sustainability Article An Evaluation of Dryland Ulluco Cultivation Yields in the Face of Climate Change Scenarios in the Central Andes of Peru by Using the AquaCrop Model Ricardo Flores-Marquez 1,* , Jesús Vera-Vílchez 2, Patricia Verástegui-Martínez 2 , Sphyros Lastra 1 and Richard Solórzano-Acosta 1,3 1 Centro Experimental La Molina, Dirección de Supervisión y Monitoreo en las Estaciones Experimentales Agrarias, Instituto Nacional de Innovación Agraria (INIA), La Molina 1981, Lima 15024, Peru; slastrapaucar@gmail.com (S.L.); investigacion_labsaf@inia.gob.pe (R.S.-A.) 2 Estación Experimental Agraria Santa Ana, Dirección de Supervisión y Monitoreo en las Estaciones Experimentales Agrarias, Instituto Nacional de Innovación Agraria (INIA), Carretera Saños Grande—Hualahoyo Km 8, Huancayo 12007, Peru; jvera@lamolina.edu.pe (J.V.-V.); patymarve@gmail.com (P.V.-M.) 3 Facultad de Ciencias Ambientales, Universidad Científica del Sur (UCSUR), Lima 15067, Peru * Correspondence: ricardo.floresm29@gmail.com Abstract: Ullucus tuberosus is an Andean region crop adapted to high-altitude environments and dryland cultivation. It is an essential resource that guarantees food security due to its carbohydrate, protein, and low-fat content. However, current change patterns in precipitation and temperatures warn of complex scenarios where climate change will affect this crop. Therefore, predicting these effects through simulation is a valuable tool for evaluating this crop’s sustainability. This study aims to evaluate ulluco’s crop yield under dryland conditions at 3914 m.a.s.l. considering climate change scenarios from 2024 to 2100 by using the AquaCrop model. Simulations were carried out using Citation: Flores-Marquez, R.; current meteorological data, crop agronomic information, and simulations for SSP1-2.6, SSP3-7.0, Vera-Vílchez, J.; Verástegui-Martínez, and SSP5-8.5 of CMIP 6. The results indicate that minimum temperature increases and seasonal P.; Lastra, S.; Solórzano-Acosta, R. An precipitation exacerbation will significantly influence yields. Increases in rainfall and environmental Evaluation of Dryland Ulluco CO2 concentrations show an opportunity window for yield increment in the early stages. However, Cultivation Yields in the Face of a negative trend is observed for 2050–2100, mainly due to crop temperature stress. These findings Climate Change Scenarios in the highlight the importance of developing more resistant ulluco varieties to heat stress conditions, Central Andes of Peru by Using the AquaCrop Model. Sustainability 2024, adapting water management practices, continuing modeling climate change effects on crops, and 16, 5428. https://doi.org/10.3390/ investing in research on smallholder agriculture to reach Sustainable Development Goals 1, 2, and 13. su16135428 Keywords: Andean tuber; Ullucus; climate change; AquaCrop; yield response Academic Editors: Ioannis Mylonas, Fokion Papathanasiou and Elissavet Ninou Received: 11 March 2024 1. Introduction Revised: 7 June 2024 Tubers are important crops within Andean societies and their agro-food systems. They Accepted: 10 June 2024 are characterized by their resilience to the region’s adverse conditions. Among these crops, Published: 26 June 2024 ulluco is prominent in food security due to its fundamental role in the Andean population’s diet [1,2]. It is cultivated primarily in Peru, Bolivia, and Ecuador and is grown between 2000 and 4000 m.a.s.l. at temperatures between 11 and 13 ◦C [1–3]. Its