to 4% for the est line. The production stility of the hyrid ppered to e etter in the nine trils. The coefficient of vrition for the hyrid rnged from 31 to 68%, wheres tht of the lines rnged from 45 to 94%. The coefficients of vrition for the hyrid were lwys lower thn those for the line. The differences were significnt for the nine loctions. The multi-loction nlysis reveled highly significnt difference etween the hyrid nd the line (F = 3.1, Pr \ 0.01), nd non-significnt interction (F = 1.4, Pr = 0.06). The men coefficient of vrition of the hyrid ws 55% s opposed to 66% for the line. Comprison of hyrids nd lines ccording to the cropping system A mixed-effect nlysis of vrince ws used to estimte yield differences etween types of cultivr under the two frming systems. The interction etween frming systems nd type of cultivr ws highly significnt (F = 77.4, P \ ). In the AS, the men hyrid yield ws 460 g yer -1 of green coffee per tree s opposed to men yield of 290 g yer -1 per tree for the lines (Fig. 4). The difference in fvor of the hyrid men ws highly significnt (P \ ) nd the yield increse mounted to 58%. In the full-sun system, the men hyrid yield ws estimted t 754 g yer -1 s opposed to 562 g yer -1 for the men line yield (Fig. 4). The difference in fvor of the hyrids mounted 34% nd ws highly significnt. Discussion In Centrl Americ, Aric coffee cn e grown t elevtions rnging from 700 to 1,600 m.s.l, in frgile mountinous ecosystems. Furthermore, the Yieldin green coffee (in g per treend per yer) Hyrids Lines Agroforestry system Full-sun system Fig. 4 Comprison of the men yield of lines nd of hyrids ccording to the cropping system. Two different letters indicte significnt differences t P \ 0.05 etween cropping systems coffee-growing re is one of the world hotspots for iodiversity. Coffee gro-forests, with shde trees interspersed mongst the coffee plnts, re often the only hitt with remining tree cover within these res for migrtory irds in the Mesomericn iologicl corridor, s lmost ll forests hve een removed t the elevtions where coffee is grown (Hrmnd et l. 2007; Jh nd Dick 2008). These AS thus enefit interntionl conservtion inititives. Regrding soil erosion, studies indicte tht shde trees cn reduce runoff y more thn hlf on coffee slopes vi nturl litter fll or pruning residues tht cover the soil, reduced impct of rindrops, improved soil structure nd enhnced infiltrtion (Beer et l. 1998; Snoeck nd Vst 2009). Furthermore, rurl households rely on fuelwood derived from trees in coffee gro-forests. Over the lst 40 yers, coffee cultivtion intensifiction hs led to the loss of more thn 50% of the tree cover (Hrmnd et l. 2007). This coffee intensifiction hs lso resulted in loss of lndscpe connectivity nd extensive loss of iodiversity in mny coffee-growing zones, nd proly n increse in grochemicl pollution of rivers nd quifers (Br nd Zk 1995). The lrge-scle doption of AS depends on the economic enefits tht they cn provide growers in vrious socio-economic nd ecologicl settings. The present study reveled tht introducing new intrspecific hyrid cultivrs in coffee-sed AS of Centrl Americ would considerly increse cropping system productivity, while lso considerly rodening the nrrow genetic se of cultivted C. ric y introducing genetic diversity from wild Ethiopin progenitors.