nutritional contribution Copyright: © 2024 by the authors. is based on its high carbohydrate content (64.96–84.2%), between 8.5 and 15.7% protein Licensee MDPI, Basel, Switzerland. content, low-fat content (0.1–1.4%), fiber ranging 0.5–5%, and vitamin C [1,4,5]. Generally, This article is an open access article tubers are boiled before eating. Nonetheless, they can also be dried to turn into chuño distributed under the terms and or milled into flour [6]. It is grown by more than 60 thousand producers in Peru, and −1 conditions of the Creative Commons during 2023, a cropping area of 42,912 ha with a 7.62 t·ha average yield was reported, Attribution (CC BY) license (https:// generating a total production of 173,116 t [7]. On average, dryland ulluco cultivation creativecommons.org/licenses/by/ represents 82% of Peru’s cropping area of ulluco, widely located in the Quechua and Sunni 4.0/). regions (94%). Monoculture patterns (86%) prevail over associations (14%) with crops such Sustainability 2024, 16, 5428. https://doi.org/10.3390/su16135428 https://www.mdpi.com/journal/sustainability Sustainability 2024, 16, 5428 2 of 22 as corn, mashua, potatoes, quinoa, etc. [8]. Even though its productive potential is less than other Andean tubers or roots [9], it is widely consumed in this region [1,10,11]. In the current context, climate change driven by carbon emissions from agriculture, livestock, industry, and households affects global agriculture through extreme weather events [12,13]. To research future impacts, the Working Group I (WGI) of the Intergov- ernmental Panel on Climate Change (IPCC) analyzed historical information and Global Climate Models (GCMs), generated as part of the Coupled Model Intercomparison Project (CMIP) to broaden our comprehension about the process and effects linked to climate change [14]. The fifth phase of the CMIP (CMIP5) analyzed future climatic projections based on Representative Concentration Pathways (RCPs), which express the increase in the most probable radiative forcing for 2100 [14]. Subsequently, the sixth phase’s report pro- vided different socioeconomic development pathways to achieve each RCP, thus proposing new levels of mitigation and adaptation actions to the effects of climate change [15]. These new scenarios are named Shared Socioeconomic Pathways (SSPs). SSP 1 and SSP 5 both propose increasing human development, differing in that the second proposes intensive fossil energy consumption as a means to achieve this; SSP 3 and SSP 4 propose pessimistic and inequitable development trends, while SSP 2 outlines continuity regarding historical patterns [14,15]. O’Neill et al. [15] identified priority analysis scenarios to update and cover information gaps of CMIP5 projections for RCPs—SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5. With all this information, it is possible to use existing models to explore potential yield changes in different crops. There are some experiences regarding modeling tuber and root growth such as potatoes, cassava, and sweet potatoes under diverse climates and water-available conditions [16–19]. Nevertheless, there is little research on crops in Andean regions [20–23] or on native crops such as ulluco [24–28], despite its importance for Andean food security and value in gourmet cuisine and industry. AquaCrop is a model that simulates crop yields under different conditions based on plant water consumption; for this, it requires information about climate, crop, soil, and management conditions without considering pests’ or diseases’ potential effects [29]. Constant efforts are made to improve AquaCrop’s simulation results while maintaining its simple modeling scope [30]. Due to its versatility, a wide range of research is linked to crop modeling through AquaCrop use for different climatic and management conditions, includ- ing climate change projections at different latitudes, longitudes, and altitudes [20,31,32]. At this point, the model’s main limitations include its single-field-scale scope, assump- tions of spatial uniformity inside the field, and the unidimensional vertical water fluxes simulated [33]. In addition, Vanuytrecht et al. [30] also outline the constant improvement of AquaCrop results to make them more accessible and understandable for agricultural extensionists, decision-makers, and stakeholders. Several studies suggest that smallholding farming productivity in the Peruvian Andes is at imminent risk of decreasing due to water availability changes and increased tempera- tures [34–36]. The susceptibility of dryland crops to rainfall variability is exacerbated by changes in rainfall patterns due to climate change [34,37] which ultimately determines poverty conditions in rural areas [38]. Additionally, rising temperatures reduce the produc- tive potential of crops adapted to cold regions [39]. This study focused on precipitation and temperature changes and their effect on ulluco yield. Agronomic factors that influence tuber development have been investigated [40–43], as well as their potential food industry use due to their nutritional benefits [10,44–46] and their importance in local culture [47,48]. Ulluco’s phenological development and productivity depend on climatic, edaphic, and genotypic factors, among others. Sowing dates are usually linked to rain presence in a crop rotation system with potatoes, oca, or barley [3,49]. Therefore, analyzing crop behavior facing extreme climate events is essential to de- velop adaptation strategies for different regional conditions [36]. This study evaluates the yield changes in Ullucus tuberosus, a key native Andean tuber, under dryland cultivation in Sustainability 2024, 16, 5428 3 of 22 the Peruvian central highlands, according to climate change scenarios from the Coupled Model Intercomparison Project (CMIP) Phase Six for the 2024–2100 period. 2. Materials and Methods 2.1. Studied Area This study was conducted in the town of San José de Apata, district of Apata, Jauja province, department of Junín, Peru. The experimental field was installed at 75◦19′30.41′′ W; 11◦47′53.73′′ S at 3914 m.a.s.l. (Figure 1). The area’s climatic conditions are average annual precipitation, 689 mm; minimum temperatures between −4 ◦C (June–July) and 4 ◦C (March); maximum temperatures between 20 ◦C (March) and 22 ◦C (November–December); and average relative humidity, 72%. Averages were calculated based on historical data Sustainability 2024, 16, x FOR PEER REVIEW 4 of 24 from Peru’s National Service of Meteorology and Hydrology (SENAMHI) [50]. FigFuirgeu 1r. eSt1u.dieSdt uadreiae;d (aa) rJeuan;ín( ad)epJaurntmínendt elopcaatritomn ewnitthlion cPaetriuo;n (bw) Jiatuhjian pProevriunc;e( blo)caJtaiounj;a (cp) rovince location; loc(act)iolno coaf tthioen evoaflutahteione vpalolut aantdio mnepteloortoalongdicaml setatetioornos luosgeidc;a (dl )s taaetriiaoln ssatuelslieted v; i(edw) oafe trhiea levsaalt-ellite view of the uation plot. evaluation plot. 2.2. Experimental Design This study