10 156 Euphytic (2011) 181: The set of tril loctions ws representtive of the rnge of productive sites in Centrl Americn coffeegrowing zone. In this region, temperture minly vries s function of the elevtion. The network of trils covered the regionl elevtion rnge for coffee growing, i.e. from 750 to 1,580 m.s.l. With respect to rinfll, in Cost Ric, for exmple, the trils t Trrzu nd Snt Mri de Dot were under Pcific climte (reltively low rinfll with mrked dry seson). On the other hnd, the plot t Plmir ws in very wet zone without ny dry seson. The soils in which coffee plnts were grown in the trils were representtive of three mjor morphopedologicl regions of Centrl Americ (Bornemisz et l. 1999), ccording to the 1976 FAO clssifiction. In Cost Ric, these were luvisols or ndosols in the sic volcnic highlnds of the Tertiry. Such soils re highly suited for coffee growing. In Hondurs nd prt of El Slvdor, luvisols of the cid volcnic highlnds were involved, derived from sediments of the Tertiry, which re frgile with low fertility. In estern El Slvdor, ndosols re formed on the volcnic cordiller of the Quternry, i.e. excellent soils for coffee growing. This network of trils ws lso designed to represent the diversity of cropping systems in Centrl Americ, since it consisted of 10 trils in AS nd five in full-sun systems. In the ltter cse, griculturl prctices were reltively uniform, prticulrly with respect to fertilizer pplictions nd pesticide tretments. On the other hnd, the AS rnged from very low-input griculture to intensive coffee growing. As underlined previously, we decided to llow plot mngers to mke decisions ccording to their experience nd resources. The result ws network of trils with contrsting cropping systems nd yields. For exmple, in El Slvdor, the Sn Antonio tril t 820 m.s.l. represented low-intensity coffee plnttion in n AS where no grochemicls were pplied, which proly explins the low yields. Conversely, t 1,420 m.s.l., the Snill de Aljuel tril ws on frm tht prcticed intensive coffee growing, which explins the high yields. Production cycle lengths vried in the network of trils. This vriility ws linked to soil fertility, tempertures nd growers choices. In the hot, wet lowlnd climte t Plmir, the trees needed coppicing fter 3 yers of high yields. In the temperte climte t Snt Mri de Dot (1,580 m.s.l.), coffee ws produced for 7 yers efore coppicing. In the very low-intensity, hence low-yield, AS t Sn Antonio or Ahuchpán in El Slvdor, it lso took 7 yers efore the trees needed coppicing, ccording to the growers criteri. With this network set up, it ws therefore possile to compre cultivrs in contrsting situtions representtive of the soil-climte conditions nd griculturl prctices tht previl in Centrl Americ. Under these conditions, yield differences etween hyrids nd lines rnged from 8 to 127%, with medin of 48% in fvor of hyrids. This first result tllies with those otined in smll controlled trils in Ltin Americ nd Afric y Wlyro (1983) nd Cils et l. (1998), where hyrids produced etween 10 nd 200% more thn lines. Bsed on controlled trils in full sunlight, we previously estimted tht heterosis rnged from 22 to 47% y compring hyrids with their mternl lines (Bertrnd et l. 2005). Here, we compred selected hyrids with trditionl or recent lines, it is therefore norml tht even greter differences were found etween hyrids nd lines. A few trends were lso noted tht confirmed previous oservtions. Firstly, hyrids egn producing erlier, nd they ppered to e more stle over the production cycle nd more stle cross different environments. Prt of the erliness could e explined y heterosis, ut lso it my hve een due to the propgtion method used for these hyrids. Indeed, somtic emryogenesis cuses some secondry effects tht hve yet to e clrified nd properly mesured ut which seem to ply role in tissue rejuvention (Hckett Wesley nd Murry John 1993; Perrin et l. 1997). When working with the sme genotype, we found tht, for nursery coffee plnts of the sme height, plnts derived from somtic emryos were significntly more vigorous thn those derived from seeds (Menéndez-Yuffá et l. 2010). As regrds stility, most uthors gree tht hyrids re more stle thn lines (Gllis 2009). In generl, heterozygotes re more cple thn homozygotes of exploiting sptiotemporlly vrile environment. We noted this homeostsis in the coffee hyrids, which led to lower yield vritions over the cycle in comprison to the lines. This greter homeostsis in hyrids proly plyed n importnt role in the yield stility. As we hve lredy highlighted (Bertrnd et l. 2006), hyrids with greter vegettive vigor hve higher lef-to-fruit rtios nd etter crohydrte supply to erries