had a split-plot design in which the following two factors were tested: (i) sowing date (main plot with three levels) and (ii) the type of fertilization (sub-plot with two levels). Thus, six treatments with four replicates per treatment were used for 24 ex- perimental units, as described in Table 1. Sustainability 2024, 16, 5428 4 of 22 2.2. Experimental Design This study had a split-plot design in which the following two factors were tested: Sustainability 2024, 16, x FOR PEER REVIEW (i) sowing date (main plot with three levels) and (ii) the type of fe5r toilf iz24a tion (sub-plot with two levels). Thus, six treatments with four replicates per treatment were used for 24 experimental units, as described in Table 1. Table 1. Factors Taanbdl ele1v.eFlsa ctetostresda.n d levels tested. Factor Levels Factor Levels F1: 13 October 2022 Sowing date (F) Sowing date (F) F2: 28 October 2022 F1: 13 October 2022 F2: 28 October 2022 F3: 12 November 202F23 : 12 November 2022 M1: Traditional (sub-dose) * Fertilization managemFernttil i(zMat)i on management (M) M1: Traditional (sub-dose) *M2: Recommended (oMpt2i:mRaelc odmosmee) n*d ed (optimal dose) * * Details of each* fDerettialiilzsaotfioeanc hmfaenrtailgizeamtioen tm aarnea dgemscernibt aerde idne sTcaribbled oifn STeacbtlieoonf 2S.e5c.t ion 2.5. 2.3. Experiment2a.l3 P. lEoxtsp Sereitm-Uenpt al Plots Set-Up Experimental pElxoptse rwimereen tianlstpalloletsd woenr ea itnhsrteael-lyedeaor nfaallothwre fie-eyldea wr fhaelrloe wpofitaeltdoews hearde potatoes had previously beepnr egvrioowunsl.y Ubleluecnugs rtouwbenro.sUusll ucvcu. sCtaunbaerrioos uwsacsv .cChoasneanr idouwe atso cihtso wseindedsuperetaodi ts widespread cultivation in tchuel tcievnattriaoln Pienrutvhieance AnntrdaelsP [e5r1u],v siaonwnA natd ae sde[5n1s]i,tys oowf 4n1,a6t6a6 pdleannstist yhao-f1 4(10,.686 6 plants ha-1 m × 0.3 m). A n(0e.t8 pmlot× co0n.3simsti)n. gA ofn 9e tfuprlrootwcos,n 6s imsti nlogngo,f w9afsu rersotawbsli,s6hemd wlointhgi,nw eaasche setxa-blished within perimental unieta. Dchateax wpeerriem ceonllteaclteudn fitr.oDma tthaewme. rTehceo slpleacttieadl afrrroamngtehmemen.t Tohf tehsep taretiaatlmaerrnatns gement of the can be found intr eFaigtmureen 2ts. c an be found in Figure 2. Figure 2. Field dFisigtruibreut2i.oFni eolfd thdeis etrxipbeurtiimonenotfatlh ueneixtsp. erimental units. 2.4. Soil Charac2te.4ri.sStiocisl Characteristics Before installaBtieofno,r ae sinosilt asltluadtiyo nw, aass cooilnsdtuucdtyedw ians tchoen edxupcetreidmienntthale aerxepae. rFimirsetn, taa sloairle a. First, a soil pit was dug top ait dwepasthd oufg 1t.o3 amd, eapntdh, oafft1e.r3 ombs,earnvda,tiaofnte, rthorbesee hrvoartizioonn,st hwreeree hdoertiezromnisnwede,r e determined, comprising topcsoomilp (rAisPin) gantodp tswooil s(AtraPt)aa onfd ptawrtoiasltlrya dtaecoofmpaprotisaeldly pdaerceonmt mpoasteerdiapl a(rCe1n tanmda terial (C1 and C2). For each oCf 2th).eFmo,r seaamcphleosf wtheerme ,taskamenp alensdw seenret ttoa ktheen Saonidl, Wseanttetro atnhde LSeoailv,eWs aLtaebroarnad- Leaves Labo- tory (LABSAF)r—atoINryIA(L SAaBntSaA AFn)—a tIoN aInAalSyaznet abuAlkn adetonsaintya l(yBzDe)b [u52lk], dteexntsuirtey ((TBxDt)) [[5532]],, stoeixlt ure (Txt) [53], pH [54], electriscoaill cpoHnd[u5c4t]i,veilteyc (tEriCca) l[5c5o]n, dourgcatinviict ym(aEttCe)r [(5O5M], o) crgoannteicntm, naitttreorg(eOnM (N) )c,o pnhtoens-t, nitrogen (N), phorus (P), poptahsossipuhmo r(uKs),( Pa)n, dp oetxacshsiaunmge(aKb)l,ea nadluemxicnhuamng e(Aabl)le [5al3u] m(Tinabulme 2(A). lI)n[5 a3d] d(Titaibolne, 2). In addition, other soil sampoltehse rwseoriel staamkepnl east wtheer ecetanktreanl aztotnhee ocfe neatrcahl zeovnaleuoaftieoanc hunevita, lauta 3ti0o cnmu ndiet,patth3, 0 cm depth, to to measure soiml meaosiustruerseo cilomntoenistt uurseincgo nthteen gtruasviinmgetthreicg mraevthimodet [r5ic6]m. ethod [56]. Table 2. Soil profile characterization. Cod Depth pH EC OM P K Al N Txt BD m d∙Sm−1 % mg∙kg−1 mg∙kg−1 meq∙(100 gr)−1 % g∙(cm3)−1 AP 0.4 5.00 0.11 9.30 24.37 358.05 18.21 0.47 Loam 1.10 C1 0.65 4.92 0.07 1.15 2.24 54.61 19.50 0.06 Loam 1.38 C2 1.3 4.78 0.07 0.76 4.19 36.31 7.29 0.04 Sandy loam 1.80 Cod: horizon code, EC: electrical conductivity, OM: organic matter, P: phosphorus, K: potassium, Al: exchangeable aluminum, N: total nitrogen, Txt: texture, BD: bulk density. Sustainability 2024, 16, 5428 5 of 22 Table 2. Soil profile characterization. Cod Depth pH EC OM P K Al N Txt BD m d·Sm−1 % mg·kg−1 mg·kg−1 meq·(100 3 −1gr)−1 % g·(cm ) AP 0.4 5.00 0.11 9.30 24.37 358.05 18.21 0.47 Loam 1.10 C1 0.65 4.92 0.07 1.15 2.24 54.61 19.50 0.06 Loam 1.38 C2 1.3 4.78 0.07 0.76 4.19 36.31 7.29 0.04 SandySustainability 2024, 16, x FOR PEER REVIEW 16.8 0of 24 loam Cod: horizon code, EC: electrical conductivity, OM: organic matter, P: phosphorus, K: potassium, Al: exchangeable aluminum, N: total nitrogen, Txt: texture, BD: bulk density. 2.5. Crop Fertilization 2.5. Crop Fertilization Fertilization was planned according to the second factor, proposed management (Ta- ble 1)F, ecrotnilsiizdaetrioinng wthaes pphlyasnincaeld–cahcecmoridcainl sgotilo chthaerascetceorinzdatifoanct roers,upltrso fproosmed thme taonpasgoeiml (Tean-t b(Tlea b2l)e. D1)e,tcaoilnss oidf ethrien fgerthtielizpahtyiosnic apll–acnh aerme ischaol wsonil icnh Taarbaclete 3r.i zation results from the topsoil (Table 2). Details of the fertilization plan are shown in Table 3. Table 3. Fertilization plan. Table 3. Fertilization plan. Management Sheep Manure Urea DAF ClK Sulpomag Agr Lime Ammonium Nitrate Manage m ent Sheep Mt∙ahnaur−e1 kUgre∙aha−1 kg∙DhAaF−1 kg∙haC−1l K kg∙ha−S1u lpomagkg∙ha−1 Agr Lime kg∙hAam−1moniumNi trate M1: Traditional t·ha−51.00 kg·ha0− 1 kg0· ha−1 0 kg·ha−1 0 kg·ha−1 0 kg·ha−1 0 kg·ha−1 M2:M R1:eTcroadmitimonaelnded 5.005.00 901.36 222.067 138.100 55.55 0 807.14 0 91.36 0 M2: Recommended 5.00 91.36 222.67 138.10 55.55 807.14 91.36 DAF: Diammonium phosphate; ClK: potassium chloride; Sulpomag: potassium and magnesium sDuAlfFa:teD;i Aamgmr: oAngiurmicuplhtuorsaplh. ate; ClK: potassium chloride; Sulpomag: potassium and magnesium sulfate; Agr:Agricultural. 22..66.. Weeaatthheerr DDaattaa Weeaatthheerr ddaattaa weerree oobbttaaiinneeddf rfroomma aD DavavisisV VanatnatgaegeP rPor2os2t asttiaotniomn amnaangaegdebdy bAyG ARGORROU-- RRUARLAloLc altoecdataetd7 5a◦t 1795′2°81.93′228′′.3W2;″ 1W1◦;4 71′14°54.75′94′5′ .S59a″t 3S9 a50t 3m9.5a0.s .ml. .Ra.esc.lo. rRdsecinorcdlusd iendcltuhdeewdh tohlee wgrhoowlein ggrosweaisnogn s(e2a0s2o2n– 2(2002232)–(F20ig2u3r)e (F3i)g. ure 3). FFiigguurree 33.. AAppaattaa ssttaattiioonn ddaattaa ffoorr 22002222––22002233 oonn aa ddaaiillyy ssccaallee.. RRaaiinnffaallll ((rrff)),, mmaaxxiimmuumm tteemmppeerraattuurree ((maaxxtteemp)),, aand miiniimum tteempeerraatturree ((miintteemp)).. 