11 Euphytic (2011) 181: thn trditionl cultivrs, which results in etter qulity everge. We therefore conclude tht heterosis gve hyrids true dvntge over lines in terms of productivity, ut lso in terms of yield stility. Moreover, this difference in productivity ws not chieved y incresing inputs. A comprison of hyrid mens nd line mens ccording to cropping systems showed tht the yield increse in the groforestry system mounted to 58%. This increse, which seems considerle, mounted to 170 g of green coffee per tree. In the full-sun system, the reltive increse ws less (34%), ut the difference in green coffee ws 190 g. Consequently, the returns on investment with oth systems were very similr. The success of lrge-scle dissemintion of hyrid cultivr depends on the extent of control over the reproductive system. Somtic emryogenesis hs een perfectly techniclly mstered in Nicrgu. The remining question concerns economic risks ssocited with the doption of this innovtion. Investing in hyrid micropropgted plntlets is expensive. The dditionl cost per tree is currently USD, i.e. n investment of 2,500 3,000 USD/ h. However, we elieve tht hyrid cultivrs would e more cost-effective t this price. A recent study of ECOM-INCAE (dt not shown) indicted tht renovting AS with hyrid plnts is etter thn with trditionl cultivrs. After 6 yers, the difference in net present vlue etween hyrids nd trditionl cultivrs ws found to e more thn 5,000 USD/h. In Green Revolution systems gered towrds high production, irrtionl use of hyrids cn worsen the ecologicl footprint. As with ny innovtion, thought will hve to e given to the ecologicl sustinility of intensive frms. In AS, such doption will hve to e fcilitted y credit policies, nd especilly y minimum intensifiction policies, notly with respect to input use. In ddition, the investment rtionle must come with commercil policy to cpture new lucrtive mrkets. Acknowledgments This work ws finnced y FONTAGRO ( We thnk the dministrtive stff of PRO MECAFE, PROCAFE, ICAFE, nd IHCAFE for ssistnce. Open Access This rticle is distriuted under the terms of the Cretive Commons Attriution Noncommercil License which permits ny noncommercil use, distriution, nd reproduction in ny medium, provided the originl uthor(s) nd source re credited. References Ar LI, Zk DR (1995) Nitrogen cycling in coffee grosystems: net N minerliztion in the presence or sence of shde trees. Agric Ecosyst Environ 48: Beer J, Muschler R, Kss D, Somrri E (1998) Shde mngement in coffee nd cco plnttions. Agrofor Syst 38: Bertrnd B, Etienne H, Cils C, Chrrier A, Brdt P (2005) Coffe ric hyrid performnce for yield, fertility nd en weight. Euphytic 141: Bertrnd B, Vst P, Alpizr E, Etienne H, Dvrieux F, Chrmetnt P (2006) Comprison of en iochemicl composition nd everge qulity of ric hyrids involving Sudnese-Ethiopin origins with trditionl vrieties t vrious elevtions in Centrl Americ. Tree Physiol 26: Bornemisz E, Collinet J, Segur A (1999) Los suelos cfetleros en Améric Centrl y su fertilizción. In: Bertrnd B, Rpidel B (eds) Desfios de l cficultur en Centroméric. IICA, Sn José, Cost Ric, pp Cils C, Bouhrmont P, Boccr M, Eskes AB, Brdt P (1998) Prediction of genetic vlue for coffee production in Coffe ric from hlf-dillel with lines nd hyrids. Euphytic 104:49 59 Eerhrt SA, Russel WA (1966) Stility prmeters for compring vrieties. Crop Sci 6:36 40 Etienne H (2005) Protocol of somtic emryogenesis: coffee (Coffe ric L. nd C. cnephor P.). In: Jin SM Gupt PK (eds) Protocols for somtic emryogenesis in woody plnts. Series: Forestry sciences, vol 77. Springer, The Netherlnds, Hrdcover, pp , ISBN: Etienne-Brry D, Bertrnd B, Vsquez N, Etienne H (1999) Direct sowing of Coffe ric somtic emryos mssproduced in iorector nd regenertion of plnts. Plnt Cell Rep 19: Gllis A (2009) Heterosis et vriétés hyrides en méliortion des plntes. Ed Que, Versilles, Frnce Hckett Wesley P, Murry John R (1993) Mturtion nd rejuvention in woody species. In: Ahuj MR (ed) Micropropgtion of woody plnts. Kluwer, Dordrecht, pp , ISBN , Hrmnd JM, Avil H, Dmrine E, Ski U, De Miguel S, Renderos Durn RV, Oliver R, Jimenez F, Beer J (2007) Nitrogen dynmics nd soil nitrte retention in Coffe ric Euclyptus deglupt groforestry system in southern Cost Ric. Biogeochemistry 85: Jh S, Dick CW (2008) Shde coffee frms promote genetic diversity of ntive trees. Curr Biol 18 24: Menéndez-Yuffá A, Brry-Etienne D, Bertrnd B, Georget F, Etienne H (2010) A comprtive nlysis of the development nd qulity of nursery plnts derived from somtic emryogenesis nd from seedlings for lrge-scle propgtion of coffee (Coffe ric L.). Plnt Cell Tiss Orgn Cult 102(3): Perrin Y, Doums P, Lrdet L, Crron MP (1997) Endogenous cytokinins s iochemicl mrkers of ruer-tree (Heve rsiliensis) clone rejuvention. Plnt Cell Tiss Orgn Cult 47:

12 158 Euphytic (2011) 181: Snoeck D, Vst P (2009) Importnce of orgnic mtter nd iologicl fertility in coffee soils. In: Wintgens JN (ed) Coffee growing nd processing, 2nd edn. Wiley VCH, Germny Vst P, Hrmnd JM (2002) The importnce of groforestry systems for coffee production in Centrl Americ nd Mexico. In: CIRAD-CP (ed) Recherche et cféiculture. Montpellier, pp Vst P, Perriot JJ, Guyot B, Génrd M (2006) Fruit thinning nd shde improve en chrcteristics nd everge qulity of coffee (Coffe ric L.) under optiml conditions. J Sci Food Agric 86: Wlyro DJ (1983) Considertions in reeding for improved yield nd qulity in ric coffee (Coffe ric L.). Doctorl Thesis, Agriculturl University, Wgeningen