22..77.. DDaattaa CCoolllleeccttiioonn ooff MMoorrpphhoo--PPhhyyssiioollooggiiccaall VVaarriiaabblleess ffoorr MMooddeell CCaalliibbrraattiioonn To calibrate the AquaCrop model, 9 evaluations were carried out during the whole growTion gcasleibarsaotne, tshtea rAtiqnugafCroromp tmheodpehl,e 9n oelvoagliucaatliosntasg we eorfe ecmarerrigeden ocue.t dInureinagch thoen we,hfiovlee gvraorwiaibnlges sweaesroena, ssstaerstsiendg. fIrnomth tehpe hpehneonloolgoigcaiclaslt satgaeg,ed oifr eecmt eorbgseenrvcea.t iIonn esaochf polnaen, tfisvwe ivthairn- ifaubrlreosw wneurem absesres5se(tdh. eInce tnhter aplh5emnoolofgiitc)awl estraegpee, rdfiorremcte odb. sPelravnatticoannso opfy pcloavnetsr wpeitrhceinn tfaugre- rwoaws nevuamlubaetre 5d (uthsien cgenRtGraBl p5 hmo toofg irta) pwhesreta pkeernfowrmitheda. NPliaknotn cDa3n5o0p0yc caomveerr apaenrcdenptraogcee swseads ewviatlhuIamteadg uesCinagn oRpGyB3 p.6haontodgAraupthosC tAakDen2 0w2i2ths oaf tNwiakroen.DA3t5o0t0a lcaomf 1e2rap alanndt spwroecreessaesdse wssiethd Ifmroamgee aCcahnnoeptyp l3o.6t (athnrde eApultaonCtsAfDro 2m02fu2 rsroofwtwnaurme. bAe rtso2ta, l3 ,o7f ,1a2n dpl8anetasc hw)e. rTeh aespsehsosteodg rfarpohms ewaecrhe ntaekt epnloatp p(trhorxeiem paltaenlyts1 fmromab ofuvrergorwo unnudmlebveersl. 2A, 3sq, u7a, raenmd e8t aelapcihe)c.e Tmhee apsuhroitnoggr1a0pchms wonerae stiadkeenw aapspursoexdimasataelmy e1a msu aribnogvree gferoreunncde .leTvheel.n A, t shqeusearpel manettsawl peireecee xmtreaacstuedrinfrgo 1m0 cthme on a side was used as a measuring reference. Then, these plants were extracted from the ground to measure their height and root depth. Finally, plant leaves were separated and prepared for dry matter analysis [16]. At the end of the season, samples were taken from the central 9 m2 (3 m × 3 m) of the net plot to estimate the following crop-yield-associated variables: crop fresh weight, tuber dry matter, aerial and total dry matter, tuber dry matter (10 representative tubers per unit of assessment), and aerial and total dry matter (the sampled plants for yield) [16]. Sustainability 2024, 16, 5428 6 of 22 ground to measure their height and root depth. Finally, plant leaves were separated and prepared for dry matter analysis [16]. At the end of the season, samples were taken from the central 9 m2 (3 m × 3 m) of the net plot to estimate the following crop-yield-associated variables: crop fresh weight, tuber dry matter, aerial and total dry matter, tuber dry matter (10 representative tubers per unit of assessment), and aerial and total dry matter (the sampled plants for yield) [16]. 2.8. Statistical Analysis Morpho-physiological variables were analyzed using a two-way ANOVA (alpha = 0.05) to examine treatment differences. Model assumptions were verified graphically with the autoplot function from the ggfortify library for R [57]. Variables that showed significant differences were analyzed with the Least Significant Difference Test (alpha = 0.05), LSD.test function from agricolae library for R [58], corrected with the Benjamini–Hochberg proce- dure. 2.9. AquaCrop’s Crop Development Modeling AquaCrop’s climate input came from the weather station. It included rainfall data, maximum and minimum temperatures, wind speed, and relative humidity on a daily scale, which enabled the calculation of reference evapotranspiration (ETo) through the Penman– Monteith equation using the EToCalculator 3.2. The cropping module was completed with the image processing results from the canopy coverage analysis. Soil input values were determined with the information from the physical-chemical analysis of the soil pit. A management module was configured for dry conditions, weed management, fertilization, and furrowing characteristics. The iterative calibration process was based on the recom- mendations of [29], using calibrated parameters for potato cultivation from AquaCrop, the literature, and experts’ opinions. The first simulation was performed with field-collected data (soil and weather). Missing parameters (conservative and non-conservative) were completed with available values in the literature, including those from potato crop research, which is extensive. Canopy cover data were adjusted for each phenological stage, consid- ering the potential effects of heat and water stress. The impact of traditional fertilization (Table 1) was also considered. Calibration was performed considering the data from sowing dates 1 and 2 (Table 1). Model efficiency was evaluated through Pearson’s correlation index (r) and the Nash–Sutcliffe efficiency index (NSE) for plant canopy coverage and crop yield data. The model was then validated with the data from sowing date 3 (Table 1), without changing the conservative parameters but adapting the sowing date. The model’s efficiency was assessed using both previously mentioned indexes. 2.10. Climate Change Explorations Since the weather station used for AquaCrop climate input does not have data before 2019, we used data from the SENAMHI network of weather stations [50]. The criteria for choosing proper data involved proximity to the experimental location and accessibility to historical data for rainfall (rf), maximum temperature (maxtemp), and minimum tem- perature (mintemp). Records from 1986 to 2017 were found. Downloaded data from four weather stations (Jauja: 75◦29′12.73′′ W–11◦47′11.87′′ S; Ingenio: 75◦17′47.9′′ W–11◦52′30.8′′ S; Ricran: 75◦31′38.29′′ W–11◦32′24.05′′ S; Santa Ana: 75◦13′17.7′′ W–12◦0′34.4′′ S) were homogenized, and missing registers were filled by the Paulhus and Kohler [59] method included in the Climatol package in R [60]. Subsequently, homogenized weather data for the experiment’s geographical location were calculated using the weighted inverse distance interpolation method (WID) in Excel. Regarding climate change models, a total of 7 Gen- eral Circulation Models (GCMs) belonging to the Coupled Model Intercomparison Project Phase 6 (CMIP6) were chosen (Table 4). The selection criteria involved prioritizing those evaluated for Peru and South America’s conditions, in terms of their statistical downscaling adjustment [61,62]. GCM data corresponded to the historical period (1980–2014). To avoid the innate bias problems of the GCMs (over-estimation of minimum temperatures and Sustainability 2024, 16, 5428 7 of 22 rainfall, as well as under-estimation of maximum temperatures) [62], after climate data extraction for plot location (using the terra package in R) [63], bias correction by quantile mapping was performed using the Qmap package in R [64]. Given our historical data, the model efficiency of all seven GCMs was evaluated to select the best-fitting model, based on Fernandez-Palomino et al. [62]. As a result of this analysis, the EC-Earth3 model was chosen to perform future projections, based on the Shared Socio-Economic Pathways’ (SSPs) scenarios SSP1-2.6, SSP3-7.0 and SSP5-8.5, using the chosen model. The SSP1-2.6 scenario corresponds to a sustainable route, in which zero CO2 emissions for the second half of this century are achieved. The SSP3-7.0 scenario considers doubled CO2 emissions by 2100 due to regional rivalries and the absence of additional mitigation measures. The SSP5-8.5 scenario considers doubled CO2 emissions by 2050 and the heavy use of fossil fuels to achieve human development. No further political measures are taken [14]. ETo was calculated using the Hargreaves method in the EToCalculator based on maximum and minimum projected temperature data for the selected model’s future scenarios. To simulate ulluco yield for 2024–2100, rainfall, temperature, and ETo results were used as inputs for the climate module of the previously validated AquaCrop model, and atmospheric CO2 concentration corresponded to the software database. Table 4. Used CMIP6 models. Model Member Historical(rf, Mintemp, Maxtemp) IPSL-CM6A-LR r14i1p1f1 [65] CNRM-CM6-1 r1i1p1f2 [66] CNRM-ESM2-1 r1i1p1f2 [67] MIROC6 r50i1p1f1 [68] MRI-ESM2-0 r5i1p1f1 [69] MPI-ESM1-2-HR r2i1p1f1 [70] EC-Earth3 r150i1p1f1 [71] 3. Results 3.1. Tuber Yields by Sowing Date and Type of Fertilization In addition to modeling, whether the tested variables influenced ulluco’s yield was determined, as shown in Table 5. At the main effects level, the sowing date (F) had no significant effect on fresh and dry weight yield, nor the number of tubers per kilogram. On the other hand, the fertilization method significantly affected the three evaluated variables for ulluco. The recommended dose values doubled under conventional management conditions in fresh weight (FW) and dry weight (DW) cases. It was found that yield decreases the later the sowing date takes place. However, mineral fertilization efficiency also decreases because less water is available as the sowing date is delayed (Table 5). Its efficiency is similar to traditional manure fertilization (Figure 4). The cumulative rainfall for the F1 growing season was 677 mm, 657 mm for F2, and 635 mm for F3. Table 5. Ulluco tuber’s yield by sowing dates and type of fertilization. Treatment Fresh Weight Yield Dry Weight Yield Number of Tubers per(Mg·ha−1) (Mg·ha−1) kg F ns ns ns M *** ** *** F × M * ns ns Sustainability 2024, 16, 5428 8 of 22 Table 5. Cont. Treatment Fresh Weight Yield Dry Weight Yield Number of Tubers per(Mg·ha−1) (Mg·ha−1) kg Sowing Date (F) F1 10.50 ± 5.0 1.28 ± 0.6 117.7 ± 24 F2 9.53 ± 6.8 1.10 ± 0.8 102.6 ± 27 F3 8.41 ± 1.6 1.16 ± 0.3 108.2 ± 17 Fertilization Method (M) Sustainability 2024, 16, x FOR Tradition PaElE(RM R1E)VIEW 6.4 ± 2.3 b 0.77 ± 0.3 b 123.0 ± 22 a 9 of 24 Recommended (M2) 12.56 ± 4.7 a 1.59 ± 0.5 a 96.0 ± 15 b F × M MA1t the main effects l6e.8ve±l,2 t.h8ec sowing date (F0.)7 9ha±d0 n.3o significant e1ff3e5c.3t8 o±n 2f0r.e9sh and F1 dry Mwe2ight yield, nor the1 4n.2um±b3e.7r aof tubers per ki1l.o7g7r±am0..5 On the other h1a0n0.d0,6 t±he5 .f7ertiliza- tion Mm1ethod significantly5. 0a±ffe2c.t7ecd the three eva0lu.5a7t±ed0 .v3ariables for u1ll1u1c.9o6. ±Th30F2 e. 1recom- menMde2d dose values dou14b.l1e±d u6.n8daebr conventional1 .m64a±na0g.8ement conditio9n3s. 1i4n ±fre2s4h.2 weight bc F3 (FWM) a1nd dry weight (D7W.4)± ca0s.5es. It was found t0h.9a6t ±yie0.l1d decreases the 1la2t1e.7r ±th7e sowing dateM ta2kes place. Howev9e.4r, ±m1in.9 abc eral fertilization e1ffi.36ci±en0c.y3 also decreases 9b4e.c7a±us1e2 less wa- *t*e* rIm isp laievsasitlaatibstliec aalss itghneifi scoanwceinatg0 d.1a%t;e* *isim dpelliaesysetadt i(sTtiacabllseig 5n)i.fi Ictasn ceeffiatc1ie%n; c*yim ips lsieismstialatirst itcoa ltsriagdniifiticoanncael at 5%; ns: indicates not significant (p > 0.05); different letters in each testing parameter represent statistical smignainfiucarnec efearmtiolnizgagtrioounp s(Fatigpu